(Herbert S. Terrace, Janet Metcalfe) The Missing L

The Missing Link in
Cognition:
Origins of Self-Reflective
Consciousness
Herbert S. Terrace
Janet Metcalfe,
Editors
OXFORD UNIVERSITY PRESS

The Missing Link in Cognition
This page intentionally left blank
Origins of Self-Reflective
Consciousness
Edited by
Herbert S. Terrace and Janet Metcalfe
1
2005
3
Oxford New York
Auckland Bangkok Buenos Aires Cape Town Chennai
Dar es Salaam Delhi Hong Kong Istanbul Karachi Kolkata
Kuala Lumpur Madrid Melbourne Mexico City Mumbai Nairobi
São Paulo Shanghai Taipei Tokyo Toronto
Copyright  2005 by Oxford University Press, Inc.

Published by Oxford University Press, Inc.
198 Madison Avenue, New York, New York 10016
www.oup.com
Oxford is a registered trademark of Oxford University Press
All rights reserved. No part of this publication may be reproduced,
stored in a retrieval system, or transmitted, in any form or by any means,
electronic, mechanical, photocopying, recording, or otherwise,
without the prior permission of Oxford University Press.
Library of Congress Cataloging-in-Publication Data
The missing link in cognition : origins of self-reflective consciousness /
edited by Herbert S. Terrace and Janet Metcalfe.
p. cm.
Includes bibliographical references and index.
ISBN-13 978-0-19-516156-4
ISBN 0-19-516156-4
1. Self-perception. 2. Cognition in animals. 3. Psychology, Comparative.
I. Terrace, Herbert S., 1936– II. Metcalfe, Janet.
BF697.5.S43M58 2004
156′.3—dc22 2004010567
9 8 7 6 5 4 3 2 1
Printed in the United States of America
on acid-free paper
Contents
Contributors vii
1. Episodic Memory and Autonoesis: Uniquely Human? 3

Endel Tulving
2. Self-Reflective Consciousness and the Projectable Self 57

Janet Metcalfe and Hedy Kober
3. Metacognition and the Evolution of Language 84

Herbert S. Terrace
4. Emerging Levels of Consciousness in Early

Human Development 116
Katherine Nelson
5. A Continuum of Self-Consciousness That Emerges in Phylogeny

and Ontogeny 142
Marcel Kinsbourne
6. Humans as Applied Motivation Scientists: Self-Consciousness

From “Shared Reality” and “Becoming” 157
E. Tory Higgins
7. Two Normative Roles for Self-Consciousness 174

Patricia Kitcher
8. Progress in the Study of Chimpanzee Recall

and Episodic Memory 188
Charles Menzel
9. Do Nonhuman Primates Have Episodic Memory? 225

Bennett L. Schwartz
vi Contents
10. Studies of Uncertainty Monitoring and Metacognition in Animals

and Humans 242
J. David Smith
11. Can Rhesus Monkeys Discriminate Between Remembering

and Forgetting? 272
Robert R. Hampton
12. Metaconfidence Judgments in Rhesus Macaques: Explicit

Versus Implicit Mechanisms 296
Lisa K. Son and Nate Kornell
13. The Self and Other: A Missing Link in Comparative

Social Cognition 321
Josep Call
Index 343
Contributors
Josep Call
Max Planck Institute for Evolutionary Anthropology
Deutscher Platz 6
D-04103 Leipzig
Germany
call@eva.mpg.de
Robert R. Hampton
Laboratory of Neuropsychology
National Institute of Mental Health
Building 49, Room 1B-80
Bethesda, MD 20892–4415
robert@ln.nimh.nih.gov
E. Tory Higgins
Professor of Psychology and Professor of Management
Columbia University
406 Schermerhorn Hall
New York, NY 10027
tory@psych.columbia.edu
Marcel Kinsbourne
Professor of Psychology
New School University
New York, NY
Kinsbourne@aol.com
Patricia Kitcher
Professor and Chair of Philosophy
Mark van Doren Professor of the Humanities
Columbia University
Department of Philosophy
New York, NY 10027
Pk206@columbia.edu
vii
viii Contributors
Hedy Kober
Columbia University
Department of Psychology
New York, NY 10027
hedy@psych.columbia.edu
Nate Kornell
Columbia University
New York, NY 10027
nkornell@psych.columbia.edu
Charles Menzel
Language Research Center
Georgia State University
3401 Panthersville Road
Decatur, GA 30034
lrccrm@langate.gsu.edu
Janet Metcalfe
Professor of Psychology and of Neurobiology and Behavior
Columbia University
New York, NY 10027
jm348@columbia.edu
Katherine Nelson
Distinguished Professor of Psychology (Emerita)
City University of New York Graduate Center
50 Riverside Drive #4B
New York, NY 10024
knelson@gc.cuny.edu
Bennett L. Schwartz
Associate Professor
Florida International University
bennett.schwartz@fiu.edu
J. David Smith
University at Buffalo, State University of New York
346 Park Hall
Buffalo, NY 14260
psysmith@buffalo.edu
Contributors ix
Lisa K. Son
Barnard College
3009 Broadway
New York, NY 10027
lson@barnard.edu
Herbert S. Terrace
Professor of Psychology and Psychiatry
Columbia University
New York, NY 10027
terrace@columbia.edu
Endel Tulving
Rotman Research Institute of Baycrest Centre
3560 Bathurst Street
North York, Ontario, Canada M6A 2A1
tulving@psych.utoronto.ca
1
Episodic Memory and Autonoesis:

Uniquely Human?
Endel Tulving
The mental faculties of man and the lower animals do not differ in
kind, although immensely in degree.
C. Darwin, The Descent of Man
Among the games that scientists play, one of the best known is virtual
tug-of-war. Two teams position themselves at the opposite ends of an
imaginary rope that represents a continuum of nature, grab their end
of the rope, and try to pull the other team over the center line. If they
succeed, which actually seldom happens, they would declare their
end of the rope the “truth,” and themselves the winners.
One long-lasting virtual tug-of-war has to do with the nature of
the similarities and differences between the species, especially those
between humans and other animals. The existence of the basic similar-
ities has been accepted by intelligent people ever since Darwin’s the-
ory of “descent with modification” survived the harshest scrutiny that
any set of scientific ideas has had to face, but the battle over the differ-
ences continues unabated. The differences typically have to do with
humans and the species that occupy the neighboring branches on the
evolutionary tree—sometimes chimpanzees and gorillas, sometimes
all “greater apes,” sometimes all “nonhuman primates,” sometimes
even other “nonhuman animals,” or simply “brutes,” as Darwin called
all of them. And the two ends of the rope are called “qualitative”
versus “quantitative,” or, as Darwin put it, “kind” or “degree.”
The contestants at one end of the rope believe that there are no
essential differences between humans and the various “others,” and
that whatever differences may seem to exist are either minor or unin-
teresting in the broader scheme of things. Those at the other end be-
lieve that in addition to many (uninteresting or less interesting) simi-
3
4 The Missing Link in Cognition
larities between humans and others, there are some truly fundamental
differences, that there indeed may be “gaps.” So they keep tugging,
without realizing that even when one or the other team seems more
successful at a given time, the rope itself remains unchanged.
The contributors to this volume deal with the issue of similarities
and differences in the nature of consciousness, or “higher mental fac-
ulties,” of humans and other animals. Many of them will feel like
grabbing one end of the rope rather than the other. In my own case,
I take a position that others will want to interpret as revealing a pref-
erence for differences in kind. But in this case too, the rope will remain
whole.
In this essay, I argue that only human beings possess “autonoetic”
episodic memory and the ability to mentally travel into the past and
into the future, and that in that sense they are unique. Having thus
identified myself with one end of the rope, let me hasten to clarify my
position. It is important to note that my essay is not meant as delayed
rearguard action in support of the myth of scala naturae. It is not a
campaign for human superiority. It is meant simply as an acknowl-
edgement of both similarities and differences in animal kingdom, an
acknowledgment that sometimes comes hard to life scientists, as it
comes hard to those who think only of human uniqueness.
At the outset, it is useful to keep in mind the fact that uniqueness
is by no means unique to humans. Every species is unique in the sense
that it possesses features and traits that other species do not. Some of
these unique features may be more conspicuous than others, and
some may be better known than others, but the basic principle re-
mains the same: If there were no differences between Species A and
B, they would be the same species.
Scorpions locate their prey by seismic vibrations; bats can catch
small insects by echolocation in midflight in the darkness of the night;
electric fish can perform comparable feats through electrolocation, in
ways that scientists have yet to completely fathom; migrating birds
know how to reach destinations thousands of miles away. The list of
these and similar kinds of spectacular achievements by creatures of
nature can go on and on. But neither such unique features nor many
less conspicuous differences rule out the general principle of broad
phylogenetic continuity. Equally important, broad phylogenetic conti-
nuity does not rule out differences between the species, even those
that to an external observer may seem like gaps. Diversity in nature
can take many diverse forms.
The scientific question of interest is not whether the human mind
is similar to or different from those of other animals. Nor is it whether
phylogenetic evolution is continuous or whether there are gaps in it.
The answer to both questions is either “neither” or “both,” and neither
Episodic Memory and Autonoesis 5
of these answers is terribly informative. A more appropriate question

is how and in what sense are the minds of two species similar and
how and in what sense are they different, or, how and in what sense
the phylogenetic evolution is continuous and in what sense it is not.
In the context of these sorts of questions, it may even be appropriate
to speculate, as some contributors to this volume do, how, why, and
when the current situation might have come about.
It ought to be completely noncontroversial to say that nonhuman
animals are conscious, that they have memory, that they can learn,
that they can think, that they know what happens in their world—in
brief, that they are “intelligent” (Weiskrantz, 1985). It ought to be
equally noncontroversial to say that the consciousness of other species
is probably different from ours, that their awareness of the world is
different from ours, that they are more skilled at some and less skilled
at other forms of behavior than we are, and that in general their intel-
ligence is different from ours. Finally, everybody also knows, of course,
that in terms of the ultimate biological criterion, survival, an over-
whelming majority of the other species are as capable and successful
in their world as we are in ours (Darwin, 1874/1998; Macphail, 1998;
Shettleworth, 1998).
It is against the backdrop of these musings—the premises of the
argument—that I present the thesis (hypothesis) of this essay: Human
beings possess a form of memory (episodic memory) and a form of
consciousness (autonoetic consciousness, or “autonoesis”) that no
other animals do. Thus, the thesis is that these two aspects of the mind
are unique in humans, in the sense that the mental capacities that
define them do not exist in quite the same full-fledged form in other
species. They do not exist in insects, in birds, in mice or rats, in cats
or dogs, and not even in gorillas and chimps.
Some see the thesis as representing self-evident truth; others may
think of it as woefully misguided; still others view it as little more
than idle speculation that cannot possibly get us anywhere. Indeed, it
is possible to argue that no essential progress has been made toward
resolving the issues involved for more than a hundred years. So why
try again? There are at least two reasons. One, much of the previous
debate has involved issues that have been formulated too broadly to
allow an incisive approach. Narrowing it down may help. Two, today
we have a bit more evidence relevant to the issues than was available
yesterday. Therefore, it may be worth looking at it again.
In this essay, I narrow the issue down to one kind of consciousness
(autonoetic) that characterizes the workings of one kind of memory
(episodic), and discuss it in light of the empirical findings that can
be seen as pertinent to it. The essay consists of five sections, plus a
summary:
1. Thesis: Mental time travel exists in humans only.

2. Theory: This thesis is a part of a more comprehensive story
of the human brain/mind, the theory of episodic memory.
3. Data: People exist who are fully competent mentally, who
have no episodic memory, yet manage to do quite well in
staying alive.
4. Analysis: What we know about episodic memory in animals.
5. Resolution: It is in principle possible to obtain solid evidence
for the existence of episodic memory in nonverbal organisms.
THE THESIS
In 1983, I wrote a book titled Elements of Episodic Memory, in which I
discussed the possibility that episodic memory is functionally differ-
ent from other kinds of memory. The book opened with the following
paragraph:
Remembering past events is a universally familiar experience. It is also
a uniquely human one. As far as we know, members of no other species
possess quite the same ability to experience again now, in a different situa-
tion and perhaps in a different form, happenings from the past, and know
that the experience refers to an event that occurred in another time and in
another place. Other members of the animal kingdom can learn, benefit
from experience, acquire the ability to adjust and adapt, to solve problems
and make decisions, but they cannot travel back into the past in their own
minds. (Tulving, 1983, p. 1)
This paragraph set out, somewhat guardedly and rather fortuitously,
of course, the main point of the present chapter—episodic memory is
uniquely human. Reading it 20 years later, it is not difficult to see that
it shows its age—I would not say exactly the same thing now. By and
large, however, it does not seem to be terribly out of line.
Back in 1983, the claim about the differences between humans and
nonhuman animals was introduced primarily as a device of getting a
head start on the formidable task of trying to explain to the reader
what the new kind of memory—“episodic memory”—was all about.
At the time, the study of human memory was still trying to shake the
powerful behaviorist or behavioralist (Leahey, 1987) influences of the
past, influences in which the idea of unity of all learning and memory,
between and within the species, held a dominant position. The throw-
away comment on humans sharing not quite all of their learning and
memory capabilities with their furry and feathered cousins was sup-
posed to provide a graphic description of the differences between epi-
sodic and other kinds of memory.
The casual reference to “mental time travel” similarly served as a
device for distancing episodic from other forms of memory. From the
beginning (Tulving, 1972) it had been useful to contrast episodic mem-
ory with what, for historical reasons, was called “semantic” memory.
Because semantic memory has many features in common with epi-
sodic memory—as we will see in greater detail later in the chapter—it
was easy for people not to take any proposed differences seriously.
The idea of mental time travel was pressed into service as a device
that might help others see the difference: Episodic memory did,
whereas semantic memory did not, necessarily involve any such time
travel.
Finally, the paragraph mentions, even if somewhat indirectly, an
idea about episodic memory that sometimes tends to be overlooked,
an idea that turns out to be relevant to the issue of the relation be-
tween episodic and “episodic-like” memory (Clayton & Dickinson,
1998; Clayton, Bussey, & Dickinson, 2003). The idea is that a happy
owner of episodic memory can think about personal happenings
(“what”) that took place at another time (“when”) and in another
place (“where”)—that is, thinking in Place P1 about a personal experi-
ence in Place P2. The ability to “mentally travel in space” is, as we
will see, a property of semantic (declarative) memory and is therefore
a precondition for mental travel in time, in keeping with the general
idea that episodic memory “grows out of,” and represents an exten-
sion of, semantic memory (Tulving, 1984). Mental “space travel”
(imagining different spatial locations) does not require mental time
travel (imagining oneself at different times). The present point is that
mental time travel always occurs not only in subjective time but also
in mental space, and that the mental space of the remembered past
and imagined future may be different from the present space. Individ-
uals with (autonoetic) episodic memory can, if the situation calls for
it, think here and now about personal happenings in other places and
other times. Therefore, if we ask whether other animals have episodic
memory, we ask, among other things, whether they can also do so.
As it turned out, until recently not much fuss was made about what
the 1983 paragraph said about episodic memory of other animals. Not
only had similar thoughts been expressed by many others, they were
peripheral to the main message of the book. Just about the only person
who rose to defend the impugned repute of our furry friends was
David Olton. He had proposed the distinction between working mem-
ory and reference memory (Olton, Becker, & Handelmann, 1979, 1980;
Olton & Pappas, 1979), and he suggested that these two parallelled
the distinction between episodic and semantic memory (Olton, 1984,
1985). At the time it was easy enough to agree with him, as I did
(Tulving, 1983, pp. 118–119). Now, thirty years later, the issues have
become a bit more complicated, as they frequently do in living sci-
ence. Let us consider two relevant points.
First, there is no proof that the thesis is true. Indeed, because of its
logical nature, it will always remain impossible to prove its truth. The
thesis is based on the absence of evidence contrary to it, and its rela-
tively smooth integration into a larger picture for which supporting
evidence is available. The absence of evidence, as everyone knows, is
not evidence for the absence, and that is why more evidence is needed.
The fit between the thesis and the larger picture (the theory of epi-
sodic memory) can be seen as a part of “more evidence.” Science does
not progress by proving hypotheses true to begin with; it progresses
by proving hypotheses false. In the present case, too, it is possible to
prove the hypothesis false. In the final section of the essay, I make a
suggestion as to what would constitute evidence that the hypothesis
is untenable.
Second, the neuropsychological and developmental findings that I
summarize in this chapter are relevant to the argument in at least
three senses. First, they help delineate more precisely what the mental
faculties are that, according to the thesis, nonhuman animals do not
have. Second, they illustrate the procedures used to distinguish be-
tween different kinds of learning and memory. If there were no way
to separate (autonoetic) episodic memory from other forms of mem-
ory, the claim that nonhuman animals do not have it would be hol-
low. No one has much doubted the usefulness of the concept of epi-
sodic memory for classificatory (“bookkeeping”) purposes, but there
has been considerable resistance to the idea that it represents anything
special in biological reality. Nevertheless, steady progress in the accu-
mulation of pertinent evidence makes the claim that only humans
have episodic memory more readily tractable. Third, the findings are
also relevant insofar as they help support the idea that episodic mem-
ory is not necessary (even) for humans. If so, the proposition that it
does not exist in other animals is less shocking than it would be other-
wise. It also would make more acute the question as to why episodic
memory may have evolved.
It is reasonable to assume that episodic memory evolved recently,
although there is as yet no way to find out whether the assumption
is true and, if so, how recently. Because chimpanzees, our closest rela-
tives on the evolutionary tree, do not have humanlike episodic mem-
ory, it makes sense to assume that our common ancestors did not
have it either. The alternative possibility, that it had evolved in other
species too but was somehow lost in the course of evolution, is diffi-
cult to imagine. We will return to this issue presently.
THEORY: EPISODIC MEMORY AND AUTONOETIC CONSCIOUSNESS

Most readers of this volume know that episodic memory has to do
with remembering personally experienced events. But there are varia-
tions on the theme. Therefore, before we can undertake the task of
evaluating the presence of episodic memory in nonhuman animals,
the concept needs to be sharpened. Just saying that it concerns per-

sonal happenings is too vague and leaves too much room for nonfo-
cused debate. We need to be as clear as possible about the kind of
memory that I am denying to our feathered and furry friends.
Let us begin with a thumbnail sketch, or definition, of episodic mem-
ory. Because definitions do play a role in the study of nature, even in
today’s dominant Zeitgeist of “exploratory science,” and because defini-
tions have a habit of changing, it is helpful to identify definitions in a
way that sets them apart from others in their class. This is why I refer
to the present definition as “episodic memory—2004”:
Episodic memory is a recently evolved, late developing, and early dete-

riorating brain/mind (neurocognitive) memory system. It is oriented to the
past, more vulnerable than other memory systems to neuronal dysfunc-
tion, and probably unique to humans. It makes possible mental time travel
through subjective time—past, present, and future. This mental time travel
allows one, as an “owner” of episodic memory (“self”), through the me-
dium of autonoetic awareness, to remember one’s own previous “thought-
about” experiences, as well as to “think about” one’s own possible future
experiences. The operations of episodic memory require, but go beyond,
the semantic memory system. Retrieving information from episodic mem-
ory (“remembering”) requires the establishment and maintenance of a spe-
cial mental set, dubbed episodic “retrieval mode.” The neural components
of episodic memory comprise a widely distributed network of cortical and
subcortical brain regions that overlap with and extend beyond the net-
works subserving other memory systems. The essence of episodic memory
lies in the conjunction of three concepts—self, autonoetic awareness, and
subjective time.
Note that episodic memory is conceptualized as a (hypothetical)

brain/mind (neurocognitive) system, as indeed is semantic memory,
mentioned here in passing. The concept of “memory system” as such
has been discussed elsewhere (Nadel, 1994; Schacter & Tulving, 1994;
Sherry & Schacter, 1987), and attempts to describe episodic memory
in system terms have been made previously (Tulving, 1985a, 1998,
2002b; Tulving & Markowitsch, 1998; Wheeler, Stuss, & Tulving, 1997;
see also Nyberg et al., 2000).
Given, then, that the term episodic memory in this essay refers to the
“episodic memory system,” it is important to note what the term does
not refer to. It does not refer to a particular kind of memory task, or
a particular kind of performance measure in a task, or a particular
kind of stored information, or a particular kind of phenomenal experi-
ence, or to any alternative sense in which it sometimes appears in the
contemporary literature. All these other senses are related to episodic
memory, but they are not identical with it (Tulving, 2000). A great
deal of confusion and futile debate can be avoided by keeping the
distinction between episodic memory and related terms firmly in mind.
Like all other systems, episodic memory consists of a number of

interacting neural and cognitive components which together are capa-
ble of operating in a manner that the same components in isolation,
or in different combinations, cannot. Like all other neurocognitive sys-
tems, it is complex and not easily summarized in a few words. It is
defined in terms of criteria such as the system’s function—what the
system does, how it works, the kind of “information” that it deals
with, its relations to other systems, and its neural substrates (Schacter
& Tulving, 1994; Schacter, Wagner, & Buckner, 2000; for a different
conceptualization of memory systems, see Eichenbaum, 2000; Gaffan,
2001). Convergent dissociations, observed across tasks or subjects,
play an important role in the identification of the properties of a mem-
ory system (Nyberg & Tulving, 1997), but they are not the only fea-
tures of the definition.
An alternative to the classification of memory in terms of multiple
systems is classification in terms of “kinds” (or “varieties”) of memory
(Roediger, Marsh, & Lee, 2002). The concept of “kinds” of memory,
as dealt with by Roediger and his colleagues, is not the same as the
concept of memory “systems” (Schacter & Tulving, 1994). For exam-
ple, Roediger and his colleagues discuss episodic memory, as one of
the different “kinds” of memory, primarily in terms of retention per-
formance of healthy adults in experiments involving verbal to-be-
remembered materials. In their conceptualization of episodic memory,
it would be difficult to raise coherent questions about episodic memo-
ry’s existence in animals, or to claim that animals do not have it.
Properties of Episodic Memory

Because the search for the identity of the constituent components of
episodic memory has barely begun, not much is yet known about
these components. An initial step in the appropriate direction, how-
ever, has been taken in the form of postulation of certain “functional
properties” of episodic memory. An early, not completely successful
attempt at such a formulation was made in Tulving (1983, table 3.1),
at a time when few research findings were available to distinguish
between episodic and other varieties of memory.
A somewhat more (empirically) disciplined list of the properties,
or features, of episodic and semantic memory systems, modified from
Wheeler et al. (1997), Tulving and Markowitsch (1998), and Griffiths,
Dickinson, and Clayton (1999), is presented in table 1.1. The list con-
sists of a number of key terms or brief phrases, each of which could
be greatly extended. It is organized into two sublists.
Both sublists are necessary to convey the full scope of episodic
memory. The sublist on the left gives the properties of episodic mem-
ory that it has in common with semantic memory, whereas the sublist
Table 1.1 Features of Episodic Memory
Semantic/Episodic Memory—Common Features Episodic Memory—Unique Features
• Key function: Knowing—Registering, stor- • Key function: Remembering—

ing, and using sharable knowledge of the Conscious awareness of hap-
world penings in subjective time
• Multimodal input (chronesthesia)
• Transmodal storage • Makes possible mental time

travel in both temporal direc-
• Fast encoding operations—single-trial
tions, past and future
learning possible
• Operations accompanied by
• Large, complex, highly structured storage
autonoetic conscious aware-
• Stored information is representational— ness
isomorphic with what is, or could be, in
• Operations depend on a re-
the world
membering self
• Stored information is propositionalizable
• More recently evolved than
• Stored information has truth value other memory systems
• Stored information can be used as a basis • Ontogenetic development lags
of inferences behind other memory systems
• Information processing is highly sensitive • More vulnerable to disease, in-
to context jury, and aging
• Stored information can be accessed flexibly • Operations require the estab-
• Stored information is expressed symboli- lishment and maintenance of a
cally special neurocognitive set—
• System is cognitive—contents can be episodic retrieval mode
thought about • Operations depend on seman-
• Behavioral expression is optional and not tic memory
obligatory • Episodic remembering implies
• Operations do not require awareness of semantic knowing, but seman-
time tic knowing does not imply re-
membering
• Operations accompanied by noetic con-
scious awareness • Dependent on prefrontal cor-
tex and other neocortical re-
• System interacts closely with other neuro- gions in a way that other sys-
cognitive systems, such as those involved tems are not
in language, affect, and reasoning
• Probably unique to humans
• Dependent on widely distributed cortical
and subcortical neural networks, including
temporal lobe and diencephalic structures
• Present in a wide range of animals; highly
evolved in mammals and birds
11
on the right gives “unique” properties of episodic memory, properties

shared neither by semantic memory nor by any other memory system.
The (partial) overlap between the properties of episodic and semantic
memory reflects the idea that episodic memory evolved “out of” se-
mantic memory (Tulving, 1984).
The two sublists imply that it is possible to define episodic memory
broadly or narrowly. The broad definition comprises all the properties
listed in table 1.1, both those on the left and those on the right. The
narrow definition includes only the properties listed on the right.
The thumbnail sketch of “episodic memory—2004” presented ear-
lier corresponds to the narrow definition. It is important to note that
the thesis of this essay also applies only to the narrow (“unique”)
definition of episodic memory. It does not apply to the properties that
episodic memory shares with semantic memory. Quite to the contrary,
the shared properties, those that define semantic memory, do very
well in describing learning and memory capacities possessed by many
animals, including birds and mammals. Indeed, an important part of
the argument here is that many behavioral feats in nonhuman animals
that tend to be attributed to humanlike episodic memory are, accord-
ing to theory, manifestations of a highly complex and powerful mem-
ory system, namely semantic (or declarative) memory.
It has been suggested elsewhere (Tulving & Markowitsch, 1998)
that “declarative memory” (Squire, 1992; see also Squire & Kandel,
1999) can be thought of as representing features common to episodic
and semantic memory, that is, features listed on the left in table 1.1.
Defined along these lines, semantic (declarative) memory is a large,
complex, multimodal system capable of fast (“single-trial”) encoding
operations. Information that it handles is representational and can be,
even if it need not be, described in propositional format. The represen-
tational information has truth value: it corresponds to objects, events,
relations, and states of the world. The information is accessible and
expressible flexibly, that is, through different input and output routes,
and it can serve as a basis of inferences about other objects, events,
relations, and states of the world, those that may not have been di-
rectly experienced. The operations of semantic (declarative) memory
are cognitive (not rigidly tied to behavior) and sensitive to context.
These characteristics of semantic (declarative) memory fit reason-
ably closely descriptions of declarative memory in the literature (Co-
hen, Poldrack, & Eichenbaum, 1997; Eichenbaum, 1997; Knowlton &
Squire, 1995; Squire, 1987).
It is important to note that neither semantic nor episodic memory
as defined here depends on language or any other symbol system for
its operations, although both systems in humans can greatly benefit
from language. “Semantic memory” is a designation that is what it is
for purely historical reasons; a more fitting term for it is “knowledge
of the world.” Language may have played an important role in the

evolution (or coevolution) of human semantic memory, and probably
even more so in the evolution of episodic memory, and it can greatly
facilitate the operations of memory and learning systems, but it is not
necessary for such operations.
Relations Between Episodic and Semantic Memory

The two sublists of table 1.1 not only distinguish between two defini-
tions of episodic memory, broad and narrow, they also help describe
the relation between episodic and semantic memory. The thumbnail
sketch earlier said that “the operations of episodic memory require,
but go beyond, the semantic memory system.” We can now say that
episodic memory represents an extension of semantic memory, both
in terms of its emergence in the course of evolution and in terms of
its operations. This kind of relation between the two systems, initially
sketched in terms of the “embeddedness” hypothesis (Tulving, 1984,
p. 260), was later reformulated as the serial parallel independent (SPI)
model (Tulving, 1993a, 1995).
The structure of the SPI model reflects the assumed sequence of the
two kinds of memory in both phylogenetic evolution and ontogenetic
development: semantic memory precedes episodic memory (Nelson,
1993, see also chapter 4, this volume; Perner & Ruffman, 1995; Sud-
dendorf & Corballis, 1997; Wheeler et al., 1997). The importance of
this feature of the model lies in the fact that it allows those organisms
that have no episodic memory to acquire knowledge of the world,
but it rules out acquisition of episodic information without semantic
memory. That is, episodic memory is not necessary for the operation
of any other memory and learning system. This is why nonhuman
animals can do perfectly well without episodic memory.
The functional relations between episodic and semantic memory in
the SPI model are process-specific, that is, the relations depend on
the processes. Encoding processes are organized serially (S), storage
processes in parallel (P), and retrieval processes in the two systems
are independent (I). The seriality of encoding of information “into”
episodic memory means that the information must be first processed
in semantic memory and that, therefore, impairments of semantic
memory have consequences for the operations of episodic memory.
However, episodic memory is not required for encoding information
into semantic memory: organisms with impaired or totally missing
episodic memory (nonhuman animals, young children, some brain-
damaged patients) are capable of acquiring (complex) knowledge
about the world even if they cannot remember (autonoetically recol-
lect) anything of their own past lives. In other words, seriality of en-
coding implies that only single dissociations are possible in encoding
(Tulving, 1995). Parallellity of storage means that corresponding infor-
mation may be stored in both systems. And independence of retrieval

means that stored information can be retrieved from one system or
the other system, or both, thus allowing for both single and double
dissociations between retrieval of episodic and semantic information
(for more details on SPI, see Tulving, 2001a).
Uniqueness of Episodic Memory

The features listed in the right-hand column in table 1.1 define the
uniqueness of episodic memory. Four of them can be thought of as
being central: (1) episodic memory’s function, namely that of mental
time travel, or “remembering”; (2) episodic memory’s dependence on
a remembering “self”; (3) the expression of such remembering
through the self’s autonoetic consciousness, or autonoesis; and (4)
episodic memory’s relation to subjectively apprehended time, or
“chronesthesia” (Tulving, 2002a).
First, episodic memory’s function is to enable mental time travel or
remembering, that is, to make it possible for the rememberer to travel
back in his or her mind to an earlier occasion or situation in the re-
memberer’s life, and to mentally relive the experienced and thought-
about happenings. Semantic memory cannot do so, at least not with a
comparable efficacy. Episodic memory allows the individual to re-
member now, at Time 2, something about what happened at an earlier
Time 1, as well as know what happened. Because the output of epi-
sodic memory can serve as an input into semantic as well as episodic
memory, the remembering individuals can also know what happened
on an occasion, even if they do not remember the happening or the
occasion.
Only individuals who have episodic memory are capable of re-
membering past events, in the sense of mental time travel. Other cog-
nitive memory systems may provide access to the past, including
one’s personal past, but they do so in the absence of autonoesis. For
example, both healthy people and those with certain kinds of brain
damage can recite autobiographical facts about themselves without
necessarily remembering anything. Amnesic patients can tell stories
about themselves that have to do with the past, but in doing so they
rely on semantic memory (“personal semantic memory” as suggested
by Cermak & O’Connor, 1983).
Sherry and Schacter (1987) have proposed that new memory sys-
tems evolve when existing systems cannot satisfy the adaptive needs
of the organism. The proposal that episodic memory evolved to serve
a function not served by other systems, including semantic memory,
fits well into Sherry and Schacter’s scheme.
Second, episodic memory differs from other kinds of memory in
that its operations require a self. It is the self that engages in the men-
tal activity that is referred to as mental time travel: there can be no
travel without a traveler. If it is not self that does the traveling, then
who, or what? “Self” and “self-awareness” are among those terms
that are indispensable for discussing phenomena of the mind, yet
have many meanings that are difficult to define and explicate (Kircher
& Leube, 2003). We can think of self as the traveler who engages in
mental time travel. Like other components of the system that is epi-
sodic memory, self too is defined in terms of its properties, and in
terms of its relations to other components of the system.
Some thinkers prefer a philosophical framework for the scientific
approach to mental life in which the phenomena to be explained are
expressions of processes, but in which the entities that do the process-
ing (agents) are not permitted. Thus, thinking occurs without thinkers,
knowing without knowers, and consciousness without anyone being
conscious. The idea is to avoid using structural terminology, because
it is not fashionable these days. Besides, “self” sounds like the
dreaded homunculus that needs to be exorcised by all means possible
(Noelle, 2001). But until such time that we have better ways of ex-
plaining the phenomenal existences of things such as pain, smell, and
recollection of the past, we need an agent such as self for the sake of
the completeness of the story (theory). Eventually, self may turn out
like phlogiston or aether—a convenient temporary prop. But the
problem today is that the story of the mind is incomplete and awk-
ward to tell if a concept like “self” is omitted from it. This is why it
should not be exorcised yet.
Third, the conscious awareness that characterizes remembering is
different from other forms of conscious awareness, and different in an
unmistakable manner. When you remember an event, however
vaguely, you are aware that the present experience is related to the
past experience in a way that no other kind of experience is. You do
not confuse it with perceiving, or imagining, or dreaming, or halluci-
nating, or having thoughts about what is or could be in the world.
In order to be able to refer to the kind of conscious awareness that
characterizes episodic remembering, the term “autonoetic conscious-
ness” has been proposed (Tulving, 1985b; Wheeler et al., 1997). The
term “autonoesis” was introduced later as a (shorter) synonym for
“autonoetic consciousness” (Tulving, 2001b). Autonoesis refers to the
kind of conscious awareness that characterizes conscious recollection
of personal happenings. It refers to what William James (1890) proba-
bly had in mind when he talked about the “warmth and intimacy” of
remembering one’s past experiences. (The relation between conscious-
ness and awareness is discussed in Tulving, 1993c; for a somewhat
different conceptualization of consciousness and awareness, see Chal-
mers, 1996, especially pp. 28–29.)
Many organisms without episodic memory possess a highly func-
tional semantic memory. Retrieval (use) of information from semantic
memory occurs whenever an organism thinks about, or mentally rep-

resents, something that exists or could exist in the world. Such re-
trieval (“knowing”) is accompanied by a form of conscious awareness
that I refer to as noetic awareness. Both people and animals are fully
capable of knowing things about the world, including things that oc-
curred in their own past. The essential difference is between knowing
that something is such and such, or occurs so and so, on the one hand,
and remembering that one had a particular experience (witnessed, or
felt, or thought something) in a particular place at a particular time.
Neither the knowledge of the facts nor the experiencing of past events
needs to be complete or accurate in any sense, of course. Nor is it
necessary to assume that the distinction between knowing and re-
membering is always sharp. In real life, many thoughts a person has
may have elements contributed by both the semantic memory (know-
ing) system and the episodic memory (remembering) system.
Episodic Memory, Autonoesis, and Time

The fourth central feature of episodic memory, and most relevant in
the present context, has to do with episodic memory’s unique relation
to time. Unlike all other forms of biological memory, or memory sys-
tems, episodic memory is oriented to time. All other forms of memory
operate in time, as does everything else in life, but only episodic mem-
ory allows people to consciously reexperience past experiences. Its
special relationship to time is neither widely known nor adequately
appreciated. Most people naturally associate all kinds of memory with
the past and are surprised to learn that this is not so.
Episodic remembering is oriented to the past. When William James
(1890) and Henri Bergson (1911) wrote about the “pastness” of mem-
ory, they had in mind what we now refer to as episodic memory, even
if they did not use the term. At the time, there was no need for it,
because what writers meant by “memory” referred either to episodic
memory or to the combination of episodic and semantic memory, that
is, declarative memory (Squire & Kandel, 1999). Note that Bennett
Schwartz, in chapter 9 in this volume, uses the term “palinscopy” to
refer to the same idea of “pastness.”
The time in which episodic memory operates is the same in which
all physical and biological events occur, physical time. But the time in
which remembered events occur is different. We can call it subjective
time. It is related to but not identical with physical time. The relation
between physical time and subjective time is analogous to the relation
between the physically present stimulus energy and the psychologi-
cally experienced awareness of the corresponding aspects of this en-
ergy, as studied in the venerable science of psychophysics. A falling
tree in the forest will produce a physical disturbance in the air even
if no living thing is nearby, but it produces a sound only if a living
thing hears it. Sound exists as surely as do moving air molecules, but
it exists solely by virtue of the interaction between the moving air
molecules and brains equipped with auditory systems. The same is
true of subjective time—it too exists, but only by virtue of the interac-
tion between physical time and the part of the brain/mind that we
call (autonoetic) episodic memory.
Philosophers and others (Dalla Barba, 2000, 2001) sometimes write
about the “paradox” that phenomenal experience of pastness emerges
from an aftereffect of the past (engram, or memory trace) that itself
exists in the present. How is this possible, they wonder? How is it
possible for what was to be present in what is?
From the perspective of episodic theory, there are two points to be
made about the paradox. First, it exists, or may appear to exist, only
for episodic memory. Because there is no pastness in any other mem-
ory system in humans, and no pastness in any memory system in all
other creatures, the paradox, even if it were “real” otherwise, would
not be a paradox of memory in general. Second, human episodic past-
ness does not reside in memory traces as such; it emerges as the
phenomenally apprehended product of the episodic memory system,
autonoetic consciousness, in ways that are as mysterious as the emer-
gence of other kinds of consciousness from brain activity. Dalla Barba
(2000, 2001) similarly attributes the feeling of pastness to a special
kind of “temporal consciousness.”
Autonoetic Future
Subjective time not only covers the past; it also extends into the future.
The forward-looking sense of subjective time, or “proscopic chronesthe-
sia” (Tulving, 2002a), is especially noteworthy, because it represents a
key feature of autonoesis, which plays such an important role in the
human condition. Anticipation of, thinking about, and explicitly or
implicitly preparing and planning for the personal future seems to be
a thoroughly quintessential human activity. It pervades an individu-
al’s daily life, from early in the morning till late at night, as it pervades
his or her entire life course, from childhood to old age. If we retained
all our other marvelous mental capacities but lost the awareness of
the future time in which our lives are going to be played out, we
might still be radically different from other animals, but we would no
longer be human as we understand humanness.
This remarkable ability of humans to be aware of our own future
has already been subjected to thought and study (Atance & O’Neill,
2001; Clayton et al., 2003; Haith, 1997; Haith, Benson, Roberts, & Pen-
nington, 1994; Ingvar, 1985; Klein, Loftus, & Kihlstrom, 2002; Sudden-
dorf & Corballis, 1997), but nowhere yet near to the extent that it
deserves. A major stumbling block in this study might have been the
absence of a clear conceptual distinction between future-oriented be-
havior and future-oriented thought (cognition). Frequently, the two

have been treated as directly related, future-oriented behavior being
seen as based on and reflecting future-related thought. Although such
concordance between behavior and cognition, and behavior and con-
sciousness, undoubtedly occurs, it does not follow that it always oc-
curs. It is quite possible to imagine that some (even highly complex)
future-related behaviors, or behaviors that can be seen as such, occur
quite independently of any deliberate conscious activity. As in many
other instances of apparent concordance between behavior and cogni-
tion, the question about the nature of the relation between behavior
and thought will have to be answered on the basis of empirical study;
it cannot be solved by assumption (Tulving, 1989a).
Semantic memory, as defined by the features listed in the left-hand
column of table 1.1, cannot serve the same function of remembering,
in the sense of mental time travel, as does episodic memory. Unlike
episodic memory, semantic memory has no special relation to time.
Indeed, as already stated, and popular misconceptions notwithstand-
ing, no kind of memory other than episodic has any special relation
to time. Semantic memory allows the individual to know, at Time 2,
something about what happened at an earlier time, Time 1, but it
does not allow the individual to remember what happened. Semantic
memory also allows an individual to construct possible future worlds,
but since it is lacking autonoetic capability, it would not allow the
individual to mentally travel into his own personal future.
Those without episodic memory can learn, retain what they have
learned, make use of what they have retained—and thus they have
memory in this sense—but they cannot remember. An event happens
and is registered at Time 1 that may have consequences for behavior
or cognition at a subsequent Time 2. But this temporal sequence of
stages of knowledge acquisition is in no way special to memory. Tem-
poral sequencing of this kind does not distinguish semantic memory
from other temporal cause-and-effect sequences in, say, eating a meal,
or being vaccinated against malaria, or growing old as one goes
through life. There is no need whatsoever for any temporal marker to
be attached to and retained about the learned facts of the world. The
facts can be put to good use by the owner of a semantic memory
system regardless of whether episodic memory is functional or not. It
is crucial to distinguish between the consequences of an experienced
event and the remembering of the experienced event (Lockhart, 1984).
Darwin’s Moral Beings

When Charles Darwin contemplated the differences in mental charac-
teristics of his subjects in Descent of Man, he did make an exception in
his insistence on total continuity otherwise:
A moral being is one who is capable of reflecting on his past actions

and their motives—of approving of some and disapproving of others; and
the fact that man is the one being who certainly deserves this designation,
is the greatest of all distinctions between him and the lower animals. (Dar-
win, 1874/1998, p. 633)
Darwin’s additional writing provides a rather detailed description

of what he conceives of as a moral being. Interestingly, this descrip-
tion includes features of such a being that correspond to some of the
central features of episodic memory: (1) conscious reflection, (2) self,
(3) personal past, and (4) ethical judgment. Thus we see that Darwin
anticipated, implicitly at least, the idea that nonhuman animals do not
have what we now call autonoetic consciousness. Indeed, by adding
man’s “ethical aspects” to the equation, he goes beyond it.
It is especially instructive to note that Darwin’s moral beings were
concerned about the future. This means, by our theory, that they must
have had autonoetic consciousness. This raises the question: Can one
be a moral being in the absence of autonoesis?
I have to leave the exercise of determining the nature of the relation
between autonoesis and moral judgment to others, although at first
glance it does not seem to be unreasonable to think of autonoesis as
a necessary condition for moral judgment, rather than vice versa. If
this is so, then the inability of Darwin’s “beasts,” like that of very
young children, to differentiate right from wrong has its roots in their
lack of ability to mentally travel into their own personal futures. Dar-
win’s position on human uniqueness also covers the major properties
of Metcalfe and Kober’s concept of projectable self—the sensing,
imagining, willing, and doubting self (chapter 2, this volume).
Why Did Episodic Memory Evolve?

Why did autonoetic consciousness and episodic memory emerge in
the process of evolution? Wherein lies their evolutionary payoff?
What can organisms with autonoetic consciousness do that organisms
without it cannot?
Robert Hampton, a contributor to this volume, and a researcher on
the opposite end of the rope, put these very questions to me in e-mail
correspondence some time ago:
Thus I would like to know what the possession of episodic memory
does for an individual. What additional capacity is provided when one has
episodic memory? This needs to go beyond the mere ability to report “I
remember that happening to me. My experience of this memory is distinct
from others.” In other words, what can an animal with episodic memory
do that gives it an evolutionary advantage over an animal without episodic
memory? Having different covert experiences by itself cannot lead to selec-
tion for episodic memory. These experiences need to be turned into adap-
tive behavior if one can specify what that adaptive behavior is—what an
animal with episodic memory can do that another cannot—then one can
do an experiment to test for that capacity. (Personal communication, July
2, 2002)
Questions of this kind—why did X evolve—of course are difficult to
answer in any case, but when asked about mental capabilities, which
do not leave any fossils, answering them becomes essentially impossi-
ble, and one can only speculate.
One possible story begins with the assumption that the common
ancestors that the species that eventually became Homo sapiens shared
with what became living pongids, some 5 or 6 million years ago, pos-
sessed capabilities that we now identify with semantic memory, but
they did not possess episodic memory. Episodic memory emerged,
presumably gradually, in the course of human evolution. It may have
grown out of a gradual extension of the human mental reach farther
and farther back into subjectively apprehended past, perhaps as a sort
of temporal stretching of the duration of the subjectively experienced
here and now. There are clear evolutionary advantages to being con-
sciously aware not only of what is happening here and now, but also
of what happened 5 seconds ago, 10 seconds, a minute, 10 minutes,
an hour, a day ago, even if the dimness of such awareness increased
monotonically with the retention interval. A special evolutionary leap
may have been necessary to produce brains that were capable of
bridging the remembering across the diurnal divide. But these unfath-
omable specifics aside, it is a fact that humans somehow acquired the
ability to remember their experienced past, in addition to the earlier
acquired skill of knowing of things in the present.
Along with such an expansion of the subjective time horizon to-
ward the past in remembering occurred a similar, even if possibly
more muted, expansion toward the future. Once the brain “discov-
ered” the trick of representing subjective time and making access to it
available to the evolving self, through similarly evolving autonoetic
consciousness, our distant forbears came to live with the capability of
awareness of subjective time in which they and their group existed,
their ancestors had existed, and their children and their children were
going to exist.
Evolutionary Payoff: Awareness of the Future
The truly momentous development was the emergence of autonoesis
that allowed individuals to imagine and worry about their own fu-
ture, because it brought with it a radical shift in humans’ relation to
nature. Instead of using their wits to adjust to the vagaries of nature,
including the uncertainties of availability of food, shelter, and protec-
tion from predators, humans began to anticipate these problems and
take steps to mitigate their unpredictability.
Other early expressions of future-oriented thought and planning

consisted of learning to use, preserve, and then make fire, to make
tools, and then to store and carry these with them. Furnishing the
dead with grave goods; growing their own crops, fruits, and vegeta-
bles; domesticating animals as sources of food and clothing; creating
the spirit world and endowing its inhabitants with powers that ex-
plained otherwise unexplainable natural phenomena, and then in-
venting ways of placating the spirits through rituals and other proper
ways of behavior—these all represent relatively recent developments
in human evolution. Every single one is predicated on the awareness
of the future.
Future-oriented (proscopic) time sense (chronesthesia) was only
one of the necessary conditions that had to be simultaneously satisfied
for human culture and civilization to come into being and then to
proceed and flourish. But, at least according to the results of the
thought experiments that I mentioned, the requirement for autonoetic
awareness of the future is one of the more stringent ones.
Hampton (personal communication, July 2, 2002) is right in sug-
gesting that if inner thoughts always remain inner thoughts, that is, if
they do not affect behavior, they will die with the one who thinks
them. They need to be translated into action to make any difference
in the world. The question therefore is, what kinds of inner thoughts,
or covert experiences, lead to what kind of action. The world is full
of living things that may be said to think, and then act on their
thoughts. Or at least it is possible to interpret many observable activi-
ties in this way. But an overwhelming majority of these thoughts and
the ensuing actions are motivated by the creatures’ current needs and
are directed at the environment that exists here and now. These
thoughts and actions do not change the world from one to which all
living creatures must adapt, to one that better satisfies their needs.
These kinds of musings do contain one possible answer to Hamp-
ton’s query about the biological utility of covert experiences. Inner
thoughts, phenomenal awareness of one’s world, need not but in fact
often do lead to overt action. One notable example of such action is
doing something now that will pay off only in the future. Humans
with autonoetic episodic memory think about the future, anticipate
what challenges and rewards it will bring, and take action now that
fit the problems and the expected rewards.
If nonhuman animals do not have autonoetic awareness that they
can direct at the imagined but physically nonexistent future, they will
not be able to create any culture in the sense of changing the natural
world. They may act in ways that a friendly anthropologist can call
“culture”—for instance, elements of deliberate instruction of the
young by their parents—but that kind of culture works using the
same tried-and-true method as does evolution, or development, or
learning: they increase the fitness of the animal by changing the ani-
mal. Nature is full of demonstrations that the method works, but it is
the plodder’s method. The autonoetic Homo sapiens invented a truly
revolutionary method: They enhanced their own fitness by altering
the Earth to suit their needs.
The idea of altering the world was unrivalled by anything that had
ever happened on Earth, and it could not have happened without
thinking about the future. Creatures lacking the ability to imagine
their possible futures might still be capable of impressive feats of in-
telligence: semantic memory, problem solving, communication, and
the like, but they would be unable to act on the basis of their knowl-
edge of the past and their expectations for the future.
To sum up this section, episodic memory evolved because au-
tonoesis, its critical constituent component, added to the already exist-
ing ability to mentally travel in space the ability to mentally travel in
time, not only into the past but, more important, also into the future.
Future-oriented consciousness (proscopic chronesthesia) made possi-
ble a feat that had no precedence anywhere in nature: individuals
intentionally, voluntarily, consciously taking action in response to
something that did not exist in the physical world. As a consequence,
humans were able to create a world to fit them, rather than live in
one into which they had to fit.
AUTONOETIC EPISODIC MEMORY: DATA

In this section, I discuss two rather different but converging lines of
evidence that speak to the issues of the episodic theory and the thesis
of this essay. One line is neuropsychological. The evidence consists of
clinical and experimental observations of memory and other cognitive
capabilities of certain brain-damaged patients who exhibit dissocia-
tions on memory tasks that vary in their assumed dependence on epi-
sodic memory. The second line consists of developmental studies that
have revealed task dissociations, as a function of children’s age, indic-
ative of episodic and semantic memory.
The Case of K.C.

In 1981, going home from work, a 30-year-old man accidentally rode
his motorcycle off the road and suffered brain damage of a highly
unusual kind. In addition to dense anterograde amnesia that fre-
quently follows traumatic brain injury, K.C. exhibits an uncommon
form of retrograde amnesia in that his previously acquired knowledge
of the world (accessibility of the “contents” of semantic memory) is
largely intact, while his ability to recollect premorbid personal, auto-
biographical events is, for all practical purposes, completely lost. [To-
day, K.C. lives with his parents in Mississauga, near Toronto, in the
family home where they have lived for over 40 years. As his mental
capabilities have not changed greatly from the time of his accident, I
describe him in the present tense, although a good deal of relevant
information is provided by studies done in the past (Hayman, Mac-
donald, & Tulving, 1993; Tulving, 1989b; Tulving, Schacter, McLach-
lan, & Moscovitch, 1988). For a recent thorough report on his case, see
Rosenbaum et al. (in press).]
In most ways, K.C.’s intellectual capabilities are comparable to
those of healthy adults. His thinking is clear, his intelligence is nor-
mal; language is normal, he can read and write; he has no problem
recognizing objects and naming them; his imagery is normal (he can
close his eyes and give an accurate visual description of the CN
Tower, Toronto’s famous landmark); his knowledge of mathematics,
history, geography, and other school subjects is about the same as that
of others at his educational level; he can define and tell the difference
between stalagmites and stalactites; he knows that 007 and James
Bond are one and the same person; he can play the organ, chess, and
various card games; his social manners are exemplary; and he pos-
sesses a quiet sense of humor.
K.C.’s memory, very broadly defined, too, is unimpaired, and can
be considered normal in a number of ways. He has no great difficulty
answering questions about semantic (public, objective, shared) aspects
of his own past life (autobiographical knowledge), such as his date of
birth, the address of his home for the first 9 years of his life, the names
of the schools he attended, the make and color of the car he once
possessed, and the fact that his parents owned and still own a summer
cottage. He knows the location of the cottage, can easily find it on the
map, and knows its distance (90 miles) from his home and how long
it takes to drive there from Toronto in weekend traffic. He also knows
that he has spent a lot of time there.
Note, however, that all this accessible factual (declarative, cogni-
tive, propositional) knowledge, even if it is about his own past, is
classified as semantic because it is impersonal, objective, public, and
shared with others. K.C. knows things about himself and his past in
the same way that he knows similar things about others, friends and
family. It is knowledge of one’s life from the point of view of an ob-
server rather than that of a participant, the same kind of knowledge
that people have about many other aspects of their world.
K.C.’s primary (short-term, or working) memory is normal: he re-
members what happened a short-term while (1 to 2 minutes) ago, and
his forward digit span is 7 to 8. He can play a whole hand of hearts, or
bridge, without any apparent memory handicap. He has no particular
problems with many perceptual-motor and cognitive skills, or with
the retrieval of premorbidly acquired general knowledge. His percep-
tual priming performance, as measured by word-fragment comple-
tion, even exceeds that of an average University of Toronto student

(Tulving, Hayman, & Macdonald, 1991). And although he is very slow
in comparison with healthy controls, he is capable of acquiring novel
semantic information (Hayman et al., 1993; Tulving et al., 1991; West-
macott & Moscovitch, 2001).
K.C.’s Episodic Memory

K.C.’s major problem is that he cannot remember anything that has
happened to him. However hard he tries, and however powerfully he
is prompted, he cannot bring into his conscious awareness a single
event, happening, or situation that he witnessed or in which he partic-
ipated. This global episodic amnesia covers the span of his whole life,
the period from his birth to the present day: he cannot recollect any-
thing from his life either before or after the accident. He knows the
address and, when standing in front of it, recognizes the house where
he lived for the first 9 years of his life, but does not remember a single
event that took place in the house. He does not remember a single
visit to the family cottage, and not a single happening there in which
he participated. Nor is he capable of remembering anything ever hav-
ing happened in the house where he has now lived for over 40 years.
He knows that he owned a black Honda, but does not remember a
single trip he ever took in it.
In the course of studying his amnesia, we collected descriptions of
a number of poignant events from his life that would be regarded as
highly memorable by everyone—a fight he had in a pub resulting in
a broken arm that took him to the hospital, a traffic accident that
caused his jaw to be wired shut for a week, the accidental death of
his older brother to whom he was close, and a huge chemical spill
near his home that caused a 10-day evacuation of over 100,000 people
in his neighborhood, including himself. The idea was to test his auto-
biographical memory with increasingly complete cues about the
events. (We also made up a collection of descriptions of otherwise
comparable events that had not happened to him, and used them as
controls.) The results were clear. Even when he was given full descrip-
tions of the real events, his response was the same as those he gave
to the fabricated events: he said he did not remember the events, and
did not feel any familiarity toward them.
Thus, although K.C. knows a great deal about the world because he
learned it before his accident, he does not remember anything from the
same period in his life (or any other period). This striking dissociation
suggests that episodic and semantic memory are subserved by at least
partially distinct sets of neural mechanisms. The brain damage that
K.C. suffered, a rare one-in-a-million accident, must have greatly im-
paired the operations of one while leaving the other largely intact.
Other Cases
Other neuropsychological cases have been described in the literature
that are similar to K.C.’s in that they exhibit a disproportionate im-
pairment of episodic memory in relation to other kinds of memory
(Calabrese et al., 1996; Giovagnoli, Erbetta, & Bugiani, 2001; Kapur,
1999; Levine et al., 1998; Markowitsch et al., 1993; Rousseaux, God-
frey, Cabaret, Bernati, & Pruvo, 1997; Van der Linden, Brédart, De-
poorter, & Coyette, 1996; Viskontas, McAndrews, & Moscovitch, 2000;
Wheeler & McMillan, 2001).
Especially striking, both with respect to their behavioral profiles
and correlated neuroanatomical pathology, are the cases of young de-
velopmental amnesia patients described by Vargha-Khadem and her
colleagues (Mishkin, Suzuki, Gadian, & Vargha-Khadem, 1997; Var-
gha-Khadem et al., 1997, 2003). Their behavioral profiles are interest-
ing inasmuch as they remember little of ongoing experiences and, like
K.C., cannot mentally relive the past, yet learn about the world with-
out undue difficulties. Their neuroanatomical damage is interesting
inasmuch as it is largely confined to the hippocampus proper and
does not extend, as it frequently does, to the subjacent cortex. Thus,
despite hippocampal pathology, these young people have managed to
acquire normal or near-normal levels of intelligence, including normal
language skills and general knowledge of the world. These fascinating
cases once again suggest that a fully functioning episodic memory
system is not required for the learning of general skills and knowl-
edge. The cases of these young patients once more suggest that epi-
sodic memory is not necessary for the learning of general skills and
knowledge.
Self-Awareness and Autonoetic Awareness
What about K.C.’s consciousness? Is he conscious? Is he reflectively
or self-reflectively conscious? Does he possess normal self-awareness?
Is he autonoetically conscious?
The answers to these questions clearly depend on what one means
by the terms. I would say that of course K.C. is conscious. Further-
more, he is self-reflectively conscious. Yet, at the same time, I would
also say that he does not have normal self-awareness, and that he is
not autonoetically conscious.
On the one hand, he is not in the least confused or uncertain about
himself as an independently functioning individual. He knows his
name, where he lives, his family, how to take care of himself, the daily
routine, how to spend time in the house, and what he does when he
goes for walks in the neighborhood. He knows about them in two
senses: he carries out relevant actions without any hesitation and usu-
ally without any prodding by anyone else, and he can reflect on what
he is doing, that is, he can answer questions appropriately about these

actions, when he is asked to do so. As a result of his accident, K.C.’s
personality was greatly changed (Rosenbaum et al., in press), but he
has even learned his new self as revealed by trait judgments (Tulving,
1993b; see also Klein, Loftus, & Kihlstrom, 1996).
When K.C. is asked about whether his “mind is clear,” he has no
hesitation in stating that it is, and nothing in his ongoing everyday
behavior would lead anyone to question his judgment in the matter.
Thus it seems reasonable to say that he has a well-developed and
properly functioning self and that he possesses self-awareness. Al-
though he has not yet been formally tested on his understanding of
the theory of mind, or metacognition (chapter 2, this volume), I and
others who know him well believe that he would have no difficulty
passing relevant tests. Indeed, it is not unreasonable to imagine that
giving K.C. a Gallup mirror test or even Povinelli’s delayed video test
(Povinelli, Landau, & Perilloux, 1996) in order to check on his self-
awareness would be equivalent to giving Columbia University profes-
sors a test of the alphabet.
On the other hand, K.C. is a densely amnesic person who remem-
bers nothing of what has ever happened to him. It is difficult to imag-
ine that these missing features have no bearing on his awareness of
himself. This awareness is not “normal.” What is missing is what I
have referred to as autonoetic awareness. He is severely deficient
when it comes to autonoesis: for all practical purposes, he has no func-
tioning autonoetic self, or Metcalfe and Kober’s (chapter 2) projectable
self. On Katherine Nelson’s (chapter 4) scale of consciousness, K.C.
seems to be missing the lower level (3–6 years of age) narrative con-
sciousness but, paradoxically perhaps, seems to have the higher level
(5–10 years of age) cultural consciousness. The apparent paradox,
however, can be resolved along the lines of the SPI model (Tulving,
1995, 2001a) by distinguishing between the processes of encoding and
retrieval.
K.C. and the Future

K.C. cannot think about his own personal future. Thus, when asked,
he cannot tell the questioner what he is going to do later on that day,
or the day after, or at any time in the rest of his life, any more than
he can say what he did the day before or what events have happened
in his life. When he is asked to describe the state of his mind when
he thinks about his future, whether the next 15 minutes or the next
year, he again says that it is “blank.” Indeed, when asked to compare
the two kinds of blankness, one of the past and the other of the future,
he says that they are “the same kind of blankness” (Tulving, 1985b).
Thus K.C. seems to be as incapable of projecting himself mentally into
his personal future as he is incapable of seeing himself in his personal

past. He lacks autonoetic awareness.
It is important to note that K.C. has no greater difficulty with the
concept of the future as an aspect of physical time than he has with
physical space. He knows and can talk about what most other people
know about physical time, its units, its structure, its measurement by
clocks and calendars, and the location of events in the world in the
past. But such knowledge of time in and of itself does not allow him
to remember events as having happened at a particular time. It is
necessary but not sufficient. Something else is needed, and this some-
thing else—the awareness of time in which one’s experiences are re-
corded—seems to be missing from K.C.’s neurocognitive repertoire.
He thus exhibits a dissociation between knowing time and experienc-
ing time, a dissociation that parallels one between knowing the facts
of the world and remembering past experiences.
When K.C. is engaged in activities that do not require mental time
travel into his own past or future, his awareness is indistinguishable
from what most neuropsychologists would consider normal. When he
is asked the name of the capital of France, or the difference between
stalagmites and stalactites, or thousands of other such facts, there is
no sign of any deficiency. In these situations, he is naturally con-
sciously aware of what he is doing, but the kind of consciousness
involved is different from autonoesis: it contains no awareness of per-
sonal time. We can describe the situation by saying that there is noth-
ing apparently wrong with or missing in his noetic consciousness.
Thus, although K.C.’s autonoesis is severely impaired, his capability
of conscious awareness of the world beyond subjective time, that is,
his noetic consciousness or noesis, is well preserved.
It was this striking pattern of K.C.’s mental life—his extensive rep-
ertoire of conscious thoughts about the impersonal world contrasted
with his essentially nonexistent conscious thoughts about his own
past and future—that first suggested the distinction between noetic
and autonoetic consciousness (Tulving, 1985b). K.C. possesses the for-
mer and does not possess the latter. Because he is perfectly well aware
of his timeless self—self in the present—it seems reasonable to attri-
bute his difficulties with personal past and personal future to defi-
cient, perhaps largely lacking, autonoesis.
This brief description of K.C. can be summarized by saying that he
“fails” the test of possession of the episodic memory capacity by all
four criteria represented by episodic memory’s defining features. He
does not remember any personally experienced events; he is not self-
aware in the dimension of time; he has no autonoetic awareness; and
he does not seem to possess any feeling of subjective time. I concede
that some of these assertions may be too strong. It is possible that he
has a little left of one or more of the properties of episodic memory

and that we are dealing with a case of severe impairment in episodic
memory rather than its total absence (Squire & Zola, 1998). However,
this possibility does not change the striking dissociation between epi-
sodic and other kinds of memory that the case of K.C. represents.
Why Is the Case of K.C. Relevant?

The case of K.C. is relevant in the context of the present essay for
several reasons. First, unlike many other living beings who have no
episodic memory (e.g., nonhuman animals and preverbal children),
K.C. is capable of producing reasonably detailed introspective reports
about his mental life. Thus he can do what these others cannot—pro-
vide direct first-person narrative evidence of the nature of the world
of an otherwise normal person without episodic memory and auto-
noetic capability.
Second, K.C.’s introspective reports are valuable in drawing a con-
nection between mental time travel into the past and mental time travel
into the future. There is no simple way of directly comparing these two
facets of time, and subjective time at that. Nevertheless, the fact that
K.C.’s ability to think about his personal future seems to be as severely
impaired as his ability to think about his personal past allows us to
assume that mental time travel is a single neurocognitive capability, one
whose domain of operations extends from the present in both possible
temporal directions. Before his case, it would have been possible to
imagine that the remembered time might be functionally different from
the imagined future time, and the two might even be subserved by
different neural substrates. The case of K.C. supports the idea of the
continuity of subjective time, in this sense paralleling physical time.
Recently, an interesting and revealing case, of patient D.B., has
been presented by Klein et al. (2002). It speaks rather directly to the
issue of autonoetic mental time travel in both directions, the past and
the future. As a result of hypoxic brain damage, D.B. suffers se-
vere retrograde episodic amnesia. Like K.C., D.B. has a relatively well
preserved general knowledge base, accumulated before the onset of
his amnesia, but his ability to recollect premorbid personal experi-
ences is severely impaired. The important new feature of the Klein
et al. (2002) study was that they formally tested D.B.’s ability to imag-
ine the future and his ability to anticipate future events. They found
that this ability, like his retrograde memory, was fractured along the
lines of “personal” versus “impersonal.” D.B. has knowledge of the
world’s past and can intelligently express opinions about the world’s
probable future. But both his ability to recollect happenings from his
own personal past and to imagine his own personal future are se-
verely impaired. The case of D.B. thus formally illustrates a dissocia-
tion between semantic (noetic) and episodic (autonoetic) mental time

travel.
Third, it is important to note that as long as K.C. continues to live
in a stable world with which he is familiar, that is, with his parents
in his home, he has no problem surviving, and surviving well. He is
not dependent on others to tell him how to behave, what to do, or
how to take care of himself. As long as somebody fills the refrigerator
and pays the bills, and water flows from the taps and his bed is ready
to be slept in in his bedroom, he manages without any difficulty. He
probably also could, if necessary, walk to the supermarket, and (if he
has written down what he needs, if he has not forgotten that he has
the list in his pocket, and if he has not forgotten to take money with
him), he could fill the basket and walk back home. The point is that
in a stable environment there is no crying need for episodic memory
to live a satisfying life. K.C.’s life may be abnormal and uninteresting
by ordinary standards that people in our world are accustomed to,
but because he himself is not aware of it, and his needs are taken care
of, he has no complaints. When he is asked to rate the quality of his
life on a five-point scale (5 “very good,” 1 “very bad”), he judges it
to be 4.
Fourth, K.C.’s case graphically illustrates the distinction between
timeless and time-based self-awareness. As mentioned earlier, his
knowledge of physical time is normal. He knows perfectly well what
time is, and because he has no problems with thinking and reasoning,
he also knows, and can articulate the fact, that he has lived in the past
and is going to live in the future. But such knowledge does not trans-
late into a felt time. He does not seem to possess what others do—an
ever-present awareness of one’s being existing in a subjective sea of
time, always in transition from what is now becoming the past to
what once was the future. K.C. possesses a noetic (knowing) self, but
lacks an autonoetic (or projectable, or time-traveling or remembering
self).
Autonoetic Memory in Children

The hypothesis that episodic memory is not present in animals re-
ceives support from the observation that episodic memory is also ab-
sent in young children. Young children can act upon their earlier
experiences, and they can even verbally describe aspects of these ex-
periences. The question is whether they also (autonoetically) remem-
ber these experiences. The proposed answer to the question is that
young children do not remember personal experiences, and that the
ability to remember events follows rather than precedes the develop-
ment of the ability to become knowledgeable about the facts of the
world (Wheeler, 2000).
The observations concerning the differentially delayed develop-

ment of episodic memory are still somewhat controversial, and not all
developmentalists are willing to interpret them along the lines sug-
gested here. The problem is rather similar to that in the interpretation
of learning in animals—many instances of learning can be (anthropo-
morphically) interpreted as revealing episodic memory, although they
need not, because nonepisodic forms of memory are sufficient. When
parents witness their 9-month-old daughter crawling to the spot
where she previously found her favorite toy, it is quite reasonable for
them to conclude that the child remembers the prior episode. Even
more convincing is a 3-year-old who verbally answers questions about
something that happened to him only a short time ago. It requires only
common sense and no specialized knowledge to attribute to the child
the ability to recollect the past. Surely an adult, answering the same
question, would rely on episodic retrieval, so why should the same
inference not apply to the child? The answer to this question is diffi-
cult to resolve conclusively, yet additional evidence now suggests that
even when young children are able to talk about the details of prior
events, it need not mean that they consciously recollect the events
(Lockhart, 1984; Nelson, 1993; Wheeler et al., 1997). It is also true, of
course, that just because an act of learning can be attributed to nonepi-
sodic forms of memory, it cannot benefit from episodic memory.
The argument here is that children up to age of 4 years or so lack
the same kind of episodic memory and autonoetic consciousness that
seem to be missing in animals. The relevant findings have been pro-
vided by comparative studies of episodic and semantic memory in
children between the ages of 3 and 6 years. During these years, chil-
dren are already highly verbal and very accomplished learners. They
can also recall novel events from their lives, sometimes across inter-
vals on the order of months (Fivush, Hudson, & Nelson, 1984). Some
of these studies show striking dissociations between episodic and se-
mantic memory. Many of the most relevant studies have examined
memory for context or for source. I provide examples below.
If one assumes, as I do, that the operations of episodic memory are
critically dependent on autonoetic consciousness, it follows that any
child who is incapable of this highest level of consciousness should be
unable to recollect the personal past. Therefore, assessing autonoetic
awareness in children should set boundary conditions on the presence
or absence of episodic memory (Wheeler, 2000).
One relevant piece of evidence comes from the behavior of infants
when they are placed in front of mirrors. Exposure to a mirror typi-
cally causes older children and adults to begin paying attention to
themselves, and it often stimulates them to begin reflecting on their
own mental states (Duval & Wicklund, 1973). Before the age of about
18 months, young children will pay attention to their mirror image
but do not realize that it represents their body, and they will not use
the mirror as a tool to think about themselves (see Lewis and Brooks-
Gunn, 1979, for an extensive discussion of children’s behavior around
mirrors).
It would be convenient to assume that the toddlers who recognize
themselves in the mirror possess self-awareness. And if self-aware-
ness were defined in terms of the mirror test, the statement would be
true, by definition. However, passing the mirror test does not mean
that the children have autonoetic consciousness, that they are aware
of themselves as individuals with a past and a future. Autonoesis
emerges only later, as an integral component of episodic memory.
Episodic memory matures gradually, and therefore it is difficult to
pin down a particular age at which young children’s episodic memory
capabilities can be said to have become fully functional. But one magic
number that is frequently used in the attempts to do so is 4 years
(Nelson, 1992). It does not mean, of course, that episodic memory
emerges suddenly around a child’s fourth birthday. Nor does it mean
that it becomes fully established shortly thereafter. What the number
designates is the common finding that most 3-year-olds fail or do
poorly on tasks that require them to bring back to mind their own
personal experiences, whereas most 5-year-olds do very much better
on the same tasks.
One of the most instructive examples of this generalization is pro-
vided by experiments reported by Gopnik and Graf (1988). They had
3-, 4-, and 5-year-olds learn about the contents of a drawer in one of
three different ways. Some of the children were told about the con-
tents without seeing them, others saw with their own eyes what was
in the drawer, and a third group was given hints so they could infer
what was there. Gopnik and Graf were interested in two things: (1)
Do the children know what is in the drawer? (2) Do they know why
they know—was it via seeing, hearing, or inferring? The first question
has to do with acquired knowledge, and, over the studied age range,
there were no age differences on this question, even on a delayed
test when recall was far from perfect. As to the second question, the
reasoning was that for the subjects to be able to answer it, they must
be able to recollect how they found out about the contents of the
drawer. While the 5-year-olds made very few mistakes on this second
question, few of the 3-year-olds could respond at levels higher than
chance. Thus, all children knew “what” (the contents of the drawer),
but only the older ones knew why they knew (remembering the event
of seeing what was in the drawer).
The general finding has been replicated in a number of different
ways (Lindsay, Johnson, & Kwon, 1991; Wimmer, Hogrefe, & Perner,
1988). It appears that young children cannot, or do not, represent their
knowledge as deriving from a particular time in the personal past
(e.g., “I saw that there are crayons in the drawer”), but rather as de-
tached, impersonal knowledge (e.g., “There are crayons in the drawer”).
The studies just described may even overestimate children’s abili-
ties to understand how they learned information. The critical test
questions used by Gopnik and Graf (1988) and others only asked chil-
dren to differentiate between different intraexperimental sources of
information (seeing, hearing, inferring). Many children cannot ac-
knowledge the simple fact that a recently learned bit of information
has been acquired recently. In one experiment (Taylor, Esbensen, &
Bennett, 1994), 4- and 5-year-old children were taught unfamiliar color
names (e.g., chartreuse, taupe). All the children learned the names
easily, and after learning were able to select items from an array of
colors according to the specified name. They were then questioned
about when they had learned the new color names. Incredibly, a large
majority of the 4-year-olds claimed to have “always” known these
names, and only a few would admit that they had been taught the
colors that day. The 5-year-olds performed markedly better in identi-
fying the source of their knowledge, although a few of them consis-
tently made this same source error. Across several experiments, the
authors showed that young children typically are unaware of recent
learning events and claim to have known recently acquired informa-
tion for a long time. Thus, it looks as if children are better at making
use of learned facts than they are of identifying the episode during
which the facts were learned. One plausible reason for such failure is
that because they lack fully fledged episodic memory, the children
actually do not remember the learning episode. This suggestion is
consistent with the idea that childhood amnesia reflects nothing more
nor less than the absence of episodic memory in young children (Per-
ner & Ruffman, 1995).
Children’s ability to remember how and when and in what setting
they learned a new fact can be assessed even more directly. When this
is done, findings again suggest a magical number of 4 as the number
of years needed to develop a nearly fully operational episodic mem-
ory system.
When young children are more directly assessed for the extent to
which they suffer from source amnesia, the findings again point to a
critical age around 4 years. Drummey and Newcombe (2002) used the
paradigm like the one introduced by Schacter, Harbluk, and McLach-
lan (1984). Children were taught new facts in the laboratory setting
(e.g., “giraffes are the only animals that cannot make a sound”) and
were later tested for (1) their knowledge of the learned facts and (2)
their recollection of the learning episode. Children’s retention of the
learned facts showed steady improvement with age from 4 years to 8
years. At the same time, however, the 4-year-olds were very much
worse than the older children in remembering where and how they
had learned the facts that they now knew.
These lines of research show what young children can, and cannot,
do. There is little question that children as young as 3 or 4 years can
talk about events from their own lives. But an important limitation
appears to be that the event must have been novel. It is a common
finding that when young children are asked to talk about novel or
interesting events that occurred several months ago, they are able to
do so (Fivush et al., 1984). Yet if they are asked in the evening what
happened at school that day, they do not know; they cannot remem-
ber. Children below the age of about 5 years can only talk about daily
events if something unusual has happened.
For a period of at least a few years, children may have what Kather-
ine Nelson (1984) has called a “general undifferentiated knowledge
base,” perhaps even at the age of 4 years. Information exists in the
knowledge base in noetic form only. Even when a child can master
semantic concepts like “time,” “future,” “yesterday,” and “me,” there
is some considerable period of time, on the order of years, before there
is a qualitative shift in thought, allowing the child to remember the
past episodically.
One example of a test that probably cannot be used with young
children is the remember/know task (Gardiner, 1988; Gardiner &
Richardson-Klavehn, 2000; Rajaram & Roediger, 1997). If children are
truly pre-episodic and have not yet developed the capacity to recollect
the past autonoetically, then it will not make any sense to ask them
to distinguish between remembering and knowing. Evidence that
these terms are problematic for children was reported by Johnson and
Wellman (1980). (It is edifying to note that this study appeared many
years before the formally articulated distinction between remember-
ing and knowing.) They found that 4-year-olds were not cognizant of
differences between cases of remembering, knowing, and guessing,
although in some cases they could distinguish their mental state from
an externally perceived state. Five-year-olds showed some ability to
differentiate these mental states, while 6-year-olds generally had a
very solid command of the terms. The authors stress that this is not a
simple vocabulary problem, as children of that age are fully capable
of acquiring and using words and concepts that, to adults, are much
more sophisticated than words like “know” or “guess.” Difficulties
likely stem from the inability to monitor one’s own mental states.
Nelson and Fivush (2004; see also chapter 4, this volume) have re-
cently proposed a major social-cultural theory of the development of
autobiographical memory, which for them represents a subset of epi-
sodic memory (for a somewhat different idea on the relation between
episodic and autobiographical memory, see Conway & Pleydell-
Pearce, 2000). Autobiographical memory is said to entail a “personal

significance that characterizes specific memories that re-appear many
times during one’s life, often decades after the event.” They are con-
trasted with “lesser episodic memories [which] tend to lead an
ephemeral existence, leaving little trace in the future.” Nelson and
Fivush contend that autobiographical memory emerges following a
number of changes developments in language, consciousness, and
self-awareness. The developing system begins with a basic, functional
declarative-like memory system that is able to retain information
about events (routines and episodes) for weeks or, eventually,
months. A critical addition, however, is the beginning of language
comprehension and expression, especially the burgeoning capacity to
provide labels to elaborate upon and retain information. Around the
age of 18 months, young children begin to establish a cognitive self
(see also Howe & Courage, 1993) which serves as a foundation for
encoding and understanding ongoing events. For the first time, a child
can begin to understand his own special place in the world (and his
own idiosyncratic feelings), as distinct from others.
An additional necessary step occurs when parents and other adults
begin to talk about past and future events in the presence of the child,
especially when the adults describe such events narratively. Such con-
versations provide the support for the child’s developing concept of
time. It is between about the ages of 2 and 5 years that developing
language and narrative skills allow children to represent and become
aware of complex sequences of events, and to think about the events
in terms of a personal past, present, and future.
In summary of this section, young children represent a prime ex-
ample of a population that possesses some critical aspects of episodic
memory but lacks others. As such, the discussion of children’s mem-
ory may be useful as a way to understand our eventual interpretation
of evidence from animal research. Children between the ages of 2 and
5 years can encode, store, and retrieve declarative-like information
and learn new facts about the world, and they are fully capable of
applying this learned information flexibly and confidently. Yet devel-
opmental limitations in their conscious awareness place boundary
conditions upon their uses of memory. It is not until children can
reflect upon subjective experiences in the past, present, and future
that they can experience the past episodically. Evidence from multi-
ple, diverse sources implies that, below the age of approximately 4 or
5 years, young children lack the linguistic, representational, and
“self”-related skills to reflect upon the personal past. The develop-
mental findings not only emphasize the role that a fully developed
human brain plays in episodic memory; they also underscore the fact
that episodic memory is not necessary in a world in which one’s needs
can be and are satisfied without remembering autonoetically what

happened in the past.
ANALYSIS: EPISODIC MEMORY IN ANIMALS

We now return to the main point of this chapter and the question: Do
animals other than humans possess episodic memory? Hundreds of
articles have been written on the topic of mental capacities of animals,
and to say that the topic is controversial is to greatly understate the
vehemence with which the issue has been pursued. By restricting our
analysis of the mental capacities to episodic memory, it is possible to
expect closer agreement.
The answer to the question is multiply determined. It depends
partly on what one means by episodic memory, partly on the kinds
of evidence one considers, and partly on how one interprets the evi-
dence. When episodic memory is defined loosely as “memory for
(specific) past events,” then the standard commonsense answer is that
of course animals have it. Every owner of a pet dog or cat, every
visitor to the monkey cages in the zoo, anyone who has ever watched
Walt Disney cartoons knows that even if animals are not like people
in every way, they are like people in many ways, and those many
ways include their memory abilities. They may not have human lan-
guage, may lack the capacity for abstract reasoning, and may fall short
on some other esoteric human traits (the likes of musicality, religios-
ity, science, literature, and love of the arts), but surely they know what
they are doing in their daily lives and they naturally remember it too.
When one sees a dog bury a bone one day and next day make a
beeline to the place, it is easy to imagine that on the first day the dog
was completely aware of his purpose in burying the bone, and that
on the second day he was completely aware of having done so the
day before.
This kind of anthropomorphically driven thinking, which Charles
Darwin put to scientific and practical use in his Descent of Man (Dar-
win, 1874/1998), and which George Romanes (1881) developed to
high art, comes naturally to human minds. Despite Lloyd Morgan’s
(1894) early strictures against it, in the form of his famous canon, and
continued urgings of caution by contemporary writers, it is very much
alive today. It is quite possible that anthropomorphizing is an evolu-
tionary adaptation, as suggested by Povinelli and Vonk (2003, p. 157)
in their musings about why the chimpanzee mind seems so suspi-
ciously human: “the human mind may have evolved a unique mental
system that cannot help distorting the chimpanzee’s mind, obligato-
rily recreating it in its own image. This idea should be taken seriously.
After all, from change-blindness, to false memories, to cognitive disso-
nance, don’t we already know the various ways in which the human
mind systematically distorts its own workings?”
Many memory researchers have adopted the commonsense notion
and have taken it for granted that animals have episodic memory. A
good example is provided by the animal model of memory and amne-
sia proposed by Squire and Zola, and their associates (Alvarez, Zola-
Morgan, & Squire, 1994; Squire, 1987; Squire & Zola, 1998). A central
tenet of the model is that the hippocampal system plays a critical role
in declarative memory. The hippocampal declarative memory system
covers both memory for events (episodic memory) and memory for
facts (semantic memory) and is the same in rats, monkeys, and human
beings (Squire, 1992). If animals have declarative memory, and if de-
clarative memory comprises both episodic and semantic memory,
then, logically, animals must have episodic memory. Comparable
views are held by Eichenbaum and his collaborators (Eichenbaum,
2000; Eichenbaum, Otto, & Cohen, 1996), although the critical neuro-
anatomy is seen somewhat differently from Squire and Zola.
The idea seems perfectly reasonable; otherwise its popularity
would not have lasted. But what it demonstrates above all is how
subtle important differences in brain and behavior can be, and how
difficult to come to grips with. On closer reflection, there are several
problems with it.
First, in common sense and theories based on it, the question of the
existence of episodic memory in animals is “solved” by postulation
instead of empirical inquiry. Memory for events and memory for facts
are lumped together as one and not treated separately. The practice
may serve for certain purposes—such as investigation of the neuroan-
atomical correlates of declarative (conjunction of episodic and seman-
tic) memory—but it possibly cannot help to answer the question I am
asking here.
Second, the kinds of tasks that have been used in evaluations of
the hippocampal declarative memory system do not, and cannot, dis-
tinguish between memory for events and memory for facts. A great
deal of laboratory research on animal memory is essentially concerned
with perceptual (recognition) memory (Aggleton & Pearce, 2001;
Wright, Santiago, Sands, Kendrick, & Cook, 1985), as is research done
with preverbal human infants (Rovee-Collier & Hayne, 2000), and re-
quires no declarative (semantic and episodic) memory. The widely
used DNMS (delayed nonmatching to sample) and other kinds of ob-
ject recognition tasks can be effectively executed by identifying the
stimulus objects on the basis of their perceived novelty or familiarity.
Moreover, there is increasing evidence that the execution of DNMS
tasks does not even depend on the integrity of the hippocampus (Ag-
gleton & Brown, 1999; Aggleton & Pearce 2001; Murray, 1996; Murray
& Bussey, 2001; Mishkin, Vargha-Khadem, & Gadian, 1998). The abil-
ity of animals to perform the task, and the effects of experimentally

induced lesions on the performance, therefore, have no bearing on
the issue of whether animals have episodic memory. Even when we
consider tasks that cannot be handled at the level of perceptual mem-
ory alone, it turns out that most of them require only semantic mem-
ory and not episodic. Young children can acquire knowledge about
the world efficiently and rapidly long before they develop the ability
to recollect specific happenings from their past (Nelson, 1993). Al-
though not all amnesic patients are capable of doing so (Kopelman,
2002; Manns, Hopkins, & Squire, 2003), some amnesic patients, who
have severely impaired or no functional episodic memory, can never-
theless acquire new semantic information (Guillery et al., 2001; Ha-
mann & Squire, 1995; Hayman et al., 1993; Kitchener, Hodges, & Mc-
Carthy, 1998; McKenna & Gerhand, 2002; Rajaram & Coslett, 2000;
Van der Linden et al., 2001; see also Verfaellie, Koseff, & Alexander,
2000).
Third, the kinds of tasks that have been typically used in laboratory
studies usually test the subjects’ knowledge of what they perceived
or what they did at the time of study (training), and may also test the
subjects’ knowledge of the spatial coordinates of the to-be-remem-
bered objects (where something was). However, they do not challenge
the subjects’ knowledge of the time of the occurrence (when) of the
event in question (Griffiths et al., 1999). Thus, one key element of
episodic memory, the temporal dimension of what is remembered,
has been lacking in laboratory experiments on animal memory.
Clayton and Dickinson and their colleagues have remedied this
shortcoming, and reported ingenious and convincing demonstrations
of memory for time in scrub jays. Scrub jays are food-caching birds,
and connoisseurs when it comes to choosing what they eat. When
they have food they cannot eat, they hide the food and recover it later.
Because some of the cached foods (such as wax worms and crickets)
are preferred but perishable and must be consumed within a few days
of caching, while other foods (such as seeds and nuts) are less pre-
ferred but do not perish easily, for efficient recovery the birds must
recall not only cache locations (What did I hide where?) but also the
time of caching (When did I do it?). Clayton and her colleagues made
clever use of these facts in designing their laboratory experiments,
whose outcomes clearly showed that, even days after caching, scrub
jays knew not only what kind of food was where but also when they
had cached it. Similar laboratory findings of these kinds have not yet
been reported for rodents or nonhuman primates.
The Fourth Problem

The three problems I have mentioned—solving a theoretical issue by
postulation rather than empirical inquiry, difficulty of distinguishing
memory for events and memory for facts, and the absence, until re-
cently, of evidence of memory for when an event occurred—are sim-
ple in comparison with the fourth. What I here call the fourth problem
is the central one.
How do we know what kinds of mental experiences accompany
memory-based activities in nonverbal organisms? How can we find
out whether there is any evidence for episodic memory in its narrow
(unique) definition in animals, given that animals do not talk, and
that even rudiments of human-taught language, known from some
very rare cases, cannot be used for the purpose? How can we find out
whether animals autonoetically remember past experiences? That they
are capable of mentally traveling in time? Do Clayton’s clever scrub
jays actually remember when they cached the wax worms and when
they cashed the peanuts, or do they know, on some other basis, how
to act at the time of the recovery of the food?
Scientists and others interested in the issue do not doubt that their
fellow human beings, like they themselves, have the capacity to men-
tally travel in time and the capacity to distinguish the kind of con-
scious awareness that accompanies mental time travel from other
forms of consciousness. Normal human beings can remember events
such as a visit to Washington, DC; they know, without necessarily
remembering any personal encounter, what the capitol in Washington
looks like; and they can perform behavioral feats, without any con-
scious awareness that they are using their memory, such as unhesitat-
ingly completing the word fragment C-P-T-L. These kinds of distinc-
tions, between autonoetic, noetic, and anoetic consciousness (Tulving,
1985b) come to humans as naturally as the ability to tell the difference
between seeing white and black, or hearing a ping and a thud, or
feeling the difference between a sharp and a dull pain. The extensive
human literature on the distinction between two kinds of conscious
awareness, remembering and knowing (Gardiner, 1988; Gardiner &
Richardson-Klavehn, 2000; Rajaram & Roediger, 1997), together with
early findings of neural correlates of subjectively experienced states
(Düzel, Vargha-Khadem, Heinze, & Mishkin, 2001; Eldridge, Knowl-
ton, Furmanski, Bookheimer, & Engel, 2000) has helped do away with
the fourth problem in the study of human memory. But as yet no one
has come up with a comparable solution to the fourth problem in the
study of animal memory.
The fourth problem as such looms large in the thinking of others
who have thought about the time sense in animals, and who have
arrived at conclusions similar to those I am arguing for here. An ex-
ceptionally thorough and lucid discussion of the topic is found in an
article by Suddendorf and Corballis (1997); another is an equally de-
tailed and scholarly treatment of the same theme by Roberts (2002;
see also Roberts & Roberts, 2002).
In an article titled “Mental Time Travel and the Evolution of the

Human Mind,” Suddendorf and Corballis (1997), in much greater
depth than I have been able to do here, have anticipated and laid bare
the kinds of arguments I have been presenting. I very much agree
with most of the things that they say, including their main point that
“the ability to travel mentally in time constitutes a discontinuity be-
tween humans and other animals,” their sober observation that “the
importance of mental time travel as a prime mover in human cogni-
tive evolution has not been adequately recognized,” and their declara-
tion that “the real importance of mental time travel applies to travel
into the future rather than into the past.” I am also in complete agree-
ment with their assessment of the development of time sense in young
children and with their idea that language, although greatly facilitat-
ing the use of the human capacity for mental time travel, is not neces-
sary for it.
Suddendorf and Corballis (1997) have done a further good deed in
bringing to the attention of Anglophone readers what they refer to as
the Bischof-Köhler hypothesis. This hypothesis is “that animals other
than humans cannot anticipate future needs or drive states, and are
therefore bound to a present that is defined by their current motiva-
tional state.” The thesis of this essay is clearly very similar to it, except
perhaps in that it is a part of a broader theory of episodic memory,
with all the ramifications that it entails. Because the original texts do
not seem to be readily accessible, a more informed comparison has to
wait for the future.
Roberts’s (2002) article is titled “Are Animals Stuck in Time?” After
a thorough review, and after appropriate caveats, he concludes that
although the picture is not entirely clear, most of the evidence sug-
gests that animals are indeed stuck in time. Even chimpanzees’ ability
to imagine their extended future is in doubt. Wolfgang Köhler (1917/
1927), famous for his discovery of “insight” in chimpanzees, noted
this kind of shortcoming in the mental makeup of his otherwise intel-
ligent subjects: “The time in which the chimpanzee lives is limited in
past and future” (p. 272). Roberts (2002) suggests that chimpanzees
are able to “think future,” in the sense of planning ahead to solve
problems, only when (1) they are hungry, and (2) they can see the
food reward they will receive for solving the problem. In other words,
solving a complex goal-oriented task seems to depend on the antici-
pated satisfaction of a current need in the very near future.
I myself stumbled across the idea that mental time travel into the
future may be more important than mental time travel into the past, in
a quest for the answer to the question of biological utility of episodic
memory. What is episodic memory good for? Given that there are
relatively few obvious advantages of episodic memory over semantic
memory—kind of “who really needs to know when or where a past
event happened; the important thing is the lessons we can learn from
the fact that it did happen”—it is easy to think of episodic memory as
an evolutionary frill, as a Stephen Jay Gould kind of spandrel. When I
first attempted to speculate about the adaptive utility of episodic
memory, all I could offer was the thought that it provides knowledge
in which one can be more confident than the knowledge that is fre-
quently obtained second hand (Tulving, 1985b). It may have been
plausible, but not convincing.
My more recent thought on the topic was that episodic memory’s
adaptive value lies in the autonoetic awareness of subjective time, and
especially future time (Tulving, 2002a). The idea is that this kind of
awareness may have been a critical driver of human cultural evolu-
tion. Culture is usually thought of as a behavior pattern of a commu-
nity. I use the term, because of the lack of a better one, to signify the
sum total of the differences between the world as it existed before
humans changed it and as it exists now (or has existed at various
times before the present). It includes changes in both of the two hu-
man realities, physical and mental.
The key notion behind human cultural evolution can be put simply:
For anyone to take steps at one point in time that would make the
unpredictable, frequently inhospitable natural environment more pre-
dictable at a future time, it is necessary to be able to be consciously
aware of the existence of a future. An animal that cannot “think fu-
ture”—cannot pre-experience possible happenings, as suggested by
Atance and O’Neill (2001)—that is, cannot think about time that has
not yet arrived, will not initiate and persist in carrying out activities
whose beneficial consequences will become apparent only in the fu-
ture, at a time that does not yet exist. For the human species, the
masters of the awareness of the future, that nonexistent time has been
and continues as a powerful determinant of higher-level behaviors,
behaviors not motivated by the satisfaction of physiological needs.
Awareness of the future has radically changed the humans’ relation-
ship with their environment. When one thinks about it, it becomes
clear that a staggeringly large proportion of human behavior today—
social, economic, political, religious, and otherwise—is governed,
both directly and indirectly, by awareness of the future. At any rate,
the point is that the ability to think about the future, as articulated by
Suddendorf and Corballis (1997; see also Suddendorf, 1994), can be
seen as all-important on a fundamentally broader scale of things.
RESOLUTION
It is commonly thought that the question of whether or not animals
can and do travel mentally in time can never be answered. Language
is necessary, the argument is, for the verification of phenomenal men-
tal processes, such as autonoetic awareness of time. Because animals

cannot talk, their mental processes will forever remain beyond the
domain of objective science. This is why researchers such as Robert
Hampton (chapter 11, this volume) suggest that we should limit our
concerns regarding animal learning and memory to observable behav-
ior and leave subjective experience out. Nicola Clayton and her col-
leagues (Clayton, Bussey, Emery, & Dickinson, 2003) have also proposed
purely behavioral test criteria—content, structure, and flexibility—but
these are for episodic-like memory rather than episodic memory, be-
cause they do not address the criterion of autonoesis.
Phenomenal Experience Without Language?

A minor obstacle to accepting the “behavioral resolution” of the prob-
lem of episodic memory in animals is that there exist people like my-
self. We are very much interested in objectively observable behav-
ior—this is, after all, the mainstay of our science—but we also wish
to reach beyond behavior and scientifically study the even more puz-
zling and fascinating domain of subjective reality. If we leave the phe-
nomenal considerations out of the equation and limit ourselves to
“mindless” behavior, we will automatically deny ourselves any hope
of getting at concepts such as autonoesis. A major obstacle lies in the
well-established fact that it is always possible to account for any form
of behavior in a number of different ways. Some of these accounts
may appear more reasonable or plausible than others, but their evalu-
ation is usually highly subjective, determined as much by the charac-
teristics of the observers as the characteristics of the observed behavior.
A useful criterion that can be applied in situations of this kind is
Lloyd Morgan’s (1894) canon. The canon does not solve the problem
of disagreements among observers, but its application does impose
constraints on the domain of possible interpretations of the observed
behavior. A typical problem situation is one in which the animal be-
haves in a way that encourages a human observer to impute mental
time travel to the animal but which also leaves open the possibility of
accounting for the behavior without the need to invoke such a com-
plex concept. Your pet dog burying the bone one day, and digging it
up the next, is an excellent example of the situation. Human beings,
lucky owners of autonoetic episodic memory, are intimately familiar
with problems of finding their car in a huge parking lot, and equally
familiar with the solution: You “just remember” where you left it. By
analogy, it makes sense to imagine that the dog’s ideas about the out-
of-sight bone are based on the same kind of a just-remember internal
mechanism. The possibility, the thesis of this essay, that just remem-
bering is an unbelievably complicated and near-miraculous invention
of nature, only available to healthy humans older than 4 years or so,
does not occur to a casual observer. But it must be considered, because
an alternative, and simpler, explanation is that the dog acts on its

noetic knowledge, acquired during a particular episode but not stored
in episodic memory, that the world is the kind of place where there’s
a bone buried in a particular place. The dog’s knowledge-guided ac-
tion is like that of any other intelligent organism who has no episodic
memory. K.C. knows that the world is a place in which he, as an
individual, owned a black Honda. He knows that there are fictional
characters like James Bond who have aliases like 007. He also knows
that “dogs confront bullfrogs,” because he learned it in an experiment
(Tulving et al., 1991). Therefore, when he is asked to complete the
sentence frame “dog confronted . . . ?” he is happy to respond with
the word “bullfrog,” without remembering, and without any need to
remember, that he himself saw the complete three-word phrase only
5 minutes ago. Similarly, the 4-year-old who has just learned the color
name “chartreuse” and now displays the knowledge, without remem-
bering the learning event that occurred 10 minutes earlier, is also do-
ing nothing more than exhibiting her noetic skills.
Thus, the difficulty in demonstrating episodic memory in animals
lies in the availability of simpler explanations of “episodic-looking”
behavior, explanations that do not require evoking episodic memory.
But if so, the logical solution to the problem is obvious. Arrange for
the animal to engage in a kind of episodic-looking behavior for which
no such simpler explanation is possible. Make the invoking of epi-
sodic memory necessary. In other words, set up a situation in which
a nonlinguistic organism exhibits behavior that can be readily ac-
counted for in terms of phenomenal ability such as autonoetic aware-
ness but cannot be accounted for in terms of mindless behavior.
One straightforward suggestion of a test designed along these lines
was proposed by Thomas Suddendorf in his MA thesis (Suddendorf,
1994), which provided the basis for the Suddendorf and Corballis
(1997) article that I referred to earlier in this essay. As the thesis is
available on the Internet, I leave it to the interested reader to get its
details. What is important are the basic principles that should be gen-
erally agreed upon: It is necessary to reliably show that animals en-
gage in an activity that (1) is not instigated and maintained by the
animals’ motivational states or the presence of discriminative stimuli
in their internal or external environment, and (2) has no immediate
relevance to satisfying present needs but has consequences that have
value to the animal at a future time. Under these conditions, it be-
comes not only meaningful but necessary to argue that the observed
behavior reflects awareness of and ability to think about one’s per-
sonal future. This then would be evidence for mental time travel, and
hence evidence for autonoetic episodic memory.
Suddendorf’s suggestion means that, at least in principle, the possi-
bility of proving the Bischof-Köhler hypothesis wrong is open already.
Given the expectation that creative experimenters in the future will

come up with additional clever designs (Morris, 2001), the possibility
of rejecting the central hypothesis of this essay becomes increasingly
probable, and the challenge that it entails practically manageable.
Herein lies the challenge to comparative psychologists. In pursuing
the challenge, they may find useful hints on how to proceed from
studies of brain-damaged patients who lack the ability to mentally
travel in time, as well as studies of young children who are extremely
capable learners in the absence of episodic memory.
I am not convinced that the kind of surprise that Zentall, Clement,
Bhatt, and Allen (2001) wrote about is necessary for the test—healthy
humans benefit from their autonoetic consciousness in many life situ-
ations that are not unexpected. And I agree with Bennett Schwartz
(chapter 9, this volume; see also Schwartz, Colon, Sanchez, Rodriguez,
& Evans, 2002; and discussion of the issue by Morris, 2001) that the
important feature of episodic memory, and hence of a test for it, need
not be built around immediate, single-trial learning. Young children,
long before they acquire episodic memory, learn not only to avoid hot
stoves but also acquire scores upon scores of new words every day
after a single exposure, in ways that still mystify developmental psy-
chologists. In addition, there is no reason to doubt that a great deal of
adults’ acquisition of new knowledge of the world occurs on a single
“trial.” It is for these reasons that “fast encoding” is assigned to the
category of properties that semantic and episodic memory systems
have in common, shown on the left-hand side of table 1.1.
Here I would like to propose another version of a definitive test of
autonoetic episodic memory that can be administered to nonlinguistic
animals such primates. (Nonprimate versions require fine-tuning of
some of the practical details.)
The Spoon Test

It is indeed difficult to see how one could differentiate between auto-
noetic remembering and noetic knowing in nonverbal organisms in
situations where both the remembered events and acquired knowl-
edge are derived from the past. But if one accepts the idea that auto-
noetic consciousness encompasses not only the past but also the fu-
ture, the problem becomes tractable. The resolution of the fourth
problem lies in autonoetic awareness of the future.
Here I propose a future-based test of autonoetic consciousness that
does not rely on and need not be expressed through language. In
order for us to be able to unambiguously refer to the kind of a test
for autonoetic episodic memory that I have in mind, and that others
would find convincing, let me refer to it as the spoon test. Because no
other spoon test exists in the literature, the use of the term gets us
immediately to the idea referred to.
In an Estonian children’s story with a moral, a young girl dreams

about going to a friend’s birthday party where the guests are served
delicious chocolate pudding, her favorite. Alas, all she can do is to
watch other children eat it, because everybody has to have her own
spoon, and she did not bring one. So the next evening, determined
not to have the same disappointing experience again, she goes to bed
clutching a spoon in her hand.
We do not know much about the heroine of the story. But we can
surmise that she is intelligent, can learn from experience, is capable of
reasoning, and can solve problems like that of the missing spoon: If
they do not give you one, bring your own. At this point in the story
in this essay, we also know that she can mentally travel in time, not
only into the past but also into the future, that therefore she must
possess autonoetic consciousness, and, furthermore, that she is proba-
bly more than 4 years old.
I would like to suggest the spoon test as a possible kind of test for
the purpose of assessing the presence of autonoetic episodic memory
in any species, or, indeed, in any individual member of a species. The
test is independent of linguistic or other symbolic abilities of the
tested species or individuals, and does not require any first-person
testimony regarding phenomenally apprehended processes or states.
Yet, if successful, it would provide evidence of the kind that could
not be readily explained without the postulation of something like
covert, neurocognitively grounded, autonoetic consciousness of one’s
existence in protracted time and the ability to make use of this con-
sciousness. Therefore, if successful, the test would force the rejection
of the hypothesis.
To pass the test, the individuals must act analogously to carrying
their own spoon to a feast that is likely to come in another place and
at another time. In practical terms, a chimpanzee or gorilla “spoon”
could be a straw for drinking water that is not otherwise accessible.
They can do so, the argument is, if and only if they possess the ability
to mentally travel into (or foresee, preexperience, anticipate) the fu-
ture. Lloyd Morgan’s canon would not apply in this case, because it
is difficult to imagine any simple explanation of the behavior in ques-
tion that does not include a reference to a neurocognitive capacity like
autonoesis.
The spoon test can be arranged in a number of specific ways, pro-
vided that three general requirements are observed. Its first, central
requirement is that the test subject must deliberately engage in behav-
ior that could not come about in the absence of mental time travel
into the future. That means, as already discussed, the behavior in
question must not be instigated by, and must not satisfy, a present
need or be governed by current physiological states.
A second important requirement is that the execution of the critical

behavior should not be triggered, evoked, or guided by specific envi-
ronmental stimuli that were present in the original learning situation.
This requirement has turned out to be the most difficult one to meet
in past research on episodic memory in animals. Yet is critical because
in its absence one cannot rule out the possibility that the behavior is
governed by nonautonoetic associative learning and retrieval mecha-
nisms. This means that the test subject must not be tested in the same
environment in which the to-be-remembered information was en-
countered or the event witnessed.
A third requirement is that the critical behavior should turn out to
satisfy a need that will be a part of the physical or psychological real-
ity of the animal on an occasion that actually will arise in the future.
Behavior serving to satisfy current needs, or triggered and maintained
by a discriminative stimulus potentiated during an earlier learning
episode, requires no mental time travel; it can occur, even if on a
particular occasion it need not occur, independently of the current,
perceptually present reminders of the learning episode. Behavior not
serving any current needs but serving future needs does so.
The spoon test has implications for the possibility of observing fu-
ture-oriented behavior in animals in their natural habitat. For exam-
ple, it is not the making and using of tools that is critical, but rather
storing them, or carrying them from one place to another, that points
to mental awareness of the future. Goodall (1986) has reported such
behavior in her chimpanzees of Gombe. If the observations could be
repeatedly confirmed, under conditions where no physiological need
for food and hence food-related behavior is present, they would con-
stitute evidence against the thesis of this essay. Conversely, using the
same line of argument, deliberate burial of the dead is a possible piece
of evidence of future thinking in early hominids, but more convincing
is the provision of the dead with grave goods.
Finally, note the requirement regarding a different place: The fu-
ture intention should be directed at something that happens in a place
other than one in which the present preparatory action is carried out.
The spoon has to be picked up here and now, with a deliberate inten-
tion of putting it to use somewhere else at a future time. The major
reason for the insistence on different place has to do with the minimi-
zation of the influence of the present situational cues on the spoon-
carrying behavior. [Keeping in mind the Clever Hans story, the re-
quirement for differences between original learning and present test
should be extended to the experimenters, as indeed has been done in
some of the well-controlled recent studies (Menzel, 1999; chapter 9,
this volume; see also Schwartz et al., 2002).] In their research on
“what, where, and when” of memory with scrub jays, Clayton and
colleagues (Clayton et al., 2003; Clayton & Dickinson, 1998) tested the
knowledge of the jays’ spatial (where) component of an earlier event
in the same environment in which the original learning took place.
Thus their “where” component of the jays’ knowledge was defined in
terms of differences in local space, within one and the same visual
environment. Testing episodic recollection of an event in a place other
than the one in which the event took place would make for a more
realistic analogue.
Is the Spoon Test Unfair?

Is the spoon test too difficult for nonhuman animals? Are we expect-
ing and demanding too much? Are all the conditions that the nonhu-
man test subjects must meet to satisfy really necessary? In general,
by repeatedly changing the rules of the game and setting the bar at
increasingly high levels, are we being unfair to our fellow travelers
on Spaceship Earth? Let me briefly note three points.
First, the spoon test is not difficult by human standards. Many, if
not most, 5-year-old children would have no trouble passing it, and
the rest of humankind takes the kind of abilities required to pass the
test so much for granted that they have not even found it necessary
to develop special terminology to discuss it objectively. Indeed, gener-
ations of humans have been taking thousands upon thousands of
spoon tests over thousands upon thousands of years, even if they did
not know they were doing so. And they have been passing them satis-
factorily, at least some of the time, even if not always with flying
colors. They have learned what the world is like from their experience
and then, being armed with the extraordinary insight that the world
in which they and their children will have to live will be there tomor-
row, and again the day after, and even after the next dry season, or
the next cold season, they have gone ahead to fashion it into a more
hospitable place to live. The discovery that one need not always adapt
to the world as it exists, and that one can change aspects of it for the
better, has been made only by a few species, scattered here and there
through the animal kingdom. In all cases other than humans, the abil-
ity to shape the world to better suit one’s needs is limited in scope
and requires no autonoetically guided learning. In the case of humans,
the discovery that one can improve one’s chances of survival by re-
shaping the world was based on the presumably gradual emergence
of a wholesale awareness of a dimension of the world that does not
exist as a component of the brain/mind in other species. Like the
electric fish with their self-generated electromagnetic fields in which
they can detect signals of which other creatures under the sea remain
unaware, and like the migrating birds that are sensitive to the mag-
netic properties of the earth which exist for others only physically
but not physiologically, human beings are blessed with a remarkable
awareness of a dimension of the world that seems to have eluded the

grasp of other nervous systems. The slow emergence of this special
kind of awareness, autonoesis, in ways that we can only guess at,
must have been one of the most momentous events in human evolu-
tion.
Second, as to the fairness of moving standards, think of the electric
fish, or bats, or barn owls, and imagine them trying to test humans
on the kinds of brain/mind achievements that they take very much
for granted. If Dr. Doolittle asks them, would they admit being wor-
ried about the tests being fair for humans? I think not. If something
is possible for one species, it is, in principle, obviously possible for
another.
Third, it is regrettable, perhaps, that our ideas about the world,
including ideas about the brain/mind and memory, do not stay put
over time. Life would be simpler if they did. For example, Nicola
Clayton’s scrub jays would have been certified as full-fledged episodic
creatures back in 1972. But whether we like or do not like the changes
in our ideas about nature, there is no other way in science. Science
lives and survives only as long as there is progress in its understand-
ing of nature, and change is a necessary condition of progress. Those
who seek and need the comfort of permanent certainty can find it in
other scripts and books composed by great authorities.
SUMMARY
Many claims have been made about human uniqueness, that is, about
human brain/mind capacities that other animals do not possess. The
list of such capacities is long. It is reasonable to assume that no single
item in the list alone is responsible for “making us what we are.”
What makes humans unique is a collection of capabilities that emerged
in human evolution some time after what became the human line
broke off from what became the chimpanzee line.
In this essay, I have discussed what I believe is a worthy candidate
on the list of unique human traits, namely autonoetic episodic mem-
ory. Autonoetic episodic memory makes it possible for people to en-
gage in a kind of conscious activity, mental time travel, that is beyond
the reach of living creatures who do not possess episodic memory.
Mental time travel takes the form of remembering personally experi-
enced and thought-about events, occasions, and situations that oc-
curred in the past, together with imagining (preexperiencing) personal
happenings in the subjectively felt future.
Autonoetic episodic memory is a singular, even if underappreci-
ated, achievement of biological evolution. Its separate existence was
largely unknown until recently, and its relationship to other kinds of
memory has not yet been thoroughly studied. We have good evidence
of its existence only in humans, but even among humans the distribu-
tion and prevalence of (autonoetic) episodic memory is not yet well
known. It is reasonable to assume that there are (large) individual
differences among humans in autonoetic episodic memory, but the
matter has not yet been studied. It is even possible to imagine that
intelligent and cognitively capable human beings live in our world
today who have no autonoetic episodic memory and who may not be
even (fully) aware that they lack a kind of consciousness that others
sometimes talk about. This issue seems to be worthy of careful study
quite independently of the central thesis of this essay.
Episodic memory is unique to humans in the sense that no other
animals have yet been reliably reported as being capable of behaviors
that require episodic memory. Many kinds of complex behaviors of
many kinds of animals can be, and have been, interpreted as manifest-
ing episodic memory, and in many cases these behaviors do have
many features in common with behaviors that are grounded in epi-
sodic memory. Practically invariably, however, the same behaviors
can also be interpreted more parsimoniously, as manifestations of se-
mantic or declarative memory, which do not provide for, and do not
require postulation of, the apprehension of subjective past or subjec-
tive future time.
The hypothesis that episodic memory does not exist in nonhuman
animals is part of, and follows from, a broader theory of episodic
memory that is meant to be (1) internally consistent, or at least as
consistent as possible at the present stage of our knowledge, and (2)
not clearly contradicted by any empirical facts, at least as they are
known to us today.
A central part of the theory is the assumption that episodic memory
“grows out of but remains embedded within” other memory systems,
in particular semantic memory (Tulving, 1984, p. 260). Semantic (or
declarative) memory temporally precedes episodic memory both in
phylogenetic evolution and in ontogenetic human development, and
it provides a foundation for the operations of episodic memory in
organisms that possess it. To accept the thesis of this essay, one must
accept the idea of the precedence of semantic over episodic memory,
along the lines indicated above, or in some improved form. This idea
is one of the linchpins of the theory.
The hypothesis of the uniqueness of episodic memory in humans,
like many other scientific hypotheses, is difficult to prove, for logical
reasons. But it is amenable to evaluation by empirical data, and it can
be proven false. I have suggested one scenario, dubbed the spoon test,
whose empirical demonstration would be contrary to the uniqueness
hypothesis. The test is independent of linguistic or other symbolic
abilities of the tested species or individuals and does not require any
first-person testimony regarding phenomenally apprehended pro-

cesses or states. Yet, if successful, it would provide evidence of the
kind that could not be readily explained without the postulation of
covert, neurocognitively grounded, autonoetic consciousness of one’s
existence in protracted time and the ability to make use of this kind
of conscious ability. That is, if successful, the outcome of the test
would force the rejection of the hypothesis as currently formulated.
Finally, what about Darwin’s differences in kind and in degree be-
tween humans and others? What about the rope of phylogenetic conti-
nuity with its two ends? How do the various claims for human unique-
ness, including autonoetic episodic memory, fit into the picture?
There is no profound mystery here, and no need to create one.
There are things that bees, birds, and humans all do. But there are
also things that bees do but birds and humans do not. There are things
that birds do that bees and birds do not. And there are also things that
humans do but bees and birds do not. All living creatures are similar
to one another in many ways, and they are also different from one
another in many ways. The differences are neither in kind nor in de-
gree; they are differences in light of similarities. Life is part of an
endless matrix of sameness and diversity that is nature.
ACKNOWLEDGMENTS I greatly appreciate the invaluable help in writing this
chapter provided by Mark A. Wheeler and fruitful discussions on the topic
with Hans J. Markowitsch. My research is supported by a grant from the
Natural Sciences and Engineering Research Council of Canada and by an en-
dowment by Anne and Max Tanenbaum in support of research in cognitive
neuroscience.
REFERENCES
Aggleton, J. P., & Brown, M. W. (1999). Episodic memory, amnesia and the
hippocampal-anterior thalamic axis. Behavioral and Brain Sciences, 22, 425–
444.
Aggleton, J. P., & Pearce, J. M. (2001). Neural systems underlying episodic
memory: Insights from animal research. Philosophical Transactions of the
Royal Society, B356, 1483–1491.
Alvarez, P., Zola-Morgan, S., & Squire, L. R. (1994). The animal-model of hu-
man amnesia—long-term-memory impaired and short-term-memory in-
tact. Proceedings of the National Academy of Sciences U.S.A., 91, 5637–5641.
Atance, C. M., & O’Neill, D. K. (2001). Episodic future thinking. Trends in
Cognitive Sciences, 5, 533–539.
Bergson, H. (1911). Matter and memory. London: Allen and Unwin.
Calabrese, P., Markowitsch, H. J., Durwen, H. F., Widlitzek, H., Haupts, M.,
Holinka, B., & Gehlen, W. (1996). Right temporofrontal cortex as critical
locus for the ecphory of old episodic memories. Journal of Neurology, Neuro-
surgery, and Psychiatry, 61, 304–310.
Cermak, L. S., & O’Connor, M. (1983). The anterograde and retrograde re-
trieval ability of a patient with amnesia due to encephalitis. Neuropsycho-
logia, 21, 213–234.
Chalmers, D. J. (1996). The conscious mind: In search of a fundamental theory.
New York: Oxford University Press.
Clayton, N. S., Bussey, T. J., & Dickinson, A. (2003). Can animals recall the
past and plan for the future? Nature Reviews Neuroscience, 4, 685–691.
Clayton, N. S., Bussey, T. J., Emery, N. J., & Dickinson, A. (2003). Prometheus
to Proust: The case for the behavioral criteria for “mental time travel.”
Trends in Cognitive Sciences, 7, 436–437.
Clayton, N. S., & Dickinson, A. (1998). Episodic-like memory during cache
recovery by scrub jays. Nature, 395, 272–274.
Cohen, N. J., Poldrack, R. A., & Eichenbaum, H. (1997). Memory for items and
memory for relations in the procedural/declarative memory framework.
Memory, 5, 131–178.
Conway, M. A., & Pleydell-Pearce, C. W. (2000). The construction of autobio-
graphical memories in the self memory system. Psychological Review, 107,
261–288.
Dalla Barba, G. (2000). Memory, consciousness, and temporality: What is re-
trieved and who exactly is controlling retrieval? In E. Tulving (Ed.), Mem-
ory, consciousness and the brain: The Tallinn Conference (pp. 138–155). Phila-
delphia: Psychology Press.
Dalla Barba, G. (2001). Beyond the memory-trace paradox and the fallacy of
the homunculus—A hypothesis concerning the relationship between mem-
ory, consciousness and temporality. Journal of Consciousness Studies, 8, 51–78.
Darwin, C. (1998). Descent of man (2nd ed.). Amherst, NY: Prometheus Books.
(Original work published 1874)
Drummey, A. B., & Newcombe, N. S. (2002). Developmental changes in
source memory. Developmental Science, 5, 502–513.
Duval, S., & Wicklund, R. A. (1973). Effects of objective self-awareness on
attribution of causality. Journal of Experimental Social Psychology, 9, 17–31.
Düzel, E., Vargha-Khadem, F., Heinze, H. J., & Mishkin, M. (2001). Brain ac-
tivity evidence for recognition without recollection after early hippocam-
pal damage. Proceedings of the National Academy of Sciences U.S.A., 98, 8101–
8106.
Eichenbaum, H. (1997). How does the brain organize memories? Science, 277,
330–332.
Eichenbaum, H. (2000). A cortical-hippocampal system for declarative mem-
ory. Nature Reviews Neuroscience, 1, 41–50.
Eichenbaum, H., Otto, T., & Cohen, N. J. (1996). The hippocampal system:
Dissociating its functional components and recombining them in the ser-
vice of declarative memory. Behavioral and Brain Sciences, 19, 772.
Eldridge, L. L., Knowlton, B. T., Furmanski, C. S., Bookheimer, S. Y., & Engel,
S. A. (2000). Remembering episodes: A selective role for the hippocampus
during retrieval. Nature Neuroscience, 3, 1149–1152.
Fivush, R., Hudson, J., & Nelson, K. (1984). Children’s long-term memory
for a novel event—an exploratory study. Merrill-Palmer Quarterly Journal of
Developmental Psychology, 30, 303–310.
Gaffan, D. (2001). What is a memory system? Horel’s critique revisited. Behav-

ioral Brain Research, 127, 5–11.
Gardiner, J. M. (1988). Functional aspects of recollective experience. Memory
and Cognition, 16, 309–313.
Gardiner, J. M., & Richardson-Klavehn, A. (2000). Remembering and know-
ing. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory
(pp. 229–244). New York: Oxford University Press.
Giovagnoli, A. R., Erbetta, A., & Bugiani, O. (2001). Preserved semantic access
in global amnesia and hippocampal damage. Clinical Neuropsychologist, 15,
508–515.
Goodall, J. (1986). The chimpanzees of Gombe: Patterns of behavior. Cambridge,
MA: Harvard University Press.
Gopnik, A., & Graf, P. (1988). Knowing how you know: Young children’s
ability to identify and remember the sources of their beliefs. Child Develop-
ment, 59, 98–110.
Griffiths, D. P., Dickinson, A., & Clayton, N. S. (1999). Declarative and epi-
sodic memory: What can animals remember about their past? Trends in
Guillery, B., Desgranges, B., Katis, S., de la Sayette, V., Viader, F., & Eustache,
F. (2001). Semantic acquisition without memories: Evidence from transient
global amnesia. NeuroReport, 12, 3865–3869.
Haith, M. M. (1997). The development of future thinking as essential for the
emergence of skill in planning. In S. L. Friedman & E. Kofsky-Scholnick
(Eds.), The developmental psychology of planning: Why, how, and when do we
plan? (pp. 25–42). Hillsdale, NJ: Erlbaum.
Haith, M. M., Benson, J. B., Roberts, R. J., Jr., & Pennington, B. F. (1994). The
development of future-oriented processes. Chicago: University of Chicago Press.
Hamann, S. B., & Squire, L. R. (1995). On the acquisition of new declarative
knowledge in amnesia. Behavioral Neuroscience, 109, 1027–1044.
Hayman, C. A. G., Macdonald, C. A., & Tulving, E. (1993). The role of repeti-
tion and associative interference in new semantic learning in amnesia. Jour-
nal of Cognitive Neuroscience, 5, 375–389.
Howe, M. L., & Courage, M. L. (1993). On resolving the enigma of infantile
amnesia. Psychological Bulletin, 113, 305–326.
Ingvar, D. H. (1985). “Memory of the future”: An essay on the temporal orga-
nization of conscious awareness. Human Neurobiology, 4, 127–136.
James, W. (1890). Principles of psychology. Cambridge, MA: Harvard University
Press.
Johnson, C. N., & Wellman, H. M. (1980). Children’s developing understand-
ing of mental verbs: Remember, know, and guess. Child Development, 51,
1095–1102.
Kapur, N. (1999). Syndromes of retrograde amnesia: A conceptual and empiri-
cal analysis. Psychological Bulletin, 125, 800–825.
Kircher, T. T. J., & Leube, D. T. (2003). Self-consciousness, self-agency, and
schizophrenia. Consciousness and Cognition, 12, 656–669.
Kitchener, E. G., Hodges, J. R., & McCarthy, R. (1998). Acquisition of post-
morbid vocabulary and semantic facts in the absence of episodic memory.
Brain, 121, 1313–1327.
Klein, S. B., Loftus, J., & Kihlstrom, J. F. (1996). Self-knowledge of an amnesic

patient: Toward a neuropsychology of personality and social psychology.
Journal of Experimental Psychology: General, 125, 250–260.
Klein, S. B., Loftus, J., & Kihlstrom, J. F. (2002). Memory and temporal experi-
ence: The effects of episodic memory loss on an amnesic patient’s ability
to remember the past and imagine the future. Social Cognition, 20, 353–379.
Knowlton, B. J., & Squire, L. R. (1995). Remembering and knowing: Two dif-
ferent expressions of declarative memory. Journal of Experimental Psychol-
ogy: Learning, Memory and Cognition, 21, 699–710.
Köhler, W. (1927). The mentality of apes (E. Winter, Trans.). London: Routledge
and Kegan Paul. (Original work published 1917)
Kopelman, M. D. (2002). Disorders of memory. Brain, 125, 2152–2190.
Leahey, T. H. (1987). A history of psychology (2nd ed.). Englewood Cliffs, NJ:
Prentice-Hall.
Levine, B., Black, S. E., Cabeza, R., Sinden, M., McIntosh, A. R., Toth, J. P.,
Tulving, E., & Stuss, D. T. (1998). Episodic memory and the self in a case
of isolated retrograde amnesia. Brain, 121, 1951–1973.
Lewis, M., & Brooks-Gunn, J. (1979). Social cognition and the acquisition of self.
New York: Plenum Press.
Lindsay, D. S., Johnson, M. K., & Kwon, P. (1991). Developmental changes
in memory source monitoring. Journal of Experimental Child Psychology, 52,
297–318.
Lockhart, R. S. (1984). What do infants remember? In M. Moscovitch (Ed.),
Infant memory (pp. 131–143). New York: Plenum.
Macphail, E. M. (1998). The evolution of consciousness. Oxford, UK: Oxford Uni-
versity Press.
Manns, J. R., Hopkins, R. O., & Squire, L. R. (2003). Semantic memory and
the human hippocampus. Neuron, 38, 127–133.
Markowitsch, H. J., Calabrese, P., Liess, J., Haupts, M., Durwen, H. F., & Geh-
len, W. (1993). Retrograde amnesia after traumatic injury of the frontotem-
poral cortex. Journal of Neurology, Neurosurgery, and Psychiatry, 56, 988–
992.
McKenna, P., & Gerhand, S. (2002). Preserved semantic learning in an amnesic
patient. Cortex, 38, 37–58.
Menzel, C. R. (1999). Unprompted recall and reporting of hidden objects by a
chimpanzee (Pan troglodytes) after extended delays. Journal of Comparative
Psychology, 113, 426–434.
Mishkin, M., Suzuki, W. A., Gadian, D. G., & Vargha-Khadem, F. (1997). Hier-
archical organization of cognitive memory. Philosophical Transactions of the
Royal Society, B352, 1461–1467.
Mishkin, M., Vargha-Khadem, F., & Gadian, D. G. (1998). Amnesia and orga-
nization of the hippocampal system. Hippocampus, 8, 212–216.
Morgan, C. L. (1894). An introduction to comparative psychology. London: Scott.
Morris, R. G. M. (2001). Episodic-like memory in animals: Psychological crite-
ria, neural mechanisms and the value of episodic-like tasks to investigate
animal models of neurodegenerative disease. Philosophical Transactions of
the Royal Society, B356, 1483–1491.
Murray, E. A. (1996). What have ablation studies told us about the neural
substrates of stimulus memory? Seminars in Neuroscience, 8, 13–22.
Murray, E. A., & Bussey, T. J. (2001). Consolidation and the medial temporal
lobe revisited: Methodological considerations. Hippocampus, 1, 1–7.
Nadel, L. (1994). Multiple memory systems: What and why, an update. In
D. L. Schacter & E. Tulving (Eds.), Memory systems 1994 (pp. 39–63). Cam-
bridge, MA: MIT Press.
Nelson, K. (1984). The transition from infant to child memory. In M. Moscovi-
tch (Ed.), Infant memory: Its relation to normal and pathological memory in hu-
mans and other animals (pp. 103–130). New York: Plenum.
Nelson, K. (1992). Emergence of autobiographical memory at age 4. Human
Development, 35, 172–177.
Nelson, K. (1993). The psychological and social origins of autobiographical
memory. Psychological Science, 4, 7–13.
Nelson, K., & Fivush, R. (2004). The emergence of autobiographical memory:
A social-cultural developmental theory. Psychological Review, 111, 486–511.
Noelle, D. C. (2001). Exorcising the homunculus: There is no one behind the
curtain. Free Inquiry Magazine, 21(2).
Nyberg, L., Habib, R., Tulving, E., Cabeza, R., Houle, S., Persson, J., & McIn-
tosh, A. R. (2000). Large scale neurocognitive networks underlying epi-
sodic memory. Journal of Cognitive Neuroscience, 12, 163–173.
Nyberg, L., & Tulving, E. (1997). Searching for memory systems. European
Journal of Cognitive Psychology, 9, 121–125.
Olton, D. S. (1984). Comparative analysis of episodic memory. Behavioral and
Brain Sciences, 7, 250–251.
Olton, D. S. (1985). Memory: Neuropsychological and ethopsychological ap-
proaches to its classification. In L.-G. Nilsson & T. Archer (Eds.), Perspec-
tives on learning and memory (pp. 95–113). Hillsdale, NJ: Lawrence Erlbaum.
Olton, D. S., Becker, J. T., & Handelmann, G. E. (1979). Hippocampus, space,
and memory. Behavioral and Brain Sciences, 2, 313–365.
Olton, D. S., Becker, J. T., & Handelmann, G. E. (1980). Hippocampal function:
Working memory or cognitive mapping? Physiological Psychology, 8, 239–
246.
Olton, D. S., & Pappas, B. C. (1979). Spatial memory and hippocampal func-
tion. Neuropsychologia, 17, 669–682.
Perner, J., & Ruffman, T. (1995). Episodic memory and autonoetic conscious-
ness: Developmental evidence and a theory of childhood amnesia. Journal
of Experimental Child Psychology, 59, 516–548.
Povinelli, D. J., Landau, K. R., & Perilloux, H. K. (1996). Self-recognition in
young children using delayed versus live feedback: Evidence of a develop-
mental asynchrony. Child Development, 67, 1540–1554.
Povinelli, D. J., & Vonk, J. (2003). Chimpanzee minds: Suspiciously human?
Rajaram, S., & Coslett, H. B. (2000). New conceptual associative learning in
amnesia. Memory and Language, 43, 291–315.
Rajaram, S., & Roediger, H. L., III. (1997). Remembering and knowing as
states of consciousness during recollection. In J. D. Cohen & J. W. Schooler
(Eds.), Scientific approaches to the question of consciousness (pp. 213–240).
Hillsdale, NJ: Erlbaum.
Roberts, W. A. (2002). Are animals stuck in time? Psychological Bulletin, 128,
473–489.
Roberts, W. A., & Roberts, S. (2002). Two tests of the stuck-in-time hypothesis.
Journal of General Psychology, 129, 415–429.
Roediger, H. L., III, Marsh, E. J., & Lee, S. C. (2002). Kinds of memory. In H.
Pashler & D. Medin (Eds.), Stevens’ handbook of experimental psychology: Vol.
2. Memory and cognitive processes (3rd ed., pp. 1–41). New York: Wiley.
Romanes, G. J. (1881). Animal intelligence. London: Kegan Paul.
Rosenbaum, R. S., Kohler, S., Schacter, D. L., Moscovitch, M., Westmacott, R.,
Black, S. E., Gao, F., & Tulving, E. (in press). The case of K.C.: Contribu-
tions of a memory-impaired person to memory theory. Neuropsychologia.
Rousseaux, M., Godfrey, O., Cabaret, M., Bernati, T., & Pruvo, J. P. (1997).
Retrograde memory after rupture of aneurysms of the anterior communi-
cating artery. Revue de Neurologie, 153, 659–668.
Rovee-Collier, C., & Hayne, H. (2000). Memory in infancy and early child-
hood. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory
Schacter, D. L., Harbluk, J. L., & McLachlan, D. R. (1984). Retrieval without
recollection: An experimental analysis of source amnesia. Journal of Verbal
Learning and Verbal Behavior, 23, 593–611.
Schacter, D. L., & Tulving, E. (1994). What are the memory systems of 1994?
In D. L. Schacter & E. Tulving (Eds.), Memory systems 1994 (pp. 1–38). Cam-
Schacter, D. L., Wagner, A. D., & Buckner, R. L. (2000). Memory systems of
1999. In E. Tulving & F. I. M. Craik (Eds.), The Oxford handbook of memory
Schwartz, B. L., Colon, M. R., Sanchez, I. C., Rodriguez, I. A., & Evans, S.
(2002). Single-trial learning of “what” and “who” information in a gorilla
(Gorilla): Implications for episodic memory. Animal Cognition, 5, 85–90.
Sherry, D. F., & Schacter, D. L. (1987). The evolution of multiple memory
systems. Psychological Review, 94, 439–454.
Shettleworth, S. (1998). Cognition, evolution, and behavior. New York: Oxford
University Press.
Squire, L. R. (1987). Memory and brain. New York: Oxford University Press.
Squire, L. R. (1992). Memory and the hippocampus: A synthesis from findings
with rats, monkeys, and humans. Psychological Review, 99, 195–231.
Squire, L. R., & Kandel, E. R. (1999). Memory: From mind to molecules. New
York: Scientific American Library.
Squire, L. R., & Zola, S. M. (1998). Episodic memory, semantic memory, and
amnesia. Hippocampus, 8, 205–211.
Suddendorf, T. (1994). Discovery of the fourth dimension: Mental time travel and
human evolution. Unpublished master’s thesis, University of Waikato, New
Zealand.
Suddendorf, T., & Corballis, M. C. (1997). Mental time travel and the evolu-
tion of the human mind. Genetic and Social General Psychology Monographs,
123, 133–167.
Taylor, M., Esbensen, B. M., & Bennett, R. T. (1994). Children’s understanding
of knowledge acquisition: The tendency for children to report that they
have always known what they have just learned. Child Development, 65,
1581–1604.
Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Don-

aldson (Eds.), Organization of memory (pp. 381–403). New York: Academic
Press.
Tulving, E. (1983). Elements of episodic memory. Oxford, UK: Clarendon Press.
Tulving, E. (1984). Relations among components and processes of memory.
Behavioral and Brain Sciences, 7, 257–268.
Tulving, E. (1985a). How many memory systems are there? American Psycholo-
gist, 40, 385–398.
Tulving, E. (1985b). Memory and consciousness. Canadian Psychology, 26, 1–12.
Tulving, E. (1989a). Memory: Performance, knowledge, and experience. Euro-
pean Journal of Cognitive Psychology, 1, 3–26.
Tulving, E. (1989b). Remembering and knowing the past. American Scientist,
77, 361–367.
Tulving, E. (1993a). What is episodic memory? Current Perspectives in Psycho-
logical Science, 2, 67–70.
Tulving, E. (1993b). Self-knowledge of an amnesic individual is represented
abstractly. In T. K. Srull & R. S. Wyer, Jr. (Eds.), The mental representation of
trait and autobiographical knowledge about the self (pp. 147–156). Hillsdale, NJ:
Erlbaum.
Tulving, E. (1993c). Varieties of consciousness and levels of awareness in
memory. In A. Baddeley & L. Weiskrantz (Eds.), Attention: Selection, aware-
ness and control. A tribute to Donald Broadbent (pp. 283–299). London: Ox-
ford University Press.
Tulving, E. (1995). Organization of memory: Quo vadis? In M. S. Gazzaniga
(Ed.), The cognitive neurosciences (pp. 839–847). Cambridge, MA: MIT Press.
Tulving, E. (1998). Study of memory: Processes and systems. In J. K. Foster &
M. Jelicic (Eds.), Memory: Systems, process, or function? (pp. 11–30). Oxford,
UK: Oxford University Press.
Tulving, E. (2000). Concepts of memory. In E.Tulving & F. I. M. Craik (Eds.),
The Oxford handbook of memory (pp. 33–43). New York: Oxford University
Press.
Tulving, E. (2001a). Episodic memory and common sense: How far apart?
Philosophical Transactions of the Royal Society B, 356, 1505–1515.
Tulving, E. (2001b). The origin of autonoesis in episodic memory. In H. L.
Roediger, J. S. Nairne, I. Neath, & A. M. Suprenant (Eds.), The nature of
remembering: Essays in honor of Robert G. Crowder (pp. 17–34). Washington,
DC: American Psychological Association.
Tulving, E. (2002a). Chronesthesia: Awareness of subjective time. In D. T.
Stuss & R. C. Knight (Eds.), Principles of frontal lobe functions (pp. 311–325).
Tulving, E. (2002b). Episodic memory: From mind to brain. Annual Review of
Tulving, E., Hayman, C. A. G., & Macdonald, C. A. (1991). Long-lasting per-
ceptual priming and semantic learning in amnesia: A case experiment.
Journal of Experimental Psychology: Learning, Memory and Cognition, 17, 595–
617.
Tulving, E., & Markowitsch, H. J. (1998). Episodic and declarative memory:
Role of the hippocampus. Hippocampus, 8, 198–204.
Tulving, E., Schacter, D. L., McLachlan, D. R., & Moscovitch, M. (1988). Prim-
ing of semantic autobiographical knowledge: A case study of retrograde
amnesia. Brain and Cognition, 8, 3–20.
Van der Linden, M., Brédart, S., Depoorter, N., & Coyette, F. (1996). Semantic
memory and amnesia: A case study. Cognitive Neuropsychology, 13, 391–413.
Van der Linden, M., Cornil, V., Meulemans, T., Ivanoiu, A., Salmon, E., &
Coyette, F. (2001). Acquisition of novel vocabulary in an amnesic patient.
Neurocase, 7, 283–293.
Vargha-Khadem, F., Gadian, D. G., Watkins, K. E., Connelly, A., Van Paes-
schen, W., & Mishkin, M. (1997). Differential effects of early hippocampal
pathology on episodic and semantic memory. Science, 277, 376–380.
Vargha-Khadem, F., Salmond, C. H., Watkins, K. E., Friston, K. J., Gadian,
D. G., & Mishkin, M. (2003). Developmental amnesia: Effect of age at in-
jury. Proceedings of the National Academy of Sciences U.S.A., 100, 10055–10060.
Verfaellie, M., Koseff, P., & Alexander, M. P. (2000). Acquisition of novel se-
mantic information in lesion location. Neuropsychologia, 38, 484–492.
Viskontas, I. V., McAndrews, M. P., & Moscovitch, M. (2000). Remote episodic
memory deficits in patients with unilateral temporal lobe epilepsy and ex-
cisions. Journal of Neuroscience, 20, 5853–5857.
Weiskrantz, L. (1985). Animal intelligence. Oxford, UK: Oxford University
Press.
Westmacott, R., & Moscovitch, M. (2001). Names and words without mean-
ing: Incidental postmorbid semantic learning in a person with extensive
bilateral medial temporal damage. Neuropsychology, 15, 586–596.
Wheeler, M. A. (2000). Episodic memory and autonoetic awareness. In E. Tul-
ving & F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 597–608).
Wheeler, M. A., & McMillan, C. T. (2001). Focal retrograde amnesia and the
episodic-semantic distinction. Cognitive, Affective, and Behavioral Neurosci-
ence, 1, 22–37.
Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic
memory: The frontal lobes and autonoetic consciousness. Psychological Bul-
letin, 121, 331–354.
Wimmer, H., Hogrefe, G.-J., & Perner, J. (1988). Children’s understanding of
informational access as source of knowledge. Child Development, 59, 386–
396.
Wright, A. A., Santiago, H. C., Sands, S. F., Kendrick, D. F., & Cook, R. G.
(1985). Memory processing of serial lists by pigeons, monkeys, and people.
Science, 229, 287–289.
Zentall, T. R., Clement, T. S., Bhatt, R. S., & Allen, J. (2001). Episodic-like
memory in pigeons. Psychonomic Bulletin and Review, 8, 685–690.
2
Self-Reflective Consciousness
and the Projectable Self
Janet Metcalfe & Hedy Kober
In 1619, while secluded in his poele, Descartes undertook to discover

the indisputable source of all knowledge, the unshakable foundation
upon which he could base his philosophy with certainty. Many years
later, in his 1637 Discourse on Method, he reported that the one thing
that he was unable to doubt, which became this foundation, is some-
thing that today we might dub self-reflective consciousness, metacog-
nition, secondary representation, or autonoetic consciousness—the
reflection of the self upon its own thoughts, memories, mental pro-
cesses, and other possible worlds, including the ability to mentally
project oneself outside the boundaries of one’s immediate stimulus
environment and thereby entertain counterfactuals. He affirmed that
in order to have self-reflection of this sort, one must have a self (cf.
Russell, 1945/1972). “But what then am I? A thing that thinks. What
is that? A thing that doubts, understands, affirms, denies, wills, re-
fuses, and that also imagines and senses” (p. 66). In this chapter, we
refer to the kind of self-reflective consciousness so aptly captured in
Descartes’s meditations and discourse as the projectable self. It can, as
Descartes noted, perceive the present and understand, affirm, and
deny; but it can also be projected into the past to allow episodic mem-
ory (Tulving, 2002; Wheeler, Stuss, & Tulving, 1997; chapters 1 and 7,
this volume). It can imagine the future and itself in it. It can imagine
other possible perspectives, including, importantly, the points of view
of other people, both physically and also, even more interestingly,
mentally and emotionally. The potential to project oneself in outer
and inner space and time allows capabilities familiar to people, but
perhaps rare or unknown in other animals: reminiscence, planning and
scheming, connection to other people through empathy and under-
standing, and manipulation of other people through lies and deception.
57
Descartes equated this self-reflective ability with the soul and ar-
gued, primarily on religious grounds (but with the uniqueness of lan-
guage being his empirical evidence), that only humans have it. With-
out it, a creature is a “mere” mechanism, regardless of the complexity
of the behavior. While taking no stand on the ineffability of the soul,
we here argue that the projectable self is the singular evolutionary
adaptation underpinning the most advanced achievements of humans
including our culture—without doubt what Dennett (1991) would call
a Good Trick.
Humphrey (2003) advanced that we humans have what he un-
ashamedly called an inner eye. This inner eye could look down on
more basic cognitions and interpret them in a user-friendly way, that
is, as motivations, feelings, goals, hopes, intentions, fears, thoughts,
memories, and so on (rather than as, say, p300s, serotonin imbalances,
hippocampal activation, or reverberating Hebb nets). He was specific
about noting that this self-reflective capability is an evolutionary ad-
vance (though he did not specify exactly at what point it came into
being). He allayed concern about the problem of having to postulate
a homunculus within a homunculus within a homunculus, saying:
“The problem of self-observation producing an infinite regress is, I
think, phony. No one would say that a person cannot use his own
eyes to observe his own feet. No one would say, moreover, that he
cannot use his own eyes, with the aid of a mirror, to observe his own
eyes. Then why should anyone say a person cannot, at least in princi-
ple, use his own brain to observe his own brain?” (p. 11).
Humphrey (2003) noted that one could exhibit many skills without
this kind of self-reflective consciousness, as in the top panel of figure
2.1. For example, one might well be able to drive, play the piano,
protect one’s young, or do many of the things that people have been
shown to be able to do automatically without input from the inner
eye. An issue that is investigated in much of this book is whether or
not much highly complex behavior of nonhuman primates might be
possible without this kind of consciousness and what behavior actu-
ally requires self-reflective consciousness. On the phenomenology—
the transcendent or otherworldly qualities of this kind of conscious-
ness—Humphrey demurs, alluding only to “one curious feature: the
output of the inner eye is part of its own input. As I expect you know,
a self-referential system of this sort may well have strange and para-
doxical properties—not least that so-called ’truth functions’ go awry”
(p. 12).
So why would such an inner eye evolve? The answer given by
Humphrey, much like that of Terrace, Nelson, Higgins, and others in
this volume, is that people are intensely social, as a species. Having
an inner eye—which gives a quick and accessible description of how
one feels and thinks oneself, what one wants, plans, and fears—may
Self-Reflective Consciousness and the Projectable Self 59
BRAIN
SENSE MOTOR
ORGANS SYSTEMS
BRAIN
SENSE MOTOR
ORGANS SYSTEMS
Figure 2.1 A. Diagrammatic representation of an entity lacking in-

sight. B. Diagrammatic representation of an entity possessing insight.
Illustrations courtesy of N. Humphrey.
allow one to make similar attributions to other people, and quickly.

Facility at predicting the behavior of others, by getting inside their
minds, provides an enormous social advantage. A person with this
ability highly honed is attractive as a mate and formidable as a foe.
This social aspect of the projectable self has come to be known as
theory of mind, and some of the literature concerned with this specific
issue is reviewed below.
A related just-so story about the evolution of the projectable self is
that “one can let one’s hypotheses die instead of oneself.” The ability
to imagine possible future situations with oneself in them in consider-

able detail, and to play out possibilities in the mental rather than the
physical world, allows a person to consider disastrous alternatives
without physical consequences, and hence to find new solutions that
might otherwise be impossible. Thinking, in this sense of simulating
events—both social and physical—has clear advantages for a species
that can do it, and some explorations of this are given in chapter 8.
Impulsiveness due to the lack of projecting the self in this way (i.e.,
acting without first thinking) in humans is considered a mental dis-
order.
Finally, the ability to look back on the details of one’s personal past,
to be able to revisit what happened to oneself in the past via episodic
memory (rather than merely having one’s semantic memory or auto-
matic responses changed, without the possibility of reflection on the
earlier events that caused the changes), also may enhance survival.
For example, one could update current judgments and courses of ac-
tion in the light of new information about the causes of events in the
past. Such malleability seems highly adaptive and also entertaining.
In the print shown in figure 2.2, by the Japanese master Hiroshige, a
powerful samurai is seen in three panels: the past, the present, and
the future. He is portrayed as looking fiercely into his own personal
past (but gesturing toward and guarding his much-hoped-for future
protectively). He contemplates the ghosts of his now-vanquished ene-
mies, which are multistable figures—at one moment innocuous snow-
laden trees, at the next fearsome skulls that come back to haunt him.
This time travel into his past, and the flexibility of the interpretation of
that past from different vantage points, as is characteristic of episodic
memory, enhances the beauty of his imagined future, portrayed as the
beyond-lovely courtesan housed in the exquisite palace in the right
panel. Episodic memory (see Nelson, 2000; chapter 4, this volume)
constitutes the foundation for one’s personal self, with all three pan-
els, as well as the narrative that joins them. Such time travel into one’s
own personal past, also, according to Tulving (chapter 1, this volume),
provides the basis for all future planning and is the mental foundation
of human culture.
Formally, the idea that people have an inner eye that can look at
other cognitive functions and content is the same as the description,
agreed upon by the field, given by Nelson and Narens (1994) for meta-
cognition (and see chapter 12, this volume). They propose that there
are at least two levels, one of which is considered to be basic cogni-
tion, and the other “meta”—at a higher level looking down upon and
making judgments about the events happening at the basic level.
When they are judgments about events in the world, of brightness or
numerosity, say, rather than about entities of the mind, such as memo-
ries or beliefs, then they are not properly called metacognitions. To be
Figure 2.2 Taira Sees the Ghosts of His Enemies, by Ando Hiroshige. From the collection of J. Metcalfe. See color
insert.
in a position to have metacognitions, an animal must first have cogni-

tions (i.e., mental representations). But the issue at stake here is not
whether other primates have cognitions (which we assume they do)
but rather whether they can reflect upon them. In Nelson and Nar-
ens’s scheme, there is a feedback loop whereby the metacognitive
level can change or modulate what goes on at the basic level. This
feedback allows the person to have control over basic-level thoughts.
Some kinds of metacognition need not require a full-blown project-
able self. For instance, metacognitive judgments might be made about
semantic knowledge (bona fide representations, but not personal
ones). The inner eye could stay stationary, in space and time, to make
such judgments. Metacognitions might also be made about events in-
volving the self over time past and future, however, or involving what
it is like to be another person, and such judgments would involve
projection. Insofar as having an inner eye that looks at one’s own
cognitions is true self-reflection, investigation of metacognition, even
in the nonprojective form, can be informative concerning the evolu-
tion of the projectable self. Presumably the ability to make even sim-
ple metacognitive judgments is a precursor to full-blown projective
consciousness. It may, indeed, be this particular capability in nonhu-
man primates that provides the most incisive clues to the evolution of
our human consciousness, since it may have been the first to manifest
itself (see also chapters 3 and 12, this volume).
In this chapter, we propose that the emergence of the projectable
self is the central evolutionary advance culminating in human con-
sciousness. For this conjecture, we owe much to Tulving (2002;
Wheeler et al., 1997), to Suddendorf and Corballis (1997) and Sudden-
dorf and Whiten (2001), to Humphrey (2003), and to Donald (1991).
This projectable self, full blown, is the capability that makes us
unique, but it is a very new adaptation. We seek to find evidence of
where, in the evolutionary tree, this capability originated. The word
“consciousness” is very slippery, insofar as there are undoubtedly
many kinds and levels of consciousness. While the projectable self
indisputably involves consciousness at the very highest level, there
are many other lower forms of consciousness sometimes referred to
as awareness, phenomenology, noetic and anoetic consciousness, sen-
tience (see especially chapters 1, 4, and 5, this volume). Beings that do
not have a fully developed sense of self-reflective projectable con-
sciousness may nevertheless have these other kinds of consciousness,
and furthermore, these are sufficient to underlie much clever behavior
and learning (see chapter 12, this volume, for an argument that they
may, indeed, underlie metacognition itself). The central tenet of this
chapter is that while there is considerable evidence for these other
kinds of consciousness in nonhuman primates, there are also some
indications of at least some characteristics of the projectable self. For
example, there is now some suggestion that monkeys may have the
beginnings of metacognition (see chapters 10, 11, and 12, this volume),
though the flexibility of this capability is far from impressive. Chimps,
bonobos, gorillas, and orangutans show greater metacognitive capa-
bilities (see chapter 13, this volume). Glimmerings of episodic mem-
ory, it is asserted by some, may be within the range of chimps and
gorillas (chapters 8 and 9, this volume;cf. chapter 1). At the human
end of the scale, there is also an indication of the newness of this
capability. Evidence from studies of autistic humans suggests that the
fully articulated projectable self may have not yet fully saturated the
human gene pool.
SELF-RECOGNITION
To have self-reflective consciousness, one must have a self. Thus, tasks
that tap into a person’s feeling of self, or self-recognition, provide an
obvious empirical starting point. The most famous task of this ilk is
the dye or rouge test, devised by Gallup (1970). Unbeknownst to the
person or animal, a spot of dye or rouge is put on his or her face, in
a location that cannot be seen without a mirror. Once the person or
animal looks in a mirror, the question is, does he or she self-refer
concerning the rouge? Gallup has argued that such self-reference indi-
cates that the creature has a sense of self (though a number of other
researchers, see Heyes 1994, 1998; Povinelli, 1993; Tomasello & Call,
1997, have suggested that self-awareness, in the Cartesian sense,
might not be implied because the animal might simply be using the
mirror to gain visual access to a part of the body previously perceived
only tactually or proprioceptively, or it might be recognizing only its
own behavior rather than its psychological or Cartesian self).
Humans
By the age of around 18 to 24 months (Amsterdam, 1972; Lewis &
Brooks-Gunn, 1979), human babies pass the rouge test. Asendorpf,
Warkentin, and Baudonniere (1996) note that at around the same time,
and in synchrony, they show evidence of recognition of mind in oth-
ers—showing empathy, communication through synchronic imita-
tion, and cooperation. If this correspondence, and the implied cross-
task dependence, turns out to be correct, then mirror self-recognition
may indeed be a marker of a projectable self. Although young chil-
dren without experience with mirrors showed some deficits in relat-
ing reflection to location in space, they showed no impairment in self-
recognition (Priel & De Schonen, 1986).
Although most autistic children recognize themselves, some do not.
Spiker and Ricks (1984) reported that only 36 of 52 autistics in their
sample, between the ages of 4 and 12 years, showed evidence of self-
recognition. Failure to self-recognize was predicted by language im-

pairment and level of functioning. Another study (Neuman & Hill,
1978) found that 1 out of 7 autistics between the ages of 5.5 and 11
years failed to recognize themselves. Last, Ferrari and Matthews
(1983) found that those autistics who did not clearly recognize them-
selves (46.7% of their sample) had the lowest mental age in the sample
(22 months) and were rated by their teachers as significantly lower-
functioning on behavioral observations of affect, attentional skills, lan-
guage, and interpersonal skills.
Nonhuman Primates
There is considerable evidence that chimps, so long as they are given
experience with mirrors, and so long as they are socially raised, can
pass the rouge test (Gallup, 1970; Gallup, McClure, Hill, & Bundy,
1971; Tomasello & Call, 1997). In Gallup’s (1970) seminal studies with
chimps and macaques, the chimps initially displayed aggressive be-
havior toward their own mirror image—as if the reflection were an-
other, and threatening, chimp. Within a few days, however, they
started using the mirror in a self-directed way—to pick food out of
their teeth, and so on. When the chimps were anesthetized and
marked with dye on a spot that they could not see directly without
the use of the mirror, they would touch the spot on their own face—
much as young children do when they are surreptitiously marked
with a trace of rouge on their face (figure 2.3). Chimps, then, pass this
Figure 2.3 Megan looking in the mirror. Photo courtesy of the Cognitive Evolution
Group, University of Louisiana at Lafayette.
test of self (see Tomasello & Call, 1997, for a review), as do orangutans
(Suarez & Gallup, 1981), and some gorillas. Koko, for example, the
famous human-reared gorilla (Patterson & Cohn, 1994), and King, the
circus gorilla studied by Schwartz (chapter 9, this volume), both
passed.
Despite extensive testing and extensive use of mirrors, there is still
scant evidence for self-recognition in monkeys. An exception was re-
ported for six cotton-topped tamarins (Hauser, Kralik, Botto, Garrett,
& Oser, 1995), but the results did not replicate (Hauser, Miller, Liu, &
Gupta, 2001). The fact that monkeys will look away rather than staring
back at another monkey may qualify the results of mirror testing. No-
vak (1996, cited in Tomasello & Call, 1997) found that when monkeys
were first trained to look at one another, they showed some evidence
of mirror self-recognition. Overall, though, clear evidence for monkey
self-recognition is lacking.
THEORY OF MIND
Gallup (1982) has argued that awareness of self implies awareness of
other minds. The self, by this perspective, is an essentially social being
who gains his or her definition from the social matrix (see also chap-
ters 3, 4, and 6, this volume). However, one might imagine that a
mind could exist in a solipsistic world, in which there is self-reflection,
but no other-reflection. This may be an empirical question though,
and, as mentioned above, self-recognition and empathy seem to de-
velop at about the same time. Furthermore, Frith and Happe (1999)
have shown an association between impairment on theory of mind
tasks and an inability to introspect, reflect on one’s own actions, or
anticipate one’s own actions. Even so, self-recognition typically devel-
ops earlier (it is usually in place by age 18 months) than does theory
of mind (which is only entrenched by about 4 years). Insofar as an
individual might have a mind, but not extend the attribution to other
beings, a failure to show evidence of theory of mind does not neces-
sarily rule out the possibility of self-reflective consciousness of some
sort. But if a nonhuman primate showed evidence of having theory of
mind, this in itself would be proof positive of self-reflective conscious-
ness of a highly projectable form.
Humans
For most adult humans, theory of mind is a ubiquitous attribution to
other people. It is often so overextended that we also attribute mind—
intentions, goals, thoughts, memories, plans, emotions—to animals,
indiscriminately, and even to inanimate objects. People will anthropo-
morphize even to moving abstract shapes, ascribing volitional mental
states to them. For example, Adolphs (1999) reviewed a replication of
a classic study by Heider and Simmel (1944) in which people viewed

a short film of an open square and some triangles and circles moving
in various directions. A typical description was, “And then the big
triangle chased the little triangle around. Finally, he went in, got in-
side the box to go after the circle, and the circle was scared of him
. . . and they went off on their way, and the big triangle got upset and
started breaking the box open” (p. 473). Contrast Klin’s (2000, p. 840)
transcription of an autistic’s description of the same film: “The big
triangle went into the rectangle. There were a small triangle and a
circle. The big triangle went out. The shapes bounce off each other.
The small circle went inside the rectangle. The big triangle was in the
box with the circle.” It is the rare person who can stick to the kind of
strict operational description of the patterns of movement given by
the autistic, stripped of intentional terminology. When they do so,
they sound very odd indeed.
Wellman, Cross, and Watson (2001) concluded from a meta-analy-
sis of experiments on theory of mind capabilities in human infants
that the development of theory of mind begins at around 2 years of
age, and continues until about 4 years of age (see also Nelson, 2000;
Perner, 1991). Of course, insofar as social intelligence is highly related
to theory of mind, the development of this capability may continue
well into adolescence and beyond. There are some cases of patients,
however, who apparently lack the ability.
Stuss, Gallup, and Alexander (2001) described a patient with le-
sions to the right frontal cortex (or with bifrontal lesions) who failed
an inference-based theory of mind task. The patient was to indicate
which of two graduate students (both of whom had previously dem-
onstrated their trustworthiness) he wished to show where a prize,
which was concealed from his own view by a barrier, was hidden.
One student sat by the subject in front of the barrier as the experi-
menter hid the prize, while the other student sat next to the experi-
menter, behind the barrier, and could clearly see where the prize was
hidden. When normal people, or patients with nonfrontal or left fron-
tal lesions are asked, in this situation, they consistently choose the
student who sat next to the experimenter, the one who could see
where the prize was hidden. Presumably they choose this student be-
cause they imagine what they would see if they were him, and realize,
effortlessly, that only the person behind the barrier could actually see
where the prize was hidden. Without theory of mind, or the realiza-
tion that another’s experience is like their own, however, they might
not make this inference. The right frontal patient chose at random
between the two students.
In contrast to Stuss’s finding, which implicates the right prefrontal
cortex in theory of mind, Happe and colleagues (1996), in a positron
emission tomography study using normals, found that a highly cir-
cumscribed region of the left medial prefrontal cortex was activated

during story comprehension that involved taking another’s perspec-
tive. This area was not activated in Asperger’s autistics, however, who
(as we shall review), have great difficulty with theory of mind tasks.
Rowe, Bullock, Polkey, and Morris (2001) found theory of mind deficits
in patient groups with either left or right unilateral frontal lesions, and
Stone, Baron-Cohen, and Knight (1998) found deficits, similar to those
of Asperger patients, in patients with bilateral orbitofrontal lesions.
It has been proposed (Baron-Cohen, Leslie, & Frith, 1985; Baron-
Cohen, 1995) that the fundamental impairment in autism is a lack of
theory of mind. Although numerous other idiosyncratic behaviors
and perceptions are associated with autism (for example, stereotypy
of movement, compulsions, extremely repetitious behavior, lack of ap-
propriate voice modulation, heightened perceptions, pain tolerance,
language deficits, etc.), the inability to understand another person’s
perspective is a theme that underlies much research (see, for example,
Adolphs, Sears, & Piven, 2001; DeLong, 1992; Hobson, 1990; Hughes
& Russell, 1993; Leslie, 1987; Ozonoff, Pennington, & Rogers, 1991, for
other theories of the dysfunction). The lack of theory of mind would
result in a social blindness that allows the highly developed Asper-
ger’s autistic to follow rules of behavior well, but without the sensitiv-
ity to nuance that we take for granted as part of a person’s social
skills. Such people often behave inappropriately without realizing it.
Two tasks, both with a number of variants, have come to be classic
tests of theory of mind research. The first (Wimmer & Perner, 1983),
often called the Sally-Anne task, is one in which a person (Sally), or
sometimes a puppet, sees some kind of reward being hidden and then
leaves the room. A second person (Anne) then changes the location of
the reward. When Sally returns, the child being tested is asked where
she will look. Normals typically say that she will search in the original
location in which she left the reward, but autistic children (and nor-
mal children younger than age 4) typically say that she will look
where the reward is actually currently located (Baron-Cohen et al.,
1985). The second task is one in which a package that typically con-
tains candy (i.e., M&M’s) is opened and revealed to contain pencils.
The child is asked what a new person will think is in the package.
Normals say M&Ms, but autistic children frequently say pencils—
revealing an inability to put themselves in the epistemic position of
the other person (Perner, Frith, Leslie, & Leekam, 1989).
It would appear that, insofar as autism is a genetically predisposed
condition, and insofar as the theory of mind hypothesis is correct, this
very high level of self-reflective consciousness may not have yet fully
saturated the human gene pool, or that it is, at the least, fragile.
Nonhuman Primates
The first exploration of theory of mind in nonhuman primates was
done by Premack and Woodruff (1978), who gave their chimpanzee,
Sarah, a series of problems to solve that they claimed involved infer-
ences about the goals, wants, needs, desires, and knowledge of other
people, that is, which implied theory of mind. For example, a video-
tape may have shown a human trying to get some inaccessible food,
or trying to extricate himself from a locked cage, or shivering from a
nonfunctioning, unplugged heater, or trying to play an unplugged
record player. Sarah was then given a forced-choice test in which one
of the alternatives was a solution to the human’s problem. That she
picked the correct choices most of the time suggested to the authors
that she empathized with the human’s problem, exhibiting theory of
mind. Humans, in this situation, would solve the problem, essentially,
by asking themselves what they would do if they were in the other
person’s shoes, and the claim was that Sarah did the same thing.
This interpretation, while intriguing, might be disputed. It is possi-
ble that the chimps simply gave the solution as a learned response, or
remembered the correct sequence from having seen it before, rather
than projecting themselves into the position of the other. The authors
argued against mere familiarity as a possibility, since the chimps
would have had the opportunity to view unplugged as well as
plugged-in plugs, or burned wicks as well as new ones. They also
argued for intentionality insofar as the responses varied depending
upon whether Sarah liked the humans involved or not. Keith was
Sarah’s favorite keeper. When he was the actor in the problems, she
solved them correctly eight out of eight times. However, when Bill,
an acquaintance whom Sarah disliked, was the actor, she was right
only two out of eight times.
The nature of the problems posed clouds a straightforward inter-
pretation. These were not classic theory of mind problems, such as
the Sally-Anne task or the M&M’s task, but rather sequences that cul-
minated in a solution. Notably, in the research on autism, three types
of sequence problems are typically given: (1) mechanical, which are
understood by all autistic participants; (2) behavioral, which can be
done without reference to mental states; and (3) mentalistic, which
require knowledge of the depicted person’s state of belief, and which
typically autistics cannot do. Only the third type provides information
about theory of mind, though all three types seem to have been in-
cluded in the Premack and Woodruff set. Thus, although Premack
and Woodruff’s contention is intriguing, their conclusions are not be-
yond dispute.
A second approach was taken by Povinelli and his group. A num-
ber of researchers have shown that chimps, like children, follow a
person’s gaze. When people do this, presumably it is because they are

looking to see what the other person sees, and it implies theory of
mind. However, based on a series of clever experiments, Povinelli
(2000; Povinelli & Eddy, 1996a, 1996b, 1996c) argued that this interpre-
tation does not necessarily follow with chimps, insofar as they do not
distinguish between people who can see or cannot (because they ei-
ther have a mask on or a bucket over their heads, say). They seem to
respond, instead, to body orientation, and conform to what Povinelli
thinks is a lower order rule rather than to a notion about what the
other person or animal is perceiving. He concludes that gaze follow-
ing does not necessarily imply theory of mind.
Interestingly enough, Premack and Premack (1983), years earlier,
noted that Jessie was the only one of the four non-language-trained
juvenile chimps who, in trying to get a trainer to move across the
room to help her with a task, without hesitation and on the first trial,
removed a blindfold from the eyes of the trainer—apparently realiz-
ing that a blindfold on the eyes impaired vision. Jessie did not remove
the blindfold when it was over the mouth. It may be the rare chimp
who actually understands the relation between eyes and seeing, and
Povinelli may not have had such a rare chimp in his cohort. It is possi-
ble, though, that despite this lack of realization, chimps might still
understand that another creature has a mental life similar to their
own. Even so, gaze-following by chimpanzees can no longer be taken
as unequivocal evidence that chimps have theory of mind.
A third approach was taken by Call and Tomasello (1999), who
devised a nonverbal false belief (Sally-Anne) task. One adult human
(the hider) hid a reward in one of two identical containers. Another
adult (the communicator) attempted to help the subject by placing a
sticker on the container that she believed to hold the reward. Both 4-
and 5-year-old children and the apes (chimps and orangutans) were
able to use the stickers to locate the rewards in the control trials. In
the critical trials, though, the communicator left the room, and the
hider switched the location of the reward. When the communicator
came back, she marked the location at which she had last seen the
reward. The hider then gave the child or the ape the chance to look
for the reward. The question was whether the subject went for the
container that had the sticker. The children tended to discount the
sticker—revealing that they knew that the communicator did not
know; but the apes went for it—revealing what appears to be a lack
of appreciation of the communicator’s knowledge, and a lack of the-
ory of mind.
Although these results mitigate against chimps having theory of
mind, more recently Tomasello, Call, and Hare (2003) have revised
their negative conclusions, and demonstrated what may be a genuine
case of chimpanzees having theory of mind. The mind that they have
a theory of, though, has to be one they care about. Tomasello et al.
used pairs of dominant and subordinate chimps, in competition for
food. Some of the time, the subordinate chimp could see the food and
could also see that the dominant could not see it, by virtue of a barrier.
The subordinates took advantage of their knowledge of what the
dominant did not know in a variety of flexible ways that suggested
that they knew what he knew. In a second set of experiments, the
subordinate watched a human hide the food and also observed
whether or not the dominant also saw the hiding. The researchers
found that when the subordinate had observed that the dominant had
not observed the hiding, he behaved quite differently than when he
knew that the dominant had seen the hiding. Povinelli and Vonk
(2003) have criticized these experiments on the grounds that the
chimps might have had past experience or innate wiring in such situa-
tions, and that, to test the notion of knowledge of mental states, one
needs to use situations that could not possibly have innate or experi-
ential precedents (which might allow “behavioral abstraction”) for the
subject chimps:
Any experiments that rely upon a behavioral abstraction will be of little

use, especially when this invariant is one the subject has previously wit-
nessed, or that they are likely to have evolved to detect or exploit. Indeed,
contrary to recent speculations, behavioral interactions that make the most
ecological sense to the organism are precisely the ones that will be least
diagnostic about whether the organism is reasoning about mental states
and behavior or behavior alone. (p. 159)
But if, following Povinelli and Vonk, we grant that tests for theory of
mind are valid only if the minds and situations have no importance
or meaning (that could be generalized from any past experience or
innate predispositions) or social importance to the person or animal
being studied, it seems doubtful that anyone, including normal adult
humans, would ever show positive results. Povinelli and Vonk (2003)
concluded that “the idea that theory of mind is the ‘holy grail’ of
comparative cognition needs to be abandoned” (p. 160). We are more
inclined to agree with Tomasello et al.’s (2003) conclusion: “The stakes
here are large. At issue is no less than the nature of human cognitive
uniqueness” (p. 156).
In summary, then, there is no completely undisputed evidence that
any nonhuman primate has theory of mind, though there is a sugges-
tion that in some circumstances that are deeply social, chimps may
have it. There is considerable evidence that some humans—those with
autism—do not have theory of mind, or at least have deficits in this
area. As mentioned above, however, this is an extreme form of projec-
tion. Not only does the individual have to have self-reflective con-
sciousness, but he or she also has to be able to attribute it to others
and to correctly make inferences from that attribution. Some kinds

of self-reflective consciousness might well exist prior to this highly
articulated form.
DECEPTION
Humans
The ability to deceive is an interesting one, from the perspective of
the projectable mind, because the tactical deceiver—to be effective—
must be able, at least to some extent, to second-guess the target of his
or her deception, understanding the thoughts, feelings, and inferences
that will be made. It implies a glimmering of theory of mind—though
perhaps short of that required for the tasks detailed in the previous
section. Without an ability to, at least partially, project oneself into the
other person’s mind, deception would be ineffective. Furthermore, the
deception has to be subtle enough to be undetected, since a detected
deception can be disastrous for the perpetrator (see Cosmides &
Tooby, 1992, for evidence and arguments concerning the evolution of
so-called cheater-detector mechanisms in humans). People are ex-
tremely good at deception, as well as at the detection of potential
deception. Novels and histories revel in it.
However, as with the other indications of social understanding, au-
tistics appear to have particular problems both in deceiving and in
detecting deception. Frith (1989) noted that the profiles given of many
of the hermit saints suggest autism. These saints are often known for
their simplicity and lack of deceit. Their truthfulness has historically
been taken as virtue, but this characterization seems altogether unde-
served since, if Frith is correct, it results not from self-restraint and
goodness under pressure from the dark side, but rather from of a
simple lack of understanding of the other’s point of view.
Several studies have investigated deception in autistic populations,
and all have shown them to be guileless. For example, Russell,
Maunthner, Sharpe, and Tidswell (1991) compared autistic subjects,
children with Down’s syndrome, and normal 3- and 4-year-olds. The
children learned that it was in their interest to tell the experimenter
to look into an empty box for a chocolate, rather than into the box
that actually contained the chocolate. Both the 4-year-olds and the
children with Down’s syndrome used the deceptive strategy, but the
autistic children and the 3-year-olds consistently went with the box
that actually contained the chocolate, failing to inhibit the knowledge
of their own epistemic state. Similarly, Yirmiya, Solomonica-Levi, and
Shulman (1996), using a procedure in which a doll creates a false trail
of footprints, showed that while the autistic children could use the
deceptive method as well as mentally retarded children (but not
nearly as well as normals), they did not realize that by so doing they
manipulated the belief of the other person involved. Other researchers
have shown deficits in deceptive tasks such as penny hiding (Baron-
Cohen, 1992; Oswald & Ollendick, 1989) in autistic participants. Thus,
deception, like other tasks requiring a projectable self, appears to
show as a deficit in autism.
Nonhuman Primates
The literature on tactical deception in nonhuman primates is largely
anecdotal. For example, de Waal (1992) cites a case in which a young
female chimp was aggressively chased by an older female, but man-
aged to escape. Ten minutes later, the older female made reconcilia-
tory gestures, approaching the younger with an open hand and mak-
ing soft panting noises (the usual chimp prelude to a kiss). However,
when the younger chimp came close, the older one lunged and bit her
fiercely, before she was able to free herself. Presumably, the sweet
gestures were just a deceptive attempt to lure the foe in close enough
for the bite. Many such tales of chimp subterfuge have been recorded
by primatologists and tabulated, systematized, and categorized in an
enormous study by Byrne and Whiten (1992; Whiten & Byrne, 1988)
who requested all reports of observed tactical deception among pri-
mate researchers. The kinds of observations reported were like that of
Coussi-Korbel (1994), in which a subordinate young male would
move in an indirect route toward a food goal to mislead a dominant
male (and thus get the food for himself), or in which monkeys would
point to the wrong location for hidden food (Mitchell & Anderson,
1997), or in which females would use their charms to distract a male
in order to get food. “One of the female baboons at Gilgil grew partic-
ularly fond of meat, although males do most hunting. A male, one
who does not willingly share, caught an antelope. The female edged
up to him and groomed him until he lolled back under her attentions.
She then snatched the antelope carcass and ran” (Jolly, 1985, p. 412).
A standardized request, which included a computation of hours of
observation, species, and so on, resulted in 253 such records, which
were then analyzed, categorized by deception type, and classified by
species.
The most common reports of deception came from our nearest rela-
tives, genus Pan—chimpanzees and bonobos, followed closely by ba-
boons. Some primates—lemurs in particular—appear not to deceive
at all, despite the fact that it would presumably be much to their ad-
vantage. Deception, among monkeys, was rare. Whiten and Byrne
(1988) noted some caveats to their results. First, the species reported
with such a high rate of deception is also one studied by researchers
who have been vocally antibehaviorist. But even leaving aside the
possibility that some researchers might see mind to a greater extent
than others, and while acknowledging that a more experimental ap-

proach would be desirable, it is difficult not to be convinced (and
amused) by the many anecdotes provided in this massive work. While
formal tests of theory of mind in nonhuman primates are equivocal,
insofar as that capacity is reflected in deceptive behavior, we humans
do not appear to be the only ones who have it.
METACOGNITION
Humans
A large and rapidly developing literature is investigating the meta-
cognitive capabilities of adults (see Metcalfe, 1996, 2000, for fairly re-
cent reviews of data and theory). Judgments about what one knows
are used to guide problem-solving behavior in humans (Metcalfe &
Wiebe, 1987; Simon, 1979; Simon & Reed, 1976) and to indicate how
close to the solution to a problem one is, or how near one is to remem-
bering a forgotten memory (Metcalfe, 1986a). Such judgments serve
as controls of problem-solving search processes and also of memory
retrieval efforts (Miner & Reder, 1994). Adults are capable of highly
refined judgments of confidence, both prospectively and retrospec-
tively (Morris, 1990). People are even able to predict, with high ac-
curacy, what they will be able to remember later, even though they
cannot remember the answer at the time they make such “feeling-of-
knowing” judgments (Blake, 1973; Butterfield, Nelson, & Peck, 1988;
Costermans, Lories, & Ansay, 1992; Cultice, Somerville, & Wellman,
1983; Gruneberg & Monks, 1974; Hart, 1965, 1967; Hertzog & Dixon,
1994; Lachman, Lachman, & Thronesbury, 1979; Leonesio & Nelson,
1990; Metcalfe, 1986a, 1986b; Nelson, Leonesio, Shimamura, Land-
wehr, & Narens, 1982; Schacter, 1983). Adults are not perfect, how-
ever, in their metacognitions, and some of their inaccuracies about
their own cognitive processes and capabilities have garnered a great
deal of attention (e.g., Bjork, 1994; Nelson & Dunlosky, 1991). These
biases and inaccuracies notwithstanding, however, it is safe to say that
normal adult humans are able to make remarkably accurate judg-
ments about what they currently know and what they will know, and
they put those judgments to use, either implicitly or explicitly (Koriat
& Goldsmith, 1996), in their behavior.
There is consensus in the field that the judgments are made by
monitoring mental contents, in a manner that is consistent with Hum-
phrey’s (2003) inner eye. For example, in making judgments of learn-
ing, people are thought to attempt to retrieve whatever they can,
given the retrieval cues available to them, then to mentally look at the
content of what they retrieved, as well as at the characteristics of the
process in which they just engaged during the retrieval (was it easy,
fluent, and fast, or labored?). Then, based on the results of this inner
looking, they give a judgment, numerical or otherwise, about how well
they have learned the targeted item. In making feeling-of-knowing
judgments, they look at the quantity and familiarity of all the informa-
tion they have, including the cue and partial information about the
target (Koriat, 1993; Metcalfe, 1993; Metcalfe, Schwartz, & Joaquim,
1993) and they assess this to make their judgment.
The monitoring and control involved in metacognition appears to
be associated with the last-developed area of the brain, namely the
frontal cortex. As with the frontal patient reported by Stuss et al.
(2001) who showed deficits in theory of mind, deficits in metacogni-
tion are also associated with frontal damage (Janowsky, Shimamura,
& Squire, 1989). Insofar as the frontal lobes mature late, it is not sur-
prising that metacognitions also appear developmentally late.
There has been very little research on metacognition in autistics, so
no firm conclusions can yet be reached. One study (Farrant, Boucher,
& Blades, 1999) investigated whether children knew about strategies
for doing things like enhancing their memory span, or whether or not
retrieval cues or verbalization enhanced performance, and found no
differences among autistic and other children. Another study (Farrant,
Blades, & Boucher, 1999) showed some impairment in the correspon-
dence between predictions and later performance between autistic
and other children, but they may have stemmed from an underlying
memory problem rather than from an inability of the autistics to pre-
dict what they should be able to remember. Clinical and self-reports
of autistics, though, suggest that they may be able to make rather
refined metacognitive judgments in some areas. They can, for exam-
ple, understand others’ behavior by constructing a theory based on
their experience of contingencies (Frith, 1989; Sacks, 1995). Insofar as
the same mechanisms might be used in some metacognitive judg-
ments, and it is not necessary to mentally project to do the tasks, one
might not expect deficits in autistics. Certain metacognitive tasks may
thus be the simplest of tasks involving an inner eye, and may be possi-
ble where more complicated tasks that require not only an inner eye,
but that the individual be able to project it in space and time, may be
impossible.
Nonhuman Primates
There are now a number of indications that nonhuman primates, in-
cluding even monkeys, may be capable of making some metacognitive
judgments. Researchers (Shields, Smith, & Washburn, 1997; Smith,
Shields, Allendoerfer, & Washburn, 1998; Smith, Shields, & Washburn,
2003; chapter 10, this volume) have shown that monkeys can make
uncertainty judgments, although it is not generally agreed upon
whether uncertainty judgments, in and of themselves, are metacogni-
tive (Metcalfe, 2003). Hampton (2001; chapter 11, this volume) has
given an impressive demonstration of metacognition, with judgments
about the contents of working memory, in monkeys. Finally, Son
and Kornell’s data (chapter 12, this volume) also indicate that even
rhesus monkeys are able to perform metacognitive tasks. Insofar as
metacognitions, however simple, may be the germ of full-blown self-
reflection and may be the first indication of an inner eye, the discov-
ery of these capabilities in species other than ourselves may be the
strongest primordial indication of self-reflective capability in other
primates.
Call (chapter 13, this volume) documents a particularly advanced
metacognitive/control capability in chimps and other primates. Not
only do they appear to know what they do not know (see also Son &
Metcalfe, 2004, for a similar capability in humans), but they also seek
to remedy their lack of knowledge. Call shows that several species
of primates (unlike lower animals such as dogs, say), in the face of
uncertainty, will actively seek information in a effort to remedy their
ignorance.
EPISODIC MEMORY
Tulving and his colleagues (Tulving, 2002; Wheeler, Stuss, & Tulving,
1997; chapter 1, this volume) have made a compelling case concerning
the close relation between episodic memory and autonoetic conscious-
ness. We claim that both are closely related to what we, in this chap-
ter, refer to as the projectable self. It follows that the pattern seen for
other markers of the projectable self, outlined above, should also be
manifested with episodic memory tasks. Nelson (2000; chapter 4, this
volume) provides an authoritative description of the development of
episodic memory in early childhood and illustrates how it relates to
the development of self-reflection, self-concept, consciousness, and in-
fantile amnesia. The timing of the onset of episodic memory seems to
correspond well with that of other projectable-self capabilities. Whether
or not nonhuman primates show any signs of episodic memory is a
topic of focal concern in other chapters (chapters 8 and 9, this volume;
and see Schwartz & Evans, 2001). Since these chapters deal with this
issue specifically, and there is much discussion throughout the book,
we defer to them.
The one domain in the literature that bears on the relation between
the projectable self and episodic memory that is not covered in other
chapters in this volume—the missing piece to the puzzle—is the
memory capability of autistics. As reviewed above, autistics have
demonstrated deficits in self-recognition, in theory of mind, and in
deception. If autistics lack a projectable self, then this should also be
manifested by impairments in episodic memory.
The few studies that we have been able to find on this topic sug-
gest, though with some equivocation, that autistics do indeed have
difficulties with just this kind of memory, though not with all kinds
of memory. For example, Boucher (1981a) reported that recall of re-
cent events in autistic children was inferior to that of both normal
children and ability-matched retarded control children. In a different
study, she (Boucher, 1981b) reported similar overall free recall of word
lists, but the autistic children relied more heavily on recalling the most
recently presented items and reported fewer of the earlier words—
perhaps suggesting less “time travel” to the earlier parts of the list.
Tager-Flusberg (1991) found that autistic children were not different
from mentally retarded and normal children in free recall of an unre-
lated list of words, or in their use of semantic or rhyme cues to re-
trieve unrecalled words from memory. However, they were impaired
in their free recall of the semantically related list.
Boucher and Warrington (1976) compared autistic children to con-
trols in a variety of tasks. Tasks that we would classify as implicating
episodic memory, namely, recall tasks (in this case of pictures, written
words, and spoken words), revealed memory deficits. In tasks that
involved less of an autonoetic component because they were cued, the
autistic children were not impaired. They were not impaired on cued
recall tasks or a test of unrelated paired associates. Oddly, though, the
autistic children were impaired on a forced-choice recognition task, a
task that might presumably be done on the basis of familiarity, and
that need not entail autonoetic involvement. Overall—though the
data are not entirely consistent—it would seem that the memory tasks
on which the autistic children were impaired were those that may
require autonoetic consciousness. The authors of these studies sug-
gested a parallel between autism and amnesia.
Similarly, Bennetto, Pennington, and Rogers (1996) found that au-
tistics were impaired on source memory, temporal order memory, free
recall, and working memory (though the last need not involve a pro-
jective self, and its impairment is also not a consistent finding in the
autism literature). They were unimpaired on short- and long-term rec-
ognition, cued recall, and new learning ability. Boucher and Lewis
(1989) have pointed out that autistic children have difficulties answer-
ing questions about their own past activities.
Finally, Klein, Chan, and Loftus (1999) conducted an interesting
case study investigating semantic and episodic self-knowledge in a
high-functioning autistic individual. At the time of the study, R.J. was
a 21-year-old whose autistic symptoms dated back to about 8 months
of age. His immediate memory span was normal, as was his verbal
fluency in generating category exemplars (measuring a kind of seman-
tic memory). Like other autistics, R.J.’s free recall of unrelated nouns
was impaired. Interestingly, R.J. was found to have rather accurate
knowledge of his own personality traits, as measured by both his own

test-retest reliability and also by the concordance of his self-ratings
with his mother’s ratings of him. However, his accurate semantic as-
sessment of himself was in stark contrast to his inability to retrieve
autobiographical events from his own past, when given those traits as
cues. Normal subjects were able to do this retrieval 10 out of 10 times,
meeting all criteria of episodic scoring. R.J. attained a score of only 2
out of 10, and then only when the criterion for acceptance was ex-
tremely lenient. An example: “Tester: Can you remember a time when
you acted friendly toward someone? R.J.: Mmm . . . when . . . when
people were nice to me. Tester: Was this a particular person you re-
member being nice to you? R.J.: Anyone” (p. 422). As the authors put
it, “Apparently, R.J. did not need to remember how he had behaved
in the past to know what he was like” (p. 425).
In summary, then, autistic children appear to be selectively im-
paired on memory tasks that are episodic in nature and that would
appear to involve autonoetic consciousness. These findings suggest
that they may be unable to project themselves into their own pasts, as
is required by such tasks. Both theory of mind tasks (requiring projec-
tion to another’s point of view), and episodic memory tasks (requiring
projection into one’s own personal past) appear to be impaired in
autism.
CONCLUSION
The conclusion that seems to be emerging, but which is debated and
disputed more fully in the other chapters in this volume, is that there
are some indications of self-reflective consciousness in some primates
other than humans. There are indications of self-recognition in the
great apes. There are suggestions of some fragmentary episodic (-like)
memory in some of the great apes. Other primates, though hardly
candidates for CEO of a major corporation, have the beginnings of an
ability to deceive. The great apes, and even monkeys, appear to have
some metacognitive capabilities. So far, the evidence that any pri-
mates other than humans have full-blown theory of mind is still under
dispute, but then very little research has been conducted on this in-
triguing topic, and it may turn out that with further investigation, a
consensus will emerge in favor of humans not being alone in this
regard. Adult humans project their consciousness of themselves into
their own past and future, and into the minds of others, with remark-
able ease. Investigation of the emergence of these capabilities, in a
primordial form in other primates, as well as the acknowledgment of
their fragility, as shown in autistic and frontal lobe patients, provides
insight into this most quintessentially human kind of knowledge—
our consciousness of ourselves.
REFERENCES
Adolphs, R. (1999). Social cognition and the human brain. Trends in Cognitive
Sciences, 3, 469–479.
Adolphs, R., Sears, L., & Piven, J. (2001). Abnormal processing of social infor-
mation from faces in autism. Journal of Cognitive Neuroscience, 13, 232–240.
Amsterdam, B. K. (1972). Mirror self-image reactions before age two. Develop-
mental Psychobiology, 5, 297–305.
Asendorpf, J. B., Warkentin, V., & Baudonniere, P. (1996). Self-awareness and
other-awareness II: Mirror self-recognition, social contingency awareness,
and synchronic imitation. Developmental Psychology, 32, 313–321.
Baron-Cohen, S. (1992). Out of sight or out of mind: Another look at deception
in autism. Journal of Child Psychology and Psychiatry, 33, 1141–1155.
Baron-Cohen, S. (1995). Mindblindness: An essay on autism and theory of mind.
Cambridge, MA: MIT Press.
Baron-Cohen, S., Leslie, A. M., & Frith, U. (1985). Does the autistic child have
a “theory of mind”? Cognition, 21, 37–46.
Bennetto, L., Pennington, B. F., & Rogers, S. J. (1996). Intact and impaired
memory functions in autism. Child Development, 67, 1816–1835.
Bjork, R. A. (1994). Memory and metamemory considerations in the training
of human beings. In J. Metcalfe & A. P. Shimamura (Eds.), Metacognition:
Knowing about knowing (pp. 185–206). Cambridge, MA: MIT Press.
Blake, M. (1973). Prediction of recognition when recall fails: Exploring the
feeling-of-knowing phenomenon. Journal of Verbal Learning and Verbal Be-
havior, 12, 311–319.
Boucher, J. (1981a). Memory for recent events in autistic children. Journal of
Autism and Developmental Disorders, 11, 293–301.
Boucher, J. (1981b). Immediate free recall in early childhood autism: Another
point of behavioural similarity with the amnesic syndrome. British Journal
of Psychology, 72, 211–215.
Boucher, J., & Lewis, V. (1989). Memory impairments and communication in
relatively able autistic children. Journal of Child Psychology and Psychiatry,
30, 99–122.
Boucher, J., & Warrington, E. K. (1976). Memory deficits in early infantile
autism: Some similarities to the amnesic syndrome. British Journal of Psy-
chology, 67, 73–87.
Butterfield, E. C., Nelson, T. O., & Peck, V. (1988). Developmental aspects of
the feeling of knowing. Developmental Psychology, 24, 654–663.
Byrne, R. W., & Whiten, A. (1992). Cognitive evolution in primates: Evidence
from tactical deception. Man, 27, 609–627.
Call, J., & Tomasello, T. (1999). A nonverbal false belief task: The performance
of children and great apes. Child Development, 70, 381–395.
Cosmides, L., & Tooby, J. (1992). Cognitive adaptation for social exchange. In
J. H. Barkow, L. Cosmides, & J. Tooby (Eds.), The adapted mind (pp. 163–
228). New York: Oxford University Press.
Costermans, J., Lories, G., & Ansay, C. (1992). Confidence level and feeling of
knowing in question answering: The weight of inferential processes. Jour-
nal of Experimental Psychology: Learning, Memory and Cognition, 18, 142–150.
Coussi-Korbel, S. (1994). Learning to outwit a competitor in mangabeys. Jour-

nal of Comparative Psychology, 108, 164–171.
Cultice, J. C., Somerville, S. C., & Wellman, H. M. (1983). Preschoolers mem-
ory monitoring: Feeling of knowing judgments. Child Development, 54, 1480–
1486.
DeLong, G. R. (1992). Autism, amnesia, hippocampus, and learning. Neurosci-
ence and Biobehavioral Reviews, 16, 63–70.
Dennett, D. C. (1991). Consciousness explained. Toronto, Canada: Little, Brown.
Descartes, R. (1999). Discourse on method. London: Penguin Books. (Original
work published 1637)
de Waal, F. B. M. (1992). Intentional deception in primates. Evolutionary An-
thropology, 1, 86–92.
Donald, M. (1991). Origins of the modern mind. Cambridge, MA: Harvard Uni-
versity Press.
Farrant, A., Blades, M., & Boucher, J. (1999). Recall readiness in children with
autism. Journal of Autism and Developmental Disorders, 29, 359–366.
Farrant, A., Boucher, J., & Blades, M. (1999). Metamemory in children with
autism. Child Development, 70, 107–131.
Ferrari, M., & Matthews, W. S. (1983). Self-recognition deficits in autism: Syn-
drome-specific or general developmental delay? Journal of Autism and De-
velopmental Disorders, 13, 317–324.
Frith, U. (1989). Autism: Explaining the enigma. Cambridge, MA: Basil Black-
well.
Frith, U., & Happe, F. (1999). Theory of mind and self-consciousness: What is
it like to be autistic? Mind and Language, 14, 1–22.
Gallup, G. G. (1970). Chimpanzees: Self-recognition. Science, 167, 86–87.
Gallup, G. G. (1982). Self-awareness and the emergence of mind in primates.
American Journal of Primatology, 2, 237–248.
Gallup, G. G., McClure, M. K., Hill, S. D., & Bundy, R. A. (1971). Capacity for
self-recognition in differentially reared chimpanzees. Psychological Record,
21, 69–74.
Gruneberg, M. M., & Monks, J. (1974). “Feeling of knowing” and cued recall.
Acta Psychologica, 38, 257–265.
Hampton, R. R. (2001). Rhesus monkeys know when they remember. Proceed-
ings of the National Academy of Sciences U.S.A., 98, 5359–5362.
Happe, F., Ehlers, S., Fletcher, P., Frith, U., Johansson, M., Gillberg, C., Dolan,
R., Frackowiak, R., & Frith, C. (1996). “Theory of mind” in the brain: Evi-
dence from a PET scan study of Asperger syndrome. NeuroReport, 8, 197–201.
Hart, J. T. (1965). Memory and the feeling-of-knowing experience. Journal of
Educational Psychology, 56, 208–216.
Hart, J. T. (1967). Memory and the memory-monitoring process. Journal of
Verbal Learning and Verbal Behavior, 6, 685–691.
Hauser, M. D., Kralik, J., Botto, C., Garrett, M., & Oser, J. (1995). Self recogni-
tion in primates: Phylogeny and the salience of species-typical traits. Pro-
ceedings of the National Academy of Sciences U.S.A., 92, 10811–10814.
Hauser, M. D., Miller, T. C., Liu, K., & Gupta, R. (2001). Cotton-top tamarins
(Saguinus oedipus) fail to show mirror-guided self-exploration. American
Journal of Primatology, 53, 131–137.
Heider, F., & Simmel, M. (1944). An experimental study in apparent behavior.

American Journal of Psychology, 57, 243–259.
Hertzog, C., & Dixon, R. A. (1994). Metacognitive development in adulthood
and old age. In J. Metcalfe & A. P. Shimamura (Eds.), Metacognition: Know-
ing about knowing (pp. 227–252). Cambridge, MA: MIT Press.
Heyes, C. M. (1994). Reflection on self-recognition in primates. Animal Behav-
ior, 47, 909–919.
Heyes, C. M. (1998). Theory of mind in nonhuman primates. Behavioral and
Hobson, P. R. (1990). On acquiring knowledge about people and the capacity
to pretend. Psychological Review, 97, 114–121.
Hughes, C., & Russell, J. (1993). Autistic children’s difficulty with mental dis-
engagement from an object: Its implications for theories of autism. Develop-
mental Psychology, 29, 498–510.
Humphrey, N. K. (2003). The uses of consciousness. In N. Humphrey (Ed.),
The mind made flesh: Essays from the frontiers of psychology and evolution (pp.
65–85). New York: Oxford University Press.
Janowsky, J. S., Shimamura, A. P., & Squire, L. R. (1989). Memory and meta-
memory: Comparisons between patients with frontal lobe lesions and am-
nesic patients. Psychobiology, 17, 3–11.
Jolly, A. (1985). The evolution of primate behavior (2nd ed.). New York: Mac-
millan.
Klein, S. B., Chan, R. L., & Loftus, J. (1999). Independence of episodic and
semantic self-knowledge: The case from autism. Social Cognition, 17, 413–
436.
Klin, A. (2000). Attributing social meaning to ambiguous visual stimuli in
higher-functioning autism and Asperger syndrome: The social attribution
task. Journal of Child Psychology and Psychiatry, 7, 831–846.
Koriat, A. (1993). How do we know that we know? The accessibility model
of the feeling of knowing. Psychological Review, 100, 609–639.
Koriat, A., & Goldsmith, M. (1996). Monitoring and control processes in the
strategic regulation of memory accuracy. Psychological Review, 103, 490–
517.
Lachman, J. L., Lachman, R., & Thronesbury, C. (1979). Metamemory through
the adult life span. Developmental Psychology, 15, 543–551.
Leonesio, R. J., & Nelson, T. O. (1990). Do different metamemory judgments
tap the same underlying aspects of memory. Journal of Experimental Psychol-
ogy: Learning, Memory, and Cognition, 16, 464–470.
Leslie, A. M. (1987). Pretense and representation: Origins of “theory of mind.”
Psychological Review, 94, 412–426.
New York: Plenum Press.
Metcalfe, J. (1986a). Feeling of knowing in memory and problem solving. Jour-
nal of Experimental Psychology: Learning, Memory, and Cognition, 12, 288–294.
Metcalfe, J. (1986b). Premonitions of insight predict impending error. Journal
of Experimental Psychology: Learning, Memory, and Cognition, 12, 623–634.
Metcalfe, J. (1993). Novelty monitoring, metacognition, and a control in a
composite holographic associative recall model: Implications for Korsakoff
amnesia. Psychological Reviews, 100, 3–22.
Metcalfe, J. (1996). Metacognitive processes. In E. L. Bjork & R. A. Bjork (Eds.),

Memory: Handbook of perception and cognition (pp. 381–407). San Diego, CA:
Academic Press.
Metcalfe, J. (2000). Metamemory: Theory and data. In E. Tulving & F. I. M.
Craik (Eds.), The Oxford handbook of memory (pp. 197–211). New York: Ox-
Metcalfe, J. (2003). Drawing the line on metacognition. Behavioral and Brain
Sciences, 26, 350–351.
Metcalfe, J., Schwartz, B. L., & Joaquim, S. G. (1993). The cue-familiarity heu-
ristic in metacognition. Journal of Experimental Psychology: Learning, Memory,
Metcalfe, J., & Wiebe, D. (1987). Intuition in insight and non-insight problem
solving. Memory and Cognition, 15, 238–246.
Miner, A. C., & Reder, L. M. (1994). A new look at feeling of knowing: Its
metacognitive role in regulating question answering. In J. Metcalfe & A. P.
Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 47–70). Cam-
Mitchell, R. W., & Anderson, J. R. (1997). Pointing, withholding information,
and deception in capucin monkeys (Cebus apella). Journal of Comparative
Morris, C. C. (1990). Retrieval processes underlying confidence in comprehen-
sion judgments. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 16, 223–232.
Nelson, K. (2000). Memory and belief in development. In D. Schacter & E.
Scarry (Eds.), Memory, brain, and belief (pp. 259–289). Cambridge, MA: Har-
vard University Press.
Nelson, T. O., & Dunlosky, J. (1991). When people’s judgments of learning
(JOLs) are extremely accurate at predicting subsequent recall: The “de-
layed-JOL-effect.” Psychological Science, 2, 267–270.
Nelson, T. O., Leonesio, R. J., Shimamura, A. P., Landwehr, R. S., & Narens,
L. (1982). Overlearning and the feeling of knowing. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 8, 279–288.
Nelson, T. O., & Narens, L. (1994). Why investigate metacognition?. In J. Met-
calfe & A. P. Shimamura (Eds.), Metacognition: Knowing about knowing (pp.
1–25). Cambridge, MA: MIT Press.
Neuman, C. J., & Hill, S. D. (1978). Self-recognition and stimulus preference
in autistic children. Developmental Psychobiology, 11, 571–578.
Oswald, D. P., & Ollendick, T. H. (1989). Role taking and social competence
in autism and mental retardation. Journal of Autism and Developmental Dis-
orders, 19, 119–127.
Ozonoff, S., Pennington, B. F., & Rogers, S. (1991). Executive function deficits
in high-functioning autistic individuals: Relationship to theory of mind.
Journal of Child Psychology and Psychiatry, 32, 1081–1105.
Patterson, F. G. P., & Cohn, R. H. (1994). Self-recognition and self-awareness
in lowland gorillas. In S. Taylor Parker, R. W. Mitchell, & M. L. Boccia
(Eds.), Self-awareness in animals and humans: Developmental perspectives (pp.
273–290). Cambridge, MA: Cambridge University Press.
Perner, J. (1991). Understanding the representational mind. Cambridge, MA: MIT
Press.
Perner, J., Frith, U., Leslie, A., & Leekam, S. (1989). Exploration of the autistic
child’s theory of mind: Knowledge, belief, and communication. Child De-
velopment, 60, 689–700.
Povinelli, D. J. (1993). Reconstructing the evolution of mind. American Psychol-
ogist, 48, 493–509.
Povinelli, D. J. (2000). Folk physics for apes. New York: Oxford University Press.
Povinelli, D. J., & Eddy, T. J. (1996a). Chimpanzees: Joint visual attention.
Psychological Science, 7, 129–135.
Povinelli, D. J., & Eddy, T. J. (1996b). Factors influencing young chimpanzees’
recognition of “attention.” Journal of Comparative Psychology, 110, 336–345.
Povinelli, D. J., & Eddy, T. J. (1996c). What young chimpanzees know about
seeing. Monographs of the Society for Research in Child Development, 61, record
#247.
Povinelli, D. J., & Vonk, J. (2003). Chimpanzee minds: Suspiciously human?
Premack, D., & Premack, A. J. (1983). The mind of an ape. New York: Norton.
Premack, D., & Woodruff, G. (1978). Does the chimpanzee have a theory of
mind?. Behavioral and Brain Sciences, 4, 515–526.
Priel, B., & De Schonen, S. (1986). Self-recognition: A study of a population
without mirrors. Journal of Experimental Child Psychology, 41, 237–250.
Rowe, A. D., Bullock, P. R., Polkey, C. E., & Morris, R. G. (2001). “Theory of
mind” impairments and their relationship to executive functioning follow-
ing frontal lobe excisions. Brain, 124, 600–616.
Russell, B. (1972). A history of western philosophy. New York: Simon and Schus-
ter. (Original work published 1945)
Russell, J., Mauthner, N., Sharpe, S., & Tidswell, T. (1991). The “window task”
as a measure of strategic deception in preschoolers and autistic subjects.
British Journal of Developmental Psychology, 9, 331–349.
Sacks, O. W. (1995). An anthropologist on Mars: Seven paradoxical tales. New
York: Knopf.
Schacter, D. L. (1983). Feeling of knowing in episodic memory. Journal of Ex-
perimental Psychology: Learning, Memory, and Cognition, 9, 39–54.
Schwartz, B. L., & Evans, S. (2001). Episodic memory in primates. American
Shields, W. E., Smith, J. D., & Washburn, D. A. (1997). Uncertain responses by
humans and rhesus monkeys (Macaca mulatta) in a psychophysical same-
different task. Journal of Experimental Psychology: General, 126, 147–164.
Simon, H. A. (1979). Information processing models of cognition. Annual Re-
view of Psychology, 30, 363–396.
Simon, H. A., & Reed, S. K. (1976). Modeling strategy shifts in a problem-
solving task. Cognitive Psychology, 8, 86–97.
Smith, J. D., Shields, W. E., Allendoerfer, K. R., & Washburn, D. A. (1998).
Memory monitoring by animals and humans. Journal of Experimental Psy-
chology: General, 127, 227–250.
Smith, D. J., Shields, W. E., & Washburn, D. A. (2003). The comparative psy-
chology of uncertainty monitoring and metacognition. Behavioral and Brain
Sciences, 26, 317–339.
Son, L. K., & Metcalfe, J. (2004). Judgments of learning: Evidence for a two-stage
process. Manuscript submitted for publication.
Spiker, D., & Ricks, M. (1984). Visual self-recognition in autistic children: De-
velopmental relationships. Child Development, 55, 214–225.
Stone, V. E., Baron-Cohen, S., & Knight, R. T. (1998). Frontal lobe contribution
to theory of mind. Journal of Cognitive Neuroscience, 10, 640–656.
Stuss, D. T., Gallup, G. G., & Alexander, M. P. (2001). The frontal lobes are
necessary for theory of mind. Brain, 124, 276–286.
Suarez, S. D., & Gallup, G. G. (1981). Self-recognition in chimpanzees and
orangutans, but not gorillas. Journal of Human Evolution, 10, 175–188.
tion of the human mind. Genetic, Social, and General Psychology Monographs,
123, 133–167.
Suddendorf, T., & Whiten, A. (2001). Mental evolution and development: Evi-
dence for secondary representation in children, great apes, and other ani-
mals. Psychological Bulletin, 127, 629–650.
Tager-Flusberg, H. (1991). Semantic processing in the free recall of autistic
children: Further evidence for a cognitive deficit. British Journal of Develop-
Tomasello, M., & Call, J. (1997). Primate cognition. New York: Oxford Univer-
sity Press.
Tomasello, M., Call, J., & Hare, B. (2003). Chimpanzees understand psycho-
logical states—the question is which ones and to what extent. Trends in
Tulving, E. (2002). Episodic memory: From mind to brain. Annual Review of
Wellman, H. M., Cross, D., & Watson, J. (2001). Meta-analysis of theory-of-
mind development: The truth about false belief. Child Development, 72, 655–
684.
Wheeler, M. A., Stuss, D. T., & Tulving, E. (1997). Toward a theory of episodic
memory, the frontal lobes, and autonoetic consciousness. Psychological Bul-
letin, 121, 331–354.
Whiten, A., & Byrne, R. W. (1988). Tactical deception in primates. Behavioral
and Brain Sciences, 11, 233–273.
Wimmer, H., & Perner, J. (1983). Beliefs about beliefs: Representation and
constraining function of wrong beliefs in young children’s understanding
of deception. Cognition, 13, 103–128.
Yirmiya, N., Solomonica-Levi, D., & Shulman, C. (1996). The ability to manip-
ulate behavior and to understand manipulation of beliefs: A comparison
of individuals with autism, mental retardation, and normal development.
Developmental Psychology, 32, 62–69.
3
Metacognition and the Evolution

of Language
Herbert S. Terrace
Psychology is the only life science that has yet to assimilate the theory
of evolution. Language provides a clear example. A child’s vocabulary
increases dramatically from approximately 18 months to 4 years, from
a few words to thousands, reaching a rate that has been estimated to
be one new word per hour. Equally impressive is a child’s mastery of
complex grammatical rules without any formal instruction. These
facts, which have been confirmed in all human cultures, suggested to
Chomsky (1965) that children are born with a language acquisition
device (LAD), an innate structure of the brain that contains the ab-
stract rules of a universal grammar.
Chomsky and other linguists have asked, quite reasonably, how
could so major a structural change have evolved by natural selection
during the approximately 6 million years that have elapsed since hu-
mans and chimpanzees split from a common ancestor? To be sure, the
point of Chomsky’s question is not whether language evolved but
how natural selection could account for the sudden appearance (as
measured in evolutionary time) of the most complex form of natural
communication known to man (Chomsky, 1986). Was there a stage at
which grammar was a quarter evolved, two thirds evolved, and so
on? If there was, what kind of grammar would that be?
Because language is so radically different from other forms of com-
munication, Chomsky argued that the LAD is an exaptation of a struc-
ture that evolved by natural selection for some other function, such
as navigation (Hauser, Chomsky, & Fitch, 2002). In this view, the LAD
does not owe its existence to natural selection because it was already
in place when the need to generate and comprehend grammatical ut-
terances arose. A number of biologists, citing Darwin’s concept of
“pre-adaptation” (Gould, 1977; Gould & Vrba, 1982; Williams, 1966),
have argued that many structures first appeared as exaptations. Wings,
84
Metacognition and the Evolution of Language 85
for example, are considered to be exaptations of structures whose

original functions were predation and/or thermoregulation. It is im-
portant to recognize, however, that exapted structures are just as sus-
ceptible to environmental pressures as structures that evolved by nat-
ural selection. Primitive wings, for example, which could propel birds
in flight for only very short distances, were modified by natural selec-
tion for long-distance flying.
In this chapter, I show how natural selection can provide a plausi-
ble explanation of the necessary conditions for the origin of language.
My explanation is constrained by known facts about the development
of language in children and about differences in the cognitive abilities
of human infants and nonhuman primates. It also explains how lan-
guage evolved as rapidly as it did by building upon a prelinguistic
developmental stage that is also uniquely human.
The rapid onset of language in human infants appears to differ
radically from other developmental processes only if one ignores the
social skills from which language emerges. Prominent among those
skills is the ability to “read” another person’s mind, that is, to detect
similarities and differences between another person’s perception of
the world and one’s own. That ability is an example of metacognition,
the central concept of this book. The simplest forms of metacognition,
which appear during the first year of life, do not require language.
In agreement with Chomsky, I argue that language is uniquely hu-
man and that its grammatical structure distinguishes it from all other
forms of animal communication. However, I also argue that the fun-
damental divergence between human and nonhuman minds resulted
from selection for metacognitive skills. Indeed, it seems inconceivable
that language could have evolved without the ability to perceive that
other individuals think as well as act.
The plan of this chapter is as follows. To highlight the importance
of social skills in the evolution of language, I contrast two types of
research that investigate cognitive precursors of language in nonhu-
man primates. The traditional approach has been to determine the
complexity of cognitive processes that can be trained in individual
laboratory-housed subjects, for example, problems administered on a
Wisconsin General Test Apparatus, tool use, list-learning, and so on.
The success of this approach, in particular Skinner’s imaginative ap-
plications of the principles of operant conditioning (Skinner, 1953),
fostered a climate in which it seemed possible to teach an animal to
master any skill, including language. It was in that Zeitgeist that proj-
ects were begun whose goal was to provide evidence that an ape
could learn the rudimentary features of language.
A more recent approach focuses on social intelligence, in particular,
the ability of nonhuman primates to infer what their conspecifics can
and cannot perceive in a particular situation (Povinelli, 2001; Povinelli
& Prince, 1999; Povinelli, Rulf, & Bierschwale, 1994; Tomasello, 1999;
Tomasello, Hare, & Fogleman, 2001). Although there have been far
fewer experiments on the social intelligence of nonhuman primates
than on the intelligence of individual subjects, it appears that our un-
derstanding of the social intelligence of our nearest relatives will
prove to be at least as important as our understanding of an animal’s
ability to master complex skills that can be trained individually.
As I hope to make clear in the remainder of this chapter, research
on the cognitive abilities of individual subjects can provide interesting
information about the comparative psychology of problem solving, but
even the most complex problem-solving skills are not sufficient for
learning a language. What is also needed is a comparative psychology
of social intelligence. Instead of viewing language as an extension of
the ability to understand and produce complex sequences of arbitrary
symbols, it is also necessary to understand the social contexts in which
symbols can, and cannot, be used by nonhuman primates. In this re-
gard, it is encouraging that questions about the social functions of lan-
guage have begun to supersede more narrowly based questions about
the linguistic abilities of apes. A similar shift of emphasis can be seen
in research on the development of language in children. In addition to
measuring vocabulary size and the ability to use particular grammatical
rules, researchers have begun to focus on social skills that emerge be-
fore a child learns to speak (Bruner, 1983; Scaife & Bruner, 1975).
THE EVOLUTION OF PROBLEM-SOLVING SKILLS

An unintended consequence of the cognitive revolution was its posi-
tive influence on the study of animal cognition. In particular, the con-
trast between Chomsky and Skinner’s views of language generated
interest in the acquisition of language and whether it was uniquely
human. During the late 1960s, two independent projects sought to
demonstrate that chimpanzees could acquire the rudiments of human
language, grammar in particular (Gardner & Gardner, 1969, 1975a,
1975b; Premack, 1970, 1971). Though differing in details of methodol-
ogy, both projects attempted to reverse earlier failures to teach chim-
panzees to communicate with spoken words (Hayes, 1951; Hayes &
Hayes, 1951; Kellogg & Kellogg, 1933/1967; Ladygina-Kohts, 2002) by
shifting from a vocal to a visual medium of communication. At the
time, it seemed reasonable to attribute previous failures to significant
differences in the human and chimpanzee vocal apparatus, the posi-
tion of the larynx in particular, that made it impossible for a chimpan-
zee to articulate human phones (Lieberman, 1968, 1975).1
The Gardners endeavored to teach an infant female chimpanzee
named Washoe to use American Sign Language (ASL), a natural lan-
guage used by thousands of deaf Americans (Gardner & Gardner,
1969). Washoe was trained in a home environment in which the pre-

dominant medium of communication was ASL. Premack trained the
principal subject of his study, a juvenile female chimpanzee named
Sarah, to use an artificial visual language consisting of plastic chips
of different colors and shapes (Premack, 1970). Rather than wait for
language to emerge spontaneously, as one might with a child or with
a home-reared chimpanzee, Premack devised specific training proce-
dures for teaching Sarah various “atomic” components of language.
In an early diary report, the Gardners noted that Washoe used “two
or more signs . . . in 29 different two sign combinations and four dif-
ferent combinations of three signs” (quoted in Brown, 1970, p. 211).
That report prompted Brown, an eminent psycholinguist, to comment,
“It was rather as if a seismometer left on the moon had started to tap
out ‘S-O-S’” (p. 211). Indeed, Brown compared Washoe’s sequences
of signs to the early sentences of a child and noted similarities in
the structural meanings of Washoe’s and children’s utterances (agent-
action, agent-object, action-object, and so on).
Other projects reported similar combinations of two or more sym-
bols. Sarah produced strings of plastic chips such as Mary give Sarah
apple (Premack, 1976). Lana, a juvenile female chimpanzee, was trained
to use an artificial visual language of “lexigrams.” Each lexigram, a
combination of a particular geometric configuration and a particular
colored background, was presented on the keys of a computer console
or on a large visual display. After learning to use individual lexi-
grams, Lana learned to produce sequences of lexigrams such as Please
machine give M&M (Rumbaugh, Gill, & von Glasersfield, 1973). Two
young gorillas (Koko and Michael) and a young male orangutan
(Chantek) were trained to use signs by Patterson (1978; Patterson &
Linden, 1981) and Miles (1983), respectively. Both investigators re-
ported that their subjects produced many combinations containing
two or more signs. By the late 1970s, a considerable amount of evi-
dence suggested that apes could create new meanings by combining
words according to grammatical rules.
That claim was challenged by two independent analyses of pur-
ported grammatical sequences. One was based on data collected on
Project Nim (Terrace, 1979b) a replication of the Gardners’ project in-
volving Washoe. The goal of Project Nim was to obtain nonanecdotal
evidence of the grammatical structure of sequences of signs produced
by a chimpanzee. Nim Chimpsky was a young male chimpanzee that,
like Washoe, had been reared by his human surrogate parents in an
environment in which ASL was the major medium of communication.
Over the course of 4 years, Nim’s teachers recorded more than 20,000
combinations of two or more signs. Superficially, many of Nim’s com-
binations appeared to be generated by simple finite-state grammatical
rules (e.g., more + x; transitive verb + me or Nim). Many of Nim’s multi-
sign utterances resembled a child’s initial multiword utterances (cf.

Braine, 1976; Bretherton, McNew, Snyder, & Bates, 1983; Nelson, 1981).
All told, the corpus of Nim’s utterances provided the strongest evi-
dence to date of an ape’s ability to create sentences.
As clear as the combinatorial evidence seemed, closer inspection
showed otherwise. A frame-by-frame analysis of videotapes of Nim’s
signing revealed that his signs were nothing more than elaborate re-
quests for rewards that he could not otherwise obtain and that there
was no basis for interpreting the various sequences of signs he pro-
duced as sentences (Terrace, 1979b; Terrace, Petitto, Sanders, & Bever,
1979). The vast majority of Nim’s signs, both in his single- and multi-
sign utterances, occurred when his teachers withheld rewards until he
signed. They were also full or partial imitations of signs that his teach-
ers used to prompt Nim to sign. Virtually none of Nim’s sequences
were spontaneous. When, for example, Nim wanted to play with a cat
his teacher was holding, he might sign Nim cat, me cat, hug cat, Nim
hug cat before his teacher handed him the cat. Even then, videotape
analyses showed that the teacher prompted Nim continuously as he
was signing, for instance, Who that? or Nim want cat? (see figure 3.1).
In hindsight, the meanings of Nim’s signs were projections of his
teachers, who directed all of their attention to capturing what they
thought he was saying. So focused were they on Nim’s signs that they
were unaware of their nonspontaneous and imitative nature. It is, of
course, true that young children imitate many of their parents’ utter-
ances. But the relative frequency of a child’s imitated utterances is
substantially lower than that of a chimpanzee. Further, although the
imitative phase in children is transitory, Nim never moved beyond
that phase (Terrace et al., 1979). Unlike a child’s speech at the end of
Stage I of language acquisition (cf. Bloom, Rocissano, & Hood, 1976;
Brown, 1973), Nim’s signing remained predominantly nonspontaneous
and imitative. Analyses of available films of other signing apes revealed
similar patterns of nonspontaneous and imitative discourse (e.g., Was-
hoe signing with the Gardners (Gardner & Gardner, 1973) and her other
teachers, and Koko signing with Patterson (Schroeder, 1977). Miles
(1983) performed a discourse analysis of videotapes of the orangutan
Chantek’s signing with his teachers and reported, “there is no evidence
. . . that Chantek’s multi-sign combinations . . . are sentences” (p. 53).
My conclusions that the signs uttered by an ape are imitative of the
trainer’s signs and that their sole function is to request rewards (Ter-
race, 1985a) has yet to be challenged by evidence to the contrary, in
particular by unedited videotapes or films of an ape signing.
Different considerations led to a rejection of the view that Lana’s
and Sarah’s sequences were sentences. Thompson and Church (1980)
analyzed a corpus of approximately 14,000 of Lana’s combinations
that were collected by a computer. They concluded that the vast ma-
Figure 3.1 Nim signing me hug cat. In panel A, Nim signs me in response to his teacher
signing you; in panel B, Nim signs hug in response to his teacher beginning the sign for
Nim; in panel C, Nim signs cat in response to his teacher signing who? Nim’s signing
would have been of greater interest had his teacher not signed to him while he was
signing me hug cat. Photo by H. Terrace.
jority of those combinations resulted from paired-associate and condi-

tional discrimination learning. First, Lana learned that particular lexi-
grams were associated with particular incentives (e.g., lexigramapple →
apple). She then learned conditional discrimination rules that speci-
fied which of six stock sequences she should produce. For example, if
the incentive was in view, the stock sequence was, Please machine give
piece of X. (X refers to the symbolic member of the paired associate,
e.g., the lexigrams apple, music, banana, and chocolate.) Similarly, if
there were no incentive in view, the stock sequence was, Please put
into machine X. The symbol for the incentive was typically inserted in
the last position of the stock sentence. Although Lana clearly under-
stood the meanings of those lexigrams (in the sense that she could
use lexigrams contrastively to request particular rewards), there is no
evidence that she understood the meanings of the other lexigrams she
used to assemble stock sentences (e.g., Please, machine, give, put, and
piece of). Similar arguments have been made about the plastic symbols
that Sarah used to produce her sequences (Terrace, 1979b).
Like the sequences produced by apes that were trained to sign, the
sole function of the sequences reported by Rumbaugh and Premack
was to obtain rewards. Consider, for example, the sequence glossed
as Please machine give food. That sequence could just as well been de-
scribed as a sequence composed of four arbitrarily selected letters,
say, X → N → F → G. That a nonhuman primate could be taught to
touch symbols that its trainer glossed as please, machine, give, and food
is of interest not because of the purported meanings of those symbols,
but because it raises an important evolutionary question about cogni-
tive precursors of language. How does a nonhuman primate memo-
rize arbitrary sequences without language, in particular, without un-
derstanding the meaning of each symbol?
To explore that question, I performed experiments in which pi-
geons and monkeys were trained to produce rote sequences (Straub
& Terrace, 1981; Swartz, Chen, & Terrace, 1991). The sequences in
question were lists of arbitrarily selected photographs. Although there
are fundamental differences in the manner in which pigeons and
monkeys represent an arbitrary sequence (Terrace, 1993), there is no
reason to assume that either species required linguistic ability of any
kind to learn those sequences. In the following section, I summarize
the remarkable ability of monkeys to develop expertise at learning
arbitrary lists (Terrace, Son, & Brannon, 2003). Although the lists on
which the monkeys were trained were longer by a considerable mar-
gin than any list mastered in previous experiments by a nonhuman
primate, including sequences learned by “linguistic apes,” they were
as meaningless as a list of nonsense syllables of the type used in ex-
periments on human memory.
Simultaneous Chains
The lists on which the monkeys were trained differed radically from
traditional tasks that have been used to train animals to learn an arbi-
trary task (e.g., running through a maze). To navigate a maze, a sub-
ject must learn how to respond at successive choice points. For exam-
ple, to find one’s way while driving through a strange town, one must
remember to turn left at the library, right at the bank, right at the gas
station, and so on—in short, a series of stimulus-response associa-
tions. Because the choice points of a maze are experienced in isolation
from one another, there is no need to represent one’s position in the
sequence while navigating the maze. It is sufficient to recognize each
choice point as a particular stimulus of a successive chain and to then
respond appropriately.
Instead of training monkeys to learn successive chains, we trained
them to learn simultaneous chains, a much more difficult serial task.
In contrast to a successive chain, all of the choice points that comprise
a simultaneous chain are displayed at the same time, typically on a
touch-sensitive video monitor (Terrace, 1984). Another important fea-

ture of a simultaneous chain is that the configuration of the choice
points can be changed randomly from trial to trial. These features of
a simultaneous chain force a subject to maintain a representation of its
position in the sequence before making each response. The following
thought experiment, based on the seven-item simultaneous chains on
which the monkeys were trained, shows why.
Imagine trying to enter your seven-digit personal identification
number (PIN) at a cash machine, say 9-2-1-5-8-4-7, on which the posi-
tions of the numbers were changed each time you tried to obtain cash.
You could not enter your PIN by executing a sequence of distinctive
motor movements, that is, first pressing the button in the lower right
corner of the number pad to enter 9, then the button in the upper
middle position to enter 2, and so on. Instead, you would have to
search for each number and keep track mentally of your position in
the sequence as you pressed different buttons. As difficult as this task
may seem, it would be far more difficult if you did not know your PIN
and you had to discover it by trial and error. Any error ended that trial
and resulted in a new trial on which the digits were displayed in a
different configuration. Thus, to determine your PIN, you would have
to recall the consequences of any of the 21 types of logical errors you
could make while attempting to produce the required sequence. Fur-
ther, you would have to determine the first six digits without getting
as much as a penny from the cash machine. This is precisely the type
of problem the monkeys had to solve at the start of training on each 7-
item list. Instead of numerals, the monkeys had to respond to photo-
graphs. Instead of cash, they were given banana pellets.
Figure 3.2 illustrates the trial structure of the seven-item lists on
which the monkeys were trained. Each list was composed of different
photographs that were selected arbitrarily from a library of more than
2,500 photographs of natural or artificial objects. The monkey’s task
was to respond to each item in a particular order (figure 3.2A), regard-
less of its spatial position (figure 3.2B). The simultaneous display of
all of the items ensured that the monkey’s task was at least as difficult
as the paradigms used by Premack and Rumbaugh. In those para-
digms, all of the symbols the apes had to manipulate or touch were
available at the same time, but their positions were fixed.
The monkeys had to learn each list by trial and error. As shown in
figure 3.2C, any error terminated the trial and produced a brief time-
out during which the computer monitor was dark. All correct re-
sponses allowed the trial to continue and also produced brief auditory
and visual feedback. A food reward, in the form of small banana-
flavored pellets, was provided only after the subject completed the
entire sequence correctly. Thus, to learn the ordinal positions of the
first n − 1 items of each list, subjects had to rely exclusively on second-
Figure 3.2 Simultaneous chaining paradigm. A. Example of a seven-item
list. Subjects are required to respond to each color photograph in a
prescribed order (A → B → C → D → E → F → G). B. Variation of the
configuration of list items. List items were presented on a touch-sensitive
video monitor in randomly configured displays. The task was to touch
the items in the prescribed order irrespective of their position on the
monitor. Prior to each trial, a new configuration was selected at random
from the more than 5.8 million configurations that could be generated
by presenting seven items in any of 16 positions. C. Types of forward
and backward errors. The shaded boxes show the types of logical errors
that could occur before a subject discovers the correct sequence. See
text for additional details. D. Determining the ordinal position of a list
item. The probability of guessing A by making logical errors to eliminate
the A− items. See text for additional details. See color insert.
92
ary reinforcement (the positive or negative feedback that followed

their responses to each of the first n − 1 items).
The bottom half of figure 3.2C shows the different types of logical
error a subject can make while determining the ordinal position of
each item at the start of training on a new seven-item list. A logical
error is the first incorrect guess a subject makes to a particular item at
a given position of the list (e.g., responding to G at the second posi-
tion). By definition, each type of logical error can occur only once.
Although logical errors are necessary for discovering the ordinal posi-
tion of an item, repetitions of the same error are not. Logical errors
are made to obtain information by virtue of their consequences; for
example, G cannot be the second item because the trial was termi-
nated. Repetitive errors occur because the subject has forgotten the
consequences of an earlier logical error.
The subjects of this experiment were trained to memorize 7 three-
item lists, 11 four-item lists, and 4 seven-item lists. The goal of three-
and four-item training was based on Harlow’s (1949) classic study of
learning sets. Instead of achieving a high level of proficiency on each
list, the goal was to have the monkeys learn a trial-and-error strategy
for determining the order of items from a new list. Thus, on three-
and four-item lists, subjects were given a maximum of 3 days of train-
ing and then moved onto a new list. If they responded at better than
chance levels of accuracy before 3 days, a new list was introduced at
the start of the next session. On seven-item lists, subjects were trained
to a stringent criterion of completing at least 65% of the trials correctly
during a 60-trial session.
Learning New Lists

Subjects made progressively fewer errors during the course of learn-
ing new three- and four-item lists. However, the most dramatic evi-
dence of expertise at acquiring new lists was subjects’ performance on
seven-item lists. The likelihood of guessing the correct order in which
to respond to the items of a novel seven-item list (.0005) was less than
1/1000th the likelihood of guessing the correct order on a novel four-
item list (.04). Each monkey nevertheless needed progressively fewer
sessions to satisfy the high accuracy criterion on these lists (65% cor-
rectly completed trials in one session). As can be seen in figure 3.3,
subjects needed on average 31.5, 17.5, 13, and 12.25 sessions to master
seven-item lists 1, 2, 3, and 4, respectively (ranges: 21–55, 11–25, 11–
19, and 7–17, respectively).
Knowledge of the Ordinal Position of List Items

Monkeys are not only adept at learning new lists but can also apply
their knowledge of each list to solve new problems. Consider the fol-
lowing example. You are given the name of two baseball players from
100
90
80
70
60
50
40
List 1
30
List 2
20 List 3
List 4
10
0
2 4 6 8 10 12 14 16 18 20 22 24 26 28 30 32
SESSION NUMBER
Figure 3.3 Learning seven-item lists. Each function shows the mean accuracy of respond-
ing on each seven-item list during even-numbered sessions. The probability of executing
a new seven-item list correctly by chance is 1/7! = 1/5,040 (assuming no backward errors).
your local team and asked which one bats first, Smith or Jones. You
solve that problem by recalling each player’s position in the batting
order. Since Jones bats fourth and Smith seventh, you answer that
Jones bats first. You’re then asked the same question about players
from different teams: Thompson, the fourth batter from Team B, or
Armstrong, the second batter from Team C. Using the same rule of
choosing the batter whose position is lowest in the batting order, you
answer Armstrong. Suppose, however, that, like monkeys, you never
learned numbers. To come up with the correct answer, you would
have to go through each team’s batting order mentally and stop when
you came to one of the players.
Surprisingly, monkeys excelled on a similar test based on the
seven-item lists they learned without the benefit of any knowledge of
numbers or names. The monkeys were presented with all of the two-
item subsets that could be derived from the 28 items used to compose
those lists: 84 within-list subsets and 252 between-list subsets. Within-
list subsets were composed of items from a particular list (e.g., the
subsets A3B3, A3C3 . . . A3G3; B3C3, B3D3 . . . F3G3, from List 3). Between-
list subsets were composed of items drawn from different lists (e.g.,
the subsets A2B4 from Lists 2 and 4, C3F5 from Lists 3 and 5, E1G3 from
Lists 1 and 3, and so on). Within- and between-list pairs were inter-
spersed randomly throughout the subset test. Subjects were rewarded
if they responded in the order specified by their ordinal positions on
the original list. For example, subjects were rewarded for responding
to items C and F, in that order, whether or not they came from the
same or different lists (e.g., C3F3 or C2F6). Excluded from the subset
test were the 21 pairs that could be composed of items occupying the
same ordinal positions on the 4 seven-item lists (e.g., the subsets A1A2,
A1A3 . . . G3G4).
Subjects’ accuracy to the first item of both within- and between-list
subsets was almost perfect on the first presentation of each subset
type. On average, subjects responded correctly on 94% of the trials on
the first occasion on which a particular token of a within-list subset
was presented, and on 91% of the trials on the first occasion on which
a particular token of a between-list subset was presented. These data
are significant because they show that each monkey represented, in
long-term memory, the ordinal position of the items of each seven-
item list and that they were able to compare, in working memory, the
ordinal positions of any two items drawn from those lists.
Representation of the Ordinal Position of List Items

How might a monkey compare the ordinal positions of subset items
from the same or from different lists without the benefit of numerical
tags such as first, second, and so on? An analysis of subjects’ reaction
times (RTs) to the first item of each subset suggested that they repre-
sented each item’s ordinal position spatially along a linear continuum
and then used the distance from the beginning of the list to decide
which item came first. For this analysis, subsets were assigned to one
of six categories that were defined by the distance between their ordi-
nal positions on the original lists (e.g., pairs of items separated by a
distance of 1: the subsets A1B1, B2C2 . . . A1B2; B2C3, C3D4 . . . F3G4; a
distance of 2: A1C1, B2D2 . . . A1C2; B2D3, C3E4 . . . E3G4; a distance of 6:
A1G1, A2G1 . . . A2G1, A2G2, A2G3 . . . A4G4). As can be seen in figure 3.4,
reaction time to the first item of a subset decreased linearly as distance
increased. At a distance of 1, reaction time increased linearly as the
ordinal position of the first item increased. This was true for both
within- and between-list subsets.
The functions shown in figure 3.4 suggest that subjects represented
the ordinal position of each item in working memory and that they
then compared those positions on a linear scale to see which item was
closer to the beginning of the list. The more advanced the first item
on the original list, the longer it took to determine its position. Simi-
larly, the greater the distance between the two ordinal positions, the
less time it took to discriminate their ordinal positions. Analogous
5000
Distance: 1 2 3 4 5 6
1.0 4500
0.9 4000
0.8
Mean Reaction Time

3500
Mean Accuracy
0.7
3000
0.6
2500
0.5
2000
0.4 Chance Accuracy
1500
0.3
Accuracy 1000
0.2
Reaction Time
0.1 500
0.0 0
AB
BC
CD
BD
EG
DG
CG
BG
AG
AC
DE
FG
CE
AD
DF
BE
AE
EF
CF
BF
AF
Subset
Figure 3.4 Accuracy and reaction times on subset test. The upper functions show mean
accuracy to each type of within- and between-list subsets at distances 1–6. The entries
on the abscissa are generic in that they refer to the types of within- and between-list
subsets that are represented at each position. For example, AB refers to A1B1, A2B2, A3B3,
A4B4 in the case of within-list subsets and to A1B2, A1B3, . . . A4B1 subsets in the case of
between-list subsets. The lower functions show mean RTs to each type of within- and
between-list subsets.
reaction time functions have been obtained from human subjects in

experiments in which they were asked to discriminate pairs of letters
whose position in the alphabet varied from trial to trial (e.g., G vs. B,
M vs. R, C vs. J; Hamilton & Sanford, 1978) or between adjectives
(Potts, 1974).
Serial Expertise Without Language

The lists trained in this study are by far the most difficult lists mas-
tered by a nonhuman primate. These include the production of sym-
bolic and numerical lists (Matsuzawa, 1985; Premack, 1976; Rum-
baugh, 1977) and arbitrary lists (Colombo, Eickhoff, & Gross, 1993)
and the recognition of the ordinal position of items on arbitrary lists
(Carpenter, Georgopolous, & Pellizzer, 1999; Orlov, Yakovlev, Hoch-
stein, & Zohary, 2000). It is doubtful, however, that the sequential
skills observed in this study reflect the upper limit of a monkey’s
serial expertise. The ease with which monkeys learned seven-item lists
and the steady decrease in the number of sessions they needed to
master new lists suggests that they could learn such lists more rapidly
and that they could also master longer lists. Subjects were also able to
instantly apply their knowledge of seven-item lists to determine the
correct order in which to respond to novel pairs of items that were
selected at random from four different lists.
The Gap Between Serial Expertise and Language

The serial expertise of monkeys undermines claims that the sequences
of symbols produced by language-trained apes are meaningful or
grammatical. It also serves as a reminder that serial expertise per se
is not a sufficient condition for language. Just as the ability of a human
savant who can tell you instantly the day of the week on which Sep-
tember 3 fell in 1445 does not generalize to mathematical ability, the
ability to memorize arbitrary sequences quickly does not generalize
to grammatical ability. Serial skills are crucial for grammatical ability,
and grammatical skills are necessary to use language. It does not fol-
low, however, that our ancestors’ first linguistic utterances were se-
quential or grammatical.
The suggestion that language did not start out with grammatical
structure is not to deny that grammar is the sine qua non of language,
at least as we know it. The point is that language could only have
evolved from ancestors who had already been selected for their ability
to share and exchange knowledge nonverbally. Although it seems un-
controversial that the social function of language antedated its seman-
tic and grammatical functions, it has been argued that there can be no
serious discussion of the order in which different aspects of language
emerged in the absence of physical (paleological) evidence of the evo-
lution of language (Chomsky, 1986).
If physical evidence of language use by our ancestors is a require-
ment for theorizing about the evolution of language, a glorious mo-
ment of human history will remain forever shrouded in mystery. It is
not clear, however, why psychologists cannot emulate other sciences
that have overcome similar problems. Astrophysics provides a clear
example. Although astrophysicists lack physical evidence as to just
when and how the universe was created, their hypotheses are con-
strained by empirical knowledge about the current state of the uni-
verse.
Recent research on the verbal and nonverbal cognitive abilities of
human infants and apes suggest analogous constraints with respect to
language. Below, I consider two important examples of such con-
straints; one centered on the issue of ape “language” I discussed ear-
lier, and the other on the ability of human infants to share attention
toward a particular object with another person. Both examples illus-
trate why language presupposes metacognition.
APE LANGUAGE PROJECTS

The goals of recent ape language projects were defined by the Chom-
skian Zeitgeist in which grammar was regarded as the essential fea-
ture of language. As noted earlier, claims that apes can produce gram-
matical sequences (Gardner & Gardner, 1975b; Premack, 1976;
Rumbaugh, 1977) have been widely criticized (Pinker, 1994; Terrace
et al., 1979). It was, however, generally assumed that apes can acquire
sizeable vocabularies (Gardner & Gardner, 1973; Patterson & Linden,
1981; Premack, 1976; Rumbaugh, 1977). Because that assumption
seems less controversial, the vocabularies attributed to various apes
have undergone less scrutiny than claims about grammar. It is easy
to show, however, that the contents of ape vocabularies have been
greatly exaggerated. There is no evidence that apes used any of the
symbols they learned to refer to objects or events or that those symbols
had any function other than to request whatever food or drink its
trainer could provide (e.g., please, me, more, etc.). This was true even
of the symbols that apes used to obtain particular rewards (e.g., apple,
Coke, banana, etc.).
To see why, it is important to recognize a fundamental limitation of
association theory as an explanation of vocabulary development (e.g.,
Mowrer, 1954; Winitz, 1969). Such explanations have been justified on
the grounds that words are learned individually (unlike sentences) and
that a child’s use of words can be characterized as associations between
particular conditioned stimuli and particular conditioned responses.
Skinner (1957), for example, distinguished between two kinds of con-
ditioned verbal responses. Requests for primary reinforcers, called
mands, are reinforced in whatever context they occur (e.g., requests
for food, comfort, a smile, etc.). Skinner referred to verbal behavior
that is rewarded in a specific context and followed by a secondary
reinforcer as a tact. For example, a child might say look, or a butterfly
in the presence of a butterfly, after which a parent might respond,
good, I see it too, and so on.
Mands are egocentric. An infant mands whether or not his or her
caretaker is paying attention. As illustrated in the following diagram,
mands are composed of two events: the child’s utterance and a pri-
mary reinforcer. Tacts, which require the listener’s attention and are
only reinforced in particular contexts, consist of three events: the con-
text, the child’s utterance, and a secondary reinforcer.
Rmand → primary reinforcer (e.g., food)
Specific context: Rtact → secondary reinforcer (e.g., verbal ac-
knowledgement)
Tacts are metacognitive in the sense that the speaker expects the
listener to attend to the referent of the speaker’s tact. For example,
when a child says blue jay, and then points to a blue jay, a listener
might acknowledge what the child said with verbal praise, such as
Yes! It is a blue jay, or with some nonverbal acknowledgment that the
child’s message was received (e.g., a smile, nodding the head, etc.).
Tacts are not followed by primary reinforcement. Their sole function
is to share knowledge with another person for its own sake.
It is clear that apes can learn to mand. There is, however, no evi-
dence that they can learn to tact. Examples of mands include sponta-
neous requests for a reward, for example, a chimp signing eat, drink,
tickle, and so on, or touching the lexigrams eat, drink, chase, tickle, and
so forth, on a computer keyboard. Chimpanzees can also learn to re-
quest food by using a particular symbol in response to a stimulus
presented by their trainer (e.g., the sign for dog in American Sign Lan-
guage in response to a picture of a dog). It should be clear, however,
that the function of the utterance dog is to obtain some reward and
not to initiate a conversation with the trainer about the presence of a
dog. To think otherwise is sheer projection. Conceptually, the chim-
panzee’s response dog is a mand that occurs in a particular context:
Picture of a dog: Rdog → primary reinforcer (e.g., food)
Chimpanzees can be trained to produce mands in a wide variety
of contexts. Consider, for example, the kinds of tests that Premack
(1976) administered to Sarah, the main subject of his project. When
tested on her knowledge of names and colors, Sarah might be asked
what color? in the presence of an apple. If Sarah selected the symbol
glossed as red (as opposed to the symbol for some other color), her
trainer gave her a small morsel of food. If Sarah were asked what
name?, she was rewarded for selecting the symbol glossed as apple (as
opposed to the symbol for some other object).
To solve these problems, Sarah simply had to learn to attend to the
color of an object on trials on which the symbol glossed as color was
presented, and to the type of object on trials on which the symbol
glossed as name was presented (Terrace, 1979a). If the same test were
given to a child, the child would expect verbal praise and not a candy
or some similar reward for her answers. But, unlike a child, Sarah
would stop answering questions if her trainer stopped supplying the
tidbits of food and drink to which she had become accustomed.
Do Bonobo Chimpanzees Have the Capacity to Learn Language?

In a more recent study, Savage-Rumbaugh described the remarkable
ability of Kanzi, a young male bonobo chimpanzee (Pan paniscus), to
learn lexigrams by observational learning (Savage-Rumbaugh, 1994).
The ease with which Kanzi added new lexigrams to his vocabulary
surpassed, by a considerable margin, that observed in other studies
of ape language. Kanzi’s model for using lexigrams was his foster
mother Mutata. Like other chimpanzees trained by Savage-Rum-
baugh, Mutata learned, by trial and error, to press particular lexi-

grams to obtain a particular reward. Solely by observing Mutata,
Kanzi learned to use lexigrams in a similar manner. Without any for-
mal training, his vocabulary increased so rapidly that it exceeded the
number of lexigrams that the computer could accommodate at a par-
ticular time (256 lexigrams).
Kanzi was also encouraged to explore a 50-acre forest (with his
caretakers) containing 16 food sites. Each site, which was designated
by a particular lexigram, contained a particular food item (e.g., straw-
berries, M&M’s, kiwis, etc.). Thus, it was not unusual for Kanzi to
select both the lexigram for a particular site (e.g., the A-frame house)
and the food that could always be found there (e.g., strawberries)
when he asked a trainer to take him on a trip.
Most surprising was Kanzi’s ability to respond to spoken English.
For example, he could carry out the appropriate actions in response
to a command to perform a novel act, such as put the collar in the water,
pour the lemonade in the Coke, get the balloon that’s in the microwave, and
so on. On a test of his English comprehension, he responded correctly
to 445 out of 660 different sentences, each containing three to five
words. On the basis of the size of Kanzi’s vocabulary and his ability
to carry out multiword commands in English, Savage-Rumbaugh con-
cluded that “Kanzi could understand language at least as well as a
child of two to three years old” (Savage-Rumbaugh, Shanker, & Tay-
lor, 1998, p. 67).
There is no evidence that other animals, or 2- to 3-year-old children
for that matter, could solve problems as complex as those that Kanzi
solved. It does not follow, however, that Kanzi’s performance quali-
fies as language. All of the problems presented to Kanzi, which are
undeniably difficult, can be construed as conditional discrimination
problems whose solution is devoid of any intentional meaning (Ter-
race, 1979a, 1985b, 2002). Consider, for example, the command put the
collar in the water. This command was part of a series of commands in
which Kanzi was instructed to perform some action with an object,
either by having that object do something to another object or by hav-
ing that object interact with some agent. On each trial, the object and
any potential agents were in plain view. Kanzi’s task was simply to
understand the meanings of any of the symbols that were associated
with an object, an action, or an agent, and to then follow a simple rule
such as “the first object acts on the second object,” “retrieve object
that Kanzi has recently put away,” and so on.
The absence of intentional meaning in an ape’s use of symbols is
especially evident in various double-blind tests of vocabulary that
have been administered to rule out covert cues from the trainer. Con-
sider, for example, a widely cited “vocabulary” test given to Washoe,
the first chimpanzee trained to use signs of American Sign Language
(Gardner & Gardner, 1974). Trainer 1 showed Washoe a photograph

of a familiar object. Washoe’s signed answers were evaluated by
Trainer 2, who could not see the photographs. Throughout the test,
Trainer 2 sat impassively and gave Washoe a small piece of food
whenever Washoe’s sign matched the sign that appeared on her an-
swer sheet. Even though Washoe responded at high levels of accu-
racy, it should be clear that she was not attempting to converse with
Trainer about the test items.
After more than 30 years of research on ape language, it is clear
that apes are remarkably astute at learning symbols that are related
to various rewards but that they do not use those symbols to share
information. Evidence of symbol use for the purpose of having a con-
versation with a trainer (or with another ape) is glaringly absent from
any report of language use by apes. By contrast, there are abundant
examples of the remarkably intelligent ways in which chimpanzees
learn to use symbols to solicit rewards. This killjoy interpretation of
an ape’s use of symbols can be easily tested. Suppose that the consid-
erable range of foods, drinks, and physical games that function as
rewards for a chimp’s use of symbols were eliminated (or reduced to
the low level at which they are given to children during the normal
course of a day). I predict that symbol use would gradually disappear
under those conditions.
THE EVOLUTION OF VOCABULARY AND GRAMMAR

The ability of apes to use arbitrary symbols to solve conditional dis-
crimination problems and to learn arbitrary lists of symbols is an im-
pressive intellectual achievement, especially because it is not medi-
ated by language. But questions about an ape’s grammatical ability
are premature if, as all of the available evidence suggests, they do
not use symbols referentially. The same can be said of theories of the
evolution of language that rely exclusively on grammatical ability as
a criterion for distinguishing between communication by human and
nonhuman primates.
Paradoxically, the hard question about the evolution of language is
not at the level of syntax or serial expertise, but at the level of mean-
ingful words. A grammar provides speakers with a mechanism for
creating particular meanings by arranging meaningful symbols into
particular sequences. Grammar also provides listeners with a mecha-
nism for comprehending the meaning of those sequences (Chomsky,
1957, 1965).
Like grammatical ability, referential ability is uniquely human. It is
important to recognize, however, that those abilities emerged at dif-
ferent times during the evolutionary progression from apes to hu-
mans. Grammatical ability is the end point of a long process of cogni-
tive development that was set in motion when some ancestral group
of primates began to communicate in a manner that differed radically
from the ritualized calls they made to one another while foraging,
grooming, mating, sensing danger, and so forth. At some critical
point, particular utterances acquired an intentional function in the
sense that one individual sought to influence another individual’s
mental state, rather than behavior.
The addition of intentional communication to apes’ already exten-
sive repertoire of social skills set the stage for an explosive expansion
of their considerable cognitive abilities. Most likely, intentional com-
munication began modestly in small groups of our ancestors, whose
intentional vocabularies were initially limited. From such ordinary be-
ginnings, however, it is easy to see how intentional communication
took root and how it could escalate. For the first time, it was possible
for primates to share their thoughts, perceptions, and desires and to
comment about events and individuals of mutual interest, such as
feeling happy, a particular patch of fruit, someone who broke a rule,
and so on. It was also possible to maintain small cultures, engage in
simple pedagogy, plan trips, and so forth (Donald, 1991). Each of
these examples has obvious adaptive value. In combination, there
would be strong selective pressure for larger and larger vocabularies
of intentional utterances.
Empirical Evidence of the Development of Intentional Skills

Although less complex than grammatical ability, the origins of even
the simplest forms of intentional communication pose questions that
are similar to those raised by the origins of grammatical ability. What
types of evidence are needed to attribute intentional communication
to a nonhuman primate? Is intentional communication an innate abil-
ity? What were its immediate precursors?
Unlike questions about the origins of grammatical ability, questions
about the origins of intentional communication are more amenable to
research, both with human and nonhuman subjects. One important
line of research was inspired by Premack and Woodruff’s (1978) influ-
ential article on a chimpanzee’s “theory of mind” (TOM), specifically,
a chimpanzee’s ability to “impute mental states to [one]self and to
others” (p. 515). To provide evidence for a TOM, Premack and Wood-
ruff designed a paradigm in which Sarah (their star subject) was
shown a videotape of an actor in a stressful situation. Sarah was then
asked to report the actor’s state of mind by choosing one of two pho-
tographs. One depicted an appropriate course of action, the other an
inappropriate alternative, such as a picture of a key on a trial on
which the actor was shown trying to open a door, as opposed to a
photograph of the actor eating, sleeping, or some such. Because Sarah
reliably chose the photograph showing the appropriate course of ac-
tion, Premack and Woodruff concluded that she could infer the actor’s
state of mind.
Premack and Woodruff’s article on TOM attracted much attention
from psychologists and philosophers (Carruthers & Smith, 1996;
Heyes, 1993; Hobson, 1991). Some commentators questioned the va-
lidity of Premack’s evidence that Sarah had a TOM on the grounds
that it took many trials to learn to perform a TOM task (Povinelli,
1998). That would provide ample time for her to respond correctly by
learning particular associations (e.g., door-key). It was also suggested
that Sarah could have based her answers on the solution to problems
she mastered in earlier experiments.
Objections to Premack and Woodruff’s conclusions inspired a vari-
ety of new paradigms, many of which have been used in experiments
on human infants, children, and nonhuman primates (Moore & Dun-
ham, 1995; Tomasello & Call, 1997). But, methodological problems
aside, Premack and Woodruff’s article exerted a revolutionary influ-
ence on cognitive research. Of greatest significance is the foundation
TOM provided for comparing the intentional states of chimpanzees
(our closest living relative) with those of human infants. In particular,
the concept of a TOM calls attention to two preverbal stages of meta-
cognitive cognitive development: the ability to perceive another indi-
vidual’s intentional state and the ability to communicate about that
state.
Although psychologists disagree as to what constitutes a proper
test of TOM, and how the results of such tests should be interpreted,
their differences center on evidence for TOM in children aged 2 to 5
years, who are well on their way toward mastering language. Those
and related issues concerning TOM have been discussed in a number
of excellent reviews (Baron-Cohen, Tager-Flusberg, & Cohen, 2000;
Byrne, 2000; Dunbar, 1998; Gopnik, 2002). In the remainder of this
chapter, I focus exclusively on some experiments on the perception
of intentional states in human infants younger than 2 years and by
nonhuman primates. I then use that evidence to support my claim
that the perception of intentional states plays a crucial role in the tran-
sition from nonintentional to intentional communication and, ulti-
mately, language.
JOINT ATTENTION
At about the same time that Premack introduced the concept of TOM,
Bruner independently introduced the concept of joint attention as an
example of nonverbal intentional communication between an infant
and her caregiver (Bruner, 1975, 1983). Bruner regarded joint attention
as an essential component of what he referred to as the language ac-
quisition support system (LASS), a cognitive system that had to be in
place before the LAD proposed by Chomsky could start operating.

Joint attention is an offshoot of two unremarkable and common be-
haviors: attending to a particular object and learning an instrumental
act to recruit another individual’s help in obtaining that object. An
infant, for example, might cry repeatedly and look at a toy she desired
until the caretaker moved it within reach. This type of behavior,
which has been referred to as “protoimperative,” also occurs in many
nonhuman species (Bates, 1976).
Between the ages of 6 and 9 months, a new type of behavior
emerges, but only in human infants. The infant again tries to get a
caretaker to attend to some object, often by crying, looking back and
forth between the object and the caretaker, or simply by pointing at
it. The infant’s only discernible goal is to announce that she has per-
ceived an object and that she wants her caretaker to perceive it. When
she succeeds, she stops crying and smiles, to acknowledge the caretak-
er’s attention. This type of behavior, which has been referred to as
“protodeclarative,” has not been observed in nonhuman primates.2
The transition from protoimperatives to protodeclaratives is a re-
markable cognitive leap, one that appears to be uniquely human. Pro-
todeclaratives are guided solely by an infant’s wish to share his
knowledge of the world with another person. As the infant gets older,
his protodeclaratives become shorter and less variable. He might sim-
ply point to an object or hold it up for an adult to see, or offer it to
the adult with nothing more than the expectation that it will be re-
turned. If the relevant object is not in sight, the infant might try to
lead the adult to a place where it can be seen.
By definition, protodeclaratives presuppose the ability to perceive
the caregiver as a social agent, an agent whose perception of the world
may or may not be the same as the infant’s. Once an infant can engage
an adult’s attention with a protodeclarative, it is but a small step to
use a particular word instead of a gesture. As the child gets older, she
learns more and more words that function as protodeclaratives. As
her vocabulary expands, she will eventually need rules for combining
words in a consistently meaningful manner. Only then, Bruner ar-
gued, is the infant’s LAD activated.
Aside from a few anecdotal observations, there is no evidence that
nonhuman primates use protodeclaratives in their natural environ-
ment. There is, however, evidence that chimpanzees that have been
socialized in a human environment will point to an object, on occa-
sions on which they do not appear to be interested in actually gaining
possession of it. This degree of overlap is to be expected. Like all new
behaviors, protodeclaratives did not evolve suddenly. It is therefore
not surprising to see precursors of protodeclaratives in our nearest
living relative. Indeed, similar behaviors have been identified in re-
cent experiments on TOM in laboratory-housed chimpanzees.
Experiments on TOM in Nonhuman Primates

The current literature on TOM in nonhuman primates is too large
(and contentious) to summarize in this chapter. Leading investigators
who had initially interpreted various behaviors as evidence of a TOM
have since adopted a more skeptical view of that evidence (Povinelli,
2000; Premack, 1988). In this section, I describe an experiment that
illustrates the difficulties of investigating TOM in a nonhuman pri-
mate and the elaborate control conditions that are needed to rule out
simpler interpretations of their behavior.
There is little question that a nonhuman primate can follow eye gaze.
Less clear is what it sees. Imagine, for example, two chimpanzees, both
of which hear an unfamiliar sound that evokes an orienting response.
Chimpanzee 1 has a direct line of sight to the source of the sound.
Chimpanzee 2 does not, but Chimpanzee 2 can observe the direction
of Chimpanzee 1’s orienting response. How should we interpret Chim-
panzee 2’s state of mind? Does Chimpanzee 2 try to imagine what
Chimpanzee 1 observed? Or does Chimpanzee 2 simply interpret Chim-
panzee 1’s orienting reflex as a discriminative stimulus for vigilance?
In an ingeniously simple experiment, Povinelli and Eddy (1996) in-
vestigated the relationship between eye gaze and seeing. Chimpan-
zees were first trained to reach through a hole in a Plexiglas wall to
solicit, from their trainer, food that was too far away to reach on their
own. The trainer was seated on the other side of the wall. The subjects
were then given a choice of soliciting food from either of two trainers,
only one of whom could see the food that the chimpanzee was seeking
(Trainer 1). As shown in figure 3.5, Trainer 2 was prevented from
seeing the food by a variety of obstacles, some obvious, others more
subtle. In Figure 3.5A, Trainer 2’s head is covered with a bucket; in
figure 3.5B, Trainer 2 covers her eyes with her hands; in figure 3.5C,
with a blindfold. In figure 3.5D, Trainer 2 simply sits with her back to
the chimpanzee.
Povinelli and Eddy asked what, at first glance, appears to be a triv-
ial question: would a chimpanzee solicit food from Trainer 1, who
could see the food, or would it choose randomly between Trainer 1
and Trainer 2? Intuitively, we would guess that the chimpanzee
would choose Trainer 1 on the grounds that only Trainer 1 could
know what the chimpanzee was requesting and why. Surprisingly,
the opposite was true, with but one exception. In the conditions
shown in figure 3.5A–C, the chimpanzee directed its requests for food
to both trainers at random. The one exception was the condition in
which Trainer 2 had her back to the subject (figure 3.5D).
To give their subjects the benefit of the doubt, Povinelli and Eddy
hypothesized that chimpanzees used either of the following simple
rules to obtain food: respond to the trainer who faces forward; do not
Figure 3.5 Conditions presented to the chimpanzee in the original seeing/not-seeing experiment, where the chimpanzee
decides from which trainer to solicit food. Sight may be blocked by A, having the back turned; B, bucket; C, hands over
eyes; and D, blindfolds (adapted from Povinelli & Eddy, 1996).
respond to a trainer who presents her back. However, further varia-

tions of the procedure for preventing Trainer 2 from seeing the food
revealed the shortcomings of those rules. For example, when one of
the trainers sat with her back to the subject and, at the same time,
looked over her shoulder (figure 3.6A), subjects reverted to the prac-
tice of randomly asking either trainer for food. Similar results were
obtained when Trainer 2 simply placed an opaque screen in front of
her face (figure 3.6B) or looked off to the side (figure 3.6C).
Being the intelligent creatures they are, these chimpanzees eventu-
ally learned to respond correctly on all of the tests. Again, Povinelli
and Eddy sought to rationalize their subjects’ failure to perceive im-
mediately what Trainer 1 and Trainer 2 could see by assuming that
they simply needed a few examples to learn the concept “ask the
trainer who can see the food.” That explanation also failed as the
chimpanzees reverted to random responding on each new occasion
on which they encountered a new condition that prevented Trainer 2
from seeing the food, for example, Trainer 1 facing the chimpanzee
with her eyes closed and Trainer 2 holding a mask in front of her face
with her eyes open (figure 3.7A), or when both trainers held masks in
front of their faces: Trainer 1 with her eyes open; Trainer 2 with her
eyes shut (figure 3.7B). Additional evidence that the chimpanzees had
not really learned the concept “solicit food only from the trainer who
can see the food” came from trials on which a previously trained con-
dition was repeated, for example, when Trainer 2 had a bucket over
her head. On those trials, the chimpanzees reverted to a trial-and-
error strategy of determining which trainer could provide food. Over-
all, the results of this experiment suggest that chimpanzees are unable
to infer what the trainer could see and, instead, based their requests
on external cues that they had to learn one by one.
Interestingly, Povinelli and Eddy obtained diametrically opposite
results when they tested children with the same paradigm. Even the
youngest subjects, of a group of 21⁄2- to 3-year-olds, responded cor-
rectly on most (and in, some instances, all) of the conditions, from the
very first trial under each condition. The contrast between the results
obtained from chimpanzees and children illustrate how difficult it is
to imagine how a child and a chimpanzee could look at the same
stimulus, have the same incentive to make the correct response, and
yet behave so differently. The child was able to respond to all varia-
tions of the experimental conditions on the basis of his sense of what
each trainer could perceive. The chimpanzees were able to succeed
only by identifying specific features of each situation.
“Natural” Environments for Inferring Intentionality?

In the Povinelli and Eddy experiment, the chimpanzees may have
failed to infer what the trainer perceived because the conditions on
Figure 3.6 Conditions used to confirm results of experiment
shown in figure 3.5. A. Looking-over-the-shoulder condition. B.
Screen-versus-no-screen condition. C. Attending-versus-distracted
condition (adapted from Povinelli & Eddy, 1996).
Figure 3.7 Conditions used to distinguish the relative importance of the eyes versus the
face in the appreciation of “seeing.” A. Eyes closed, face revealed versus eyes open, face
masked. B. Face masked, eyes closed versus eyes open (adapted from Povinelli & Eddy,
1996).
which they were tested were unnatural. When foraging for food,
chimpanzees do not normally encounter other members of their group
who are blindfolded. Nor do they have to depend on a trainer to
pass them food from inaccessible sources. To avoid those and similar
problems, Hare and colleagues (Hare, Call, Agnetta, & Tomasello,
2000; Hare, Call, & Tomasello, 2001) designed an experiment that,
prima facie, seemed to simulate natural feeding situations.
When foraging for food, a dominant chimpanzee has the peroga-
tive to begin eating before a subordinate does. Hare et al. reasoned
that a subordinate chimpanzee would track the extent to which a
dominant chimpanzee could perceive items of food that were attrac-
tive to both chimpanzees. If the food were in plain view of both chim-
panzees, the subordinate chimpanzee would defer to the dominant
chimpanzee. But if the subordinate chimpanzee sensed that the domi-
nant chimpanzee was unable to see the food, it might go after the
food before the dominant chimpanzee could react.
In Hare et al.’s (2001) experiment, a pair of chimpanzees were
housed in separate rooms; the dominant in one room, the subordinate
in the other. The experimenters placed tidbits of food in a third room
that was situated between the rooms in which the chimpanzees were
housed and observed the interactions between the dominant and sub-
ordinate chimpanzees under three conditions: (1) the dominant chim-
panzee could (or could not) see where the food was hidden in the
third room, (2) the subordinate chimpanzee was (or was not) given
an opportunity to track what the dominant chimpanzee could see of
the third room, and (3) the dominant chimpanzee did (or did not)
have access to the third room. Without any prior training on these
conditions, the subordinate chimpanzee responded appropriately un-
der all three conditions from the very first trial. That is, the subordi-
nate chimpanzee only went after food when it was clear that the dom-
inant chimpanzee could not see where it was hidden or when the
subordinate chimpanzee could perceive that the path of the dominant
chimpanzee was blocked. As in other experiments on primate cogni-
tion, the results of this experiment suggest that the paradigm used to
investigate a particular ability is crucial to the outcome.
BECOMING HUMAN: WHY TWO MINDS ARE BETTER THAN ONE

Since Descartes, language has been the main basis for distinguishing
between human and nonhuman animals. Language may be the most
obvious difference but, as I have tried to show in this chapter, the
cognitive leap from chimpanzees, our nearest living ancestor, to Homo
sapiens required at least one other major change. Before language
evolved, our ancestors had to develop nonverbal skills for reading
another individual’s mind. By taking those skills into account, the
problem of the evolution of language is reduced to a more manage-
able scale. In particular, it draws attention away from formal models
of grammar (an undeniable universal of language) to metacognitive
precursors of language that can be investigated in apes and human
infants. Even metacognition is too complex a metacognitive skill to
have emerged full-blown in one of our ancestors. Metacognition as
we know it is undoubtedly a blend of many precursors. Whether we
can identify all of those precursors and the order in which they
evolved is a question that cannot at present be answered. We can,
however, try to examine nonverbal examples of the metacognitive
skills of infants and apes. The fruitfulness of that approach has been
confirmed in experiments on joint eye gaze and joint attention in chil-
dren and apes, and on the ability of a subordinate chimpanzee to
sense whether the coast is clear to grab some food behind the domi-
nant chimpanzee’s back.
The results of those experiments allow us to frame the broad ques-
tion of when and how language evolved as a narrower and more man-
ageable question of when and how nonverbal sharing of information
evolved. The narrower question suggests an era in human evolution
during which the benefits of having a nonverbal means of sharing

information within a social group could establish an advantage over
groups lacking that ability. Adding the spark of language to those
exchanges would lead to the explosion of cognitive skills that we take
for granted in human adults.
NOTES
1. A study by Nishimura, Mikami, Suzuki, and Matsuzawa (2003) showed that

the larynx of an infant chimpanzee and a human infant were morphologically similar.
Thus, the inability of a chimpanzee to produce humanlike sounds cannot be attrib-
uted to anatomical differences of the articulatory apparatus.
2. Protoimperatives and protodeclaratives are conceptually similar to the catego-
ries of mands and tacts, respectively, described earlier.
REFERENCES
Baron-Cohen, S., Tager-Flusberg, H., & Cohen, D. H. (Eds.). (2000). Under-

standing other minds: Perspectives from developmental cognitive neuroscience.
Bates, E. (1976). Language and context: The acquisition of pragmatics. New York:
Academic Press.
Bloom, L. M., Rocissano, L., & Hood, L. (1976). Adult-child discourse: De-
velopmental interaction between information processing and linguistic
knowledge. Cognitive Psychology, 8, 521–552.
Braine, M. D. S. (1976). Children’s first word combinations. Monographs of the
Society for Research in Child Development, 41, 1–96.
Bretherton, I., McNew, S., Snyder, L., & Bates, E. (1983). Individual differences
at 20 months: Analytic and holistic strategies in language acquistition. Jour-
nal of Child Language, 10, 293–320.
Brown, R. (1970). Psycholinguistics: Selected papers by Roger Brown. New York:
Free Press.
Brown, R. (1973). A first language: The early stage. Cambridge, MA: Harvard
University Press.
Bruner, J. S. (1975). From communication to language: A psychological per-
spective. Cognition, 3, 255–287.
Bruner, J. S. (1983). Child’s talk: Learning to use language. New York: Norton.
Byrne, R. W. (2000). Evolution of primate cognition. Cognitive Science, 24, 543–
570.
Carpenter, A. F., Georgopolous, A. P., & Pellizzer, G. (1999). Motor control
encoding of serial position in a context-recall task. Science, 283, 1752–1757.
Carruthers, P., & Smith, P. K. (Eds.). (1996). Theories of theory of mind. Cam-
bridge, UK: Cambridge University Press.
Chomsky, N. (1957). Syntactic structures. The Hague: Mouton Publishers.
Chomsky, N. (1965). Aspects of the theory of syntax. Cambridge, MA: MIT Press.
Chomsky, N. (1986). Knowledge of language. Wesport, CT: Praeger.
Colombo, M., Eickhoff, A. E., & Gross, C. G. (1993). The effects of inferior
temporal and dorsolateral frontal lesions on serial-order behavior and vi-
sual imagery in monkeys. Cognitive Brain Research, 1, 211–217.
Donald, M. (1991). Origins of the modern mind. Cambridge, MA: Harvard Uni-
versity Press.
Dunbar, R. (1998). Theory of mind and the evolution of language. In J. R.
Hurford, M. Studdert-Kennedy, & C. Knight (Eds.), Approaches to the evolu-
tion of language (pp. 92–110). New York: Cambridge University Press.
Gardner, B. T., & Gardner, R. A. (1969). Teaching sign language to a chimpan-
zee. Science, 162, 664–672.
Gardner, B. T., & Gardner, R. A. (1973). Teaching sign language to the chimpan-
zee: Washoe. University Park, PA: The Psychological Cinema Register.
Gardner, B. T., & Gardner, R. A. (1974). Comparing the early utterances of
child and chimpanzee. In A. Pick (Ed.), Minnesota symposia on child psychol-
ogy (pp. 3–23). Minneapolis: University of Minnesota Press.
Gardner, B. T., & Gardner, R. A. (1975a). Early signs of language in child and
chimpanzee. Science, 187, 752–753.
Gardner, B. T., & Gardner, R. A. (1975b). Evidence for sentence constituents
in the early utterances of child chimpanzees. Journal of Experimental Psy-
chology, 104, 244–267.
Gopnik, A. (2002). Theory of mind. Retrieved September 12, 2002, from cognet.-
mit.edu/library/erefs/mitecs/gopnik.html.
Gould, S. J. (1977). Ontogeny and phylogeny. Cambridge, MA: Harvard Univer-
sity Press.
Gould, S. J., & Vrba, E. S. (1982). Exaptation: A missing term in the science of
form. Paleobiology, 8, 4–15.
Hamilton, J. M. E., & Sanford, A. J. (1978). The symbolic distance effect for
alphabetic order judgements: A subjective report and reaction time analy-
sis. Quarterly Journal of Experimental Psychology, 30, 33–43.
Hare, B., Call, J., Agnetta, B., & Tomasello, M. (2000). Chimpanzees know
what conspecifics do and do not see. Animal Behaviour, 59, 771–785.
Hare, B., Call, J., & Tomasello, M. (2001). Do chimpanzees know what conspe-
cifics know? Animal Behavior, 61, 139–151.
Harlow, H. F. (1949). The formation of learning sets. Psychological Review, 56,
51–65.
Hauser, M., Chomsky, N., & Fitch, W. T. (2002). The faculty of language:
What is it, who has it, and how did it evolve? Science, 298, 1569–1579.
Hayes, C. (1951). The ape in our house. New York: Harper.
Hayes, C., & Hayes, K. (1951). The intellectual development of a home-raised
chimpanzee. Proceedings of the American Philosophical Society, 95, 105–109.
Heyes, C. (1993). Anecdotes, training, trapping, and triangulating: Do animals
attribute mental states? Animal Behavior, 46, 177–188.
Hobson, R. P. (1991). Against the theory of “theory of mind.” British Journal
of Developmental Psychology, 9, 33–51.
Kellogg, L. A., & Kellogg, W. N. (1967). The ape and the child: A study of environ-
mental influence and its behavior. New York: McGraw-Hill. (Original work
published 1933)
Ladygina-Kohts, N. N. (2002). Infant chimpanzee and the human child: A classic
1935 comparative study of ape emotions and intelligence. Oxford, UK: Oxford
University Press.
Lieberman, P. (1968). Primate vocalizations and human linguistic ability. Jour-
nal of the Acoustical Society of America, 44, 1574–1584.
Lieberman, P. (1975). The evolution of speech and language. In J. F. Kavanagh
& J. E. Cutting (Eds.), The role of speech in language (pp. 83–106). Cambridge,
MA: MIT Press.
Matsuzawa, T. (1985). Use of numbers by a chimpanzee. Nature, 315, 57–59.
Miles, H. L. (1983). Apes and language: The search for communicative compe-
tence. In J. de Luce & H. T. Wilder (Eds.), Language in primates: Perspectives
and implications (pp. 43–61). New York: Springer-Verlag.
Moore, C., & Dunham, P. J. (Eds.). (1995). Joint attention: Its origins and role
and development. Hillsdale, NJ: Lawrence Erlbaum.
Mowrer, O. M. (1954). The psychologist looks at language. American Psycholo-
gist, 9, 660–694.Nelson, K. (1981). Individual differences in language devel-
opment: Implications for development and language. Development of Psy-
chology, 17, 170–187.
Nishimura, T., Mikami, A., Suzuki, J., & Matsuzawa, T. (2003). Descent of the
larynx in chimpanzee infants. Proceedings of the National Academy of Sciences
U.S.A., 100, 6930–6933.
Orlov, T., Yakovlev, B., Hochstein, S., & Zohary, E. (2000). Macaque monkeys
categorize images by their ordinal number. Nature, 404, 77–80.
Patterson, F., & Linden, E. (1981). The education of Koko. New York: Holt, Rine-
hart, and Winston.
Patterson, F. G. (1978). The gestures of a gorilla: Language acquisition by an-
other pongid. Brain and Language, 12, 72–97.
Pinker, S. (1994). The language instinct. New York: William Morrow.
Potts, G. R. (1974). Storing and retrieving information about ordered relation-
ships. Journal of Experimental Psychology, 103, 431–439.
Povinelli, D. J. (1998). Can animals empathize? Maybe not. Scientific American,
9(4), 72–75.
Povinelli, D. J. (2000). Folk physics for apes: The chimpanzee’s theory of how the
world works. New York: Oxford University Press.
Povinelli, D. J., & Eddy, T. J. (1996). What young chimpanzees know about
seeing. Monographs of the Society for Research in Child Development, 61(3).
Povinelli, D. J., & Prince, C. G. (1999). When self met other. In M. Ferrari &
R. J. Sternberg (Eds.), Self-awareness: Its nature and development (pp. 37–107).
New York: Guilford.
Povinelli, D. J., Rulf, A. B., & Bierschwale, D. T. (1994). Absence of knowledge
attribution and self-recognition in young chimpanzees (Pan troglodytes).
Journal of Comparative Psychology, 108, 74–80.
Povinelli, D. M. (2001). On the possibilities of detecting intentions prior to
understanding them. In B. F. Malle, L. J. Moses, & D. A. Baldwin (Eds.),
Intentions and intentionality: Foundations of social cognition (pp. 225–248).
Premack, D. (1970). A functional analysis of language. Journal of the Experimen-
tal Analysis of Behavior, 4, 107–125.
Premack, D. (1971). Language in a chimpanzee? Science, 172, 808–822.
Premack, D. (1976). Intelligence in ape and man. Hillsdale, NJ: Lawrence Erl-
baum.
Premack, D. (1988). “Does the chimpanzee have a theory of mind?” revisited.
In R. Byrne & A. Whiten (Eds.), Machiavellian intelligence (pp. 160–179).
Oxford, UK: Oxford University Press.
mind? Behavioral and Brain Sciences, 1, 515–526.
Rumbaugh, D. M. (1977). Language learning by a chimpanzee: The Lana project.
New York: Academic Press.
Rumbaugh, D. M., Gill, T. V., & von Glasersfeld, E. C. (1973). Reading and
sentence completion by a chimpanzee. Science, 182, 731–733.
Savage-Rumbaugh, E. S. (1994). Kanzi: The ape at the brink of the human mind.
New York: Wiley.
Savage-Rumbaugh, S., Shanker, S. G., & Taylor, T. J. (1998). Apes, language,
and the human mind. New York: Oxford University Press.
Scaife, M., & Bruner, J. (1975). The capacity for joint visual attention in the
infant. Nature, 253, 265–266.
Schroeder, B. (Producer) (1977). Koko: The talking gorilla. Home Vision.
Skinner, B. F. (1953). Science and human behavior. New York: Macmillan.
Skinner, B. F. (1957). Verbal behavior. New York: Appleton-Century-Crofts.
Straub, R. O., & Terrace, H. S. (1981). Generalization of serial learning in the
pigeon. Animal Learning and Behavior, 9, 454–468.
Swartz, K. B., Chen, S., & Terrace, H. S. (1991). Serial learning by rhesus mon-
keys. I: Acquisition and retention of multiple four-item lists. Journal of Ex-
perimental Psychology: Animal Behavior Processes, 17, 396–410.
Terrace, H. S. (1979a). Is problem solving language? A review of Premack’s
Intelligence in Apes and Man. Journal of the Experimental Analysis of Behavior,
31, 161–175.
Terrace, H. S. (1979b). Nim. New York: Knopf.
Terrace, H. S. (1984). Simultaneous chaining: The problem it poses for tradi-
tional chaining theory. In M. L. Commons, R. J. Herrnstein, & A. R.
Wagner (Eds.), Quantitative analyses of behavior: Discrimination processes (pp.
115–138). Cambridge, MA: Ballinger.
Terrace, H. S. (1985a). In the beginning was the name. American Psychologist,
40, 1011–1028.
Terrace, H. S. (1985b). On the nature of animal thinking. Neuroscience and Bio-
behavioral Reviews, 9, 643–652.
Terrace, H. S. (1993). The phylogeny and ontogeny of serial momory: List
learning by pigeons and monkeys. Psychological Science, 4, 162–169.
Terrace, H. S. (2002). Serial expertise and the evolution of language. In J. H. A.
Wray & F. J. Newmeyer (Eds.), The transition to language (pp. 64–90). New
York: Oxford University Press.
Terrace, H. S., Petitto, L. A., Sanders, R. J., & Bever, T. G. (1979). Can an ape
create a sentence? Science, 206, 891–902.
Terrace, H. S., Son, L., & Brannon, E. (2003). The development of serial exper-
tise by rhesus macaques. Psychological Science, 14, 66–73
Thompson, C. R., & Church, R. M. (1980). An explanation of the language of
a chimpanzee. Science, 208, 313–314.
Tomasello, M. (1999). Having intentions, understanding intentions, and un-

derstanding communicative intentions. In J. W. Astington, D. R. Olson, &
P. D. Zelazo (Eds.), Developing theories of intention: Social understanding and
self-control (pp. 63–76). Mahwah, NJ: Lawrence Erlbaum.
sity Press.
Tomasello, M., Hare, B., & Fogleman, T. (2001). The ontogeny of gaze follow-
ing in chimpanzees, Pan troglodytes, and rhesus macaques, Macaca mulatta.
Animal Behavior, 61, 335–343.
Williams, G. C. (1966). Adaptation and natural selection. Princeton, NJ: Princeton
University Press.
Winitz, H. (1969). Articulatory acquisition and behavior. New York: Appleton-
Century-Crofts.
4
Emerging Levels of Consciousness

in Early Human Development
Katherine Nelson
In this chapter, I am concerned with tracing the emergence during

ontogenesis of a specifically human kind of consciousness from a
more primitive awareness. This view of emergent levels of conscious-
ness builds on recent work in the evolution of human cognition, par-
ticularly that of Merlin Donald (1991, 2001). What I propose here is,
first, that consciousness is not unitary but that we can discern levels
of consciousness from simple awareness of feelings and perceptions
to complex extended states of consciousness (Damasio, 1999). A dis-
tinction can therefore be made between consciousness in animals and
humans, and particularly between evidence of very simple kinds of
metacognition and full self-knowing consciousness. Tulving (1983)
made the distinction between semantic (noetic) and episodic (autonoe-
tic) memory, which brought the issue of a specifically human self-
knowing memory into focus in cognitive psychology, and raised im-
portant issues for research on memory development as well. In this
light, I propose that full human consciousness can be understood only
in terms of both its evolutionary and developmental history. Thus,
before turning to the proposed levels of development and evidence
for them in the developmental literature, I provide some background
on the developmental and evolutionary perspectives.
DEVELOPMENTAL PERSPECTIVE
A developmental perspective necessarily concerns dynamic change in
processes as they occur over time. For this reason, many develop-
mental theorists have adopted the perspective of dynamic systems
analysis, conceptualizing human development as a self-organizing
system involving brain, body, and mind functioning together in inter-
action with both the internal and external environment, and arising
116
Emerging Levels of Consciousness 117
from this process successive periods of organization. In a larger per-

spective, the individual (brain, body, mind) and the social world form
an interdependent transactional system that is in a constant process
of organization. We can consider the “elements” of the system sepa-
rately at any point in time, but only with the caveat that they are not
in fact separable, that the context, especially the social context, is al-
ways part of the process to be traced.
From this developmental perspective, it is misleading to hunt for
origins or for a missing link, which implies a search for one specific
point where something is added such that the system takes on a new
form and subsequently takes off in a trajectory that results in human
adult competence. Admittedly, many developmental psychologists
take exactly this stance, so it is not surprising that the general field of
cognitive science does as well. The problem is that the search for the
missing link, in development or in evolution, is likely to fail because
the particular capacity we are looking for will not be found in a recog-
nizable form, whether we call it metacognition or reflective conscious-
ness or something else. Rather, because complex systems are consti-
tuted of many different subcomponents that may recombine in different
ways, self-reflective knowing may initially appear as a small kernel of
something else that eventually finds itself coopted as a function in a
system that takes on new powers (Fischer & Bidell, 1991). Under these
conditions, we need to study the development of the system on its
own terms, recognizing transformations and the emergence of new
capacities from the combinations and recombinations of old forms and
functions. Just such a process is now recognized as an essential part
of phylogenetic evolution (DePew & Weber, 1999). Thus, instead of
proposing a theory of origins, my proposal here is in terms of devel-
opment from a less complex to a more complex organization that
functions effectively for the individual in a social environment at a
particular moment in time, and that in turn also leads to new levels
of organized function.
As part of a developing functional system, it is obvious that what
is relevant to the experiencing individual changes over time with changes
in capacities, needs, and external supports. In consequence, the child’s
capacity for experiencing the complexity of the world and making
sense of it changes, not only with brain development but also with
body development and the potentials these offer for experiencing in
new ways. These considerations make it imperative to take an experi-
ential perspective on the course of development. Experience is con-
scious, the source of memory, knowledge, and belief, both present
and past experience, and serves to guide and support action in both
familiar and unfamiliar situations.
Like consciousness, everyone’s experience is uniquely personal and
unavailable to objective observation. As adults, we use speech and
writing to convey aspects of personal experience and mediated knowl-

edge to one another, as well as to construct aspects of knowledge that
exist only in the abstractions of language (e.g., in academic discus-
sions). But for the infant and young child, experience is not only per-
sonal but private in an extreme sense: the child cannot share her
thoughts with others. This condition is, I believe, central to under-
standing the development of consciousness. It also requires us to
imagine what it is like to be an infant or child in the same sense that
we need to imagine what it is like to be a bat (Nagel, 1974) or like a
chimpanzee. I will take up this challenge in the section on the devel-
opment of consciousness. I first consider some general characteristics
of human development that differ in important ways from other ani-
mals studied, including nonhuman primates.
The first point is that for each human infant the world is unique
and unpredictable in all but the most basic ways, because each infant
is born into a unique cultural arrangement. This is not a claim of radi-
cal cultural difference, but rather of unpredictability from the point of
view of any knowledge that evolution might have provided through
genetic or more general prenatal developmental mechanisms. (This
point deserves to be especially emphasized, but limitations of space
prevent dwelling on it here at length.)
The second point to be noted is the rapid change of perceptual and
motor capacities over the course of infancy and early childhood. Un-
like other primates, human infants are born in a fetal state, that is, a
state of development that most other mammals complete prior to
birth. They are uniquely helpless, totally dependent upon caregivers.
Most of the first few months are spent in a prone or supine position,
unless propped or held in a more upright state. Thus the view of the
world is highly restricted; even parts of the self, such as hands or
feet, await further motor development before they can be seen and
examined. So also, the view of others is at the disposal of those others
to make themselves seen in particular perspectives. By the midpoint
of the first year, however, the infant has learned to sit up and usually
to move about by creeping or crawling. Thus, some control over what
can be seen and handled has been gained. When the infant pulls her-
self up to standing position and begins to cruise along the edge of a
piece of furniture, a new viewpoint is gained, and when that is suc-
ceeded by independent walking around toward the end of the first
year, the sequence of motor development and its accompanying
broadening of perspectives and experience of the surroundings is al-
most complete (Campos, Bertenthal, & Kermoian, 1992).
The third point complements these developments: the infant and
young child is always in and dependent upon the social world, de-
fined by the cultural choices that particular social others have made
in terms of dwellings, furnishings, foods, friends, toys, activities, and
myriad other things. Although the infant’s experience from the begin-
ning thus depends totally upon the social arrangements of the cultural
world, the social world and its choices also change with the child’s
own development, both guiding experience and leading the way,
sometimes blocking particular pathways as well. Thus the child’s ex-
perience in the world is almost completely at the disposal of the social
others in the early years, primarily of course parents and other family
members, but in different cultures and social arrangements a variety
of people may serve in important roles as guides and caretakers. In
sum, the child’s experience is at once both utterly private and totally
social, while constantly changing within a stable background.
EMERGENCE OF CONSCIOUSNESS IN EVOLUTION AND DEVELOPMENT

Gould (1977) cautioned against the idea of the recapitulation in ontog-
eny of a phylogenetic sequence in evolution for a number of reasons,
among them its basis in the mistaken assumption that there is in fact
a sequence from lower to higher animals culminating in the human
species, rather than a contingent process of organism-environment ad-
aptation leading to an infinite number of possible solutions.1 But ac-
ceptance of this caution does not rule out the recognition of similari-
ties in developmental and evolutionary problems and processes. The
particular issue in focus here, the origins of self-knowing conscious-
ness, is a case in point. The “origins” may escape our grasp for the
very reasons that Gould warned about, namely that two species from
different phylogenetic branches might exhibit different aspects of cog-
nition that in a third unrelated species might appear to be combined
into a “higher” level of cognitive functioning, yet this level in turn
might be unrelated to our own branch of the phylogenetic bush. Even
looking backward through our primate roots faces us with the di-
lemma that our closest relatives, the chimpanzee, bonobo, and gorilla,
display similar cognitive skills in infancy appearing in different devel-
opmental sequences (Tomasello & Call, 1997). When we look back-
ward from the adult human to the human infant or to earlier species
in the hominid or primate line, we are always looking for the earliest
appearance of some component skill or capacity that the adult person
displays. But the capacity or skill in question may be the product of
the organization of some components that in isolation bear no func-
tional relation to the finished product.
Studies of children’s theory of mind are instructive here. Work on
the development of theory of mind in children actually began in re-
sponse to Premack and Woodruff’s (1978) questioning of whether the
chimpanzee had a theory of mind. (Their answer was yes, but Pre-
mack has since retreated from this conclusion; see Premack, 1988.)
For many years, studies were carried out within a few experimental
paradigms to establish when children might be said to have a theory

of mind, according to the logical requirement that understanding oth-
ers’ minds involves understanding that they may hold beliefs that are
false with respect to states of the world (for lack of access to relevant
information, for example; see Astington, Harris, & Olson, 1988, for an
early compendium of this work).
Some attempts to explain the origins of this capacity searched for
the beginnings of a theory and eventually settled on the idea that
children begin with a built-in theory and then subject the theory to
successive revisions as disconfirming data are encountered (Gopnik,
Capps, & Meltzoff, 2000; Gopnik & Wellman, 1994). Another effort
presumed that the capacity involved a brain module that must have
evolved in humans for this purpose (Leslie, 1987, 1991) and thus was
present at birth and evident in behavior by 1 year but depended on
additional cognitive skills to become fully operative at about 4 years.
Both of these efforts were hampered by the assumption that the capac-
ity they were viewing at 4 years would be visible in the same form at
some earlier point in development. It is notable that both ran into the
same dead end, namely the conclusion that the capacity must be built
into the human brain/mind to begin with. By similar reasoning, virtu-
ally all of the child’s and adult’s cognitive capacities must be innately
given. This is a position that is quite in accord with Chomsky’s (1965)
claims about language and Fodor’s (1975) claims about the language
of thought, but not one congenial to the developmental point of view.
However, this route is not inevitable. Other theorists have looked
for the roots of theory of mind in earlier social relations and skills.
For example, Harris (1992) found evidence in children’s empathy for
others and in their pretense for a simulation account of theory of mind
development. Tomasello (1999a, 1999b) focused on the establishment
of joint attention with a caretaker toward the end of the first year as
the seed from which theory of mind later emerged, after due nurtur-
ing through further interaction, both linguistic and nonlinguistic. Both
of these accounts reflect a developmental perspective. In addition,
they shed light on the difference in theory of mind between humans
(at least by 4 years) and other primates by virtue of their focus on
specifically human social conditions of development, rather than on
the cognitive givens.
Most of the great apes (chimpanzees, gorillas, bonobos) live in so-
cial groups and display social intelligence, so that many theorists now
propose that the necessity of solving social problems (e.g., of competi-
tion for mates) led to the increase in cognitive power displayed by
these creatures in comparison with other animals.2 However, the so-
cial conditions of human infancy differ from those of other primates
in terms of the intimate caretaking and social involvement of parents,
particularly mothers, over an extended period of time—for 3 to 7
years or more (Hrdy, 1999). It is within this intimate caretaking circle

that joint attention (not evident in the same way in other primates)
becomes established and soon thereafter supports the beginnings of
language in first word learning.
This approach to development—examining development in situ—
focuses on tracing what conditions and experiences obtain and how
they enter into new relations and skills and become transformed over
time. The often convoluted pathways that lead toward emerging new
skills are the target of discovery, recognizing that many components
may be recombined to produce a new level of functioning.
Evolution of Consciousness
Some of the difference between the other primates and humans is
evident in the comparison of brain size and particularly in the expan-
sion of tertiary cortex. But such physical changes in themselves do not
explain how and why cognitive differences in functioning occurred
and what the critical changes in hominid evolution were. Donald
(1991) proposed a model of the “hybrid” modern mind that involved
three major transitions beyond a general primate brain/mind such as
the modern great apes exhibit.
The first major transition that Donald identified was from a strictly
episodic mind to mimetic functioning. In his conception, the episodic
mind—the general primate model—is confined to the here and now
in the sense that memory is evoked in relevant circumstances but is
not accessible to voluntary recall. Both past and future are confined
to concerns of the present situation (where “present” may occupy a
reasonably extended episodic period of time). Thus social life flour-
ishes, organized around ongoing needs and actions, but culture as
such does not. Planning (as for a hunt or an attack) is short-lived;
memory is instrumental, not reflective. Although Donald named this
level of cognitive functioning “episodic,” this should not be confused
with Tulving’s use of the term to distinguish among kinds of memory.
It is unlikely that Tulving’s sense of episodic memory could exist at
this level; rather, memory could be either procedural or semantic in
the terms of his system.3
Mimesis emerged, according to Donald (1991), during the period
of Homo erectus about 2 million years ago, evident in the appearance
of deliberately constructed flaked stone tools and the use of fire. More
recent evidence suggests a more complex cultural mix during the pe-
riod before Homo sapiens (Donald, 2001). The cultural advances of this
level seem to signify a new capacity for planning and the practice of
skills, such as those involved in tool making. But for Donald, these
capacities reach much farther, in terms of reflective cognition, and are
based in a new use of precise imitation, of both social models and
self-repetitions. The former allow deliberate teaching; the latter enable
practice and reflection on actions outside of their practical application.

A common example is that of a young boy practicing the throwing of
stones, the only aim of which is to improve that particular skill, based
on an internalized ideal throw. Other primates have not been ob-
served to practice skills in this way; rather, when practice—and the
limited form of imitation or emulation—exists, it is always in the in-
terest of obtaining some goal, such as food or a mate. Mimesis, more-
over, enabled new forms of gestural communication (and perhaps
verbal communication as well) and the creation of communally shared
rituals, dance, and other group practices, supporting a new level of
cultural organization. A summary of these potentials includes the ca-
pacity for reenactment, imitation, skill development, and whole-body
gesture. Each of these, Donald points out, involves the direction of
attention from the world to the self, to consciously monitor one’s own
actions toward the accurate achievement of an internal model, derived
from action in the world, one’s own or another’s. However, important
as this transition was, it was also limited.
Little change in cultural practices or anatomical form, including
brain size, was observed for over 2 million years before the appear-
ance of Homo sapiens about 500,000 years ago. Dramatic changes are
evident from the arrival of humans in modern anatomical form about
125,000 years ago, particularly with the invention of complex spoken
language.4 There followed a rapid expansion of culture, quickly eclips-
ing that of the earlier mimetic period. Donald calls this transition
“mythic” to emphasize the critical role of language in constructing
oral cultural narratives that provide a common history and a common
cultural understanding. That is to say that in Donald’s terms the move
to language was in the service of communication and culture, drawing
people into communities of understanding. The cognitive foundation
for the emergence of language must have been established through
mimesis, but symbolic language itself was a product of cultural com-
munication within “communities of mind,” and in turn it enabled cul-
tures to take off in ways that were not possible without it. “Language
emerges only at the group level and is a cultural product, distributed
across many minds” (Donald, 2001, pp. 11–12). From this perspective,
“cultural mind sharing” is the unique contribution of human life, the
by-products of which are communally shared systems such as mathe-
matics, athletics, music, and literacy. In turn, the nature of the individ-
ual human mind was changed as symbolic capability became part of
its structure and function.
The third transition, to external memory representations, is now
traced by Donald (2001) to its beginnings about 40,000 years ago (e.g.,
in paintings, stone markings, emblems, and the like), although its
most extraordinary effects have emerged in more recent historical
time, after the invention of written language about 4,000 to 5,000 years
ago, and the subsequent spread of literacy through print, as recently

as 500 years ago. Donald’s conception of how this changed the func-
tion of individual memory is important, involving, as he put it, the
transferring of memory from mind to external forms of print, graph-
ics, and so on.
This entire evolutionary/historical sequence challenges “naturalis-
tic” and universalist conceptions of modern adult cognitive function-
ing and also provides an important perspective on the development
of cognition in human childhood. In his most recent book, Donald
(2001) extended his analysis to the consideration of the evolution of
human consciousness. He argued there that human conscious capacity
“provides the biological basis for the generation of culture, including
symbolic thought and language.” He posited a process of “brain-culture
symbiosis” termed “deep enculturation,” such that in human evolu-
tion, “our brains coevolved with culture and are specifically adapted
for living in culture” (p. 11). Perhaps the most provocative claim that
Donald makes, and the one with the most resonance for a develop-
mental perspective on consciousness, is the following: “The ultimate
irony of human existence is that we are supreme individualists, whose
individualism depends almost entirely on culture for its realization. It
came at the price of giving up the isolationism, or cognitive solipsism,
of all other species and entering into a collectivity of mind” (p. 12).
The challenge I am taking up here is in part how to track this exchange
in the ontogenetic process.
DEVELOPMENT OF CONSCIOUSNESS
What emerges from Donald’s work is a view of consciousness that is
active in directing attention and activity, in keeping track over time,
holding multiple ends in view, and in other ways augmenting and
directing cognitive functions (see also Damasio, 1999). But conscious-
ness is not in this view a unitary state; rather, it expands in both evo-
lution and development, where levels of consciousness emerge over
time, but where earlier levels continue to function as subcomponents
or alternative modes to more recently developed levels. The develop-
ment of levels of consciousness has been proposed as well by Lewis
and Ramsay (1999) and by Zelazo (1999), and the scheme described
here shares some features with these others. The present proposal was
worked out in light of Donald’s theory and in relation to the emer-
gence of self-understanding revealed in children’s use of narrative
and the beginnings of autonoetic memory (Nelson, 2001, 2003). Thus
I begin with a brief overview of early memory development, specifi-
cally concerned with the emergence of autobiographical memory.
Research on memory in infants and young children is a surpris-
ingly recent enterprise, with few serious efforts undertaken prior to
the 1970s. Experimental work in this period typically revealed impov-

erished memory through the preschool period (Flavell, Friedrichs, &
Hoyt, 1970). Renewed interest in the memory of children during the
preschool period came from a number of researchers who worked
with children in natural everyday settings in the home or in pre-
schools (Perlmutter, 1980). My research at that time focused on chil-
dren’s memory for everyday events, in particular their scripts for fa-
miliar routines (Nelson & Gruendel, 1981; Nelson & Ross, 1980). We
were intrigued to find that when 3-year-old children were asked to
tell what had happened at dinner the night before, or last week at a
birthday party, they responded with a general script for the event
rather than with a specific episodic account. For example, when asked
to tell what happened at dinner, they replied as though asked what
happens at dinner: “You eat and then you go to bed.” These children
were representing a whole event, and they typically reported the cor-
rect sequence of actions, but they represented it as a semantic schema,
generalized knowledge without particular details or personal reac-
tions. Of special interest was the finding that these young children
used the objective “you” form rather than the subjective “I” and the
generalized present tense of the verbs rather than the specific past,
suggesting that this knowledge form was neither personal nor tempo-
rally located in the past.
On the basis of these findings, we initially hypothesized that young
children had general semantic but not specific episodic memory. This
conclusion seemed to provide a satisfying explanation for the phe-
nomenon of infantile amnesia, the inability of adults to recall episodes
from their childhood before the age of about 3 years (Nelson, 1988).
Subsequent research has revealed complexities that have obscured the
clarity of this original hypothesis. We now know that children do
have some memories of specific episodes from the age of 2 or 3 years,
including memory fragments that seem to involve the self in a reflec-
tive sense. Such fragments have appeared in the context of “memory
talk” with parents, which has been studied extensively by researchers
over the past 20 years (Nelson & Fivush, 2000, 2004), where the frag-
ments may be incorporated by the parent into whole narratives about
the remembered event, which the child is not yet able to organize. It
is interesting that young children’s fragments appear similar to many
reports from this period of childhood by adults who remember a
scene or a bit of action but cannot recall the entire episode (Pillemer
& White, 1989).
The implication of this line of memory research is that children
learn to talk about memories, both as present events are happening
and in reflecting on past events and on future happenings, beginning
at around 2 years of age (in our society). Evidence for this is that the
amount and kind of memory talk at 4 years is correlated with the
amount and specificity of children’s episodic memories 2 years later

(Reese, Haden, & Fivush, 1993). It is important to note that significant
cross-cultural differences have been found in each of these factors,
with Asian (Chinese, Japanese, and Korean) adults having fewer and
later early childhood memories than American adults, Asian mothers
engaging in less memory talk than American mothers, and Asian chil-
dren reporting fewer and less detailed episodic memories (Han, Leicht-
man, & Wang, 1998; Mullen, 1994, Mullen & Yi, 1995). Cultural differ-
ences of this kind support the conclusion that conversational experience
contributes to episodic memory development, with different cultural
practices resulting in different developmental trajectories.
This brief summary of the work on memory in early childhood
provides only a glimpse of what has become an extremely active re-
search domain, with some of the most provocative questions emerg-
ing around the issues of explaining the phenomenon of infantile am-
nesia and the appearance of episodic memory during childhood. Each
of these questions is also clearly related to the topics covered in this
volume in terms of self-knowing consciousness, metacognition, and
autonoetic memory processes. The social-cultural approach to these
issues that Robyn Fivush and I have elaborated (Nelson & Fivush,
2000, 2004) supports the proposal that consciousness of self in time
emerges through autobiographical (autonoetic) memory and is fos-
tered through the same kinds of social conversational experiences that
influence the development of the 4-year-old’s theory of mind. Emer-
gence of levels of consciousness based on this approach is detailed in
what follows.
I propose that six levels of consciousness emerge during the first
years of life, with implications for self, social, and cognitive function-
ing. These levels are schematized in table 4.1 and described in more
detail in the following sections. To a large extent, the levels of con-
sciousness outlined here reflect those that Donald laid out as stages
in human evolution; however, the process envisioned in development
is somewhat different. The process of emergence in development de-
pends on the interaction of the self and the social world. As the child
functions at one level, social experiences made possible at that level
enable movement to the next. Thus, what is envisioned here with re-
gard to emerging consciousness is consistent with social theories of
the development of self (e.g., Mead, 1934).
Conciousness Levels in Infancy

As pointed out earlier, to track the experience of consciousness in in-
fancy requires imagining the perspective of the infant beginning in
the first encounter with the world. The neonate seeing the world for
the first time is undergoing a truly unique experience. Although Wil-
liam James (1890/1950) thought the experience must be of a “booming
Table 4.1 Levels of Consciousness
Level of Self-Other Self-World

Consciousness Onset Age Memory Contrast Contrast
Physical Postnatal Perceptual Physical Physical

boundary boundary
Social 6–12 mo. People, Social shar- Routines,
schemas ing, com- object kinds,
munication words
Cognitive 18–24 mo. Events, I-you per- Conscious
routines, spective self/objec-
categories, tive self
words
Representational/ 2–4 years Episodes My self-expe- Social and
Reflective shared rience/ physical
with others other self- causality
experience
Narrative 3–6 years Episodic, Stories of Past-future,
autobio- me/stories real and
graphical of others, imaginary
memory autonoesis worlds
Cultural 5–10 years External Cultural Cultural
represen- roles knowl-
tations, edge, in-
textual stitutions
buzzing confusion,” psychologists now know that it is quite orga-

nized, although slowly unfolding as visual acuity increases and shapes
come into sharper focus. Recent research on infant perception shows
that infants either come into the world with highly organized expecta-
tions about the nature of three-dimensional objects, human faces, and
voices, or very quickly gain information about the structure of these
basic features of the world. We should not have been surprised by
these findings, Piaget to the contrary notwithstanding. After all, hu-
man infants are descended from a long line of mammals, including
the higher primates, most of which very soon after birth become inde-
pendently mobile and thus would be extremely vulnerable to fatal
accidents if they were not equipped by evolution with the ability to
negotiate their way in the world of objects and animate creatures, and
to recognize conspecifics by sight and sound.
What is of most interest to human infants throughout the first year
are the sights, sounds, touch, and movement of people. Over this pe-
riod, the infant’s consciousness grows rapidly from fleeting levels of
awareness interspersed with long periods of sleep to sustained peri-
ods of engagement with the physical and social world. The first level
of experiential awareness is that of a physical self that distinguishes
the boundary between self and other. Over the first few months, in-
fants attend to many different aspects of the world that are changing
around them; shifts in attention, in fact, are the basis for much of
the experimental research on infant cognition, using the paradigms of
habituation and preference for novelty. We can assume that much
experiential organization is taking place during this period, although
the phenomenology of that experience is difficult to fathom. The men-
tal content of this period seems to concern primarily emotional attach-
ment to the important social figures in the infant’s world.
Beginning midway in the first year, a new level of consciousness
emerges, as almost all students of infant development agree. The
physical boundary between self and other, and self and object, is now
extended to a three-way relation between self, other, and object,
where self and other share attention to the same object. Thus a new
consciousness of self and other comes into play, where the tracking of
the other’s attention (that is, attention to attention) is of interest. It
may be that prior to this point, the infant implicitly assumed that she
was the sole object of the other’s attention; now a new realization that
the other may attend to some other point of interest (another person,
another object) may bring on a crisis of confidence in the other, mani-
fested through such new behaviors as fear of strangers, looking to the
other for clues to how to react in new situations, and generally follow-
ing closely the behavioral clues of the other.
But this crisis also provides the incentive for new organizations of
experience. Tracking attention merges into tracking intentions, of both
self and other. Sharing attention to objects leads to drawing the other’s
attention to objects through pointing and other means of gestural
communication. Expressing one’s own wants and needs, for example,
by raising arms to indicate a desire to be picked up, becomes part of
the repertoire. A newly emerging ability to use imitation as a learning
tool, as well as an enjoyable activity, comes into play. Tracking the
sequence of routines, games, and caretaking experiences such as a
bath or a feeding provides the basis for predicting and reacting to
change, as well as providing the basis for taking reciprocal roles in
the activity. This capacity also indicates an expanding working mem-
ory that is capable of extending over a sequence of actions and interac-
tions, keeping track of where the point of action is at present and
what is to come next.
All of these well-documented developments make it clear that a
revolution in consciousness takes place in the last six months of the
first year, culminating in the possibility of responding to and repro-
ducing the sounds of language in ways consistent with the acquisition
of actual words, an achievement transitional to the next level. The
onset of the following period is initiated as the contrast between the

social possibilities of the nonsymbolic, nonlinguistic exchanges and
the linguistic world that is being carried on around the infant (but
without her participation) becomes apparent, motivating the develop-
ment of “cognitive consciousness.”
Cognitive Consciousness
As suggested in table 4.1, cognitive consciousness is the first level at
which the child begins to take an objective view of the self in relation
to others and the world. This level has been recognized as a significant
advance by many theorists (e.g., Lewis & Ramsey, 1999; Neisser, 1997)
and has been seen as the beginning of the autobiographical self by the
neuropsychologist Damasio (1999) and the developmental psycholo-
gists Howe and Courage (1993). The onset of this level of conscious-
ness is gradual (and variable among infants), becoming apparent over
the course of the second year. It is coincident with the acquisition of
language, specifically the learning of first words and the construction
of simple sentences. In conjunction with that advance in communica-
tion, one can observe related moves on the cognitive level, recognized
by Piaget (1962) as the emergence of the semiotic function.
By late in the second year, children provide good evidence of rec-
ognizing the self and differentiating the perspective of self and other.
The mirror test devised originally for gorillas and chimpanzees has
been used to test self-recognition in young children as well. In this
test, a mark is surreptitiously placed on the child’s forehead while he
or she is otherwise engaged, and then the child is allowed to view
himself or herself in a mirror. The question is whether children recog-
nize that the mirror image reflects themselves, or whether it is treated
simply as that of a child. The critical observation is whether they point
to the mark on their own forehead and perhaps try to rub it off, or
point to the mark on the mirrored forehead, or ignore the mark.
Touching one’s own forehead is taken as evidence for recognizing the
“objective” self in the mirror. Children pass this test usually sometime
between 16 and 24 months (about the same time that they typically
acquire a first 50-word vocabulary).
The capacity for delayed imitation, that is, the imitation of an action
or gesture observed but not imitated on a prior occasion, comes into
play toward the end of the first year and has been used in recent
research on children’s developing memory during this period. Bauer
and others have found that children’s memory for an original experi-
ence may be retained over weeks and up to 6 months by 13-month-
olds, with longer extensions typical of older toddlers (Bauer, Wenner,
Dropik, & Wewerka, 2000; Fivush, 1994). Other studies have revealed
the abilities of toddlers (1–2-year-olds) to organize sequential knowl-
edge of familiar events in terms of causal relations (Bauer & Mandler,
1989a) and, using a form of sorting, to construct well-organized cate-

gories of objects (Bauer & Mandler, 1989b; Gopnik & Choi, 1990; Gop-
nik & Meltzoff, 1986).
Evidence from language use during this period indicates that chil-
dren are capable of distinguishing the perspective of the other from
their own, although one would not want to say that they are capable
of taking another’s perspective, a later achievement. One good clue is
that they usually quite easily acquire the pronouns “I” and “you” and
use them distinctively in conversation. Children often begin by label-
ing themselves with their own name or by using “me” or “mine” for
first-person address (Budwig, 1995; Nelson, 1989). Careful analysis of
these uses suggests that children are struggling with the varied per-
spectives on person (agent, actor, experiencer, object, possessor) that
are embedded in the language uses that they are in the process of
acquiring. That understanding the objective status of the self is deeply
interconnected to that of others is revealed in the evidence of a new
self-consciousness, embarrassment, and sense of shame or guilt that
becomes evident at around 2 years of age (Kagan, 1991; Lewis, 1997).
This interconnectedness reflects the social dependence of the process
of self-knowing described by Mead (1934).
All of these cognitive and interpersonal developments add up to a
substantial advance from infancy into early childhood, and in the
terms in which we are interested here, in self-knowing consciousness.
Most of these transformations can be seen as resulting from the new
awareness that emerged at the prior level of social consciousness com-
pounded by the new contrasts made possible through the medium of
words as they come into use by the self and understanding of the uses
by others. As Gopnik and her colleagues have argued, using words
advances cognition in this period in a number of ways (Gopnik &
Meltzoff, 1986). They make explicit concepts that have begun to
emerge through experience, but they also can guide the distinctions
among concepts and categories that were not previously recognized.
As Bauer’s work (Bauer & Wewerka, 1997) has shown, memory for
events does not depend upon language, but the ability to use lan-
guage in reference to the objects or actions in an event leads to reten-
tion of the memory for a longer period of time (see also Simcock &
Hayne, 2002). Thus, early language and the advances of this period
appear to be interconnected, but it is not clear that one is necessary to
or sufficient for the other. Rather, it appears that the social, cognitive,
and linguistic abilities mutually support each other.
Is this, then, the dawn of self-knowing consciousness? Is it the miss-
ing link we are looking for? Is it the point of origin of autonoesis?
Considering the latter, I believe that the answer is no. Clearly, the
child of 2 years has advanced, as documented, toward a new under-
standing of self and other. Moreover, language is becoming a useful
tool for both communication and cognition. However, despite the

claims made by Howe and Courage (1993) and others, there is no
evidence in all of these new skills and capabilities that children at 2
years have a sense of self in the past of the kind that Tulving refers
to. Indeed, as I have argued (Nelson, 1988, 1989, 2001), the child at
this age has no conception of the past or future; in Donald’s sense,
she is living in an episodic present. It is true that retention of memory
for an event may last for many months, but there is no evidence that
such memory is anything but noetic; that is, it is a representation of
knowledge about this kind of event. As I noted previously, the evi-
dence from 3-year-old children’s reports on their experience of events
(e.g., a birthday party) tend to be very general and are reported in the
present tense. This evidence is not definitive, of course, because chil-
dren might have a sense of self in the past but not be able to express
it. However, I believe that the bulk of the evidence indicates that 1-
or 2-year-old children are not yet tuned in to the significance of the
specific past or the possible future and have no sense of themselves
in past experience.
This is not to say that children at this age are confined to the pres-
ent moment. Referring to Donald’s concept of mimesis, it seems ap-
propriate to think of the 2-year-old as a mimetic child, capable of re-
playing in consciousness, as well as in external play with toys or
substitute objects, the actions of herself and others. Thus, a certain
degree of reflection on states of self and other can be achieved through
the imagination, either in thought or in play. But this is play with
what is and what can be, not with what was and what may yet be. It
is not using objects in play to stand for real things, or to symbolize
the self in another place or time (Nelson, in press).
Reflective Consciousness
The cognitive consciousness of the previous level, I argued, does not
imply a level of self-reflective consciousness, but it leads toward that
level, through the potentials it opens up in the cognitive contents and
social interactions involved. One of the most significant of those po-
tentials is engagement in conversations made possible by the acquisi-
tion of beginning language skills, which at the same time contributes
to the advancement of those skills. What conversations involve is rep-
resentational language, the use of symbolic communication to represent
a thought, a state of the world from the past or the future, an explana-
tion, a description, a hypothesis or prediction, an invitation to engage
in pretense, and so on. These possibilities are not available without
the considerable resources of language, not just words for objects and
actions, but the complexity of the symbolic systems that people create
to make such communications feasible. Both Merlin Donald and I
have talked about this potential of language in similar ways as “com-
munities of mind” (Donald, 2001) or as “entering a community of

mind” (Nelson, in press; Nelson, Henseler, & Plesa, 1999; Nelson et
al., 2003).
Researchers have found that 3-year-olds are aware of how others
feel, of their visual perspectives on a visual scene, of how others’
wants and tastes may differ from their own. Such contrasts may be
evident in nonverbal social situations, but expressing them in linguis-
tic form makes them more explicit. My colleagues and I were able to
observe the process involved in coming into this level in detail in the
study of a single child (Emily) over the course of the crucial age from
2 to 3 years, because her practice of talking to herself at bedtime pro-
vided a unique perspective on the emerging thought of the young
child (Nelson, 1989). These monologues were supplemented by re-
cordings of dialogues between father and daughter prior to his leav-
ing the room at bedtime. The conversations included much talk about
what would happen in the coming day or days, talk that Emily was
highly attuned to and often attempted to replicate in a subsequent
monologue.
Several observations derived from this study are notable. First, at
the beginning of the third year, the focus of her talk was almost en-
tirely on how the events of her life were organized, for example, the
morning routine, the day care routine, the afternoon play, the trips to
the store, and so on. She began to focus on singular events during this
time period, but at first with an emphasis on their organization, not
their “meaning” or place in relation to herself. In the latter half of the
year, she showed a new focus on self, and on people’s reactions to
herself, as well as her own reactions to events. Moreover, she began
to use talk as a means of imagining different storylike possibilities.
Previously, she had used her monologues as a means of raising ques-
tions and problems, such as what was likely to happen at the doctor’s
or who would bring a book to day care the next day. These questions
focus on the organization of not quite familiar events or open slots
within events. New questions—“why” questions—opened up toward
the end of the year. One notable question revolved around why her
father could not run in a race, notable both for the question of cause
and for its focus on another person. Although there was always talk
about what she wanted or did not want at bedtime, Emily’s discussion
of her own feelings or of others’ reactions to her emerged only toward
the end of this third year. Overall, for this child the questions of de-
sires and emotions seemed less in focus in the talk around bedtime
than questions of what, how, and why.
These observations add concrete details to the general description
of notable advances in reflective consciousness between 2 and 5 years.
I have labeled this level of consciousness “reflective” because verbally
interacting with others in discussions about feelings, thoughts, and
experiences evokes reflection on these aspects, as well as on current

activities. As our evidence from Emily’s monologues indicated, the
capacity for using language in talk to oneself enables the same kind
of individual reflection through private speech. This invokes the pro-
cess that Vygotsky (1986) proposed as involving the internalization
of what is first experienced on an intermental level through verbal
interaction. His account of verbal thought, a beginning of which we
were observing in Emily’s bedtime talk, posited the move from social
speech to private speech and finally to inner speech or internalized
verbal thought. Emily’s monologues suggest that many children may
take on the intrapersonal reflective role much earlier than implied in
Vygotsky’s discussion of the emergence of inner speech (or verbal
thought) from private speech, but that the process is likely the same
as he proposed.
Observational studies of young children have long noted the onset
of “why” questions early in this period, when a child suddenly ap-
pears to shift from a focus on “what” to why things are the way they
are. It appears that after a long period during which children have
focused only on understanding how things are arranged in the world
(and on keeping them in that particular arrangement), they suddenly
open up to the possibility that things might be some other way, but
happen to be the way they are, so the question that emerges is why?
A recent diary study of children’s why questions found that psycho-
logical and physical causation questions were asked by preschoolers
about equally often (Callanan & Oakes, 1992).
It is also during this period that parents and children begin engag-
ing in talk about the past and the future. As previously noted, this
talk has been extensively studied, and the gradual development of
the child’s role from listener to contributor of fragments and fuller
sequences of action up to full narratives has been documented. The
influence of parental framing and scaffolding has also been docu-
mented in important longitudinal research by Reese (2002) and her
colleagues. They found that children’s cognitive sense of self in the
toddler period and parental talk about the past during the 2- to 3-
year-old period both contribute independently but interactively to the
child’s interest in and ability to engage in autobiographical memory
discussions at 3 to 4 years.
Yet despite the good evidence that 3-year-olds in general have be-
gun to understand a “continuing me”—a sense of self that extends
from the past to the future—there is also convincing evidence that
this is a transitional stage, that they do not yet have a well-defined
distinction between the experience of the self and the experience of
others. This evidence comes from their vulnerability to false sugges-
tions about what happened (Ceci & Bruck, 1993), source errors, and
intriguing evidence of the appropriation of others’ memories (Miller,
Potts, Fung, Hoogstra, & Mintz, 1990). This evidence suggests that
when it first becomes accessible to the child, representational lan-
guage appears to become part of the general noetic knowledge sys-
tem, joined with experientially based knowledge gained from direct
experience in the world. Thus, no fine distinctions are made between
the child’s own memory/knowledge and that of another. There are
no noetic/autonoetic and no mine/yours distinctions in memory at
this time.
The conversations that parents have with children about episodes
from their past or what they can expect in the future, or how they
feel about happenings in their lives, what they like and what others
like—these may all incorporate distinctions between the child’s expe-
rience and others’ way of experiencing the same thing. An important
part of the present proposal is the claim that conversational interac-
tions that involve the child in reflection on self, the past, feelings, and
thoughts provide the context for the representational and reflective
use of language by the child, either through external private speech,
or internal verbalization that makes such reflection possible. The as-
sumption here is that verbal representing invokes a cognitive object
that is accessible to cognitive reflection. In some sense, a thought, a
feeling, an experience from the past becomes “visible” for examination
by the thinking child when it is represented in verbal form. This be-
comes possible when it is externalized in verbal form, first by the
parent, then by the child, and eventually when the function is inter-
nalized. The process is similar to that involved in using language to
remember some part of an activity sequence that is not evident in
the present context—the word holds its place in memory. A relevant
question to be addressed is this: to what extent is this kind of “off-
track” reflection possible without language or some other symbolic
system? Can nonhuman primates reflect on past events or past feel-
ings?
In any case, what takes place during the transitional period of the
reflective level of consciousness is an introduction to the uses of ver-
bal communication and thought as both representational and reflec-
tive mediums. To the extent that children take up the representational
and reflective functions, they are led further into the mysteries of the
mental world through the potentials unleashed by narrative thought.
Narrative Consciousness
In this scheme, narrative consciousness is the center of the action, just
as Donald (1991, 2001) proposes that mythic culture is the initial flow-
ering of all modern human cultural and cognitive advances. Adults
surround children with stories and narratives, mythic and experien-
tial, from the beginnings of their lives, but it is only as children be-
come competent at verbal representations (both receptive and produc-
tive) that they are able to participate in narrative making and narrative
comprehension. As I have long argued, children are very good at re-
membering event representations for familiar events and also at re-
taining the sequence of short novel events. However, narrative is not
simply a presentation of events. It involves personal perspectives, mo-
tivations, temporal and spatial locations, evaluations—all of the as-
pects that Bruner (1990) calls “the landscape of consciousness” that
together with “the landscape of action” constitute a narrative. Chil-
dren’s play may come to invoke aspects of this landscape as children
begin to collaborate on imaginative play scenarios, but play also be-
gins as the simple playing out of everyday events.
Between age 3 and 5 years, children come to contribute more to the
construction of accounts of past experience and to participate in the
making of narratives about their own lives. They also begin to listen
with understanding and to follow the story line from storybooks, vid-
eos, television, and other media. Out of these experiences come a
plethora of transformations in the child’s consciousness, among them
the following (for details see Nelson, 2001):
• An explicit awareness of the contrast between other people’s
stories and one’s own experiential story, as well as “made-up”
stories, although the understanding of the difference between
a real and a fictional story may be long in coming (Applebee,
1979; Fontaine, 2002)
• Past, present, and future contrasts with an explicit concept of
the location of events at past times and temporal locational
language such as “yesterday” and “tomorrow”
• A sense of the “continuing me”—the idea that the self existed
in the past as a baby and will continue to exist in the future
as an adult
• The contrast between other people’s minds and one’s own
mental states—a theory of mind
This level thus realizes the potential inherent in the transitional reflec-
tive level and goes beyond it. Most notable is the newly emerging
sense of self that is situated in time. These two constructs—self and
time—go together and are undoubtedly specifically human and made
specifically possible through the symbolic capacities of human lan-
guage used communicatively and cognitively. It is at this level that
true autobiographical memory begins and infantile amnesia is over-
come. The new sense of self is reflected in the autonoesis of episodic
memory, which is the basis for the development of autobiographical
memory. In this theory narrative, autobiographical memory (stories
of the self) and self-concept are interdependent in development.
It hardly needs to be stressed that the achievement of this level of
consciousness is a true landmark, one that no other animal could
reach, as it depends not only on the capacity for acquiring symbols,

but on the surround of stories produced by members of the culture,
including the child herself. It is of interest that toward the end of her
third year, Emily began producing made-up stories in her monologue,
a creative move that signifies a distinctive sense of “otherness” and
“other experience” to be explored through linguistic means.
In what sense is this a new level of cognitive consciousness beyond
the reflective/representational level? Phil Zelazo (1999), who has fo-
cused on cognitive tasks that require switching between different clas-
sifications of the same stimulus cards (e.g., between color and iden-
tity), emphasizes the move to a new level of executive function during
this age period. Could maturation of the brain areas that subserve
executive control functions explain the developments at this level?
Certainly the capacity to keep different perspectives in mind, to switch
between story lines while keeping a whole narrative in mind, may be
seen as an advanced executive skill. But note also that these skills
can only be exercised through symbolic means. Thus in some sense,
executive control and language are intertwined (a point that Zelazo
recognizes).
The more important aspect of the transition at this level is, I believe,
that signaled in Donald’s conception that human minds are cultural
constructions. It is at the level of the narrative mind, which is able to
take in a whole story about a piece of the world that has never been
experienced, that the self is truly opened to “cultural programming”
for better or worse. Prior to this level, children may know how to do
things in the world, but they are ignorant about the complexities and
proliferation of “possible worlds,” of cultural roles that people play,
thoughts that people may have, true and false, and of their own possi-
ble futures. It is at this level that children finally emerge from their
private mind state and become part of the community of minds. Is
there more? Of course, the view from within the cultural community
includes all of the possibilities that lie ahead, including most immedi-
ately the possibility of school, literacy, and its rewards. Obviously, the
extension of consciousness does not stop at the preschool level.
Cultural Consciousness
The level of cultural consciousness takes us far beyond the achieve-
ment of metacognitive thinking or autonoesis, but I would maintain
that it is a legitimate extension of consciousness to a new level, as
implied in Donald’s conception of the “theoretical” transition in hu-
man history. At the beginning, this move reveals to the child (at about
school age) the existence and significance of cultural roles and of cul-
tural rules invoked by institutions, moral, ethical, legal, or simply ad-
ministrative. The beginnings of cultural identity—national, religious,
ethnic—are established at this level. And of course, the acquisition of
cultural knowledge systems—literary, mathematical scientific, philo-

sophic, artistic—becomes possible as literacy consciousness is estab-
lished and nurtured. This level is where mature human minds dwell
much of the time, although of course all the previous levels of con-
sciousness are in play as well.
SUMMARY AND CONCLUSION: LANGUAGE, SELF, AND CONSCIOUSNESS

This excursion into the development of human consciousness has ob-
viously taken us far beyond the questions of whether some or any
animals display metacognitive awareness. One may question whether
the term “consciousness” is being misused when it extends so far.
However, part of my point here is that there is no firm line to be
drawn and no point of origin when we examine consciousness in its
human developmental path. Rather, what we find is a continuous en-
largement and extension, taking awareness to new levels both verti-
cally (categorical dimension) and horizontally (temporal dimension).
In human development, social interaction is crucial to this enlarge-
ment, and language practices provide the grounds on which children
find their way into human ways of knowing self and others within
cultural groups. Developments in infancy awaken the initial sense of
self, but its further development at the reflective and narrative levels
are constrained by the use of symbolic representations among social
groups. Reflective consciousness relies on the kind of social interac-
tions made possible by simple language and gesture or mime. These
in turn lead in normal development in natural language-using envi-
ronments to more complex language and abstract symbolic social in-
teractions that construct concepts like mind and shared representa-
tions of past and future as well as facts and fiction.
Because these language-based levels are unique to humans, we
should look to the earlier levels—perhaps to cognitive conscious-
ness—for the closest relation to our own experience among nonhu-
man animals, including nonhuman primates. Self-awareness in epi-
sodic memory—autonoesis—does not appear at this level for children,
and it seems likely that it does not appear for any other animal. Thus,
Tulving (1983) was correct when he stated that humans are the only
creatures who can travel backward in time to relive their own prior
experiences. Some reflective consciousness might be achieved by “lan-
guage learning” enculturated primates, but not narrative conscious-
ness; no evidence of understanding anything like a story or even an
account of the past has been reported from the studies of these pri-
mates.
However, I would also note again that human infancy is also
unique in its social dimensions, which set up the achievements of the
cognitive consciousness of the 1- to 2-year-old. Thus, this level of con-
sciousness may also be unique to humans. But the conclusion should

not be to rule out anything except basic awareness for other animals
(on the level of the newborn infant). Rather, the evolutionary lesson is
that other animals may evolve their own different levels of conscious
awareness, including levels of metacognition, which the chapters in
this volume appear to document for some species but not for others.
NOTES
1. Gould also wanted to destroy the idea of successive stages of phylogenesis as

an adding-on process, emphasizing instead different developmental timing changes,
for example, the possibility of neoteny.
2. This thesis has its flaws (see Nelson et al., 2000), but that is not the issue here.
3. The terminology used by both Tulving and Donald is unnecessarily confusing.
“Semantic” was originally chosen presumably because of the assumption that general
knowledge existed in some semantic form, where the term implies a relation to lan-
guage. This made sense in 1972, when all of the literature being reviewed was verbal,
and when “episodic” referred to learning lists of words. The attempt to relate these
terms to nonhuman or infant learning and memory encounters the infelicity of both.
The use of “noetic” and “autonoetic” would avoid this, but these terms are not gener-
ally familiar to readers.
4. See Donald (1991) for a defense of the proposition that language as we know
it arrived late as a human cultural invention.
REFERENCES
Applebee, A. N. (1978). The child’s concept of story. Chicago: University of Chi-
cago Press.
Astington, J. W., Harris, P. L., & Olson, D. (Eds.). (1988). Developing theories of
mind. Cambridge, UK: Cambridge University Press.
Bauer, P. J., & Mandler, J. M. (1989a). One thing follows another: Effects of
temporal structure on one- to two-year-olds’ recall of events. Developmental
Bauer, P. J., & Mandler, J. M. (1989b). Taxonomies and triads: Conceptual
organization in one- to two-year-olds. Cognitive Psychology, 21, 156–184.
Bauer, P. J., Wenner, J. A., Dropik, P. L., & Wewerka, S. S. (2000). Parameters
of remembering and forgetting in the transition from infancy to early
childhood. Monographs of the Society for Research in Child Development, 65(4,
Serial No. 263).
Bauer, P. J., & Wewerka, S. S. (1997). Saying is revealing: Verbal expression
of event memory in the transition from infancy to early childhood. In P. W.
van den Broek, P. J. Bauer, & T. Bourg (Eds.), Developmental spans in event
comprehension and representation (pp. 139–168). Mahwah, NJ: Lawrence Erl-
baum.
Bruner, J. S. (1990). Acts of meaning. Cambridge, MA: Harvard University
Press.
Budwig, N. (1995). A developmental-functionalist approach to child language. Mah-
wah, NJ: Lawrence Erlbaum.
Callanan, M. A., & Oakes, L. M. (1992). Preschoolers’ questions and parents’

explanations: Causal thinking in everyday activity. Cognitive Development,
7, 213–233.
Campos, J. J., Bertenthal, B. I., & Kermoian, R. (1992). Early experience and
emotional development: The emergence of wariness of heights. Psychologi-
cal Science, 3, 61–64.
Ceci, S. J., & Bruck, M. (1993). Suggestibility of the child witness: A historical
review and synthesis. Psychological Bulletin, 113, 403–439.
Chomsky, N. (1965). Aspects of a theory of syntax. Cambridge MA: MIT Press.
Damasio, A. (1999). The feeling of what happens: Body and emotion in the making
of consciousness. New York: Harcourt.
Depew, D. J., & Weber, B. H. (1995). Darwinism evolving: Systems dynamics and
the genealogy of natural selection. Cambridge MA: MIT Press.
Donald, M. (1991). Origins of the modern mind. Cambridge MA: Harvard Uni-
versity Press.
Donald, M. (2001). A mind so rare: The evolution of human consciousness. New
York: Norton.
Fischer, K. W., & Bidell, T. (1991). Constraining nativist inferences about cog-
nitive capacities. In S. Carey & R. Gelman (Eds.), The epigenesis of mind:
Essays on biology and cognition (pp. 199–236). Hillsdale, NJ: Lawrence Erl-
baum.
Fivush, R. (Ed.). (1994). Long-term retention of infant memories [Special is-
sue]. Memory, 2, 353–382.
Flavell, J. H., Friedrichs, A. G., & Hoyt, J. D. (1970). Developmental changes
in memorization processes. Cognitive Psychology, 1, 324–340.
Fodor, J. A. (1975). The language of thought. New York: Crowell.
Fontaine, R. G. (2002). Preschoolers’ understanding of story books: The influence of
story genre, affect, and language. Unpublished PhD dissertation, City Univer-
sity of New York, New York.
Gopnik, A., Capps, L., & Meltzoff, A. N. (2000). Early theories of mind: What
the theory theory can tell us about autism. In S. Baron-Cohen, H. Tager-
Flusberg, & D. J. Cohen (Eds.), Understanding other minds: Perspectives from
developmental cognitive neuroscience (pp. 50–72). Oxford, UK: Oxford Uni-
versity Press.
Gopnik, A., & Choi, S. (1990). Language and cognition. First Language, 10,
199–216.
Gopnik, A., & Meltzoff, A. (1986). Words, plans, and things: Interactions be-
tween semantic and cognitive development in the one-word stage. In S.
Kuczaj & M. Barrett (Eds.), The development of word meaning (pp. 199–223).
New York: Springer-Verlag.
Gopnik, A., & Wellman, H. (1994). The theory theory. In L. A. Hirschfeld &
S. A. Gelman (Eds.), Mapping the mind (pp. 257–293). New York: Cam-
bridge University Press.
Gould, S. J. (1977). Ontogeny and phylogeny. Cambridge, MA: Harvard Univer-
sity Press.
Han, J. J., Leichtman, M. D., & Wang, Q. (1998). Autobiographical memory
in Korean, Chinese, and American children. Developmental Psychology, 34,
701–713.
Harris, P. L. (1992). From simulation to folk psychology: The case for develop-
ment. Mind and Language, 7, 120–144.
Howe, M. L., & Courage, M. L. (1993). On resolving the enigma of infantile
amnesia. Psychological Bulletin, 113, 305–326.
Hrdy, S. B. (1999). Mother Nature: A history of mothers, infants, and natural selec-
tion. New York: Pantheon.
James, W. (1950). The principles of psychology. New York: Dover. (Original work
published 1890)
Kagan, J. (1991). The theoretical utility of constructs for self. Developmental
Review, 11, 244–250.
Leslie, A. M. (1987). Pretense and representation: The origins of “theory of
mind.” Psychological Review, 94, 412–426.
Leslie, A. M. (1991). The theory of mind impairment in autism: Evidence for
a modular mechanism of development? In A. Whiten (Ed.), Natural theories
of mind (pp. 63–78). Oxford, UK: Basil Blackwell.
Lewis, M. (1997). The self in self-conscious emotions. In J. G. Snodgrass &
R. L. Thompson (Eds.), The self across psychology (pp. 119–142). New York:
New York Academy of Sciences.
Lewis, M., & Ramsey, D. (1999). Intentions, consciousness, and pretend play.
In P. D. Zelazo, J. W. Astington, & D. R. Olson (Eds.), Developing theories
of intention: Social understanding and self-control (pp. 77–94). Mahwah, NJ:
Lawrence Erlbaum.
Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago
Press.
Miller, P. J., Potts, R., Fung, H., Hoogstra, L., & Mintz, J. (1990). Narrative
practices and the social construction of self in childhood. American Ethnolo-
gist, 17, 292–311.
Mullen, M. K. (1994). Earliest recollections of childhood: A demographic anal-
ysis. Cognition, 52, 55–79.
Mullen, M. K., & Yi, S. (1995). The cultural context of talk about the past:
Implications for the development of autobiographical memory. Cognitive
Development, 10, 407–419.
Nagel, T. (1974). What is it like to be a bat? Philosophical Review, 83, 435–450.
Neisser, U. (1997). The roots of self-knowledge: Perceiving self, it, and thou.
In J. G. Snodgrass & R. L. Thompson (Eds.), The self across psychology: Self-
recognition, self-awareness, and the self concept (pp. 19–33). New York: New
York Academy of Sciences.
Nelson, K. (1988). The ontogeny of memory for real events. In U. Neisser
& E. Winograd (Eds.), Remembering reconsidered: Ecological and traditional
approaches to the study of memory (pp. 244–276). New York: Cambridge Uni-
versity Press.
Nelson, K. (1989). Monologue as the linguistic construction of self in time. In
K. Nelson (Ed.), Narratives from the crib (pp. 284–308). Cambridge MA: Har-
vard University Press.
Nelson, K. (2001). From the experiencing I to the continuing me. In C. Moore
& K. Skene (Eds.), The self in time: Developmental issues (pp. 15–34). Mah-
wah, NJ: Lawrence Erlbaum.
Nelson, K. (2003). Making sense in a world of symbols. In A. Toomela (Ed.),
Cultural guidance in the development of the human mind (pp. 139–162). West-
port, CT: Greenwood.
Nelson, K. (in press). Language pathways to the community of minds. In
J. W. Astington & J. Baird (Eds.), Why language matters to theory of mind.
Nelson, K., & Fivush, R. (2000). Socialization of memory. In E. Tulving & F.
Craik (Eds.), Handbook of memory (pp. 283–295). New York: Oxford Univer-
sity Press.
Nelson, K., & Fivush, R. (2004). Emergence of autobiographical memory: A
social-cultural theory. Psychological Review, 111, 486–511.
Nelson, K., & Gruendel, J. (1981). Generalized event representations: Basic
building blocks of cognitive development. In M. Lamb & A. Brown (Eds.),
Advances in developmental psychology (Vol. 1, pp. 131–158). Hillsdale, NJ:
Lawrence Erlbaum.
Nelson, K., Henseler, S., & Plesa, D. (2000). Entering a community of minds:
Feminist perspective on theory of mind development. In P. Miller & E.
Scholnick (Eds.), Toward a feminist developmental psychology (pp. 61–84).
New York: Routledge.
Nelson, K., Plesa, D., Goldman, S., Henseler, S., Presler, N., & Walkenfeld,
F. F. (2003). Entering a community of minds: An experiential approach to
theory of minds. Human Development, 46, 24–46.
Nelson, K., & Ross, G. (1980). The generalities and specifics of long term mem-
ory in infants and young children. In M. Perlmutter (Ed.), Children’s mem-
ory: New directions for child development (Vol. 10, pp. 87–101). San Francisco:
Jossey-Bass.
Perlmutter, M. (Ed.). (1980). Children’s memory. (Vol. 10). San Francisco: Jos-
sey-Bass.
Piaget, J. (1962). Play, dreams, and imitation in childhood. New York: Norton.
Pillemer, D. B., & White, S. H. (1989). Childhood events recalled by children
and adults. In H. W. Reese (Ed.), Advances in child development and behavior
(Vol. 21, pp. 297–340). New York: Academic Press.
In R. Byrne & A. Whiten (Eds.), Machiavellian intelligence (pp. 160–179).
Oxford, UK: Oxford University Press.
mind? Behavioral and Brain Sciences, 1, 515–526.
Reese, E. (2002). A model of the origins of autobiographical memory. In J. W.
Fagen & H. Hayne (Eds.), Progress in infancy research (Vol. 2, pp. 215–260).
Mahwah, NJ: Lawrence Erlbaum.
Reese, E., Haden, C. A., & Fivush, R. (1993). Mother-child conversations about
the past: Relationships of style and memory over time. Cognitive Develop-
ment, 8, 403–430.
Simcock, G., & Hayne, H. (2002). Breaking the barrier? Children fail to trans-
late their preverbal memories into language. Psychological Science, 13, 225–
231.
Tomasello, M. (1999a). The cultural origins of human cognition. Cambridge, MA:
Harvard University Press.
Tomasello, M. (1999b). Having intentions, understanding intentions, and un-
derstanding communicative intentions. In P. D. Zelazo, J. W. Astington, &
D. R. Olson (Eds.), Developing theories of intention: Social understanding and

self-control (pp. 63–76). Mahwah, NJ: Lawrence Erlbaum.
sity Press.
Tulving, E. (1983). Elements of episodic memory. New York: Oxford University
Press.
Vygotsky, L. (1986). Thought and language. Cambridge, MA: MIT Press.
Zelazo, P. D. (1999). Language, levels of consciousness and the development
of intentional action. In P. D. Zelazo, J. W. Astington, & D. R. Olson (Eds.),
Developing theories of intention: Social understanding and self-control (pp. 95–
118). Mahwah, NJ: Lawrence Erlbaum.
5
A Continuum of Self-Consciousness That

Emerges in Phylogeny and Ontogeny
Marcel Kinsbourne
The beast does but know, but the man knows that he knows.
John Donne, sermon, 1628
IS SELF-CONSCIOUSNESS A MATTER OF DEGREE?

Are only humans capable of self-reflection? Descartes and his follow-
ers, as well as Tulving (chapter 1, this volume) are among those who
subscribe to this “discontinuity” theory. Or did our highly developed
facility for self-reflection emerge from humbler origins in less behav-
iorally sophisticated creatures, as Hampton (chapter 11) and Terrace
(chapter 3) propose? The argument for discontinuity echoes that for
language, which is considered barely to exist in nonhuman animals.
That proposition is still widely contested. However, it is not known
whether language is essential for self-reflection, or merely puts into
words what the brain has already accomplished preverbally or even
preconsciously. My discussion does not purport to resolve this sweep-
ing issue, any more than previous discussions have. Rather, I use the
premise of continuity theory to present some critical ideas about test-
ing for self-awareness in animals and also in human infants, and
about how conscious and self-conscious processes may be instantiated
in the brain.
I favor the continuity assumption because it offers a better fit with
our current understanding of the evolution of species and also with
the functional anatomy of the brain. I offer alternative explanations
for why animals and human infants frequently fail on tests that are
designed to probe the ability to self-reflect. I take as my premise the
view that consciousness is the subjective aspect of the patterned acti-
vation of the forebrain. If so, self-awareness may reflect the ability to
maintain multiple representations in the network at the same time.
142
A Continuum of Self-Consciousness 143
Self-awareness would then be one of many abilities that a highly dif-

ferentiated brain can master. It need not be regarded as the specialized
product of a uniquely constituted neural module. Self-awareness may
be unique to humans without being modular. However, arguments
for human uniqueness are often presented in terms of modularity. I
take a module to refer to a uniquely specialized part of the brain,
regardless of whether its components are concentrated within a dis-
crete area or distributed as a system.
THE MODULARITY ASSUMPTION

Chomsky’s concept of a “language acquisition device,” which humans
have and other animals do not, exemplifies a modular conception of
brain organization. An engineering notion, modularity is a gloss on
the exquisite degree of regional specialization that has long been
known to characterize the human cerebral cortical mantle. It implies
discontinuity, even the ultimate discontinuity, that hypothetical extra
module that makes us human. However, the functional neuroanatomy
of the human cerebral cortex, though highly differentiated, does not
feature discontinuity between modules. The cerebral neural network
is completely continuous, with only a few degrees of synaptic separa-
tion between any neuron and any other (Braitenberg & Schuz, 1992).
Overused as an explanatory concept, modularity has fueled “neuro-
logizing,” instant explanations for why any focal lesion might disable
a complex set of mental operations. “It takes something that clearly is
a very complex consequence of underlying mechanisms, and simply
plugs it in as an explicit structure, bypassing the question of what
these underlying mechanisms might be” (Hofstadter, 1985, p. 642).
Complex activities emerge from the coordination of multiple underly-
ing component processes, and these are unlikely to be dedicated ex-
clusively to any single purpose. It has long been taken for granted in
neuropsychology that one has to determine whether a disabled mental
operation actually is impaired due to a more general, basic underlying
cognitive deficit before one can conclude that its constituent mental
operations are uniquely dedicated to a single purpose. This presump-
tion motivates a search for the component skills that interact to enable
the performance under scrutiny. Is the neural basis of self-reflective
behavior a network of component operations that are also involved in
other complex cognitions, some of which are plausibly available to
other animals also?
SHARP INEQUALITIES IN COGNITIVE SKILLS ARE COMPATIBLE

WITH CONTINUITY THEORY
The term “missing link,” with which this book is titled, has an intrigu-
ing implication. It suggests continuity but expresses surprise about
the size of disparities between species. I argue that even in the face of
enormous disparity, a continuity interpretation is viable. There need
not be a missing link.
It is uncontroversial that the cognitive differences between our spe-
cies and nonhuman primates are enormous. A behavioral parameter,
customarily referred to as the “ascent to man,” is graphically repre-
sented along an upward sloping linear gradient, the coordinate ex-
pressing complexity and the abscissa representing time. On it, various
creatures, all facing rightward and looking wistfully upward, are ar-
ranged left to right from the lower to the upper end, in order of their
apparent behavioral sophistication, the human at the right upper
end. On this linearly conceived time scale, chimpanzees crowd in
rather closely behind us—after all, they are supposedly our cousins.
The implication is that any differences between nonhuman primates
and humans are rather limited. However, continuity does not imply
linearity, and this linear presumption is not supportable. The
straight line is an artifact of human invention. In nature, the values
of variables rise and fall nonlinearly: exponentially or even hyper-
bolically. I am not aware of any adaptive characteristic that varies
linearly across species, nor can I imagine one. The gradient of ascent
in behavioral sophistication may be sharply positively accelerated,
originating at a minimal level of the aptitude of interest. At the ori-
gin, the gradient may be so gradual that it is indeterminate at what
point back in phylogenesis the cognitive ability originated. At the
endpoint, where the positive acceleration is extreme, disparities be-
tween neighboring species can be enormous, without the existence
of a missing link. Self-awareness could have evolved along a sharp
curve.
It is tantalizing to note how similar our genomes are to those of the
other primates, barely differing by 1% or 2%. However, it is simplistic
to compare genomes arithmetically. Genes are not created equal. For
instance, consider that genes control how many times neurons divide;
when they stop dividing, the species’ full complement of brain cells
is present. Now imagine a single mutation that relaxes the control on
cell division to the extent that the neurons divide one more time be-
fore they rest. One mutation could create a forebrain twice as large,
with twice the computing space within which to be smart. Pertinent
to this claim is a gene called ASPM that is involved in regulating brain
size. Null mutations of this gene lead to a 70% decrease in brain vol-
ume (Evans et al., 2004). The claim that the genomes of humans and
chimpanzees are closely similar is an intuition pump, in Hofstadter
and Dennett’s (1981) sense. It sounds like it should mean something,
even though it does not. In nature, differences may be great and yet
reside on a continuum.
FALSE NEGATIVE DEMONSTRATIONS OF METACOGNITIVE SKILLS

Unbridgeable gaps in behavioral potential between humans and other
animals may exist due to insurmountable constraints in animals’ ner-
vous systems, or because animals lack the type of socialization that
Terrace (chapter 3, this volume) posits as being essential for metacog-
nitive functioning. But when animals fail on tests that are designed to
elicit self-awareness, this may not prove that they lack self-awareness.
Task-irrelevant features that contaminate the test situation can mask
cognitive competence.
Escape From Stimulus Control

Extraexperimental salient stimuli can mask potentially successful
performance. Self-awareness calls for the ability to escape external
stimulus control, so that attention is released and can shift internally
to representations about the self, its history, attributes and experi-
ences, that is, to the “self-archive.” Metcalfe and Kober’s (chapter 2,
this volume) “projectable self” cannot project unless the individual
can disengage his attention from the external here and now, and
access his self-archive by introspection. A similar constraint limits
episodic remembering. While nonhuman primates no doubt lack the
extreme flexibility with which people can disengage their attention
from the here and now, if the situation enables them to disengage
from ambient stimulus control, they may be able to admit aspects of
past experiences to the present moment to some extent.
Episodic memory calls for the ability to disengage attention from
the salient stimulus field, so as to permit recollected associates of the
cue to replace external stimuli in the phenomenal present. In “time
travel,” it is the past that travels into the present. There it is updated
automatically and outside awareness before it becomes conscious.
When the memory of the event is next retrieved, it is the updated
version that is recollected, updated yet again. So a superseded exter-
nal reality is replaced by a present but internalized reality.
Problem solving also calls for disengagement. In this instance, the
problem solver has to disengage from his own previous thoughts as
well as from the outside world. He has to be able to disengage from
a superficial but incorrect solution, to clear the way for a less obvious
but perhaps more apt alternative to rise into awareness. If the problem
is complex, several successive disengagements from several failed so-
lutions may be necessary before it is solved.
Disengaging from salient appearances is necessary for success on
tests of conservation and tests for theory of mind, which implies an
awareness of the self as distinct from other selves. These constructs
both require the subject to qualify her percepts by factoring in an
event in the past. If the child cannot disengage from the immediate
percept, she will respond as though she cannot conserve, or as though
she lacks theory of mind. Yet, operating outside her boundary condi-
tions, she may have failed to mobilize knowledge that she could have
expressed under different conditions, because it was overridden by
salient appearances. Escaping stimulus control is difficult for young
children. It becomes easier as they mature. However, even the adult
human brain does not always reconcile contradictory beliefs or im-
pressions but may let the currently more immediately applicable be-
lief overshadow its less salient contradictory counterpart.
In Piaget’s nonconservation demonstrations, the child views two
equivalent quantities—mass, number, and so on. One display is then
manipulated, so that its quantity (volume, number, extent, etc.) now
looks greater to the child than that of the other. The child indicates
that this display is indeed greater, although she had seen these two
displays looking equal a moment earlier, and all that had happened
was that one had been displaced (spread out, poured into a differently
shaped container, etc.). Does the child not realize that manipulating
an array, for instance a row of tokens, cannot make its constituents
fewer or more? Perhaps she does in principle, but she may not have
taken the antecedent maneuver and the conservation principle into
account when she made the judgment that reflects “nonconservation.”
In general, past events and impressions are less salient than ongoing
ones. How things look in the present outweighs how they might be
expected to look, in view of what happened just before. Salience ob-
scures the less attention-grabbing memory of a prior manipulation
and overshadows the fact that an appearance conflicts with a rule.
Manipulating the salience of stimuli can unmask an unsuspected
ability in a relatively simple nervous system. I draw an example from
my own work. Lashley (1938) opined that rats could not tell oppo-
sitely inclined obliques apart, because when tested in a jumping stand,
they could not perform the discrimination. Kinsbourne (1967) made
the direction of slope of the stimulus stripes salient. He outfitted a Y-
maze with a vivid array of alternating black and white oblique stripes,
all sloping in the same direction, along the two walls of one arm, and
stripes inclined in opposite directions along each of the two walls of
the other arm. The rats learned to discriminate the direction of slope
with ease. As in the previous studies, the perceptual problem was to
discriminate obliques. But the difference between directions of slope
had been rendered salient, and therefore was now discriminated.
Conversely, an abnormal shift of attention away from task-relevant
cues may render them less salient. Autistic children (though not ado-
lescents or adults) typically fail the false belief test. Consequently,
they are held not to appreciate that other people have minds or points
of view, and it is assumed that they relate to people as though they
were machines. But if autistic children equate people with machines,

why do they actively avert gaze from people but not from machines?
Perhaps the opposite is the case. Autistic children are only too aware
that people have minds of their own, and therefore may do something
unpredictable any time. This threatens to infringe autistic children’s
intense need for sameness. They may be so averse to social interaction
that they won’t even adopt an internalized perspective on behalf of
someone else. Similarly, gaze avoidance may be why some autistic
children do not recognize themselves in a mirror (chapter 2, this vol-
ume). Given the powerfully inward bent of autistic attention, does the
child even orient to his reflection in the glass?
Working Memory Constraints

An individual manages to safeguard information held in or recycled
through working memory from habituating, by continually refocusing
attention on it, and inhibiting orienting responses to salient but task-
irrelevant distracting stimuli (Estes, 1972). If certain animals, or chil-
dren at a young age, fail on tests that require self-awareness, their
working memory may not have been up to the demands of the partic-
ular tasks given. According to Case (1974), only by age 3 years can a
child hold two representations in mind at a time, and an additional
one every 2 years up to about seven representations. So if a younger
child is given a task that calls for comparing two representations, he
is likely to fail, regardless of whether self-reflection or an unrelated
ability is being tested. In effect, the growing child perfects the ability
to keep unrelated or conflicting representations simultaneously in
mind, or hold one in check while letting the other control behavior.
Inhibitory surrounds presumably save them from collapsing into each
other.
The ability to suppress one representation so as to attend to another
can generalize to enabling episodic remembering. The child inhibits
the influence of the salient here and now, whereupon cued memory
surfaces into recollection as episodic remembering by vector comple-
tion. From a selection of episodic memories, the individual constructs
the parameters of a self-concept. Neither episodic remembering nor
self-concept is necessarily a modular acquisition, perhaps unique to
humans. Such mental activity may be available to any brain that has
sufficient inhibitory and working memory capacity. In other words,
episodic remembering and self-concept may be constrained by the
power of general-purpose cognitive processes rather than lack of spe-
cial-purpose mechanisms.
Boundary Conditions
Although it is usually couched in yes/no format, the question “Are
nonhuman animals capable of episodic remembering?” does not nec-
essarily call for a yes or no answer. The ability might be available, but
only be accessible within restricted boundary conditions.
That an individual is able to apply a form of reasoning or computa-
tion in a given instance is no guarantee that he or she can apply it to
every performance that would benefit from the same approach. In-
deed, the opposite extreme is familiar in the term “expert system.”
This is a special-purpose mechanism that applies logic that the animal
or device is unable to mobilize for any other purpose. The general
purpose/special purpose distinction should not be considered a di-
chotomy but a continuum, or matter of degree. “The hallmark of the
evolution of intelligence . . . is that a capacity first appears in a narrow
context and later becomes extended into other domains. This seems to
be quite a reasonable interpretation of cognitive development” (Rozin,
1975, p. 262).
As I write, I see through the window a seagull raise a clamshell
high in the air, to let it fall against a rock and crack open (Beck, 1976).
How did the seagull know to do this, and can it also solve similar
problems that do not occur in its ecology? A more elaborate example
of restricted boundary conditions is offered by the following tale
about little fishes and the big one that feeds on them (Les Kaufmann,
2001, personal communication). The little fishes live in the mud in
shallow waters, in holes that are too small for the big one to enter.
Therefore the big fish can eat the little ones only when they issue forth
from their hideouts to feed. But while he eats one, the others would
scuttle back to safety. The big fish comes up with what, among hu-
mans, would pass for an ingenious solution. He seals the openings to
the holes by flattening the mud. Then he can pick off fish after fish at
his pleasure. He has planning capability (the tactic), impulse control
(not diving for the first fish that swims into range), and theory of
mind (inferring what the prey would do under given circumstances).
Regrettably, this maneuver exhausts his tactical repertoire. He can
apply his demonstrated problem-solving skills to no other situation.
Wooldridge (cited in Hofstadter, 1985) has described a dramatic
instance of apparently intentional behavior that turns out to be due to
a special-purpose mechanism. The wasp Sphex paralyzes a cricket
with its sting and drags it into a burrow, to leave beside her eggs. The
hatched wasp grubs later feed on the cricket. Before the wasp drags
the cricket in, she enters the burrow and inspects it for intruders. Then
she drags the cricket in. However, if when she reemerges, she finds
the cricket displaced a few inches from the opening, she reiterates the
dragging and the inspection. This continues as often as the experi-
menter chooses to displace the cricket. The wasp never deduces from
her recent inspections that it is safe to pull the cricket right in. The
expert system is rigid and does not profit from experience. It is not
equipped to self-monitor. The wasp represents one end of a contin-
uum. Clayton and Dickinson’s (1998) demonstration that scrub jays

have episodic memory for where food was buried may represent an
expert system. Call (chapter 13, this volume) compares the ability of
several mammalian species to override rigid rule-governed behavior
and arrive at novel solutions.
CONSCIOUSNESS AND ZOMBIES

I use “consciousness” and “awareness” interchangeably. Terrace (chap-
ter 3, this volume) draws a distinction between global awareness and
a more structured consciousness of the perceptual field. This distinc-
tion may not be categorical. It could refer to different points on a
continuum of complexity of the brain states that the neural network
can generate.
Self-reference calls for the ability to monitor one’s own brain states.
Internal monitoring, the essence of metacognition, requires that the
individual escapes from external stimulus control and is able to hold
multiple representations concurrently in mind. Can some animals do
this? Any attempt to prove that they can is restricted to demonstrating
that the animal is capable of retrieving information about itself as a
decision maker or experiencer in ways that in humans are attended
by self-awareness. The proof can never be conclusive, since we are
not, and cannot be, privy to the animal’s subjective mental state. Re-
searchers have therefore been reticent in ascribing subjectivity to ani-
mals that appear to display metacognitive behaviors (e.g., Hampton,
2001; Smith, Shields, Allendoerfer, & Washburn, 1998). By the same
logic, they could call into question what we prefer to take for granted,
that animals are conscious at all, let alone self-conscious.
The impasse is premised on a residue of the dualistic fallacy of the
Cartesian theater (Dennett & Kinsbourne, 1992). This is that awareness
is somehow separate and separable from the brain state that instanti-
ates the experience. Prominent contemporary philosophers of mind
harbor the notion that an organism may be observed to behave exactly
like a conscious person and lack only the element of consciousness. In
effect, the organism is a zombie.
So even if the animal, infant, or robot acts as though conscious,
how can we know that she really is conscious? We do assume that
other people are conscious, because they not only act as though they
were but say so. We have to give them the benefit of the doubt.
Should we extend this benefit of the doubt to any animal? Hampton
(chapter 11, this volume) points out that the “hardened skeptic” is
unlikely to do so. He would then be making a dualistic distinction
between what the brain does and what it is like for the brain to do it.
If awareness is an aspect of certain brain states and has no indepen-
dent existence, the question can be reformulated and the impasse
avoided: Does the animal, infant, or device, when generating quasi-

conscious behavior, instantiate it in neural networks more or less like
ours, albeit less elaborate and specialized?
When measuring up the human against the animal mind, one typi-
cally tries to show that an animal can succeed on a problem that bears
a family resemblance to one that a person can master only after re-
flecting on it. When nonhumans act like a human would under the
circumstance, are they thinking like humans? Kohler (1927), describ-
ing insight in chimpanzees, wrote: “Chimpanzees . . . show by their
careful looking around that they really begin with something very like
an inventory of the situation” (p. 190). Dennett (1981) has called this
pragmatically useful attitude the “intentional stance.” So, if a nonhu-
man primate thumps his chest in a courtship display, we might well
suppose that he is intentionally trying to attract the female’s attention
to himself and his assets. He appears to be capable of self-reference.
But this behavior could have been naturally selected for adaptive rea-
sons, without the animal having any introspection about what he is
doing. The question is, could the animal generalize this behavior to
other novel situations in which it would be adaptively advantageous?
HOW DOES CONSCIOUSNESS ARISE IN THE BRAIN?

According to the integrated cortical field model (Kinsbourne, 1988),
consciousness is neither a product of brain activity nor epiphenome-
nal. Consciousness is not a product, because it has no known indepen-
dent physical existence and is not separable from the functioning
brain. It is not an epiphenomenon either, like exhaust gases are to
the moving automobile, or the whistling sound is to the approaching
missile. It is not a waste product of brain activity or any sort of prod-
uct. It is an attribute that is inherent in the functioning of the neural
network itself. The neural network has subjective access to some of its
own states, and at any moment it has maximal access to the most
activated circuitry. Accordingly, consciousness would not be some-
thing a brain needs to have to perform certain mental operations.
Rather, one is conscious when, and because, one’s brain is function-
ing. Of what one is conscious reflects what the brain is currently do-
ing. For instance, the end product of an act of episodic remembering
is a conscious experience. However, it is not “consciousness” that en-
ables one to remember an episode. Rather, the attractor state that re-
sults from the act of memory retrieval persists long enough to linger
into consciousness, that is, to be subjectively accessible.
THE MODEL: DOMINANT CONFIGURATION MODEL OF CONSCIOUSNESS

Is the neural basis of consciousness centered in a specific brain area
or network, or is it “uncentered” and distributed across the cerebrum
and its connections? The model treats unconscious, conscious, and

self-conscious states not as discrete but as differing only in degree on
a continuum of differentiation of the neural activation pattern of the
forebrain.
I regard consciousness as uncentered in the brain. Integrated cere-
bral network and dominant configuration theory (Kinsbourne, 1988,
1996) assumes that there exists at any time a pattern of activation in
the forebrain that outranks its competitors in amplitude and duration,
and thereby controls the response processes of the moment. The rela-
tively most activated areas supply most of the contents of conscious-
ness. A brain that is capable only of simple mental operations (infant
or animal) would be no less conscious than one that is more complex.
The dominant configuration may represent sophisticated processing
but also simple processing. A simple brain, animal or infant, will still
have a dominant configuration, and this configuration will lend a sub-
jective aspect to the mental functioning of the moment. Processing
power does not determine whether one is conscious, but it does deter-
mine the range of experiences of which one can become conscious.
Any human or animal brain can potentially gain subjective access
to those of its states that are dominant. Whether that access includes
states that reflect self-awareness or episodic remembering depends on
whether the animal’s neural network is rich enough in excitatory and
inhibitory neurons to instantiate more than one distinct brain state at
a time within the cerebral manifold, without the states collapsing into
one another, so as to generate a simpler, albeit more stable, state. Since
only one state predominates at a time, this calls for working memory
capability to alternate the two representations, keeping them suffi-
ciently active so that they can be matched one against the other, or
associated with each other. Whether animals have this capacity is ulti-
mately an empirical issue, species by species.
Nelson (chapter 4, this volume) states that “consciousness expands
in both evolution and development, where levels of consciousness
emerge over time.” I refer this expansion to the expansion of the neu-
ral network of behaviorally sophisticated species. Its increasingly in-
tricate and nuanced functioning is reflected in an enriched sequence
of subjective states. The consciousness per se does nothing.
Relatively enduring states involved in controlled processes, de-
ployed under conditions of uncertainty (Shiffrin & Schneider, 1977),
would typically be represented in dominant configurations. Rapidly
shifting states that comprise the sequential phases of automatic se-
quential processes typically would not. The cognitive processing of
the moment is that of the activated portions of the network, and the
subjectivity of the moment is that of states which are prominent in
that cognitive processing. The processing and its subjectivity are as-
pects of those same neural events and cannot be considered sepa-
rately. Consciousness is integral to the processing. The same process-

ing, if it loses the competition for dominant configuration, would not
be conscious but could still proceed implicitly. As circumstances
change, the subdominant pattern may take the place of the dominant
configuration, or some of its elements may infiltrate the dominant pat-
tern. In the living brain, the subjective field is constantly and endlessly
reinvented.
The pattern of cortical activation takes some time to lock into a
stable (attractor) state through reentrant feedback. Temporary end
states in cognition become conscious. By the time that happens, the
next dominant configuration is already being formed. Consciousness
is always a step behind the action. It reflects, rather than informs or
controls, what the brain is doing (e.g., Kinsbourne, 1988, 1996). The
processing, that is, the settling into attractor states, is preconscious.
Smith (chapter 10, this volume) refers to instances that “inherently
elicit higher level and even conscious modes of cognition and decision
making.” From the perspective of dominant configuration theory,
what is an animal doing while hesitating? Is the animal generating an
impression or point of view, or more explicitly arriving at an opinion
(implying that it knows others might disagree)? A deliberate judg-
ment might be beyond animals that lack language. But a belief can be
derived from any of the above. More likely than any of these, the
animal is forming a sentiment, which arises from an emotional re-
sponse or feeling. In other words, it is in response conflict and is for-
mulating or selecting an action plan that feels better than the rest
(without necessarily introspecting about why it feels better). Hesita-
tion reflects relatively enduring states, which are therefore good can-
didates for contributing their representations to consciousness. The
nature of the brain state suffices to explain why it is represented in
consciousness. There is no need to credit “consciousness” with provid-
ing help.
Do animals act automatically, and therefore remain unaware of
what they are doing? Humans are only unaware of their automatic
actions when their attention is focused elsewhere, and the same
should be true of other animals. Also, behaviors do not automatize
instantly, but only as an end point of practice while dealing with rela-
tionships that are not yet fully understood. During practice, the indi-
vidual is very aware of what she is doing. Only later does the range
of cue utilization narrow to the point that little cortical territory is
implicated and attention is free to move elsewhere.
SELF-AWARENESS
Instead of being self-aware because conscious, the individual is con-
scious because he is engaged in being self-aware, which his brain does
by forming the corresponding dominant configuration. Instead of be-

ing a unique category of mental function, which an animal either has
at its disposal or does not, self-awareness becomes an accomplishment
of flexible and differentiated representing. It demands the ability to
disengage some of the neural network from stimulus control and to
shape its pattern of functioning so as to abstract from the remembered
experiences, the continuity of the invariant experiencer. The invariant
presence of one’s background body sensation may be a minimal orga-
nizing principle (Kinsbourne, 1995). Phenomenal self-awareness does
not make these operations possible. The reverse is the case. When the
mental operations are performed, they are attended by subjective self-
awareness.
I understand self-awareness to refer to an awareness of one’s per-
son as an independent experiencer and agent. This characterization,
however, is spuriously definite. Self-awareness is a matter of degree.
Its most explicit and crystallized state should be considered an end
point of a continuum that emerges from infant and animal experience.
Self-reflection appraises one’s attributes, in the context of environmen-
tal opportunities, requirements, and threats. The metacognitive abili-
ties that allegedly are predicated on the presence of self-awareness,
such as predicting one’s ability to perform a task and recollecting an
event from the past, do not necessarily call for more than minimal
explicit self-awareness. An unstructured sense of oneself as a presence
or an agent may be enough, if not more than enough.
I have suggested that self-awareness originates in the invariant per-
ceived presence of the body as somatosensory background to the
infant’s every experience and action (Kinsbourne, 1995, 2002). With
continuing human development, this diffuse feeling of continuity is
gradually elaborated into the adult’s flexible and critical self-aware-
ness, capable of being abstracted from the physical person. In animals,
self-awareness might only exist in the rudimentary body-centered
form. What is required? Sufficient computing power, that is, a suffi-
ciently extensive neural network with enough inhibitory capacity to
prevent the incipient self-construct from collapsing into undifferenti-
ated merging with the experience of the moment. Of course, at those
times when the relevant brain states are overshadowed by other ongo-
ing mental representational work, our self-consciousness will give
way to an intense engagement in the action at hand. But when we are
disengaged from interaction with the world, we can reflect on our-
selves. When we do so, the neural states that instantiate these repre-
sentations are likely to be sufficiently activated long enough to make
them viable candidates for contributing to ongoing consciousness.
The same applies to time estimation. When attention is fully en-
gaged in a task, the representations on which one relies to estimate
elapsed time are displaced from consciousness. As a result, durations
that are filled with effortful activity are underestimated (Hicks, Miller,
& Kinsbourne, 1976). We can attend interactively to the world, or in-
tra-actively to our own states. To achieve the latter, we need only
detach from the former. The thoughts, images, and fancies then come
unbidden. Similarly, episodic memories, though sometimes evoked
deliberately and with effort, at other times seem to drift into aware-
ness spontaneously.
When animals behave as a human would when she is self-aware,
would the animals also be self-aware (chapter 10, this volume), or
could they be behaving in the same way but with the use of implicit
processes (chapter 12, this volume)? If one regards consciousness as a
separate modular acquisition, then the question is reasonable, though
probably unanswerable. But based on the model of consciousness that
I have explained, there is no reason why the animal could not be self-
aware to some extent. The dominant configuration of neural pro-
cesses, that is, the one that outdoes its rivals in amplitude and dura-
tion, is ipso facto conscious. That applies whether the task is hard or
easy in an absolute sense, and whether the brain is large and complex
or smaller and simple. In this view, if the animal is using controlled
processes that in humans are conscious, then they are conscious in the
animal. Similarly, when animals retrieve information based on unique
events over long-term retention intervals (cf. chapter 9, this volume),
are they subjectively remembering these events as we human adults
would? If one dichotomizes between recollecting that requires con-
sciousness and learning that does not, then one can deny the former
to any species at will. If one regards episodic remembering as reinstat-
ing, in part, a previous brain state, then there is no reason to suppose
that this is exclusive to humans. Consciousness as such does nothing.
It is a result of the brain state, not its cause. So these animals might
well remember consciously, as we do. Like us, by remembering they
have demonstrated that they have the representational power to de-
tach (“decenter”) their attention from the current stimulus situation
and reactivate elements of a brain state from the past. Since they were
presumably conscious when they had the experience, why would they
not consciously recollect it later? They recollect by generating “domi-
nant configurations” that have awareness as an inherent and auto-
matic attribute. In this view, consciousness is not a critical condition
for, but integral to, episodic remembering.
In the uncentered brain model, there is no need to dichotomize
creatures into those that own an autonoetic module and perhaps
therefore an episodic memory module and those who do not. Ani-
mals’ varieties of self-awareness are no doubt not as explicit and
finely differentiated as those of mature humans. What they can ac-
complish in these domains must vary enormously between species
and among ontogenetic levels within species. However, nonhuman
animals could even become aware of the passing of time when it is

adaptively relevant. The tiger lurks at the watering hole. The prey
happens to be otherwise engaged. Finally the tiger picks up and re-
tires for the night. Does her mental state reflect that a long time has
passed (other than through the dimming of the light or the growling
of her stomach)? Not in terms of “how many hours” and “how bor-
ing” perhaps, but at a simple level, why not?
SUMMARY
I have advocated a continuity approach to interspecies differences in
self-awareness, which I have related to an uncentered model of con-
sciousness. This model permits representational content anywhere in
the brain to become conscious, by participating in a dominant config-
uration of the cerebral activation manifold. Awareness, consciousness,
and self-consciousness differ only in degree, and the concept of a spe-
cial-purpose module that implements self-awareness is superfluous.
Metacognition requires escape from stimulus control, adequate work-
ing memory, and operating within appropriate boundary conditions.
Any individual, human adult, child, animal, who can hold multiple
representations in working memory can be self-conscious. Contrary
to Donne’s faith-based conviction about what the “beast” knows or
does not know, for all we know, some beasts know that they know.
REFERENCES
Beck, B. (1976, June). Predatory shell dropping by herring gulls. Paper presented
to the Animal Behavior Society, Boulder, Colorado.
Braitenberg, V., & Schuz, A. (1992). Basic features of cortical connectivity and
some considerations on language. In J. Wind, B. Chiarelli, B. H. Bichakjian,
A. Nocentini, & A. Jonker (Eds.), Language origins: A multidisciplinary ap-
proach (pp. 89–102). Dordrecht: Kluwer.
Case, R. (1974). Structures and strictures: Some functional limitations on the
course of cognitive growth. Cognitive Psychology, 6, 544–573.
Dennett, D. (1981). Brainstorms. New York: Basic Books.
Dennett, D., & Kinsbourne, M. (1992). Time and the observer: The where and
when of consciousness in the brain. Behavioral and Brain Sciences, 15, 183–
247.
Estes, W. K. (1972). An associative basis for coding and organization in mem-
ory. In A. W. Melton & E. Martin (Eds.), Coding processes in human memory
(pp. 161–190). Washington, DC: Winston.
Evans, P. D., Anderson, J. R., Vallender, E. J., Gilbert, S. L., Malcolm, C. M.,
Dorus, S., & Lahn, B. T. (2004). Adaptive evolution of ASPM, a major deter-
minant of cerebral cortical size in humans. Human Molecular Genetics, 13,
489–494.

Hicks, R. E., Miller, G. W., & Kinsbourne, M. (1976). Prospective and retro-
spective judgments of time as a function of amount of information pro-
cessed. American Journal of Psychology, 89, 719–730.
Hofstadter, D. R. (1985). Metamagical themas. Toronto: Bantam.
Hofstadter, D. R., & Dennett, D. C. (1981). The mind’s I. Cambridge, MA: MIT
Press.
Kinsbourne, M. (1967). Sameness-difference judgments and the discrimination
of obliques in the rat. Psychonomic Science, 7, 183–184.
Kinsbourne, M. (1988). Integrated cortical field model of consciousness. In A. J.
Marcel & E. Bisiach (Eds.), The concept of consciousness in contemporary sci-
ence (pp. 239–256). London: Oxford University Press.
Kinsbourne, M. (1995). Awareness of one’s body: A neuropsychological hy-
pothesis. In J. Bermudez, A. J. Marcel, & N. Eilan (Eds.), The body and the
self (pp. 205–223). Cambridge, MA: MIT Press.
Kinsbourne, M. (1996). What qualifies a representation for a role in conscious-
ness? In J. D. Cohen & J. W. Schooler (Eds.), Scientific approaches to con-
sciousness (pp. 335–356). Hillsdale, NJ: Lawrence Erlbaum.
Kinsbourne, M. (2002). The brain and body awareness. In T. Pruzinsky & T.
Cash (Eds.), Body images: A handbook of theory, research and clinical practice
(pp. 22–29). New York: Guilford.
Kohler, W. (1927). The mentality of apes. New York: Harcourt Brace.
Lashley, K. S. (1938). The mechanism of vision: XV. Preliminary studies of the
rat’s capacity for detail vision. Journal of General Psychology, 18, 123–193.
Rozin, P. (1975). The evolution of intelligence and access to the cognitive un-
conscious. Progress in Psychobiology and Physiological Psychology, 6, 245–280.
Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human in-
formation processing: II. Perceptual learning, automatic attending and a
general theory. Psychological Review, 84, 127–190.
chology, General, 127, 227–250.
6
Humans as Applied Motivation Scientists:

Self-Consciousness From “Shared Reality”
and “Becoming”
E. Tory Higgins
How did self-knowing consciousness begin? One version of this ques-

tion is: What caused self-knowing consciousness to appear in the first
place? Like other evolutions, it is possible that it was simply an acci-
dent. Whatever the answer to this version of the question, what I want
to consider is the following: Once self-knowing consciousness did ap-
pear, why did it remain? What were the functional pressures on its
further development? In brief, I believe that self-consciousness has
significant advantages for goal attainment and need satisfaction in hu-
man society. Self-consciousness is a basic motivational tool for getting
along with others.
A key characteristic of humans is that they are highly dependent
on conspecifics for survival. More than any other animal, human chil-
dren need adults to provide them with nurturance and security for
many years. Even as adults, humans as individuals are relatively help-
less and survive by working together with other humans. This depen-
dency means that humans must get along with others to survive. To
a large extent, personal agency is social agency. A human needs to
predict and influence what others do. This can be done more effec-
tively by understanding how one fits into one’s social world. Self-
consciousness is a necessary ingredient for such understanding.
This chapter takes a motivational perspective on self-consciousness.
I propose that self-consciousness has remained and developed among
humans because it serves basic self-regulatory functions. I propose
that humans are applied motivation scientists. I emphasize two functions
in this chapter—shared reality and becoming. In brief, self-regulation in
a social context involves understanding what significant others hope
for and expect of you (shared reality), and understanding where you
157
are now, have been, and plan to be in relation to these hopes and
expectations (becoming). Both shared reality and becoming are central
features of human self-regulation, and it is self-consciousness that
makes them possible. The “self digest” model of self-regulation (see
Higgins, 1996a; Higgins & May, 2001) forms the underpinnings of my
motivational perspective on self-consciousness. I begin, then, with a
review of this model.
THE SELF DIGEST

The self-regulatory perspective taken here shares the common as-
sumption that human survival requires adaptation to the surrounding
environment, especially the surrounding social environment. In order
for children to obtain the nurturance and security they need to sur-
vive, they must establish and maintain relationships with caretakers
who fulfill these needs (see Bowlby, 1969, 1973). To establish and
maintain such relationships, children must learn how their appear-
ance and behaviors influence caretakers’ responses to them as objects
in the world. Children must learn to regulate their appearance and
behaviors in line with these interpersonal contingencies to increase
the likelihood that caretakers will provide them the nurturance and
security they need (cf. Bowlby, 1969; Cooley, 1902/1964; Mead, 1934;
Sullivan, 1953).
To survive and adapt to their environment, people must also learn
about their personal capabilities and, more generally, about their own
strengths and weaknesses as one kind of object in the world. Children
need to learn, for example, that certain activities, such as flying like a
bird, are beyond their capabilities and to attempt such activities is
dangerous. They must learn their limits, such as how long they can
hold their breath underwater or stay outside in freezing weather.
When choosing between alternative means to the same goal, it is
adaptive to know which means best suit one’s capabilities. When
working to attain a goal, it is adaptive to know how much effort or
personal resources should be expended, given one’s capabilities, to
attain the goal without wasting resources. Thus children must also
learn to regulate their activity engagement to fit their personal attri-
butes.
In previous works, I have introduced the notion of a self digest (Hig-
gins, 1996a; Higgins & May, 2001). A digest summarizes a body of
information, especially contingency rules and conclusions. A digest
serves regulatory functions. The notion of a self digest is meant to
capture the idea that self-knowledge summarizes information about
oneself as an object in the world in order to serve self-regulatory func-
tions. A self digest, then, is intended to highlight a new conceptualiza-
Humans as Applied Motivation Scientists 159
tion of the nature of self-knowledge—a summary of what the world is

like in relation to oneself.
It is proposed that what a person stores about himself or herself as
a distinct object in the world depends on what information is useful
for self-regulation. It was noted earlier, for example, that survival de-
pends on learning about those attributes of oneself in relation to the
world that produce beneficial interpersonal interactions and activity
engagements. The self digest represents knowledge about oneself as
an object in the world because, and to the extent that, such knowledge
facilitates adaptation to one’s environment. The self digest is a tool
for survival. It summarizes a person’s relations to the world and the
personal consequences of these relations. It increases the likelihood of
effective and efficient self-regulation in the service of survival. It is a
handy sourcebook for people about their person-environment fit that
helps them to fulfill their needs when interacting with their world.
The conceptualization of self-knowledge as a self digest involves
no claim that the cognitive properties of self-knowledge are different
from stored knowledge about other kinds of objects in the world. In
addition, there is no claim that only self-knowledge contains func-
tional information about the world. On the other hand, only knowl-
edge about oneself as an object in the world is essential to represent
because self-knowledge concerns the only object in the world that a
person must continually regulate in order to survive. Thus, self-
knowledge has a unique functional status among our stored knowl-
edge of objects in the world. Conceptualizing self-knowledge as a self
digest is meant to capture this unique functional status of self-knowl-
edge. In addition, the notion of a self digest is intended to emphasize
the self-regulatory nature of self-knowledge rather than the self-
descriptive nature of self-knowledge that has been emphasized in the
previous literature.
The actual self as one’s own beliefs about one’s stable properties as
a distinct object in the world has received more attention than any
form of the actual self in classic theories of the self (e.g., Epstein, 1973;
Lecky, 1945; Rogers, 1951; Sarbin, 1952; Snygg & Combs, 1949). The
philosopher Ryle (1949) suggested that people acquire self-knowledge
about their own personal characteristics, what they think they are,
through self-observation and inference. Bem (1967) and Andersen
(1987) have made similar suggestions (see also Markus, 1980). Self-
knowledge as self-description answers the question, “Who am I?” In-
deed, this kind of self-knowledge is explicitly elicited with “Who are
you?” and “Describe yourself” measures, such as McGuire’s “Tell me
about yourself” measure (e.g., McGuire & Padawer-Singer, 1976). In
contrast, self-knowledge as self digest answers the question, “What is
the world like in relation to me?”
Self-knowledge as self digest would contain quite different kinds

of information than self-knowledge as self-description. For example,
information about how others respond to you when you fail to be the
kind of person they want you to be would not be part of self-knowl-
edge as self-description but would be part of self-knowledge as self
digest. On the other hand, information that described oneself, such as
some physical characteristic, would be included as part of self-knowl-
edge as self-description but would not be included in self-knowledge
as self digest unless it was useful for self-regulation. In addition, self-
knowledge as self-description involves representations of decontextua-
lized “facts” about oneself in isolation, whereas self-knowledge as self
digest involves representations of oneself in relation to other persons
and activities in the world and the consequences of these relations.
The novel proposal of the self digest is that there is not a single
actual self involving a descriptive representation of personal attributes
but, rather, there are three different actual self-representations serving
three distinct self-regulatory functions—an instrumental self, a moni-
tored self, and an expectant self. Each of these self-regulatory functions
occurs in a social context and requires taking into account other peo-
ple’s judgments, hopes, and wishes for oneself. It requires developing
a shared reality with others, which in turn requires the development
of different levels of self-consciousness. Let us now consider the na-
ture of this development.
SHARED REALITY AND SELF-CONSCIOUSNESS

A fundamental characteristic of people is that they derive meanings
from the events in their lives and then respond to those meanings.
This insight into the nature of people was clearly developed at the
turn of the twentieth century in the work of Weber, Durkheim, Freud,
James, Cooley, and others. It is nicely captured in Thomas and Thom-
as’s (1928) statement, “If men define situations as real, they are real in
their consequences.” In social psychology as well, Asch (1952), Lewin
(1935), and Mead (1934) early on emphasized the importance of the
psychological situation in determining people’s motivations and ac-
tions. But what is the source of these psychological situations? Where
do the meanings assigned to life events come from? Is there a source
of meaning assignment that can account systematically for why peo-
ple from different families and different cultures respond so differ-
ently to similar events in their lives?
A major source of psychological situations, and one that produces
systematic variation in meaning assignment, is our sense of how the
world works that social interactions provide. The ways that others
respond to us communicate about the world around us and our place
in it—messages about our person-environment fit. By accepting these
messages about who we are in the world, we create a shared reality

with the people in our lives (see Hardin & Higgins, 1996; Higgins,
1996b).
Almost a century ago, the eminent social scientist Max Weber
(1967, pp. 156–157) stated, “In ‘action’ is included all human behavior
when and in so far as the acting individual attaches a subjective mean-
ing to it. . . . Action is social in so far as, by virtue of the subjective
meaning attached to by the acting individual (or individuals), it takes
account of the behavior of others and is thereby oriented in its
course.” Thus, action is “social” when its meaning and orientation
takes account of other people. Self-regulation develops from social
regulation, from children learning to take other people into account.
From this perspective, self-regulatory responses are social actions.
Socialization involves individuals adopting the values and beliefs
of their group, and when it is effective, people end up wanting to
behave as others want and expect them to behave. As Jones and Ge-
rard (1967, p. 77) point out, the “fascination stems from the fact that
most of society’s codes and values become part of the very fabric of
an individual’s personality during the process of socialization.” This
is, indeed, a fascinating issue, and it has been central to the work of
Freud, Lewin, Murray, Vygotsky, Mead, and other founders of social-
motivational science in psychology.
How does it all begin? To discuss the changes in self-regulation that
develop a shared reality with others and increasing self-consciousness, I
rely on the classic developmental literature that has described general
shifts in children’s ability to get along in the social world (e.g., Case,
1985; Damon & Hart, 1986; Fischer, 1980; Selman, 1980; see also Hig-
gins, 1989, 1991; chapter 4, this volume).
Six to Twelve Months

By the end of the first year of life, children can represent the relation
between two events, such as the relation between two successive re-
sponses their mother produces or the relation between a response that
they produce and their mother’s response to them (see Case, 1985).
Children can produce and interpret communicative signals and antici-
pate the occurrence of some event and, thus, are capable of the prelim-
inary form of role taking that Mead (1934) described—the ability to
anticipate the responses of a significant other with whom one is inter-
acting.
During this initial period, children begin to acquire the instrumental
self. Actual self-knowledge serving an instrumental function repre-
sents one’s self-attributes in relation to the responses they elicit from
other people:
Something about me → How others respond to me

This is the first step in acquiring interpersonal contingencies that will

become increasingly complex. This preliminary instrumental function
is not unique to humans. Certainly chimpanzees, for example, learn
observable behavioral contingencies concerning their behavior and
the responses of others to them (see Povinelli, 2000). What is unique
to human children during this period is social referencing or “shared
attention,” in which a child and an interaction partner will point out
to each other some interesting object or event in the immediate envi-
ronment and together pay attention to it (see Tomasello, 1999).
This joint attentional interaction provides an early example of the
distinctive human motive for shared reality. Other animals will be-
have in ways that get others to do something they want, which in-
cludes getting their attention. Such behaviors are in the service of
agency. But the joint attentional interaction of human children and
their partners is in the service of understanding. It is this sharing of an
understanding that is unique to humans. It is reflected in the unique
way in which human adults actively instruct children about how the
world works, which is the basis for cultural evolution (see Tomasello,
1999). It permits a kind of imitative learning that is unique to humans
in that the child and adult are both motivated to have a common
understanding of how to do something. It is not copying or mimick-
ing an observed behavior for the purpose of agency. It is collaborating
in order to share an understanding of how the world works and how
to get along in it (see also Pittman & Higgins, 2002).
Sixteen Months to Two Years

During the second half of children’s second year, there is a major shift
in children’s ability to represent events, which has been traditionally
associated with the emergence of symbolic representation (e.g.,
Bruner, 1964; Case, 1985; Fischer, 1980; Piaget, 1951; Werner & Kaplan,
1963). Children can now represent the bidirectional relationship be-
tween themselves and another person as an interrelation between two
distinct mental objects, self-as-object and other-as-object (see Berten-
thal & Fischer, 1978; Harter, 1983; Lewis & Brooks-Gunn, 1979). Their
ability to represent themselves and other persons as distinct objects in
the world is reflected in their use of the pronouns “I” and “you.”
Children can now represent the relation between two relations: (a) the
relation between a feature of themselves (e.g., their action, response,
physical appearance) and a particular response by another person,
and (b) the relation between a particular response by another person
and a particular state that they will experience. Children, for example,
can represent the relation between their making a mess at mealtime
and their mother’s frowning, yelling, or leaving, and the relation be-
tween their mother’s frowning, yelling, or leaving and their experienc-
ing a negative state.
This higher level of perspective taking that appears by 2 years of

age is critical for what sociologists call symbolic interaction—the ability
to anticipate the responses of others to their actions and the personal
consequences of these social responses (see Stryker & Statham, 1985).
Children’s significant others link them to the larger society by provid-
ing the shared reality or social meaning of the child’s features (i.e.,
societal responses to the child’s features), as well as the importance of
the child’s features (i.e., the state produced in the child by these soci-
etal responses). In this way, children learn the interpersonal signifi-
cance of their self-features.
The level of self-consciousness that children possess by 2 years of
age can be used by children in a more sophisticated instrumental
means-end fashion to regulate their actions, responses, or appearance.
By representing the interpersonal significance of their self-features and
using this information to plan their actions, children are better able to
delay gratification, to control their momentary impulses, and to free
themselves from the demands and forces of the immediate situation
(see Miller & Green, 1985; Mischel & Moore, 1980; Mischel & Patter-
son, 1978).
Three to Six Years

From a motivational perspective, probably the most dramatic shift oc-
curs between 3 and 6 years of age, the classic change from “egocen-
tric” to “non-egocentric” thought (see Feffer, 1970; Flavell, Botkin, Fry,
Wright, & Jarvis, 1968; Piaget, 1965; Selman & Byrne, 1974; Werner,
1957). Children can now coordinate two separate systems of interrela-
tions, such as coordinating observable responses with unobservable
responses (see Case, 1985), which underlies inferences about the
thoughts, expectations, motives, and intentions of others (see Shantz,
1983).
Children now understand that other people have different attitudes
about different types of responses, that they prefer some types of re-
sponses over others. They understand that performing the types of
responses preferred by others is related to others’ positive responses
to them, and thus they are motivated to learn the types of responses
that are preferred by their significant others. In addition to learning
about a significant other’s preferences from his or her personal reac-
tions to them, children also learn by observing the significant other
react to others (i.e., observational learning; see Bandura, 1977, 1986).
For example, children can observe how their mother reacts to their
sibling’s responses and thereby infer which types of responses their
mother prefers (see Bandura, 1986; Bandura & Walters, 1963).
The critical development during this period is that children learn
that the relation between one of their self-features and another per-
son’s response to it depends on the other person’s viewpoint or stand-
point on this feature. A standpoint on the self is defined as a point of

view from which a person can be judged that reflects a set of attitudes,
opinions, or values (see Turner, 1956). For example, they can now
represent the fact that it is the discrepancy between their behavior and
the behavior that their mother prefers that underlies the association
between their behavior and their mother’s response to it (see Higgins,
1989). That is, children now understand that social regulation of them is
mediated by others’ standpoints on them.
This is a discovery of shattering importance. Children at this age
are still totally dependent on older conspecifics for survival. It is clear
to them that these older people are different from them, especially in
their possession of essential resources that the children want to be used
for them and not against them. Their highest priority is to figure out
what determines how these resource-rich persons allocate their re-
sources. Children want to tame these powerful animals called Mommy
and Daddy by learning what motivates them.
The social-cognitive literature (in both developmental and social
psychology) for decades has emphasized people’s motivation to ex-
plain the behaviors of others (“why” questions), such as dispositional
versus situational explanations. Although explaining others’ actions is
important, I believe that the major concern of both children and adults
is to figure out what resources a target person has and whether they
are likely to be used for or against them. These resources extend well
beyond the typical traits that have been studied in the classic litera-
ture. They include the observable status and power of a person that
is reflected in how family members and visitors treat that person.
Rather than causal explanations for others’ behaviors, children are try-
ing to understand the who, what, and where of resource allocation.1 Who
makes the final decisions? Whose word is law? Does this vary by
activity? For those things I want to happen, such as getting a new toy,
who is in charge of what? For things I don’t want to happen, such as
getting punished, who is in charge of what? For things I enjoy doing,
such as playing, who is a good playmate for which activity? For those
things I don’t enjoy doing, such as going to bed early, who allows me
not to do which activity? I need to figure out who has which resources
and then figure out what motivates them to use these resources for
me rather than against me.
By the age of 6, children are much more masters of their universe
than before. They have figured out that other people have wishes,
hopes, expectancies, and demands of them. Other people like and dis-
like different things than they do. By taking others’ motivational orienta-
tions into account—a new and critical kind of shared reality—children
can plan their actions to increase the likelihood that others will use
their resources for them and not against them. This increasing ability
to take others’ motivational orientations into account is also demon-
strated in children’s increasing ability to cooperate and collaborate

with one another in a way that is not found in other animals, includ-
ing primates (see Povinelli, 2000; Tomasello, 1999).
By taking into account others’ standpoints on them, children de-
velop a new kind of self-consciousness and a new kind of actual self-
function. These standpoints become standards of self-evaluation—
self-guides (see Higgins, 1989, 1991). Children now use self-guides as
a basis for self-evaluation by assessing the amount of congruency or
discrepancy between their current self-state and the end state that a
significant other desires or demands of them (e.g., “How am I doing
in relation to what father wants or demands of me?”). They can then
respond to any perceived discrepancy by taking action to reduce it.
They can also plan action by taking potential discrepancies into ac-
count, thereby using prospective self-evaluation in the service of self-
regulation (see also Bandura, 1986; Carver & Scheier, 1990).
This self-evaluative process creates a new actual self-function—the
monitored self. The monitored self is an actual self-representation that
provides self-regulatory information regarding one’s current state in
relation to some desired end state, which together provide feedback
about how well one is doing in meeting standards and pursuing de-
sired goals:
Something about me → How well am I doing?
For young children, the desired end state is typically a shared reality
with a significant other about what the child will hopefully accom-
plish (ideals) or what are the child’s duties and responsibilities (oughts).
Eight to Ten Years

Between 8 and 10 years of age, children become capable of coordinat-
ing values along two distinct dimensions (see Case, 1985; Fischer, 1980;
Piaget, 1970). When comparing their performance to that of another
child, for example, children can now consider simultaneously the dif-
ference in actors’ outcomes and the difference in actors’ effort, thus
making ability inferences possible (see Ruble, 1983). To be good at
something now means to be better than others at something that is
worthwhile, something that matters, as defined by a represented stan-
dard of value. Now the value of a target of comparison, such as per-
formance on some task, depends on its relation to a represented goal
or self-guide. Being “bad” or “good” at something thus acquires new
significance.
The social comparison processes that develop during this period
are a new way in which children take into account other people in
their self-regulation. Through social comparison, children learn what
is distinctive about themselves in relation to the world around them.
They learn what to expect about their own reactions to the world that
need not be the same as other people’s reactions. When doing certain
things, not everyone has the same experience. By comparing their ex-
perience of an activity to the experience of other children, a child
learns what the world is like for them in particular. The expectant self
represents this relation between self-attributes and activity experi-
ences:
Something about me → How I experience the world
Whereas the instrumental actual self answers the question “How

does the world respond to me?,” the expectant actual self answers the
question “How do I respond to the world?” The expectant self in-
cludes personal preferences as well as personal skills or talents. The
expectant self also develops from a shared reality. Significant others,
both adults and peers, contribute to the construction of the expectant
self by pointing out to children what is special and distinctive about
them. Those distinctive self-attributes that are positive, such as talents,
are perceived as resources by children’s significant others. The signifi-
cant others support (or pressure) the children to develop their talents
further, such as spending hours on some artistic or athletic activity.
The actual self-representations during this period also develop into
more global forms that refer to being a general type of person rather
than a particular kind of behavior—being an “honest person” rather
than just behaving honestly, being a “dancer” rather than just danc-
ing. Children not only can evaluate a particular action they just per-
formed by considering another person’s likely response to it; they can
also evaluate how the general traits they believe they possess relate
to the general traits that significant others desire or expect them to
possess. That is, they can now evaluate their success or failure at being
the “type of person” that others desire or expect them to be.
Thirteen to Fifteen Years

From a motivational perspective, a major development that occurs be-
tween 13 and 15 years of age is that children can now simultaneously
interrelate different perspectives on themselves as objects, including
conflicting perspectives on themselves (Case, 1985; Fischer, 1980;
Fischer & Lamborn, 1989; Inhelder & Piaget, 1958; Selman & Byrne,
1974). For the first time, two distinct systems of self-guides, such as
one system involving a peer standpoint and another system involving
a parental standpoint, can themselves be interrelated.
A related development during this period is that children become
capable of constructing their own standpoint, distinct from the stand-
point of others, which can function as the integrated solution to the
complex array of alternative standpoints on them. Being conscious of
having your own standpoint on yourself that is distinct from others’

standpoint on you is another level of self-consciousness. With the de-
velopment of a personal “own” standpoint, children can experience
some personal self-guides that are shared with their parents’ stand-
point on them (a high level of identification), other personal self-
guides that are not shared with their parents’ standpoint (indepen-
dent self-guides), and parental standpoints that are not shared with
the children’s own self-guides (introjects or the “felt presence of oth-
ers”; see Moretti & Higgins, 1999).
In sum, there are progressive shifts from birth to adolescence in
how children take other people into account in order to get along in
their social world. With each shift in shared reality, a new level of
self-consciousness appears. The instrumental self, monitored self, and
expectant self components of the self digest are formed from these
developments in shared reality and self-consciousness. Each of these
components of the self digest provides children with an essential mo-
tivational tool for adapting to their enviroment. These motivational
functions support, indeed press for, the maintenance and develop-
ment of different levels of self-consciousness. The next section dis-
cusses another motivational support for (or pressure on) developing
self-consciousness—“becoming.”
BECOMING AND SELF-CONSCIOUSNESS

Imagine the following scenario. A young animal is rewarded for per-
forming a certain behavior. But after a while, the young animal is no
longer rewarded for performing that behavior. In classic terms, one
would say that the animal now experiences a period of extinction trials
for that behavior. When the young animal performs a behavior that
takes more effort or greater skill than the earlier behavior, it is that
behavior which is now rewarded. But after a while, that new behavior
is also no longer rewarded. For most young animals, this would be
very confusing. Indeed, it is not clear whether any animal could figure
out the reinforcement contingency, with one key exception—human
children.
To my knowledge, this kind of “improvement” or “progress” rein-
forcement contingency has not been explicitly investigated with either
human children or young animals. Exactly when human children are
able to figure out this kind of reinforcement contingency has not been
studied. What is clear is that the development of a monitored self
described earlier, which occurs between 3 and 6 years of age, typically
involves such contingencies. As described earlier, during this period
children figure out that parents (and others) have wishes, hopes, ex-
pectancies, and demands of them. But often what parents hope and
expect of children is that they progressively become something that
the parents have in mind for them. The parents want their child to
improve, to make progress. The parents respond to children depend-
ing on whether they are or are not successful in “becoming,” over
time, what the parents have in mind. Thus, in order to control paren-
tal responses, children need to discover how the “becoming” contin-
gency works. And, indeed, they do! They demonstrate (show off?) to
parents what they can do now that they could not do earlier.
The monitored self provides feedback on how well one is doing
in meeting the goals and standards (self-guides) held by significant
others. The discovery of “becoming” contingencies creates a new
level of self-consciousness for the monitored self. Children need to
be conscious of where they are now, have been, and plan to be in
relation to these self-guides or desired end states. They now evaluate
themselves in terms of their success or failure in becoming. The posi-
tivity of the current self does not depend simply on whether it is
congruent or discrepant from a self-guide but also on whether it
is an improvement or progress from the past self (see Carver &
Scheier, 1990). But if children want in the future to have a current
self that is an improvement from their past self, then they have to
plan in the present to have a future self that is better than their
current self.
Contingencies of becoming, therefore, create motivational demands
on children’s self-consciousness to interrelate the past, present, and
future. Contrary to what is necessary for most self-regulation, self-
regulating improvement or progress requires not simply knowing in
the present what you are like now but knowing how you were in the
past and how you want to be in the future. To compare the current
self with the past self requires having specific memories of what you
were like in the past. It is no coincidence, then, that the developmental
literature reports that it is during this age 3 to 6 period that children
develop narrative self-consciousness and episodic memory (see chap-
ter 4, this volume; Tulving, 1983).
Self-consciousness of becoming is also involved in young children’s
social modeling. Children at this age are fascinated with other chil-
dren who are older than them, but not too much older. If children
were simply interested in observing advanced skills, they would be
as interested or more interested in watching adults than older chil-
dren. But they are not. Their interest in older children is because they
are fascinated with where they are going. They understand that they
have improved from the past and watch somewhat older children to
predict where they are going in the future.
Like self-consciousness in general, self-consciousness of becoming
also develops further over time. As discussed earlier, between 8 and
10 years of age children become capable of coordinating values along
two distinct dimensions, such as considering simultaneously the dif-
ference in actors’ outcomes and the difference in actors’ effort to make

ability inferences possible. Such coordination along distinct dimen-
sions can also be used by children in social comparison processes to
evaluate personal skills. Then they can compare their becoming with
the becoming of another child. “How am I doing?” becomes relative
to other peers. “Am I improving on this skill faster or slower than
other kids?” This level of self-consciousness contributes to the expec-
tant self because personal talent (or deficit) is defined less by what
you can do than by how quickly you improve relative to others.
In sum, along with the role of shared reality in providing a motiva-
tional push for developing self-consciousness, children’s use of be-
coming in their self-regulation provides an additional push. This push
involves children becoming self-conscious of others’ wanting them to
improve and make progress, which in turn requires interrelating their
present self with both their past and future selves.
CONCLUSION
Self-consciousness is a motivational tool. For children to self-regulate
in a social context, they need to understand what significant others
hope for and expect of them (shared reality) and understand where
they are now, have been, and plan to be in relation to these hopes and
expectations (becoming). Shared reality and becoming are essential
features of human self-regulation, and they both require the develop-
ment of advanced forms of self-consciousness.
Like all fundamental psychological principles, shared reality and
becoming have costs as well as benefits for humans. When individuals
share reality with multiple significant others and these significant oth-
ers have conflicting hopes or expectancies, the individuals suffer from
feeling confused, uncertain, and unsure of themselves (see Van Hook
& Higgins, 1988). When individuals have a significant other who
shares only some of their beliefs about who they are, they suffer from
alienation and loneliness (see Higgins, 1996b). Sharing reality with
their audience through message tuning causes communicators to mis-
remember their past to be more like what they said than what they
saw (see Higgins, 1992). Controlling for the current magnitude of dis-
crepancy with their hopes and aspirations (ideals), individuals who
believe that they are no closer to attaining their ideals now than they
were in the past suffer from depression-like symptoms (Strauman &
Higgins, 1993).
These and other costs need to be weighed against the obvious bene-
fits of shared reality and becoming. Without shared reality and be-
coming, there would be no human civilization. There would be no
language, no social roles and identities, no education, and no govern-
ments.
Perhaps what is unique about humans is not so much their use of

language or tools or instruction per se, but rather their motivation to
share reality with others about the past, present, and future. Perhaps
this motivation can account not only for the development of self-
consciousness, but also for humans’ use of language, tools, instruc-
tion, and so on. I began this chapter by stating that self-consciousness
is a basic motivational tool for getting along with others. The others
need not even be human. Humans are unique in trying to understand
the feelings and desires of other animals and in responding to them
in ways that cause them to be the resource humans want, whether as
vehicles, machines, weapons, food producers, or pets. This taming of
other animals to serve as resources has made a major contribution to
the development of human civilization.
During the twentieth century, the study of human motivation was
dominated by the notion that, motivationally, humans were funda-
mentally like other animals even though they had special cognitive
skills. But humans are not like other animals motivationally. They are
unique in being applied motivation scientists, constantly trying to
shape the behaviors of others by figuring out their hopes and expecta-
tions. To understand human self-consciousness and human cognition
more generally, we need to study the myriad implications of the fact
that humans, motivationally, are fundamentally different from other
animals.
ACKNOWLEDGMENTS The research reported in this chapter was supported by
National Institute of Mental Health Grant MH39429 to E. Tory Higgins.
NOTE
1. Even when children and adults do make dispositional inferences, it is often in

order to predict resource allocation rather than to predict behavior per se. Some
attributes, such as intelligence, allow inferences about the power of a person as a
resource (strength). Other attributes, such as the classic warm versus cold, allow in-
ferences about the valence of a person as a resource, that is, whether the person is
generally a positive (for) or negative (against) resource (direction). Still other attri-
butes, such as honesty, allow inferences about the certainty or consistency of a person
as a resource (stability).
REFERENCES
Andersen, S. M. (1987). The role of cultural assumptions in self-concept devel-
opment. In K. Yardley & T. Honess (Eds.), Self and identity: Psychosocial
perspectives (pp. 231–246). New York: Wiley.
Asch, S. E. (1952). Social psychology. Englewood Cliffs, NJ: Prentice-Hall.
Bandura, A. (1977). Social learning theory. Englewood Cliffs, NJ: Prentice-Hall.
Bandura, A. (1986). Social foundations of thought and action: A social cognitive

theory. Englewood Cliffs, NJ: Prentice-Hall.
Bandura, A. L., & Walters, R. H. (1963). Social learning and personality develop-
ment. New York: Holt, Rinehart and Winston.
Bem, D. J. (1967). Self-perception: An alternative interpretation of cognitive
dissonance phenomena. Psychological Review, 74, 183–200.
Bertenthal, B. I., & Fischer, K. W. (1978). Development of self-recognition in
the infant. Developmental Psychology, 14, 44–50.
Bowlby, J. (1969). Attachment and loss: Vol. 1. Attachment. New York: Basic
Books.
Bowlby, J. (1973). Attachment and loss: Vol. 2. Separation: Anxiety and anger.
New York: Basic Books.
Bruner, J. S. (1964). The course of cognitive growth. American Psychologist, 19,
1–15.
Carver, C. S., & Scheier, M. F. (1990). Origins and functions of positive and
negative affect: A control-process view. Psychological Review, 97, 19–35.
Case, R. (1985). Intellectual development: Birth to adulthood. New York: Aca-
demic Press.
Cooley, C. H. (1964). Human nature and the social order. New York: Schocken
Books. (Original work published 1902)
Damon, W., & Hart, D. (1986). Stability and change in children’s self-under-
standing. Social Cognition, 4, 102–118.
Epstein, S. (1973). The self-concept revisited: Or a theory of a theory. American
Psychologist, 28, 404–416.
Feffer, M. (1970). Developmental analysis of interpersonal behavior. Psycho-
logical Review, 77, 197–214.
Fischer, K. W. (1980). A theory of cognitive development: The control and
construction of hierarchies of skills. Psychological Review, 87, 477–531.
Fischer, K. W., & Lamborn, S. D. (1989). Sources of variation in developmental
levels: Cognitive and emotional transitions during adolescence. In A. de
Ribaupierre, K. Scherer, & P. Mounod (Eds.), Transition mechanisms in child
development: The longitudinal perspective. Paris: European Science Founda-
tion.
Flavell, J. H., Botkin, P. T., Fry, C. L., Wright, J. W., & Jarvis, P. E. (1968).
The development of role-taking and communication skills in children. New York:
Wiley.
Hardin, C., & Higgins, E. T. (1996). Shared reality: How social verification
makes the subjective objective. In R. M. Sorrentino & E. T. Higgins (Eds.),
Handbook of motivation and cognition: Vol. 3. The interpersonal context (pp.
28–84). New York: Guilford.
Harter, S. (1983). Developmental perspectives on the self-system. In P. H.
Mussen (Ed.), Handbook of child psychology: Vol. 4. Socialization, personality,
and social development (pp. 275–385). New York: Wiley.
Higgins, E. T. (1989). Continuities and discontinuities in self-regulatory and
self-evaluative processes: A developmental theory relating self and affect.
Journal of Personality, 57, 407–444.
Higgins, E. T. (1991). Development of self-regulatory and self-evaluative pro-
cesses: Costs, benefits, and tradeoffs. In M. R. Gunnar & L. A. Sroufe
(Eds.), Self processes and development: The Minnesota symposia on child psychol-
ogy (Vol. 23, pp. 125–165). Hillsdale, NJ: Lawrence Erlbaum.
Higgins, E. T. (1992). Achieving “shared reality” in the communication game:
A social action that creates meaning. Journal of Language and Social Psychol-
ogy, 11, 107–131.
Higgins, E. T. (1996a). The “self digest”: Self-knowledge serving self-regula-
tory functions. Journal of Personality and Social Psychology, 71, 1062–1083.
Higgins, E. T. (1996b). Shared reality in the self-system: The social nature of
self-regulation. European Review of Social Psychology, 7, 1–29.
Higgins, E. T., & May, D. (2001). Individual self-regulatory functions: It’s not
“we” regulation, but it’s still social. In C. Sedikides & M. B. Brewer (Eds.),
Individual self, relational self, collective self (pp. 47–67). Philadelphia: Psychol-
ogy Press.
Inhelder, B., & Piaget, J. (1958). The growth of logical thinking from childhood to
adolescence. New York: Basic Books.
Jones, E. E., & Gerard, H. B. (1967). Foundations of social psychology. New York:
Wiley.
Lecky, P. (1945). Self-consistency: A theory of personality. New York: Island Press.
Lewin, K. (1935). A dynamic theory of personality. New York: McGraw-Hill.
New York: Plenum.
Markus, H. (1980). The self in thought in memory. In D. M. Wegner & R. R.
Vallacher (Eds.), The self in social psychology (pp. 102–130). New York: Ox-
McGuire, W. J., & Padawer-Singer, A. (1976). Trait salience in the spontaneous
self-concept. Journal of Personality and Social Psychology, 33, 743–754.
Mead, G. H. (1934). Mind, self, and society. Chicago: University of Chicago
Press.
Miller, S. M., & Green, M. L. (1985). Coping with stress and frustration: Ori-
gins, nature, and development. In M. Lewis & C. Saarri (Eds.), The socializa-
tion of emotions (pp. 263–314). New York: Plenum.
Mischel, W., & Moore, B. (1980). The role of ideation in voluntary delay for
symbolically presented rewards. Cognitive Therapy and Research, 4, 211–221.
Mischel, W., & Patterson, C. J. (1978). In W. A. Collins (Ed.), Minnesota sympo-
sia on child psychology (Vol. 11, pp.199–230). Hillsdale, NJ: Lawrence Erl-
baum.
Moretti, M. M., & Higgins, E. T. (1990). Relating self-discrepancy to self-es-
teem: The contribution of discrepancy beyond actual-self ratings. Journal of
Experimental Social Psychology, 26, 108–123.
Moretti, M. M., & Higgins, E. T. (1999). Own versus other standpoints in self-
regulation: Developmental antecedents and functional consequences. Re-
view of General Psychology, 3(3), 188–223.
Piaget, J. (1951). Play, dreams and imitation in childhood. New York: Norton.
Piaget, J. (1965). The moral judgment of the child. New York: Free Press. (Origi-
nal translation published 1932)
Piaget, J. (1970). Piaget’s theory. In P. H. Mussen (Ed.), Carmichael’s manual of
child psychology (Vol. 1, 3rd ed., pp. 703–732). New York: Wiley.
Pittman, T. S., & Higgins, E. T. (2002). The human nature of agency and under-
standing. Unpublished manuscript, Princeton University.
Povinelli, D. J. (2000). Folk physics for apes: The chimpanzee’s theory of how the
world works. Oxford, UK: Oxford University Press.
Rogers, C. R. (1951). Client-centered therapy: Its current practice, implications, and
theory. Boston: Houghton Mifflin.
Ruble, D. N. (1983). The development of social comparison processes and
their role in achievement-related self-socialization. In E. T. Higgins, D. N.
Ruble, & W. W. Hartup (Eds.), Social cognition and social development: A
socio-cultural perspective (pp. 134–157). New York: Cambridge University
Press.
Ryle, G. (1949). The concept of mind. London: Hutchinson.
Sarbin, T. R. (1952). A preface to a psychological analysis of the self. Psycholog-
ical Review, 59, 11–22.
Selman, R. L. (1980). The growth of interpersonal understanding: Developmental
and clinical analyses. New York: Academic Press.
Selman, R. L., & Byrne, D. F. (1974). A structural-developmental analysis of
levels of role-taking in middle childhood. Child Development, 45, 803–806.
Shantz, C. U. (1983). Social cognition. In J. H. Flavell & E. M. Markman (Eds.),
Cognitive development, Vol. 3 in P. H. Mussen (Ed.), Carmichael’s manual of
child psychology (4th ed., pp. 495–555). New York: Wiley.
Snygg, D., & Combs, A. W. (1949). Individual behavior. New York: Harper and
Row.
Strauman, T. J., & Higgins, E. T. (1993). The self construct in social cognition:
Past, present, and future. In Z. V. Segal & S. J. Blatt (Eds.), The self in emo-
tional distress: Cognitive and psychodynamic perspectives (pp. 3–40). New
York: Guilford.
Stryker, S., & Statham, A. (1985). Symbolic interaction and role theory. In G.
Lindzey & E. Aronson (Eds.), Handbook of social psychology (Vol. 1, pp. 311–
378). New York: Random House.
Sullivan, H. S. (1953). The collected works of Harry Stack Sullivan: Vol. 1. The
interpersonal theory of psychiatry (H. S. Perry & M. L. Gawel, Eds.). New
York: Norton.
Thomas, W. I., & Thomas, D. S. (1928). The child in America. New York: Knopf.
Tomasello, M. (1999). The cultural origins of human cognition. Cambridge, MA:
Harvard University Press.
Press.
Turner, R. H. (1956). Role-taking, role standpoint, and reference-group behav-
ior. American Journal of Sociology, 61, 316–328.
Van Hook, E., & Higgins, E. T. (1988). Self-related problems beyond the self-
concept: The motivational consequences of discrepant self-guides. Journal
of Personality and Social Psychology, 55, 625–633.
Weber, M. (1967). Subjective meaning in the social situation. In G. B. Levitas
(Ed.), Culture and consciousness: Perspectives in the social sciences (pp. 156–
169). New York: Braziller.
Werner, H. (1957). Comparative psychology of mental development. New York:
International Universities Press.
Werner, H., & Kaplan, B. (1963). Symbol formation. New York: Wiley.
7
Two Normative Roles

for Self-Consciousness
Patricia Kitcher
Modern philosophers regarded self-consciousness (or capacities that

presupposed self-consciousness) as marking the fundamental di-
vide between humans and other animals; our project in this book
is to look for missing links between mere or first-order animal con-
sciousness and full-blown human self-consciousness. Seemingly,
arguments for links would be either analogical or compositional. Ani-
mals of Kind A possess something like some aspect of human self-
consciousness; animals of Kind B possess one or more of the capacities
required for some aspect of human self-consciousness. Both types of
argument are likely to encounter similar lines of rebuttal: the “self-
consciousness” of A animals is not very like human self-conscious-
ness; even if it is implicated in human self-consciousness, the capacity
of B animals does not enable them to have anything like human self-
consciousness.
Much light would be shed on the ensuing debate by a theory de-
scribing what some or several aspects of human self-consciousness
are like. I do not offer such a theory, but consider instead why self-
consciousness was regarded as so important—so quintessentially hu-
man—by modern philosophers. The hope is that if we have some
understanding of the important functions that self-consciousness per-
forms in human life, then we can appeal to those functions to provide
principled answers to questions about animal self-consciousness. Does
the capacity that has been proposed as a variety of animal self-conscious-
ness perform an analogous function in animal life? Of course, modern
philosophy may have been seriously in error about the roles of human
self-consciousness. Nevertheless, since the notion of self-consciousness
arose in the modern period, having a better understanding of what the
moderns took the functions of self-consciousness to be should illumi-
174
Two Normative Roles for Self-Consciousness 175
nate debates about “missing links” to human self-consciousness—even

if some or many of the modern views turn out to be incorrect. What
we would learn in that case is not whether some animal capacities are
very like human self-consciousness, but whether they are like what
human self-consciousness has traditionally been taken to be. Further,
when examining the motivations behind the traditional notion of self-
consciousness points to a hypothesis that is now known to be wrong
in the human case, that will also clarify the debate about animal self-
consciousness. For if an alleged feature of human self-consciousness
is rejected, then it can be ignored in debates about the aptness of an
analogy between human and animal self-consciousness.
The notion of self-consciousness had many sources in the modern
period, but I am going to focus on two central roles that it was
thought to play, one moral and one epistemological. As indicated in
my title, both roles are normative. This may seem to stack the deck,
since normativity itself is often regarded as uniquely human. I hope
the virtues of this approach will be evident in the remainder of the
chapter, but I offer three quick justifications now. First, even if full-
blown normativity only occurs among Homo sapiens, researchers are
also looking for animal links to moral behavior in human beings, so a
normative approach hardly forecloses the question of linkages. Sec-
ond, as the reader will see below, modern discussions of self-conscious-
ness were suffused with concerns about normativity. Under these cir-
cumstances, a strategy of trying to find animal links to aspects of
human self-consciousness that have no normative dimension (if there
are any) is very likely to prompt the criticism that the analogy is weak,
because something essential to human self-consciousness has been left
out. Finally, at least since Nicholas Humphrey’s (1988) seminal paper,
psychologists, anthropologists, and primatologists have been giving
serious consideration to the hypothesis that the main function of intel-
ligence in humans and animals is to navigate the complexities of the
social world. Thus the modern view that one aspect of intelligence—
the capacity to be self-conscious—enabled people to participate in
normative social contexts comports well with current directions in re-
search. The two roles I am going to consider, memory in moral re-
sponsibility and awareness of subjective limitations in what might be
termed “epistemic responsibility,”1 are fairly complex, and I have
space to consider only a few aspects of each here. Both my discussions
begin with Locke’s seminal treatments of the issues.
CONSCIOUSNESS OF A PERSONAL PAST

In An Essay Concerning Human Understanding, Locke (1690/1959) fa-
mously claimed that
since consciousness always accompanies thinking, and it is that which

makes every one to be what he calls self, and thereby distinguishes himself
from all other thinking things, in this alone consists personal identity . . .
and as far as this consciousness can be extended backwards to any past
action or thought, so far reaches the identity of that person; it is the same
self now it was then; and it is by the same self with this present one that
now reflects on it, that that action was done. (2.27.11)2
Although Locke’s understanding of “consciousness” has raised inter-

pretive controversies, the standard reading of this passage is that it
puts forward a memory criterion for personal identity: you are the
same person as someone who perpetrated a deed if and only if you
can remember doing the deed. Locke’s immediate successors objected
to both halves of this claim: memory was neither sufficient nor neces-
sary for sameness of person.
Locke was explicit about why perfect recall was essential. On Judg-
ment Day
everyone shall “receive according to his doings, the secrets of all hearts
shall be laid open.” The sentence shall be justified by the consciousness all
persons shall have, that they themselves, in what bodies soever they appear,
or what substances soever that consciousness adheres to, are the same that
committed those actions, and deserve that punishment for them. (2.27.26)
This may seem unpromising. One function of consciousness of past

deeds and thoughts is to bear witness to divine justice. Even those
subject to the torture of “every sense of the flesh and every faculty of
the soul” by fires fed by a brimstone “specially designed to burn for
ever and for ever with unspeakable fury”3 must at least admit that
they committed the offenses for which they are being punished. But
Locke also noted that “person” was a “forensic” term (2.27.26), so the
focus of concern is not just divine fairness, but the fairness of human
practices and attitudes. Society may fairly punish only where the per-
son’s own consciousness accuses him. And this forensic function of
self-consciousness persists to the present day. In 1992, then Governor
Clinton was criticized for not staying the execution of a man who
had so injured himself after sentencing that his lawyers claimed he
could not understand the connection between his crime and his pun-
ishment—even though he had been mentally competent during his
trial.
At first glance, Locke’s theory of extending consciousness back-
ward may seem to be a forerunner of Tulving’s hypothesis of “auto-
noetic” consciousness. Even the metaphors seem eerily similar: where
Tulving (chapter 1, this volume) writes of autonoetic consciousness
involving “time travel,” Locke (1690/1959) described consciousness
as being able to extend itself backward to past thoughts and deeds.
He also tried to clarify the idea of the temporal extent of a person by
an analogy with spatial extent. A person’s sense of himself stretches

over his whole body, in particular, to any part of whose good or harm
he can be conscious: “the limbs of his body are to every one a part of
himself; he sympathizes and is concerned for them” (2.27.11).4 So too
would a person’s sense of himself stretch back to any previous “tem-
poral part” of himself—say himself as a teenager—whose remem-
bered deeds were a source of sympathy or concern to present con-
sciousness (see 2.27.14, 16).
Still, there are at least two important differences between their
views. The simpler is that, as far as I can tell, Locke had no inkling of
the distinction between semantic and episodic memory that is central
to Tulving’s account. Locke thought of consciousness as extending
backward to episodes, but he regarded “memory” as covering both
experiences and facts. Where Tulving and his colleagues have elicited
interesting differences in the way subjects use “know” and “remem-
ber” (Knowlton & Squire, 1995; Rajaram & Roediger, 1997; Tulving,
1985), Locke (1690/1959) divided up the territory differently. For him,
memory was contrasted with current perception; memory was the ca-
pacity to revive previous perceptions “with this additional perception
annexed to them, that it has had them before” (2.10.2). But he made no
distinction between remembering, say, voting for Clinton and remem-
bering that Clinton was elected. On the issue that is most important
for the question of animal self-consciousness, however, they agree:
Human beings have the capacity to remember episodes, the doings of
past deeds.
The second difference is both harder to state and of direct relevance
to the question of animal self-consciousness. As we have seen, Locke
maintained that consciousness of a past thought or deed was a neces-
sary and sufficient condition for the individual who had that con-
sciousness to be identical with the person who had had the thought
or done the deed. His readers thought his view was confused. In pre-
senting memory as sufficient, Locke seemed to suggest that memory
was what made the rememberer the same person as the doer of the
remembered deed. But this could not be correct. Memory was only a
means of discovering that someone existing now is the same person as
someone who existed in the past. Derek Parfit (1984, p. 210) has intro-
duced terminology that may make this distinction clearer. In the opin-
ion of Locke’s critics (and in Parfit’s terminology), memory was reve-
latory of a “further fact,” a fact in addition to the memory itself, namely,
the fact that the rememberer and the doer of the past deed were the
same person. In Locke’s view, there was no further fact, but just one
fact that could be described in two different ways: “Individual A to-
day remembers doing something done by an individual in March
1993”; “A is the same person as someone who did that deed in that
month and year.”
Locke’s position can seem vaguely crazy and his motivations com-
pletely irrelevant to contemporary concerns. On one level, the motiva-
tions were clear: As an empiricist, he was deeply suspicious of the
notion of “substances” that endured through time but seemed to have
no properties that were available to the senses. In the Scholastic view,
substances were understood as the propertyless bearers of properties
that endured through changes in properties. Locke wanted to offer an
account of personal identity that did not depend on the (doubly) ob-
scure notion of a “spiritual substance” that endured through time car-
rying identity with it. So he offered memory or “extended conscious-
ness” as the basis of personal identity, and he persevered with that
view in the face of seemingly compelling counterexamples. Surely we
do not remember all our thoughts and deeds, yet are they any less
ours? What about the law that punishes drunkards for their actions
even in the face of plausible claims that they cannot remember what
they did?
As Edwin McCann (1999) has argued, Locke’s reply to the second
problem is particularly telling.5 In a letter to William Molyneux, he
elaborated the brief discussion of the Essay. No, Locke (1690/1959)
would not like to avail himself of a standard explanation for how the
law deals with drunkards: Since the person was responsible for the
drinking, he is held responsible for what ensues. To do so would be
to abandon his theory that personal identity and the appropriateness
of punishment was a matter of extended consciousness, for, ex hypo-
thesi, the drunk did not remember (2.27.22, note). Locke’s solution
was quite different and consistent with his view. The laws punish
“suitable to their way of knowledge” and “and in these cases they
cannot distinguish certainly what is real, what counterfeit” (2.27.22).
Since the fact can be proved against the person, but his want of mem-
ory cannot be proved for him, he is punished. That is, Locke resists
the counterexample not by weakening his theory but by claiming that,
as a practical matter, our legal punishments can be just even when
there is a danger that they err. Presumably if an infallible test for
memory failure could be developed, Locke would acquit a certain
number of drunkards of the crimes of which they were accused.
Beyond his general suspicion of the notion of “substance,” Locke
seemed to have a special motivation for clinging to his understanding
of personal identity, whatever its apparent problems. As noted, with
extended consciousness, a person’s own consciousness accuses her.
Were personal identity to be a matter of something other than con-
sciousness—call it factor X—then a present person could feel justly
punished or rewarded for a past deed only by appreciating that factor
X made her the same person as the perpetrator of the deed and, per-
haps, that her present memory was an infallible test for the presence
of X. Quite apart from worries about spiritual substances, Locke be-
lieved that we had no idea about the operation of whatever it is that

enables us to think and to be persons. In which case, we would have
no idea about what factor X might be.
From Locke’s perspective, his opponents were adding an idle and
incomprehensible piece to the mechanism. The practice of punishing
and rewarding past actions seemed fair to people—and so functioned
successfully—because they were held accountable for the deeds about
which their own consciousnesses pained them. By the view he op-
posed, there would be an additional loop: A person’s consciousness
informed her about a past deed, thereby informing her that factor X
was present between her and the perpetrator of the deed and, if she
knew what factor X was and so what its presence implied, she would
then know that she should be concerned about the remembered deed
and so act or react accordingly. But she did not have that crucial bit
of knowledge, so the concern she felt and the actions and attitudes
that concern engendered could only come along with the memory
itself. Thus, for Locke, the consciousness that made someone the per-
son she was, was only a matter of the ability to extend something like
a “consciousness with concern” back to past deeds. No other consider-
ations were relevant.
Tulving’s “autonoetic consciousness” seems more complex. Auto-
noetic consciousness involves an “autonoetic self” that is both psycho-
logically real and that accounts for the fact of memory:
When it comes to re-experiencing (remembering) of an earlier experi-

ence, an agent must exist that relates one to the other, that connects the
two events. This agent is [the] autonoetic self. Episodic remembering can
occur only if the self that does the re-experiencing of an event is the same
self that did the experiencing originally. (Tulving, 2003, p. 7)
I take it that this is not meant to be a metaphysical claim, since the

obvious counter would be that episodic memory may require nothing
more than a memory trace and some means of conscious access to
that trace. (So, for example, if time-lag mental telepathy were possible,
I could access your past experiences and so “remember” them.) Rather,
I assume that this account is offered as a description or analysis of
the psychological reality of the autonoetic self: In a case of autonoetic
consciousness, the subject not only recalls having a particular experi-
ence but implicitly recognizes that the memory is possible only if he
is the same as the self who had the original experience. This may be
correct for the human case. Perhaps whenever we have a sense of
remembering a thought or deed, we tacitly review whether that epi-
sode can be fitted into our continuing mental lives as we represent
them. So for example, a person might have a putative memory of
seeing the Mardi Gras parade, but then realize that she had never
been to New Orleans and so could not have seen it. But even a simple
version of the ability to construct a coherent narrative of the course of

your life seems far too complex to find in animals. It is not that all
animals lack the ability to recall at least relative times and places of
events—the results with scrub jays reported by Schwartz (chapter 9,
this volume) demonstrate such an ability. What is missing is a suffi-
ciently complex “theory of mind” to appreciate how mental states
should relate to each other over time in the life of an individual, with,
for example, memories following and not preceding experiences and
beliefs following and not preceding perceptions.
By contrast, the extended consciousness that Locke thought crucial
to the practice of punishment and reward might find analogues in
animals. All that would be required is the ability of an animal’s pres-
ent consciousness to extend itself backward to a past deed in such a
way that it acts appropriately to make amends for the deed or reacts
appropriately to current chastisement for it. Nor does it seem very
difficult to envision the kinds of evidence that might go some way
toward establishing this consciousness. Some is already available.
Frans de Waal (1996, p. 76) reports “thousands” of records of acts of
reconciliation following spats, if not actual transgressions. Since these
acts are much more frequent after difficulties than in times of har-
mony, they suggest that sufficient affect is connected to the past be-
havior to motivate present action—even if we might not want to go
so far as describing that affect as “concern.”
We might also be able to demonstrate that animals extend some-
thing like concerned consciousness back to past misdeeds experimen-
tally. Do animals respond differently when they are punished for past
misdeeds and when they receive the same treatment for neutral or
positive behavior? In particular, what do animals do when they have
committed no infraction and yet are punished? This would not be
easy to set up, since we lack “stooge” animals to carry out the inap-
propriate responses. Still, it might be possible for trainers to punish
the “innocent” and see whether this elicits a response that is different
from that to noxious forms of treatment unconnected with punish-
ments for past deeds. That certainly was what Locke thought hap-
pened in the human case: Unmerited punishment caused a special
kind of pain in addition to whatever the punishment itself inflicted.
As noted, Locke’s immediate successors also disagreed with his
view that memory was necessary for establishing sameness of person.
In his New Essays on Human Understanding, Leibniz (1704/1982) pro-
vided a different route:
To discover one’s moral identity unaided, it is sufficient that between
one state and a neighbouring (or just a nearby) state there be a mediating
bond of consciousness, even if this has a jump or forgotten interval mixed
into it. Thus, if . . . I did not know how I had arrived at my present state
even though I could remember things further back, the testimony of others
could fill in the gap in my recollection. I could even be punished on this

testimony. . . . And if I forgot my whole past . . . I could still learn from
others about my life during my preceding state. . . . All this is enough to
maintain the moral identity which makes us the same person. It is true
that if the others conspired to deceive me . . . then the appearance would
be false; but sometimes we can be morally certain of the truth on the credit
of others’ reports. (p. 236)
Although Leibniz appears to diverge significantly from Locke’s posi-

tion, I take the points of agreement to be more significant than the dif-
ferences. Both agree that what is central in personal identity, and so in
the continuity made possible through self-consciousness, is moral iden-
tity. Further, they agree that personal, and so moral, identity is stan-
dardly disclosed through memory or consciousness of one’s past deeds.
Leibniz also implicitly agreed that punishment and reward were justi-
fied only when the individual was morally certain (i.e., certain enough
to act) that he had perpetrated the deed. He simply suggested an alter-
native mechanism to such certainty via the testimony—and so memo-
ries—of others. In a sense, Leibniz merely offered a collective version
of Locke’s memory criterion. Deeds can be imputed (and so fairly pun-
ished or rewarded) on the basis of the individual’s memory of his past
or on the basis of other people’s memories of his past.
If a person accepts the greater reliability of collective memory, then
her own consciousness, if not memory, would accuse her of the deed
and she would be more accepting of punishment. It might be possible
to observe something like collective memory producing self-conscious-
ness among animals, though eliminating confounding factors would be
difficult. If an animal were more willing to engage in acts of reconcilia-
tion or to accept chastisement if it were collectively “accused”—treated
in the manner of a transgressor by many—then we might suspect that
collective memory can be a mechanism for self-consciousness, in the
sense of concern for past infractions, in animals as well as people.
Self-consciousness was important to modern philosophers in large
part because it was the capacity that underlay the ability of human
societies to be morally just. Through self-consciousness, someone’s
own consciousness accused him of the misdeed and recognized the
appropriateness of some ensuing decrement to his happiness (though
possibly not the size of the decrement envisioned by the church fathers).
Given this background, if some species of animals can be shown to be
sensitive to the connection between reward or punishment and past
boon or misdeed, and/or inclined to ameliorate their present situa-
tions by “canceling out” a previous misdeed, then I think there would
be a strong argument that they possess some aspects of “self-conscious-
ness,” as that expression has traditionally been understood.
On the other hand, Locke and others also fastened on self-conscious-
ness as the capacity that enabled one to behave well—or at least
against one’s immediate interests—in the first place. Locke was very
clear that self-consciousness involved not just a concern extending
back to past deeds but also forward to future happiness or misery. Of
course, there is no mystery about why he thought this. He believed
that divine and human morality could function only when individuals
were worried about the rewards and punishments that their actions
might bring. Insofar as animals have no appreciation of the future, it
might seem that no real analogue of the self-consciousness connected
to rewards and punishments is possible. In fact, I think that the animal
case suggests that modern philosophers may have been incorrect in
the prima facie plausible assumption that since the human practices
of punishing and rewarding involved both backward-looking and for-
ward-looking projections of the self, then a single mechanism of self-
consciousness underlay both. Being concerned about past action suffi-
ciently to try to make amends or to acquiesce in retaliatory attacks
might prevent a breakdown of the community even in the absence of
an ability to project into the future that would have enabled one to
foresee punishing and so not to transgress. Indeed, this kind of mech-
anism could be a link to a more symmetric type of self-consciousness
that underlies the human practices of punishing and rewarding.
CONSCIOUSNESS OF SUBJECTIVE LIMITATIONS

Locke tempered his empiricism with what might uncharitably be de-
scribed as “rationalism in empiricist clothing.” In addition to the ex-
ternal senses, (i.e., the senses as normally understood), he hypothe-
sized that we also had another source of ideas:
the other fountain is,—the perception of the operations of our own mind
within us, as it is employed about the ideas it has got;—which operations,
when the soul comes to reflect on and consider, do furnish the understand-
ing with another set of ideas, which could not be had from things without.
And such are perception, thinking, doubting, believing, reasoning, knowing,
willing, and all the different actings of our own minds;—which we being
conscious of, and observing in ourselves do from these receive into our
understanding as distinct ideas as we do from bodies affecting our senses.
This source . . . though it be not sense, as having nothing to do with exter-
nal objects is, yet is very like it, and might properly enough be called inter-
nal sense. (Locke, 1690/1959, 2.1.4)
Locke’s claim that we can be self-consciousness in the sense of being
conscious of our own mental “actings” is, to say the least, controver-
sial. Karl Lashley (1958) denied it wholesale: “No activity of mind is
ever conscious” (p. 4). More recently, Nisbett and Wilson (1977) have
reviewed a large body of literature suggesting that subjects lack privi-
leged access to their own higher and lower mental processes. It is
tempting to think that Locke and his contemporaries erred by taking
their metacognitive awareness that they were perceiving to be a per-

ception of themselves perceiving. So this aspect of the modern theory
of self-consciousness might well be rejected for the case of people,
thus taking it out of any possible analogy with animals.
Although Locke introduced the notion of an “internal sense” as an
awareness of mental actions, what he and his successors thought it
explained was our ability to scrutinize our own thoughts. In particu-
lar, the internal sense enabled cognizers to see the relations among
their ideas. To borrow one of Locke’s examples, the internal sense told
human beings that their idea of “murder” “contained” the idea of
“man” and not the idea of “sheep” (3.5.6). Despite their empiricist
credentials, both Locke (2.2.1) and Hume (1739/1978, p. 70) thought
that we could obtain knowledge by considering the relations among
our own ideas. As we can know that snow is white by observation
through our external senses, we can know—indeed know with com-
plete certainty—that triangles are three sided, murder involves killing
humans, and bachelors are unmarried through observing our own
ideas through the internal sense. In this case as well, later theorists
have cast serious doubts on Locke’s claim that we can be conscious of
our own ideas in anything like this sense. Beyond Quine’s (1953) well-
known denial of any principled distinction between claims whose
truth rests on empirical evidence and claims that are true by virtue of
the relations among our ideas, it now seems fairly clear that insofar
as we are able to determine what an idea or concept means to us, we
do so by considering how we are prone to use the concept and not by
engaging in some mysterious internal inspection. Still, even if Locke
was wrong about the mechanism by which we are able to scrutinize
our concepts, the ability to do so might still be thought of as a variety
of self-consciousness, and one not to be found in animals lacking lan-
guage.
Modern philosophers thought that self-consciousness of one’s own
ideas was important—was the hallmark of men as opposed to beasts—
because it permitted something that they took to be unique to human
beings, the capacity for epistemic self-improvement. Through being
conscious of their ideas, human beings could discard confused ideas
and clarify or patch up others. And they should. Modern philosophers
regarded epistemology as a normative discipline not just because it
was supposed to uncover standards for knowledge, but because those
standards provided what has sometimes been described as an “ethics
of belief.” Although people had the right to believe whatever they
chose, they had an obligation to do what they could to be epistemi-
cally responsible members of a mutually dependent society. From
Kant’s (1790/1987) perspective, this involved adhering to three funda-
mental epistemic maxims: “think for yourself,” “think from the stand-
point of others,” and think “in accord with yourself” (p. 294). That is,
try to ensure that your thinking was not a matter of prejudice, was
broad-minded, and was consistent.
If one normative role for self-consciousness was to enable the epi-
stemic self-improvement required for epistemic responsibility, what
are the prospects for finding animal analogues of self-consciousness?
Since Kant was giving voice (more or less) to Enlightenment bromides
about the routes to epistemic self-improvement, we can consider the
three processes captured in his maxims. The first and third look like
nonstarters for animals. How can an animal test its representations
for signs of prejudice or inconsistency without language? The second
process is often mediated by language as well. We follow the advice
“to think from the standpoint of others” in most circumstances by
asking others what they think. I should note that, for Kant, two differ-
ent norms were served by thinking from the standpoint of others: In
this way, one both showed respect for others and also lived up to
one’s obligation to make one’s beliefs as accurate as possible, by gain-
ing more information. Even in the absence of language, it might be
possible to observe rudimentary forms of both norms in animals, al-
though I consider only the second. We seem to see something like this
in cases where animals recognize the limitations of their own episte-
mic perspective in comparison with others and try to improve it by
borrowing information from others rather than acting (precipitously)
on their own limited information.
Josep Call’s very interesting experiment (chapter 13, this volume)
seems to offer some evidence that species of the great apes can both
recognize the epistemic limitation of their situation and borrow infor-
mation from others. So when food is placed in one of two tubes so that
they cannot see which has the food, they will rely on the indications of
trainers whom they recognize to have a clear view of the placements.
Still, interpretation is a difficult problem. How can we distinguish
cases where an animal is self-conscious of the limitations of its infor-
mation from cases where it looks for more information simply because
it has not got enough to proceed? The same difficulty seems to attend
the results with monkeys that Call reported. When given the opportu-
nity to opt out of a choice—when opting out provides a better reward
than answering incorrectly—rhesus monkeys will avail themselves of
that alternative, thereby significantly raising their percentage of cor-
rect choices. This implies that they choose only when they know; but
it does not show that they do so because they recognize that they
know or do not know rather than simply opting out when their infor-
mation is weak. As David Premack (1988) noted, what is really needed
is unequivocal evidence in nonhuman species that the animals “do
not act directly on the information given them, but sometimes ‘call to
question’ the information itself” (p. 173). And that sort of evidence is
hard to find.
Seyfarth and Cheney have observed that some monkeys (the vervets)
are able to ignore false information from conspecifics. They do not
always react to juveniles’ cries and, in an experiment, came to ignore
the alarms of a “boy who cried wolf” (reported by Dennett, 1987, p.
270). Here, too, there are interpretive problems, however. Did they
recognize that the information they had from immature members of
the troop might be false, or is it just an innate response to juvenile
informants? Did they come to question the unreliable alarm sounder
or simply extinguish an unrewarded response? To get a good parallel
to the human case, as the moderns envisioned it, one would like to
see them punish a giver of false alarms, at least in the adult case.
Nevertheless, if we focus on the other aspect of the ability to im-
prove one’s epistemic situation—not the recognition of the limitations
of one’s information, but the ability to borrow information from ap-
propriately placed other cognizers—then we can see a precursor not
of self-consciousness itself, but of one of the capacities that self-
consciousness enhances. When the ability to borrow information from
others is available, then it would be useful to have a mechanism that
makes one pause and evaluate one’s own information. Further, the
“disengage” mechanism, the mechanism that stops an animal from
acting when it lacks information, would seem to be an essential com-
ponent of the human ability to question our own information. Even
philosophers cannot afford to reflect on each of their beliefs without
quickly succumbing to starvation. So mechanisms must exist in the
human case that lead us to consider the quality of our information in
some cases rather than others. One obvious mechanism is the simple
insufficiency of information for action.
Since it is not easy to say what human self-consciousness is, I have
tried instead to consider what it does or, more precisely, what the
modern philosophers who introduced the notion of self-consciousness
thought it did. It seems to me that the central normative tasks as-
signed to self-consciousness by the moderns have either analogues
with animal capacities or components in common with animal capaci-
ties. This is quite a surprising result. For, as noted at the outset, mod-
ern philosophers believed that in self-consciousness they had finally
found the key to human uniqueness, that quality which, in Kant’s
(1978/1798) words, “elevated man infinitely above all other creatures
that lived on earth” (p. 127, my emphasis).
NOTES
1. I discuss epistemic responsibility at greater length below.

2. The references to Locke (1690/1959) provide book, chapter, and section num-
bers, rather than page numbers, since these are common to different editions.
3. Here I borrow Joyce’s (1976, pp. 130–131) borrowed description of hell in
Portrait. Joyce takes the description from Pinamoti’s Hell Open to Christians, to Caus-
ing them from Entering into it (1688), which itself borrowed from Suarez, Aquinas,
Loyola, and others.
4. The standard edition of the essay, A. C. Fraser’s, has a numbering problem.
There are two sections 2.27.10 and 2.27.11. This citation is to the second 2.27.11.
5. My discussion here is much indebted to McCann (1999).
REFERENCES
Dennett, D. (1987). The intentional stance. Cambridge, MA: MIT Press.
de Waal, F. (1996). Good natured: The origins of right and wrong in humans and
other animals. Cambridge, MA: Harvard University Press.
Hume, D. (1978). A treatise of human nature (P. H. Nidditch, Ed.). Oxford, UK:
Oxford University Press. (Original work published 1739)
Humphrey, N. K. (1988). The social function of the intellect. In R. Byrne & A.
Witen (Eds.), Machiavellian intelligence: Social expertise in the evolution of intel-
lect in monkeys, apes, and humans (pp. 13–26). Oxford, UK: Clarendon Press.
Joyce, J. (1976). Portrait of the artist as a young man. New York: Penguin.
Kant, I. (1987). Critique of judgment (W. Pluhar, Ed. & Trans.). Indianapolis,
IN: Hackett. (Original work published 1790)
Kant, I. (1978). Anthropology from a pragmatic point of view (V. L. Dowdell,
Trans.). Carbondale: Southern Illinois University Press. (Original work
published 1798)
Knowlton, B. J., & Squire, L. R. (1995). Remembering and knowing: Two dif-
ferent expressions of declarative memory. Journal of Experimental Psychol-
Lashley, K. S. (1958). Cerebral organization and behavior. In H. C. Solomon
& S. Cobb (Eds.), Proceedings of the Association for Research in Nervous and
Mental Disease: Vol. 36. The brain and human behavior (p. 4). Baltimore, MD:
W. Penfield.
Leibniz, G. (1982). New essays on human understanding (Abridged ed., P. Rem-
nant & J. Bennett, Ed. & Trans.). Cambridge, UK: Cambridge University
Press. (Original work published 1704)
Locke, J. (1959). An essay concerning human understanding (A. C. Fraser, Ed.).
New York: Dover. (Original work published 1690)
McCann, E. (1999). Locke on identity: Matter, life, and consciousness. In M.
Atherton (Ed.), The empiricists: Critical essays (pp. 63–88). New York: Row-
man and Littlefield.
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we can know: Verbal
reports on mental processes. Psychological Review, 84, 231–259.
Parfit, D. (1984). Reasons and persons. Oxford, UK: Clarendon Press.
In R. Byrne & A. Whiten (Eds.), Machiavellian intelligence: Social expertise
and the evolution of the intellect in monkeys, apes, and humans (pp. 160–179).
Oxford, UK: Oxford Science Publications.
Quine, W. V. (1953). Two dogmas of empiricism. From a logical point of view
(pp. 20–46). Cambridge, MA: Harvard University Press.
Rajaram, S., & Roediger, H. L., III (1997). Remembering and knowing as states
of consciousness during recollection. In J. D. Cohen & J. W. Schooler (Eds.),
Scientific approaches to the question of consciousness (pp. 213–240). Hillsdale,

NJ: Erlbaum.
Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26, 1–12.
Tulving, E. (2003). Episodic memory and autonoesis: Uniquely human? Unpub-
lished manuscript.
8
Progress in the Study of Chimpanzee

Recall and Episodic Memory
Charles Menzel
One of the central questions raised by this volume is whether nonhu-

man animals have self-reflective consciousness. Tulving’s (chapter 1,
this volume) opinion is that “no essential progress has been made
toward resolving the issues involved for more than a hundred years.”
He recommends that we focus on the following questions: “What ex-
actly makes some people believe that nonhuman primates do have
the same kind of (autonoetic, self-knowing) consciousness that hu-
mans do, and why do some others believe that they do not?” and that
we carefully examine the points of agreement and disagreement.
The idea that there is something fundamentally different about hu-
man and nonhuman memory dates back at least to the time of Aris-
totle (1990), who wrote:
many also of the other animals [as well as man] have memory, but, of all
that we are acquainted with, none, we venture to say, except man, shares
in the faculty of recollection. The cause of this is that recollection is . . . a
mode of inference. For he who endeavors to recollect infers that he for-
merly saw, or heard, or had some such experience, and the process [by
which he succeeds in recollecting] is . . . a sort of investigation. But to in-
vestigate in this way belongs naturally to those animals alone which are
also endowed with the faculty of deliberation. (p. 695)
The idea that human memory and nonhuman memory are different
is also shared by numerous philosophers throughout written history
as well as many current biologists and psychologists. The idea is
based not simply on received wisdom but also on the conspicuous
differences in what humans and nonhumans do. For example, only
humans have locked themselves away in a room for years at a time
to write their confessions and autobiographies. The classic explora-
tions of memory by St. Augustine and Proust are outstanding exam-
ples. Equally important, only humans have created libraries, muse-
188
Chimpanzee Recall and Episodic Memory 189
ums, and archives devoted exclusively to the preservation of writings,

photographs, speeches, and other media documenting events and dis-
coveries from the past.
Ruggiero and Flagg (1976), in discussing the history of the claim
that animals do not have memory in the same sense as humans do,
wrote, “Since the time of Aristotle, man has been regarded as different
from animals because he possesses a reasoning ability, based on a
spoken language. . . . Such thinking reflects the view that, man, by vir-
tue of his verbal ability, is capable of actively recalling the past (i.e.,
generating information about a past event in the absence of the origi-
nal event), whereas nonverbal organisms are capable of recognition at
best” (p. 1). Ruggiero and Flagg argue that monkeys, apes, and other
animals do possess representational, organized memory, but they
point out, “If the term ‘memory’ is restricted to instances of active
recall, we can begin to understand the . . . claim that animals do not
have memory” (pp. 1–2). Indeed, in everyday life we normally accept
a person’s recounting (via written and verbal descriptions, maps, ges-
tures, and so forth) of what he or she saw, heard, and felt on a previ-
ous occasion as prima facie evidence that he or she can recall the
event. The number of independent details about past events that a
normal adult human can recall and report exceeds by a wide margin
that demonstrated to date for any nonhuman. Until recently, however,
almost no data have been available on recall capabilities in nonhuman
animals to take the issue further.
William James’s (1890) definition of human “memory proper” also
provides a contrast with most conceptions of nonhuman memory.
James distinguished sharply between memory proper and the mere
recurrence of behaviors or images, or their dependence on past experi-
ence, and went on to say:
Memory proper . . . is the knowledge of a former state of mind after it has
already once dropped from consciousness; or rather it is the knowledge of
an event, or fact, of which meantime we have not been thinking, with the
additional consciousness that we have thought or experienced it before. . . . Mem-
ory requires more than mere dating of a fact in the past. It must be dated
in my past. In other words, I must think that I directly experienced its
occurrence. It must have that “warmth and intimacy” which were so often
spoken of in the chapter on the Self, as characterizing all experiences “ap-
propriated” by the thinker as his own. A general feeling of the past direc-
tion in time, . . . a particular date conceived as lying along that direction,
and defined by its name or phenomenal contents, an event imagined as
located therein, and owned as part of my experience. (p. 288, italics in
original)
James’s definition “implies that the memory is conscious and that the
subject has language and a formal concept of time and of Self” (E.
Menzel, 1984, p. 511). The implication that the subject has a sense of
personal agency during recall ties in closely with Tulving’s view of

episodic memory, discussed later. Although there have been hun-
dreds of impressive demonstrations of memory in nonhuman animals
(e.g., food-storing birds: Balda & Kamil, 1992; Sherry, 1984; Shettle-
worth, 1995; nonprimate mammals: Olton, 1978; Vander Wall, 1991;
chimpanzees: E. Menzel, 1973a; Tinklepaugh, 1932; other primates:
Garber, 1989; Janson, 1998; C. Menzel, 1991; E. Menzel & Juno, 1985;
review, Tomasello & Call, 1997; bees: R. Menzel et al., 1996; other
animals: Gallistel, 1990; Vander Wall, 1990), few if any of them could
be said to involve memory proper in James’s sense (Donald, 1993;
Roberts, 1996; Schacter & Tulving, 1994).
Another influential statement regarding the contrast between hu-
mans and nonhumans was Köhler’s (1925) proposal that the time in
which nonhumans live is far more limited, both in the past and in the
future. Many current investigators have echoed or extended this
theme (e.g., Bickerton, 1995; Bischof, 1978; Roberts, 2002; Suddendorf
& Corballis, 1997). For example, Tulving (1983) distinguishes between
memory of facts (“semantic memory”) and memory of particular
events that an individual has experienced personally (“episodic mem-
ory”). An organism might retain a fact about the environment (e.g.,
Object X is there) without possessing information that would enable
it to reconstruct an event (e.g., Self saw Person Y put out Object X).
Tulving (1984) acknowledges that there is much evidence that animals
can rely on information from the past, but he questions whether ani-
mals can “mentally travel back in time to recollect and reminisce the
way humans do” (p. 258). An important part of Tulving’s concept is
that humans can perform recall in different situations and in different
forms; they have a well-developed ability to “experience again now,
in a different situation and perhaps in a different form, happenings
from the past, and know that the experience refers to an event that
occurred in another time and in another place” (Tulving, 1983, p. 1).
The part of Tulving’s definition emphasizing the ability to perform
recall in different ways and from different situations than the situation
in which the event originally occurred is often overlooked in recent
discussions of “episodic-like” memory in animals.
Tulving and Markowitsch (1998) go on to say:
Episodic remembering (mental time travel) is accompanied by a special

kind of “autonoetic” conscious awareness that is clearly different from the
kind of conscious awareness (“noetic” awareness) that accompanies re-
trieval of declarative information. . . . A normal individual can distinguish
between recollecting a personal experience and recalling an impersonal
fact as readily as she can distinguish between, say, perceiving and im-
aging. . . . Many animals other than humans . . . possess well developed
knowledge-of-the-world (declarative memory) systems, and are capable of
acquiring vast amounts of flexibly expressible information. But there is no
evidence that they have the ability to autonoetically remember past events
in the way that humans do. (p. 202)
To date, few if any studies of animal memory could be said to “prove”

episodic memory as Tulving and Markowitsch (1998) fully define it
for humans, to everyone’s satisfaction. Shettleworth (1998, p. 238),
Clayton, Griffiths, Emery, and Dickinson (2002), and several contribu-
tors to this volume argue in particular that it is nearly impossible to
test for “conscious recollection” in nonverbal animals. Authors such
as Tulving and Markowitsch (1998) and Suddendorf and Corballis
(1997) definitely have a point when they say that when it comes to
the topics of “memory proper” or “true episodic memory,” most of
animal memory research is equivocal. I disagree, however, with the
opinion that animal researchers have made no essential progress on
the above topics in the past 100 years, or that no progress will be
likely in the future. Vast improvements in methods, concepts, and the
amount and quality of data have been made since the time of Thorn-
dike (1898) in the study of animal memory (Kawai & Matsuzawa,
2001), self-recognition (Gallup, 1970; E. Menzel, Savage-Rumbaugh, &
Lawson, 1985), spatial cognition (Kuhlmeier, Boysen, & Mukobi, 1999;
E. Menzel, 1973a; Tinklepaugh, 1932), and communication (Savage-
Rumbaugh et al., 1993), and there is always more to learn regarding
what is and what is not essential. Further improvements in methods
in the study of memory seem likely. To take one example, a current
limitation in the study of nonhuman memory is that there are almost
no data on recall capabilities. Shettleworth (1998) points out:
With humans, either recognition or recall of past input can be tested. That
is, a person can be presented with a stimulus and asked “Have you experi-
enced this before?” (recognition) or simply instructed to “Tell me what
you remember” (recall). Performance on recognition tasks is typically bet-
ter than on comparable recall tasks. . . . Most tests of animal memory are
tests of recognition [rather than recall]. (p. 238, italics in original)
If recall in animals is currently a fuzzy, underresearched topic,

then, by comparison, self-reflective, conscious recollection in animals
is many times as fuzzy. If we have not achieved general agreement
about the existence and form of recall capabilities in animals, then we
are unlikely to be able to agree about self-reflective, conscious recol-
lection. On the other hand, if we can identify clear signs of recall capa-
bilities in some of the everyday activities of animals, and if we can
develop methods to investigate these activities in detail, then we can
make an indirect approach to topics that are more difficult and inferen-
tial. These new methods should address the following two questions:
What are the specific differences and similarities between human and
nonhuman memory? And what is it that animals actually do?
Determining which features nonhuman animals do actually retain

about trial-unique events, and the extent to which their memory in-
volves recall as opposed to recognition, should help to clarify the simi-
larities and differences between animal and human memory systems
(Gaffan, 1994; Hampton, 2001; Kawai & Matsuzawa, 2001; Olton, 1984;
Parkinson, Murray, & Mishkin, 1988; Roberts, 1996; Schacter & Tulv-
ing, 1994; Squire, 1992; Tulving, 1984, p. 258). The features of an event
that a nonhuman animal retains might, in principle, include objective,
impersonal details, such as “what,” “where,” “when” (Clayton & Dick-
inson, 1998, 1999a, 1999b), “size,” “numerousness” (Beran & Beran,
2004), “surroundings,” and “mobility” (living versus nonliving). The
animal also might, in principle, retain more personalized details, such
as “who was involved” (C. Menzel, 1999; Schwartz, Colon, Sanchez,
Rodriguez, & Evans, 2002; Schwartz & Evans, 2001) and “why,” and
possibly “emotional impact.” Specific questions arise: How much in-
formation about an event does an animal retain in long-term memory
(Mishkin, Suzuki, Gadian, & Vargha-Khadem, 1997)? How much of
this information can the organism access and transmit from outside
the situation in which the event originally occurred (C. Menzel, 1999)?
Can the organism use indirect cues to access the information (Bayley
& Squire, 2002)? I emphasize that research on recall memory in chim-
panzees is at a preliminary stage, and it is likely that we have under-
estimated the upper limits of ability.
In the remainder of this chapter, I review several prior studies of
chimpanzee memory, with a focus on research that has used varia-
tions on delayed-response methods. I then describe my ongoing stud-
ies of memory in symbol-competent chimpanzees (genus Pan). I will
argue that at least one chimpanzee’s memory is, to a significant extent,
organized independently of stimulation from the immediate environ-
ment in question and constitutes recall as opposed to recognition. Fi-
nally, I return to the broad themes of this book.
REMEMBERING VERSUS RECOGNITION AND REPRODUCTION

Köhler (1925) made the following statement while discussing the lim-
its of his experimental findings:
though one can prove some effects of recognition and reproduction after
considerable lapses of time—as is actually the case in anthropoids—this is
not the same as “life for a longer space of time.” A great many years spent
with chimpanzees lead me to venture the opinion that, besides in the lack
of speech, it is in the extremely narrow limits in this direction that the chief
difference is to be found between anthropoids and even the most primitive
human beings. (p. 267)
Less frequently cited is the fact that Köhler went on to describe in-
stances of what he regarded as closer to real “remembering” (p. 278),
as opposed to “the plain after-effects of past experience in present

actions” (p. 277). Köhler noted, “where real remembering broadens
the scope of those conditions, which influence the life of the animal,
this real extension of the life surveyed impresses the observer consid-
erably: and ‘free,’ ‘enlightened’ does the ape then look, compared with
the ‘narrowness of time’ in which lower animals live” (p. 278). (Since
Köhler’s time, estimates of the capacities of other animals have been
drastically revised too.)
Köhler’s examples of “the plain after-effects of past experience in
present actions” included the observation that the chimpanzees imme-
diately recognized familiar people and familiar group members after
absences of several months. Köhler’s examples of “recognition and
reproduction” also included the observation that “even after long
lapses of time one finds the effect of an earlier learnt discrimination
test as strong as though only a few days had elapsed” (p. 277). Köhler
retested four chimpanzees on a variety of size and color discrimina-
tion problems, more than a year after the end of their earlier period
of learning. The chimpanzees’ choices were as accurate as they had
been previously. Köhler commented, “For these achievements, there
was certainly no necessity whatsoever for any image of the past. The
familiar situation appears again immediately in the form of apprehen-
sion acquired in the learning period, and so brings about the same
direction of choice behaviour” (p. 278).
Köhler’s examples that, following Hunter’s (1913) terminology, “in-
volved something like real ‘remembering,’ rather than the reproduc-
tion of former behaviour in a like situation” (p. 278) included the fol-
lowing experiments with a male chimpanzee, Sultan. While Sultan
watched, Köhler buried a pear in a sandy area, 1.4 meters outside
Sultan’s barred enclosure. Köhler carefully smoothed over the spot
and the surrounding area, so that no trace remained of the hole. Sul-
tan could not reach the pear directly; to obtain it, he had to reach out
with a stick. At first, no stick was available. After about 30 minutes,
Köhler moved a stick to within reach of Sultan’s enclosure. When Sul-
tan noticed the stick, he pulled it in, ran to the bars opposite the hid-
den pear, and scraped the sand away at the exact location until the
pear appeared. The field of possible error was approximately 10
square meters. Several tests with delays up to 16.5 hours and addi-
tional chimpanzee subjects yielded similar results. Köhler concluded,
“An extension of this test to longer periods, and, above all a closer
analysis of the proceedings, are both very desirable” (p. 282).
In one such variation, four chimpanzee subjects of Yerkes and
Yerkes (1928) were able to respond to food, which they had seen bur-
ied in the ground, after a delay of 48 hours. In another extension,
Tinklepaugh (1932) examined whether chimpanzees, macaques, hu-
man adults, and children could respond to a series of locations in
which they had seen objects hidden by another agent. He arranged up

to 16 pairs of containers in a circle 20 feet in diameter. The members of
each pair of containers were as close to identical as possible, with the
exception of one pair, but there were clear differences between the
pairs. A subject was brought into the test room and made to sit on a
stool in the center of the circle. The experimenter then went to each
pair of containers and, while the subject was watching, placed a piece
of banana (or some other type of lure for the humans) under a ran-
domly selected member of each pair of containers. The nonhuman
subjects were tested on a leash, and had to make their choices in a
sequence determined by the experimenter. The subject was permitted
to go around the circle to each pair and make a choice of one of the
containers, taking the lure if it was correct. The subject had to return
to the stool between each choice, regardless of whether the choice was
correct or incorrect. The chimpanzees (a female approximately 8 years
old and a male approximately 6 years old) were first tested with 8
pairs of containers (20 trials), and then the number was increased to
12 (11 trials) and 16 (18 trials).
Results were that with 8 pairs of containers, the two chimpanzees
chose the correct member of the pair on 85% and 90% of pairs, respec-
tively. With 16 pairs, the chimpanzees chose the correct member of
the pair on 79% and 78% of pairs, respectively. The chimpanzees did
as well as the average of five adult humans, who were run on a single
trial with 16 pairs of containers; the adults’ scores ranged from 44%
to 100% (mean 75%). The four children, aged 7 to 9 years, had scores
from 44% to 69% (mean 54%). The monkeys’ choices were only slightly
better than chance with 8 pairs of containers but were well above
chance with 3 pairs (77% and 73% correct, respectively).
Control procedures excluded the possibility that the chimpanzees
were simply detecting the food by odor. Other experimental varia-
tions showed that the chimpanzees were responding on the basis of
spatial position rather than by the visual features of the containers.
The focus of Tinklepaugh’s (1932) study was on the spatial and visual
cues used by the chimpanzees to locate individual hidden foods,
rather than on how the chimpanzees organized a food-harvesting
route. That is, it showed that chimpanzees could remember a large
number of food locations in a given trial but left largely unanswered
the question of whether the chimpanzees remembered the spatial rela-
tionships among these locations.
E. Menzel (reviews, 1978, 1984) conducted an extensive series of
studies of chimpanzee memory and communication in a 1-acre field.
In one study (E. Menzel, 1973a), one chimpanzee from a group of six
juveniles was carried around the experimental field and shown a
piece of fruit in each of 18 randomly selected sectors of the field. Un-
like Tinklepaugh’s study, each trial used a different set of test loca-
tions. During the cue-giving process, the chimpanzee was not allowed
to do anything other than cling to his carrier and watch the hiding of
the food. Thus, locomotor practice and primary reinforcement were
eliminated during the cue-giving phase of a trial. The field included
areas of relatively tall grass, trees, irregularities (small hills), and other
visual barriers, and not all of the 18 food locations were visible to the
chimpanzee from a single vantage point. After being shown the food,
the experimenters returned the chimpanzee to the group. The experi-
menters then ascended an observation tower and within 2 minutes
released the group into the field. The question was how the chimpan-
zee organized his route to harvest the 18 hidden food items. Four
different chimpanzees were used as test animals.
Results were that the chimpanzee that had been shown the food
obtained most of the hidden items, using a highly efficient route. In
16 trials, the informed (test) chimpanzee found 200 of 288 hidden
foods (mean 12.5 per trial), and the five control animals that had not
been shown the food found a total of 17 (0.21 per animal per trial).
Usually, the test animal ran in a direct line to the exact tree stump,
clump of grass, or hole in the ground where a piece of hidden food
lay, grabbed and ate the food, and then ran directly to the next loca-
tion, no matter how far away or obscured by visual barriers that loca-
tion was. Control animals obtained food mainly by searching near the
test animal or by begging from him. The test animal rarely searched
the ground manually farther than 2 meters from a food pile, whereas
controls did so on innumerable occasions. The itinerary that the test
animal used during the recovery of food was only 64% as long as the
mean of all possible N! itineraries on that trial and bore no relation to
the route along which the chimpanzee had been carried during cue
giving. The test animals clearly took into account more than one or
two food locations at a time, because on any given choice they not
only generally went to one of the closer locations of all those that still
contained food, but also they usually went from the outset of a trial
toward larger rather than smaller clusters.
Specifically, in a further experiment, they typically went toward
whichever half of the field contained three food locations versus two
food locations. In other tests, the experimenters hid a piece of pre-
ferred food (fruit) in nine locations and a piece of nonpreferred food
(vegetable) in nine other locations. The informed chimpanzee went
first to the fruit locations and rarely reinspected a location that he had
already emptied. Relevant to the topic of recall discussed earlier is
that on several trials, a seeming “example of sudden recall occurred
while an animal was apparently asleep. After having eaten many
pieces of food and lain supine with his eyes closed for up to 30 min-
utes, the test animal suddenly jumped to his feet and ran 10 to 30 m
straight to a hidden piece of food” (E. Menzel, 1973a, p. 944).
Other experiments in the series (E. Menzel, 1973b) showed that a

knowledgeable chimpanzee could retain and transmit information to
group members about object quality (fruit versus vegetable), the rela-
tive goodness or badness of object (food versus snake), quantity of
food, and location. The group of chimpanzees also searched very dif-
ferently after an operationally designated leader had been shown a
real snake than they did after the leader had been shown a piece of
food that was later removed. Not only would they use a stick to probe
the last location where the snake had been seen, but they would also
broaden the area searched, as if taking into account the potential of
the snake for movement. In contrast, if the leader had been shown
food, they restricted their search to the vicinity of the original location.
These studies showed not only that chimpanzees could remember
many food locations in a given trial, but also that they remembered
the spatial relationships among the locations (Gallistel, 1990; E. Men-
zel, 1973a). The studies also demonstrated transmission of information
about the nature and location of hidden objects from one chimpanzee
to another at an unexpectedly high level of efficiency (E. Menzel,
1973b).
MEMORY IN SYMBOL-COMPETENT CHIMPANZEES

An extension of these delayed-response methods for studying mem-
ory in nonhumans is to examine which features of events a subject
conveys to other beings through behavior or a symbol system (Her-
man & Forestell, 1985; E. Menzel, 1973b; Savage-Rumbaugh, McDon-
ald, Sevcik, Hopkins, & Rubert, 1986; Tulving, 1984, p. 224). Symbol-
competent chimpanzees provide a powerful method for investigating
the “time in which the chimpanzee lives” and the apes’ memory for
events. Symbol-competent apes at Georgia State University’s Language
Research Center have learned lexigrams, arbitrary geometrical forms
that refer to types of food and other objects. These animals have been
immersed in the visual symbol system and exposed to spoken English
since an early age. As a result of their rearing history, the apes can
comprehend lexigrams and can use lexigrams in a productive manner,
and some of them can understand spoken English at approximately
the level of a very young child (Savage-Rumbaugh et al., 1993). The
symbol system allows one to ask questions regarding memory that
previously had not been asked by Köhler or other researchers; it is
possible to obtain more information from the apes. Symbol-competent
apes potentially can convey the specific events and environmental fea-
tures they have seen, through use of the lexigram system. They can
be tested outside the spatial and temporal contexts in which they en-
countered the events. Some of the Language Research Center apes can
use lexigrams outside the context in which the objects were originally
encountered (Savage-Rumbaugh et al., 1986, 1993; Savage-Rumbaugh,

Pate, Lawson, Smith, & Rosenbaum, 1983).
Announcement of Travel Destinations in a Twenty-Hectare Forest

Savage-Rumbaugh et al. (1986) tested a 3-year-old bonobo, or pygmy
chimpanzee (Pan paniscus), Kanzi, in a heavily wooded 20 hectare
area. The area included 5.6 kilometers of trails. Kanzi had traveled
with human companions in the forest daily and had learned the loca-
tions of all 17 of the food stations in the forest. Kanzi’s rearing history
is reported in Savage-Rumbaugh, Rumbaugh, and McDonald (1985)
and Savage-Rumbaugh et al. (1986). Each station had a different food
type present. Four months after food was first placed in the forest, a
caregiver who had not previously entered the forest accompanied
Kanzi and allowed him to lead the way from place to place. The care-
giver did not know the location of any of the foods or the locations of
any of the trails. At each stopping point, the person presented Kanzi
with an array of 6 to 10 alternatives. The alternatives consisted of
photographs and lexigrams representing food and locations. Kanzi
made his selection from this array. When Kanzi was at a stopping
point, he used photographs on five occasions, lexigrams on seven oc-
casions, and both photographs and lexigrams on three occasions to
indicate his next destination, with 100% accuracy. After touching a
photograph or lexigram, Kanzi proceeded to guide the person to the
location of the food he had selected, sometimes traveling over 30 min-
utes to reach the food site. Distances between the successive food sites
ranged up to 600 meters. To reach his destination, Kanzi had to choose
a direction at up to seven different trail crossings. Kanzi used the
smallest possible number of turns and the shortest possible route
along the trail on all but one occasion (Savage-Rumbaugh et al., 1986).
E. Menzel (1984, p. 520) noted, “From the standpoint of ‘memory
proper,’ it would be most interesting to see if [Kanzi] can also indicate
where he has been and what he has done after such an outing.”
Use of Road Signs in a Twenty-Hectare Forest

C. Menzel, Savage-Rumbaugh, and Menzel (2002) used Kanzi’s lexi-
gram competence as a tool to study his spatial memory organization
when he was 4 years old. The focus of the study was on Kanzi as a
receiver rather than a sender of information. In the first experiment,
we showed Kanzi a “road sign” just outside his indoor sleeping area.
The sign indicated, by means of lexigrams, the location where food
was hidden. Only 2 of the 15 locations were visible from the sign.
Distances ranged up to 170 meters from the sign. In 99 of 127 test
trials, Kanzi went to the designated location on his first move. In a
second experiment, we presented Kanzi with the road sign at varied
points in the forest rather than at the original fixed place. In these
trials, the goal was a preferred toy. Kanzi’s human companions were
never informed about the location of the goal, and distances ranged
up to 650 meters. In all 12 trials, Kanzi led his companions to the
designated location using an efficient path. In sum, Kanzi appeared
to be able to move, based on the information provided by a lexigram,
from almost any arbitrarily designated starting location in his 20-hect-
are environment to any one of the numerous goal locations.
UNPROMPTED RECALL AND REPORTING OF HIDDEN OBJECTS

BY A COMMON CHIMPANZEE
Next I describe in some detail my first study (C. Menzel, 1999) on
memory in a lexigram-competent female common chimpanzee (Pan
troglodytes) called Panzee. The aim of this study was to identify some
of the types of information that Panzee could recall and convey about
hidden objects. Details of the procedure are given in C. Menzel (1999).
In brief, Panzee was 11 years old at the outset of the experiment, and
she had already learned more than 120 lexigrams (Beran, Savage-
Rumbaugh, Brakke, Kelley, & Rumbaugh, 1998). Panzee had been
reared by Sue Savage-Rumbaugh, Karen Brakke, and their associates;
she had been given intensive early exposure to spoken English and to
lexigrams that referred to foods, tools, actions, and locations in the 20-
hectare forest at the Language Research Center. Panzee was exposed
to lexigrams in the everyday contexts of food preparation, travel, and
play, rather than taught lexigrams through discrete trial training. That
is, her lexigram acquisition appeared to be based on observational
learning (Brakke & Savage-Rumbaugh, 1995, 1996; cf. Savage-Rum-
baugh et al., 1993). There had been no previous attempt to determine
Panzee’s retention time beyond about 40 seconds for visual stimuli
or locations. Panzee had not been presented with objects outside her
outdoor enclosure that she could later name or request indoors. Re-
cruitment of people to the outdoor enclosure and inducing them to
search the terrain was never a part of her experimental history. This
is not to claim, however, that she brought no relevant experience to
the experimental situation. For example, Panzee had used gestures
and lexigrams in routine interactions with caregivers to request move-
ment between cages, activities (e.g., chasing, grooming, drawing, tele-
vision), foods, and other objects (e.g., blanket, clay). As a juvenile,
Panzee also had played the game of hide-and-seek with caregivers,
and it was in the context of searching for a hidden person that she
originally used the hiding gesture described later.
During this experiment (C. Menzel, 1999), Panzee was housed in
the Lanson building, an indoor-outdoor facility. When Panzee was
indoors, she could not see the outdoor cage or the forest. The outdoor
cage was bordered by forest, and an area of forest adjacent to the
cage, measuring about 160 square meters, served as a test area for the
introduction of objects. Panzee had not entered this forest for at least
6 years prior to this study. The indoor and the outdoor cage each
contained a keyboard, each of which displayed 256 lexigrams. To ob-
tain something from outside the outdoor cage in this study, Panzee
had to recruit the assistance of a person and in effect tell them where
to go. Panzee was not required also to tell people what type of object
to look for. If she did provide this information, it would be as a result
of her own initiative or due to past experience.
Design and Procedure

This initial experiment (C. Menzel, 1999) consisted of 34 trial-unique
tests, each consisting of a cue-giving phase, a delay phase, and a re-
sponse phase. The test objects were 26 food types and 7 types of non-
food objects. During cue giving, the experimenter stood outside Pan-
zee’s outdoor enclosure, held up an object while Panzee was watching,
then walked to a predetermined hiding place in the forest and placed
the object on the ground. Each trial used a different location, and dis-
tances from the enclosure to the object ranged up to 8 meters. On 32
of the 34 trials, the experimenter placed the object under natural
cover, so that the object was concealed from view, and then smoothed
over the area. On these 32 trials, the object was buried under leaf litter
and completely concealed from view, not simply placed behind a
bush. On the remaining 2 trials the object was not concealed, and
Panzee could see it from her outdoor cage. After placing the object,
the experimenter left the area, and this began the delay phase of the
trial. On 10 trials, the experimenter imposed an overnight delay be-
tween the cue-giving phase and the response phase by showing Pan-
zee the object outdoors after all caretakers had left for the day. Panzee
did not have a chance to interact with a person until the next morning.
On the remaining 24 trials, Panzee had an opportunity to interact with
a person indoors on the same day, typically within a few minutes
after seeing the object. On any given opportunity, Panzee might or
might not choose to recruit the assistance of the person to recover the
hidden object. The delay phase lasted until Panzee initiated recruit-
ment of a person (details below). Thus, Panzee determined the exact
time of recruitment and the total length of delay.
In the response phase of a trial, Panzee interacted with a person
inside the Lanson building. This person did not know what had gone
on outside the Lanson building and did not know what the object was
nor where it was located. Three different people served as “unin-
formed persons” during the study. On at least 24 of the 34 trials, the
person did not even know in advance that a trial had been set up.
Once a person was involved in the study, he or she knew that poten-
tially a trial could happen but did not know when. Experimental trials
were interspersed with nontest days on which no object was pre-

sented, over a total study period of 268 days.
The uninformed person went about his or her routine caregiving
activities and let Panzee initiate any interactions regarding what had
gone on outside the Lanson building. During the study period, per-
sons made detailed notes regarding any occasion on which Panzee
recruited them outdoors, regardless of whether they thought a test
actually was being conducted. The person recorded on each such oc-
casion all lexigrams that Panzee touched on the indoor and outdoor
keyboards, Panzee’s gestures, and her vocalizations. Gestures re-
corded included pointing outdoors with an extended arm, beckoning
with a sweeping movement of her hand, and covering her eyes with
the palm of her hand (a gesture previously associated with the game
of hide and seek with caregivers). Spatial data recorded during each
trial included the location at which Panzee sat when she pointed man-
ually out of the outdoor enclosure (figure 8.1) and each location in
the woods that the uninformed person inspected manually. If Panzee
guided the uninformed person to a hidden food item, then the person
carried the food indoors and offered it to Panzee, regardless of
whether Panzee had touched the lexigram corresponding to the food.
Figure 8.1 After touching the lexigram on her keyboard (visible in background) corre-
sponding to the type of object hidden, Panzee points toward the location of the object.
See color insert. Photo by Charles R. Menzel.
Results
The uninformed persons found all 34 objects that were used in these
tests as the result of Panzee’s behavior. False positives occurred on
only 3 of the 268 days of the experiment. That is, Panzee almost never
recruited the uninformed person and caused him or her to search
when there was no test object in the woods. When a test object was
present outdoors, Panzee never pointed to a location that had con-
tained an object on a previous trial. In fact, in 34 recruitment se-
quences with an object present in the woods, Panzee never pointed in
an incorrect direction. In total, there were 57 days on which an object
was present. Panzee recruited a person and pointed toward a single,
specific location on 34 of these 57 days and did not recruit a person
on the remaining 23 days. (Panzee did not fail to report the objects
that were hidden on these 23 remaining days; she simply did not al-
ways recruit a person to the object at her first opportunity. She might
wait a day or more.) In contrast, there were 211 nontest days on which
no object was present in the woods, and Panzee recruited a person
and pointed toward a specific location on just 3 of these 211 days (C.
Menzel, 1999).
From Trial 1, Panzee was highly effective in attracting the unin-
formed person’s attention, in conveying the type of item hidden, and
in directing the person to the location of the hidden item. A summary
account of how the uninformed persons described Panzee’s behavior
is as follows (C. Menzel, 1999). Panzee gained the person’s attention
by vocalization, by gesturing toward the indoor keyboard, or by mov-
ing to the keyboard when the person happened to come near her cage.
She held her index finger on a lexigram until the person came over
and acknowledged her lexigram use verbally. Panzee then covered
her eyes with her hand, held her arm extended in the direction of the
tunnel leading outdoors, moved to the tunnel, and went outdoors. If
the person did not follow, then Panzee came back inside, again ges-
tured “hide,” beckoned, and pointed outdoors.
Once both Panzee and the person were outdoors, Panzee beckoned
manually and moved to the edge of the enclosure, across from the
hidden object. She sat facing the object and extended her index finger
through the cage wire in the direction of the object. She prompted the
person to search the terrain by gesturing “hide,” by pointing manually
toward the object location with her index finger extended (figure 8.1),
by giving low vocalizations, and by staring toward the location, with
interspersed looks toward the person. She might leave her position,
walk to the outdoor keyboard several meters away, touch a lexigram,
and then return to her original position and resume pointing. During
interactions outdoors, the person was outside the enclosure; Panzee
was inside the enclosure and did not have physical access to the ob-
ject. Because of the orientation of Panzee’s body and gaze and the
persistence of her pointing in a given direction, the person restricted
most of his or her searching to within 1 meter of a straight line be-
tween Panzee and the object location and found the object within a
few minutes, typically within 1 minute. The person determined the
distance to the object, in part, simply by following the direction of
Panzee’s gaze and responding to her relative degree of excitement.
Panzee kept pointing, showed intensified vocalization, shook her arm,
and bobbed her head or body as the person got closer to the site. Once
the person found the object, Panzee stopped pointing outside the en-
closure and stopped gesturing “hide.”
Lexigrams Used During Recruitment

The lexigrams that Panzee touched when she recruited a person typi-
cally corresponded to the type of object she had seen (table 8.1, table
8.2). Considering just the 33 lexigrams corresponding to the 33 types
of objects used in the experiment to be the relevant available choices,
then 84% of Panzee’s 76 lexigram touches corresponded to the specific
type of experimental object presented, rather than to one of the other
32 types of experimental objects (diagonal in table 8.2). It is obvious
that the observed frequency of correct lexigram touches far exceeded
the frequency expected by chance. Furthermore, when Panzee re-
cruited a person, she touched lexigrams that she rarely touched under
routine conditions. On 25 of 34 trials, the lexigram corresponding to
the object hidden was one that accounted for 1% or less of Panzee’s
routine lexigram use. Panzee used the correct lexigram on 21 of these
25 trials, and on the 4 remaining trials Panzee relied effectively on
gesture (C. Menzel, 1999).
Additional findings of interest were as follows. First, Panzee indi-
cated specific nonfood objects (snake, balloon, paper) as well as food
types. Second, the method did not involve any deliberate training.
Even on Trial 1, Panzee used lexigrams, the gesture “hide,” arm point-
ing, and manual pointing toward the object to recruit and direct the
person. Third, Panzee indicated object types after overnight delays;
for example, on Trial 34 she touched the lexigram corresponding to
the object without having seen the outdoor area in 15 hours. I have
replicated this specific finding in more than 30 additional trials.
Fourth, Panzee touched the lexigram stick on 14 of the 34 occasions
on which she recruited a person to a test object. The test objects in
most of the 34 trials were hidden under pine needles and sticks. Even
on the first trial (Trial 3) on which Panzee touched stick, the person
inferred correctly that “bananas were hidden under sticks” before
finding the object location. Finally, on occasion Panzee did not recruit
a person until 2 hours or more had passed, even when the object was
Table 8.1 Lexigrams Panzee Touched in the Context of Recruiting a Person

as a Function of Which Object Panzee Had Been Shown Outdoors
Lexigrams Panzee Used in Context

of Gesturing to Recruit Uninformed Person
Delay
Trial Object Shown (hours) Indoor Keyboard Outdoor Keyboard
1 Kiwi 0.17 grape, kiwi —

b
2 Monster mask 0.18 monster mask —
3 Banana 3.77 stick, banana, stick banana, banana
4 Grapesa 0.15 grape grape, grape, grape,
grape
5 Apple 3.33 hide, apple apple
b
6 Coke 1.17 Coke Coke
7 Pineapple 0.85 pineapple —
a
8 Blanket 2.52 — —
9 Orange drink (can)b 16.20 Coke orange drink, peach,
Coke
10 Strawberriesb 0.88 strawberry —
b
11 Snake 1.62 snake snake
12 String ropeb,c 0.22 surprise, surprise Coke, Coke, string
rope, Coke, stick,
string rope
13 Blueberriesb 0.07 blueberries blueberries
b,d
14 Melon 0.22 orange melon, melon
15 Raisinsa 0.03 raisin —
b
16 Peanuts and balloon 0.17 peanut peanut, peanut,
hide, balloon
17 M&M candiesa 15.62 M&M, M&M M&M, M&M
18 Popsicleb 16.35 surprise, hide, hide, hide, hide, hide,
hide, hide, sur- stick, stick
prise, hide
19 Pear 19.12 stick, hide, hide, stick, hide, stick,
stick, stick, stick stick
20 Orange 19.07 orange (×4); melon, orange
orange
21 Papera and drawing 90.17 — hide (×4); stick,
penb hide, stick, hide,
hide, hide, paper
22 Kiwi 98.85 — hide, stick, hide,
stick, stick, stick,
hide, kiwi, kiwi
(continued)
Table 8.1 Continued
Lexigrams Panzee Used in Context

of Gesturing to Recruit Uninformed Person
Delay
Trial Object Shown (hours) Indoor Keyboard Outdoor Keyboard
23 Cherriesb 0.15 stick, cherries, cherries

cherries
24 Peachesa 90.80 hide peach, hide
25 Jellyb 15.17 jelly, stick jelly
26 Cerealb 304.40 stick, stick peach, peanut, pine-
apple
27 Yogurta 22.50 stick yogurt
a
28 Egg 0.46 stick, egg stick, stick
29 Jell-O gelatinb 22.25 — Jell-O
30 Bread 0.88 bread bread
31 Sugarb 19.97 stick, stick, stick stick, stick
32 Juicea 0.05 juice, juice juice, juice
33 Raisinsa 0.35 stick, stick raisin
34 Pear 15.75 stick, stick, pear pear, pear, pear,
pear
Note. Lexigrams are those that Panzee touched prior to the person finding the object. Dashes
indicate that Panzee did not touch any lexigrams on the keyboard in question. Delay = time
elapsed from Panzee seeing the object until she began to recruit the uninformed person.
a
Lexigram accounted for 1% or less of Panzee’s total lexigram touches in the 12-day baseline
sample.
b
Lexigram rarely used by Panzee under routine conditions and never touched in the 12-day
baseline sample. Panzee had not seen strawberries, string rope, blueberries, melon, balloon,
or cherries in more than 4 months.
c
When Panzee was offered the string rope, she ignored it and again touched Coke on the
keyboard; the person interpreted this as a request for Coke.
d
Panzee traditionally used orange, melon to refer to melon.
Copyright @ 1999 by the American Psychological Association. Reprinted with permission.
a highly preferred type of food; the longest delay with correct report-
ing of object type exceeded 90 hours (table 8.1).
At least five aspects of Panzee’s performance suggested that her
memory involved a fairly rich information retrieval process (C. Men-
zel, 1999). First, Panzee, rather than the experimenters, determined
the exact time of reporting, and she reported items after extended
delays. Second, Panzee selected lexigrams from the whole keyboard
of 256 lexigrams, rather than from a small set of alternatives as in
traditional primate matching-to-sample tasks. Third, she selected the
lexigrams indoors, without an immediate view of the area in which
the object was hidden. Fourth, she did not simply touch lexigrams;
she pointed toward the outdoor area from the vicinity of the indoor
keyboard, and she persisted in the interaction using a variety of re-
sponse outputs until the person found the object outdoors. Fifth, she
used the stick lexigram when recruiting people, and many of the test
objects were covered with leaves and sticks. The findings strongly
suggest that Panzee could recall which one of several dozen types of
objects she had been shown from a distance (and had not been al-
lowed to navigate to or to touch) for overnight periods of at least 16
hours.
Sample Recruitment
The map in figure 8.2 shows an example of a recruitment by Panzee.
Panzee watched from Location 1 in her outdoor enclosure as an exper-
imenter hid M&M candies, a prized food, in the woods at the point
marked X. The experimenter then left the scene. After about 15 min-
utes, Panzee moved to her keyboard indoors (Location 2). She faced
toward John Kelley (J.K.), her favorite caregiver. She held her finger
on the M&M lexigram and vocalized; this attracted J.K.’s attention.
J.K. did not know the type or location of the object. Panzee continued
to hold her finger on the M&M lexigram until J.K. approached and
acknowledged her lexigram use verbally. At that time, Panzee moved
to Location 3 and extended her left arm toward the tunnel, looking
back toward J.K., who agreed to go outside and search for M&Ms.
Panzee went outside. Panzee held her finger on the M&M lexigram
on her outdoor keyboard (Location 4) when J.K. arrived in the woods.
When J.K. acknowledged her lexigram use verbally, Panzee moved to
Location 5 and sat and directed J.K. into the woods to the location of
the hidden M&Ms by pointing with her right hand and vocalizing.
J.K. determined the distance to the object, in part, by following the
direction of Panzee’s gaze and by responding to her relative degree
of excitement. Panzee kept pointing and vocalized more intensely as
J.K. got closer to the site. Panzee also raised her hand higher to send
J.K. further out. After J.K. found the M&Ms, he brought them indoors
and gave them to Panzee. Figure 8.1 shows a representative view of
Panzee directing J.K.
Additional Tests of Recall With Panzee

I have replicated these findings in more than 14 additional experi-
ments, including variations in type of item hidden, the number of
items hidden, the modality of cue giving (direct view versus video
representation), the modality of recruitment (direct pointing versus
reporting on a video representation), and the outcome of recruitment
(whether or not the item was removed and given to Panzee). I now
briefly describe some of the main variations. I do not give all the de-
tails; these are in preparation for publication.
Table 8.2 Lexigrams Panzee Touched in the Context of Recruiting a Person
as a Function of Which Objects Panzee Had Been Shown Outdoors
Lexigrams Panzee Touched
Trial Objects Shown KI MA BA GR AP CO PI BK OD ST SN SR BB ME RA PN
1 Kiwi 1 — — 1 — — — — — — — — — — — —
2 Monster mask — 1 — — — — — — — — — — — — — —
3 Banana — — 3 — — — — — — — — — — — — —
4 Grapes — — — 5 — — — — — — — — — — — —
5 Apple — — — — 2 — — — — — — — — — — —
6 Coke — — — — — 2 — — — — — — — — — —
7 Pineapple — — — — — — 1 — — — — — — — — —
8 Blanket — — — — — — — — — — — — — — — —
9 Orange drink — — — — — 2 — — 1 — — — — — — —
10 Strawberries — — — — — — — — — 1 — — — — — —
11 Snake — — — — — — — — — — 2 — — — — —
12 String rope — — — — — 3 — — — — — 2 — — — —
13 Blueberries — — — — — — — — — — — — 2 — — —
14 Melon — — — — — — — — — — — — — 2 — —
15 Raisins — — — — — — — — — — — — — — 1 —
16a Peanuts — — — — — — — — — — — — — — — 3
and balloon — — — — — — — — — — — — — — — —
17 M&M candy — — — — — — — — — — — — — — — —
18 Popsicle — — — — — — — — — — — — — — — —
19 Pear — — — — — — — — — — — — — — — —
20 Orange — — — — — — — — — — — — — 1 — —
21a Paper — — — — — — — — — — — — — — — —
and draw pen — — — — — — — — — — — — — — — —
22b Kiwi — — — — — — — — — — — — — — — —
23 Cherries — — — — — — — — — — — — — — — —
24 Peach — — — — — — — — — — — — — — — —
25 Jelly — — — — — — — — — — — — — — — —
26 Cereal — — — — — — 1 — — — — — — — — 1
27 Yogurt — — — — — — — — — — — — — — — —
28 Egg — — — — — — — — — — — — — — — —
29 Jell-O gelatin — — — — — — — — — — — — — — — —
30 Bread — — — — — — — — — — — — — — — —
31 Sugar — — — — — — — — — — — — — — — —
32 Juice — — — — — — — — — — — — — — — —
33b Raisins — — — — — — — — — — — — — — — —
34b Pear — — — — — — — — — — — — — — — —
Note. Scores are frequencies of Panzee’s lexigram touches within the trial. The dashes represent frequency
scores of zero. The two-letter abbreviations correspond to the objects shown.
a
Panzee was shown two objects per trial; the experimenter placed the two objects together in the same lo-
cation.
b
Panzee was tested with object for the second time.
Based on table 2, in C. Menzel (1999). Copyright @ 1999 by the American Psychological Association.
Adapted with permission.
206
Lexigrams Panzee Touched
BL MM PS PR OR PP DP KI CH PC JE CE YO EG JO NR SU JU RA PR
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — 1 — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — 1 — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
1 — — — — — — — — — — — — — — — — — — —
— 4 — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — 6 — — — — — — — — — — — — — — —
— — — — — 1 — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — 2 — — — — — — — — — — — —
— — — — — — — — 3 — — — — — — — — — — —
— — — — — — — — — 1 — — — — — — — — — —
— — — — — — — — — — 2 — — — — — — — — —
— — — — — — — — — 1 — — — — — — — — — —
— — — — — — — — — — — — 1 — — — — — — —
— — — — — — — — — — — — — 1 — — — — — —
— — — — — — — — — — — — — — 1 — — — — —
— — — — — — — — — — — — — — — 2 — — — —
— — — — — — — — — — — — — — — — — — — —
— — — — — — — — — — — — — — — — — 4 — —
— — — — — — — — — — — — — — — — — — 1 —
— — — — — — — — — — — — — — — — — — — 5
207
0 5m
JK 98º
X
5
2
3 1
0º
Figure 8.2 Map of Panzee’s recruitment of her favorite caregiver (J.K.) to M&M
candies (X) in the woods. 1 = Panzee’s location during cue giving. 2 = Her position
at indoor keyboard. 3 = Her position while arm pointing toward outdoor enclo-
sure. 4 = Her position at outdoor keyboard. 5 = Her position while pointing toward
the hidden M&M’s. Dark line is Panzee’s approximate path of travel; dotted line
is J.K.’s path. The outer box around the figure is imaginary; the woods extend a
large distance in all directions. Vertical line to 0 degrees = invisible west-oriented
vector used as reference for computing angle to the M&M’s (shown by invisible
line to 98 degrees). Across 54 sample trials in which Panzee had a direct view of
an object being hidden in the woods, the Pearson r between the “angle of the
hidden object” and the “angle of the first location that the person searched, as a
result of Panzee’s pointing” was 0.99.
Use of Indirect Cues of Object Type Panzee retained higher order, indi-
rect signs of the type of object hidden. If she saw an experimenter
hide an opaque container in the forest with a lexigram on it (corre-
sponding to the type of food inside the container), she reliably re-
tained the specific type of food for one or more days. Similarly, if the
experimenter hid an opaque container in the forest and told Panzee
in English the type of food that was inside the container, Panzee re-
tained the type of food for at least 15 minutes (Panzee understands
approximately 150 spoken English words; Beran et al., 1998). Reten-
tion of auditory cues of object type beyond even 60 seconds is consid-
ered exceptional for chimpanzees (Hashiya & Kojima, 2001a, 2001b).
Use of Televised Cues Panzee used a video representation of the forest
as a guide to locating objects in the forest. If Panzee was shown a
brief videotape of a food-hiding event that had been filmed in the

forest behind her enclosure, she reliably recruited an uninformed per-
son and directed him or her to the hidden food on the next day. Pan-
zee performed very accurately in a series of more than 30 trials, in-
cluding Trial 1. She also typically reported the type of item she had
seen on the cue tape. Each trial used a different location. On most of
the trials, Panzee was confined indoors overnight immediately after
she viewed the cue tape, and thus she did not have an opportunity to
see the forest for more than 15 hours. Furthermore, the angle of the
camera during the filming of the food-hiding event differed by up to
120 degrees from Panzee’s familiar angle of view of the forest, and on
some of the trials the video cue tape presented her with a perspective
of the forest that she could not possibly have had from her outdoor
enclosure.
Discrimination Between Relevant and Irrelevant Video Representations Pan-

zee discriminated sharply between relevant and irrelevant videotapes
of food-hiding events. When Panzee was shown a videotape of a food-
hiding event that had been filmed in a distant forest that she had
never seen, she watched the videotape with interest, but she did not
subsequently recruit a person or point toward the forest behind her
own enclosure within 40 hours. These 10 control trials were inter-
mixed with the test trials that had been filmed behind Panzee’s out-
door enclosure. As usual, the caregiver did not know whether a trial
was being conducted.
Reporting of Hidden Objects on a Video Representation Panzee could

transfer in the reverse direction, from real life to television. If she
watched the experimenter hide an object in the forest in real life, and
the experimenter then locked her indoors overnight, she was able to
report the exact location of the hidden object to an uninformed person
the next day by touching the corresponding location on a televised
image of the forest with a stick or a joystick-controlled laser pointer.
Panzee did this accurately and reliably in a series of more than 20
trials, including on Trial 1.
Use of Laser Pointer in Woods Panzee used a laser pointer to pinpoint
the location of hidden items in the woods. On overcast days, an exper-
imenter hid an object in the woods, 3 to 20 meters from Panzee’s out-
door enclosure. Each of the 20 trials used a different location. After a
delay of approximately 5 minutes, a person who did not know where
the object was hidden or the type of object gave Panzee access to a
joystick. The joystick controlled a motorized pan head, on which a
laser pointer was mounted. The laser pointer itself was located out-
side the cage and was not accessible to Panzee. Panzee controlled the
location of the laser dot in the woods by manipulating the joystick.
The area available for error was more than 100 square meters. Panzee
moved the dot to a location, then took her hand off the joystick, shook
her arm and bobbed her head while looking at the person. The person
then walked into the woods and manually inspected the location of
the dot. On all 20 trials, Panzee put the dot within 1 meter of the
hidden object (within 15 centimeters on most trials, including a trial
at 20 meters) and the person found the hidden object without further
search. Panzee also named all 20 hidden objects correctly using lexi-
grams before the person had checked the location.
Factors That Did Not Cause Panzee to Forget Hidden Objects

Panzee initiated the interactions with humans, and she did not appear
to forget her intended actions when they were interrupted. For exam-
ple, if she worked on a computer-presented cognitive task, or if she
engaged in social interactions (e.g., grooming) during a several-hour-
long retention interval, she still reported the object and directed a per-
son to it. Similarly, if she recruited a person and pointed toward the
hidden object but the person could not find the object, Panzee would
break off the interaction, but she would resume the recruitment at a
later time. (As mentioned earlier, Panzee was confined to her enclo-
sure and had to rely on a person to get the object; she could not sim-
ply get the object herself.) In several trials, Panzee continued to recruit
people and point toward a hidden object over a period of 2 weeks
until a person finally found the object. Finally, if the person found the
object but merely replaced it in the forest rather than bringing it in-
doors, Panzee continued to report the location at later times and to
different people until the object was removed. She did so with high
reliability in a series of more than 20 trials.
Inferences About Objects and Locations

Given the durability of this chimpanzee’s memory for single food lo-
cations (see also Yerkes, 1943, p. 148; Tinklepaugh, 1932), maybe a
more basic question than “Why does she remember?” is “Why does
she forget?” or “Why does she stop reporting locations?” I have iden-
tified some of the factors that cause Panzee to stop reporting locations.
If Panzee reported a hidden object and saw a person remove it from
the woods, she rarely reported that location again. That is, she rarely
showed false positives. Panzee also behaved as if she could infer
whether an object had been removed from the forest, based on events
that she had seen indoors. In one experiment, Panzee saw an experi-
menter hide a food item in the forest as usual. After Panzee had
“named” the item and initiated recruitment of a caregiver, however,
the caregiver confined Panzee indoors. Panzee then reported the loca-
tion of the item on a televised image of the woods, as described ear-
lier. The caregiver then left the building, walked to the woods, and
after several minutes reentered the building, carrying the same type
of item that Panzee had seen hidden in the forest. Under these condi-
tions (more than 20 trials), Panzee did not subsequently recruit a per-
son and point toward the location where she had seen the item hid-
den, despite opportunities to do so. It was as if Panzee assumed that
the food item the caregiver brought back was the same item Panzee
had seen hidden in the forest, even though Panzee could not see the
forest or see or hear what the person was doing outside the building.
In a control series, when Panzee initiated recruitment, the caregiver
locked her indoors, but then simply gave her the named type of item
from the refrigerator instead of going outdoors. The refrigerator was
located near Panzee’s indoor cage and was visible to her. On all 20
trials of this type, Panzee subsequently recruited a person and di-
rected him or her to the hidden item in the woods. In another experi-
ment, two different types of objects were hidden in two different loca-
tions in the woods. When Panzee came indoors, “named” one of the
objects, and initiated recruitment of a caregiver, the caregiver locked
her indoors, went outdoors for several minutes, and then reentered
the building carrying one item that was of the same type as one of
the items that Panzee had seen hidden in the woods (but not necessar-
ily the one that Panzee had named). Under these conditions, Panzee
subsequently recruited a person and directed him or her to the loca-
tion of the other type of object, that is, to the remaining type of item
that the caregiver had not brought indoors. Again, it was as if Panzee
assumed that the food item the caregiver brought back to the building
was one of the food items Panzee had seen hidden in the woods. I am
currently working with this procedure to try to determine the more
specific cues and assumptions that Panzee uses to discriminate when
a location in the forest has been emptied.
Massed Trials
Tulving suggested the following specific test to assess episodic mem-
ory in Panzee. I presented Panzee with a massed series of trials using
just two locations and two types of objects. The delay interval between
cue giving and reporting within each trial was long enough to involve
long-term memory. The test was conducted indoors, to facilitate the
presentation of many trials in succession. When humans are given a
long series of similar trials in succession with delay intervals of sev-
eral minutes, they can find it increasingly difficult to discriminate the
current trial from the preceding trials. In principle, the test places de-
mands on memory for the time of occurrence of events. The question
was whether Panzee could sustain high accuracy as the test session
progressed.
To begin a trial, the experimenter hid either a slice of apple or a
slice of banana under one of two identical opaque containers while
Panzee watched. The containers were 1.5 meters apart and 1 meter
from Panzee’s cage. The experimenter left nothing under the other
container and left the area. After a delay interval of 5 minutes, a care-
giver entered the ape area. The caregiver knew that a trial was under-
way but did not know which type of item was present or which con-
tainer was baited. Panzee indicated her selection of one of the two
containers by pointing. Before lifting the container that Panzee had
designated, the caregiver presented Panzee with a keyboard that dis-
played 256 lexigrams and noted which lexigram Panzee touched. If
Panzee selected both the correct location and the correct lexigram (cor-
responding to the food type), then the caregiver removed the food and
gave it to Panzee. Otherwise, the caregiver removed the food without
giving it to Panzee. Intertrial intervals were approximately 1 minute.
A session lasted approximately 90 minutes and consisted of 12 trials.
Eight sessions (96 trials) were conducted. The type of item and the
container baited varied across trials in an irregular, quasi-random se-
quence.
Panzee was correct on item type on 96% of trials, slightly higher
than her average in outdoor tests; she was correct on location on 88%,
and correct on both item and location on 84% of trials. On the last
trial of the session (Trial 12), Panzee was still accurate; she was correct
on Trial 12 on location on 7 of 8 sessions, correct on item on 7 of 8
sessions, and correct on both item and location on 6 of 8 sessions.
Thus, Panzee appeared to be resistant to the sort of proactive interfer-
ence that can impair performance in amnesics (Schacter, Moscovitch,
Tulving, McLachlan, & Freedman, 1986).
New Aspects of the Study

The key empirical issue of this and many other recent studies of ani-
mal memory is how information about the environment and about
the animal’s own past experiences is organized. The data strongly
suggest that, for Panzee, information is to a significant extent orga-
nized independently of stimulation from the environment in question
and constitutes recall as opposed to recognition. The “time in which
chimpanzees live” remains an open question, but Panzee could cer-
tainly recall which one of at least 30 different types of objects she had
seen outdoors, after delays of more than 48 hours.
This study presents several new features compared to other work
on animal memory. The most obvious respect is that Panzee used
lexigrams to report hidden objects, and that she did so from locations
that were visually separated from the place where the event occurred,
after substantial time delays. Panzee transmitted multiple features of
events, not just a single feature. In particular, she used lexigrams in
conjunction with a video representation of the woods to convey both
the type and the location of objects she had seen hidden in the woods
on the previous day.
Second, evidence for generalized spatial skills came from the fact
that Panzee used a video representation of the woods as a guide to
locating objects in the woods, after overnight delays. Her use of tele-
vised spatial information far exceeded that reported for other nonhu-
man organisms, including juvenile chimpanzees (E. Menzel, Premack,
& Woodruff, 1978), baboons (Vauclair, 1990), and dolphins (Herman,
Pack, & Palmer, 1993), in regard to extracting spatial information from
a video representation and using that information at a much later time
in an outdoor area. What is new is the long time span over which
Panzee retained televised spatial information, the large number of dif-
ferent locations used in the test, her ability to withstand changes in
visual perspective, her use of a video representation to report loca-
tions, and her labeling of the item type by lexigram.
Another relatively novel aspect of this study compared to other
research on animal memory concerns the physical separation between
Panzee and the physical properties of the objects. During the informa-
tion-gathering phase of each trial, Panzee was not allowed to handle
the objects herself. She had to watch the object-hiding events from a
distance. She did not always even see the hiding event directly; some-
times she saw only a video representation of the event. Later, she had
to solicit the assistance of an uninformed person and then had to
guide the person to the hidden object by pointing from a distance,
or by using a stick to touch the corresponding location on a video
representation of the forest. She was not allowed simply to take the
person by the arm to the locations. Furthermore, she had to watch the
object-removal events from a distance; and she did not always even
see the removal event directly. As stated earlier, Panzee had not en-
tered the forested area within 6 years prior to the onset of this study.
Thus, her memory of the presence, type, and location of test objects
in the forest was based on observations of events from a distance.
These aspects of her performance seem very different from memory
studies using food-storing animals. In such studies, animals usually
hide the items themselves and receive proprioceptive and close visual
feedback from the location during the information-gathering part of
the trial, both of which can improve memory. To paraphrase Nissen
(1951, pp. 379–380) and Pavlov (1904), the more attenuated the basis
of association between cues and objects, the more likely we are to
assume that some form of symbolic or sign process contributes to the
problem solving.
Finally, an aspect of the study that is relatively unique, compared
to most studies of human memory as well as animal studies, is that
Panzee’s recall and reporting of hidden objects were unprompted. In
studies of human recall memory, the experimenter typically tells the

participant when to provide information. Even in studies of “free re-
call,” the experimenter typically tells the person when to report. In
humans, the ability to recall goals, and the ability to monitor which
intended actions have and have not occurred, is somewhat fragile and
can be disrupted by aging and brain injuries (Einstein, McDaniel,
Manzi, Cochran, & Baker, 2000; Johnson, DeLeonardis, Hashtroudi, &
Ferguson, 1995). In this study, Panzee, rather than the human caregiv-
ers, initiated the exchanges of information about hidden objects, and
she persisted in reporting a given location for days, if the search was
interrupted by outside disturbances or if the person failed to find the
hidden item.
Recall Versus Recognition Memory

The sine qua non of recall in humans is often said to be verbal or
other self-generated reporting of events. In the experiments described
in this chapter, Panzee initiated the exchanges of information about
hidden objects, and she conveyed which one of at least several dozen
types of objects she had seen hidden outdoors, by touching lexigrams.
Furthermore, Panzee did not merely touch lexigrams: She used a wide
range of response outputs to convey her message, and she did so from
Trial 1 in a series of trial-unique experimental tests. The overall orga-
nization of her behavior suggested that she could retrieve information
indoors about the presence, types, and locations of objects that she
had seen outdoors hours or days earlier.
Panzee did not literally draw a lexigram or a map of the forest. She
chose lexigrams from a large set of options. Thus, the relevant visual
stimuli used in the reporting situation were not purely self-generated.
Does this mean that Panzee’s selection of lexigrams at her keyboard
could be better described as a recognition task, in which there were
256 choices, rather than as a recall task? I think not. On several hun-
dred trials, Panzee vocalized persistently, then (when the caregiver
approached) rushed to the keyboard and in less than one second tou-
ched the lexigram corresponding to the type of item hidden, and then
gestured toward and headed up to the tunnel that leads outdoors. In
light of Panzee’s nearly immediate selection of the correct lexigram
on the complex keyboard, it seems likely that she already knew the
type of item and/or the lexigram she was going for before she ap-
proached the keyboard. It seems doubtful that in these cases Panzee
knew at best that something was outdoors but not the type of item,
and that she only had this information after she had scanned the key-
board and recognized the lexigram. One might still ask whether Pan-
zee had scanned the keyboard and had recognized the lexigram just
moments before she had vocalized and attracted the caregiver’s atten-
tion. However, on more than 30 trials Panzee was locked in a special
cage overnight without any access to a keyboard. When the caretaker

entered the building in the morning, Panzee vocalized and gestured
persistently to the adjacent cage; as soon as the caregiver opened the
door between the cages, Panzee rushed to the keyboard in the adja-
cent cage and touched the correct lexigram in less than one second,
and pointed toward the tunnel. Again, Panzee’s speed of touching the
correct lexigram on these trials suggested that she already knew the
type of object and/or the lexigram before moving to the keyboard,
without having seen the outdoor area or the keyboard during the
overnight retention interval.
To be sure, Panzee did not close her eyes and give a verbal report
of what she had seen or heard. Regardless of whether her perfor-
mance satisfies everyone’s definition of “self-generated, active recall,”
what is important is the degree to which Panzee’s memory is orga-
nized independent of stimulation from the original environment in
question. The lexigrams that Panzee touched did not resemble the ob-
jects that she had seen hidden, and of course the lexigrams did not
resemble the spoken English words that Panzee had heard the experi-
menter say aloud, in some trials, as cues of object type. The video
representation of the woods on which Panzee reported object loca-
tions differed in numerous respects from a direct view of the woods
itself (e.g., it did not provide motion parallax, and it presented a dif-
ferent visual perspective). If we envision a continuum between recall
and recognition, then Panzee’s performance is much farther in the
direction of recall than any other animal performance of which I am
aware, or at least any published experimental study.
WHAT IS ESSENTIAL IN OUR FIELD?

In this volume, Tulving suggests that there has been no essential prog-
ress on the problem of self-knowing consciousness in animals in the
past 100 years. Tulving is certainly entitled to his personal opinion,
and I admire him for stating it so explicitly, but when he speaks for
psychology in general, I must differ.
James (1890) pointed out that the only meaning of “essence” is tele-
ological, in the sense that it is defined in relation to a particular set of
interests. James encouraged other researchers to look for themselves
and decide what is essential. Insofar as one remains a naturalist or
experimentalist, there is always more to learn regarding what is and
what is not essential. No finite list is ever likely to be permanently
exhaustive and definitive.
One hundred years ago, psychologists were disputing whether any
nonhuman could be said to remember anything at all, or possess
memory of any sort, or learn anything by observation, or be capable
of one-trial as opposed to slow gradual learning, or communicate, or
recognize themselves, or symbolize in any more profound a fashion

than a dog that salivates at the sound of a dinner bell, or not merely
behave but sometimes show signs of comprehension as well (Thorn-
dike, 1898). Few current scientists would hesitate to answer any of the
above questions with at least a weak affirmative. That counts as prog-
ress of sorts, especially since the debates were settled mainly by em-
pirical data rather than polemics. The issue of what types of data,
methods, and questions are permissible is currently seen in a very
different way because of the information that was obtained by animal
researchers. The amount of data on the topics listed above in nonhu-
man primates has expanded exponentially, partly as a result of re-
search by Köhler and Yerkes, the establishment of national primate
research centers, field studies, ethology, sociobiology, and cognitive
studies. Many of the advances in the study of perception and cogni-
tion in human infants have been based on methods originally devel-
oped with nonverbal animals. This is progress toward meeting Tulv-
ing’s challenge, because if we are not in agreement on the above
questions, there is no way we can reach agreement on subsidiary is-
sues that are far more specific, yet even more inferential and difficult
to operationalize.
The boundary line of “self-reflective consciousness” is vague
(Wheeler, 2000), and we cannot know in advance which phenomena
of animal behavior will shed light on the main issue at hand. Progress
on Tulving’s problem can be made through empirical research on a
broad front. Improvements in methodology will remain especially im-
portant (Rumbaugh & Washburn, 2003). Hunter’s (1913) delayed-
response procedure did not definitively answer Hunter’s original
question (whether animals have “ideas”), but it has made major con-
tributions to our understanding of observational learning, spatial and
object memory, recall, and communication among other topics. The
present chapter illustrates how delayed-response procedures have
clarified the operation of chimpanzee memory in situations that are
somewhat loosely structured but that allow the subjects adequate
space, time, environmental structures, and social options to express
their interests and abilities in a well-developed form. Additional
methodological improvements discussed here include the use of spe-
cialized rearing histories, lexigram competencies, and technologies
(keyboards, video representations, and pointing devices) to study
chimpanzee recall.
Should “Self-Knowing Consciousness” Be Taken as an Axiom,

or as a Null Hypothesis to Be Tested Experimentally?
There is a difference between the point of view of a caregiver and that
of a critic or an experimenter. We need both points of view. A human
mother does not wait for valid evidence that her infant already has
the capacity for language, memory, empathy, and self-concept before

treating her infant as if it will be able to become a real, intelligent
human being. She takes the capacity for self-reflective consciousness
as a given, not as a hypothesis to be tested. Her faith that her infant
definitely will be able to become a real, caring, intelligent human is
what prevents the infant from being treated inhumanely. How accu-
rate is a mother’s assessment of an infant’s perceptual and cognitive
capabilities, compared to the assessment of an experimentalist? His-
torically, the mother has sometimes been far more accurate. Many of
William James’s (1890) statements about the perceptual world of hu-
man infants were flatly wrong, and he would have been the first to
say that in many respects his wife had the better understanding of the
consciousness of their own infants.
This raises the question: Is there any single best way to rear chim-
panzees for the purposes of studies such as this one? I do not think
there is. The cognitive capabilities of normally reared chimpanzees in
their natural habitat are still basically unknown (Boesch & Boesch-
Achermann, 2000). With regard to laboratory studies, we have learned
a great deal since the home-rearing studies of Kellogg and Kellogg
(1933) and Ladygina-Kohts (1935/2002), but every generation and ev-
ery infant is different (Rumbaugh & Washburn, 2003). The people
who reared Panzee assumed that immersing her in an enriched com-
municative environment would support the emergence of diverse
cognitive abilities, some of which might not become apparent until
years later. Several of these people, including Sue Savage-Rumbaugh,
Rose Sevcik, and Elizabeth Pugh, had already reared and extensively
tested two other symbol-competent chimpanzees (Sherman and Aus-
tin) and two symbol-competent bonobos (Kanzi and Malika). Their
assumptions about Panzee’s developmental potential were informed
by daily interactions with her plus experience with the developmental
end points and behavior of other individual apes.
Experimental studies (Beran et al., 1998; Brakke & Savage-Rumba-
ugh, 1995, 1996; C. Menzel, 1999), conducted years later and designed
with attention to Panzee’s particular interests, preferences, and exist-
ing daily activities have provided strong support for her mother sur-
rogates’ assumptions regarding Panzee’s cognitive and developmental
potential. In particular, Panzee can comprehend spoken English
words (Beran et al., 1998), can perform recall in different situations
and in different ways (C. Menzel, 1999), and can comprehend video
representations at levels that far exceed the levels previously demon-
strated experimentally in any common chimpanzee. By comparison
with other adult chimpanzees, she seems exceptionally interested in
humans and exceptionally eager to engage them in communication
about the environment, but as yet I have not attempted to get formal
data on that.
Does Panzee Possess True (Human) Episodic Memory?

Human episodic memory, as defined by Tulving (1983) and Tulving
and Markowitsch (1998), includes the ability to perform recall in dif-
ferent situations and conscious awareness of at least some aspects of
the spatial and temporal contexts of the recalled event, or “mental
time travel.” As stated earlier, the boundary line of these capacities is
vague. The potential value of further research on nonhuman animals
should not be underestimated, and it would be premature at this early
stage of research to claim that chimpanzees lack these skills entirely.
With regard to her memory of the temporal (sequential) aspects of
events, it can be noted that Panzee kept track accurately over days
whether hidden objects had been removed from the woods, or had
been replaced in the same location, or had been moved to another
location. She showed few false positives, and she rarely appeared to
confuse the locations or the object types of the previous trial with
those of the current trial, even in a study in which she watched the
experimenter hide six different types of items in six different locations
per trial. It is as if she discriminated sharply between tasks that were
“completed” and those that were still “uncompleted” (Zeigarnik, 1927,
cited in Woodworth & Schlosberg, 1954). She also performed accu-
rately in massed trials that used just two types of items and two loca-
tions repeatedly.
Furthermore, Panzee’s use of stored information appeared to be
controlled and strategic. Her decision of whether and when to recruit
appeared to be based on a deliberate choice between known alterna-
tives; she seemed almost constantly to take into account what the care-
givers were doing, what they were likely to do in the near future, the
availability of alternative foods indoors, and the presence, type, and
location of a large number of goal objects over a large spatial extent.
For example, after seeing an experimenter hide a highly prized box of
juice in the woods in the afternoon, Panzee might recruit the assis-
tance of the caregiver to recover the juice within a few minutes after
seeing the event. If, however, she saw the experimenter hide a some-
what less preferred item (e.g., an orange), she might come indoors,
touch the sweet potato lexigram, and vocalize and point toward the
microwave in which she saw the caregiver put sweet potatoes earlier
in the day, then use lexigrams, vocalizations, and manual pointing to
request television viewing, or touch the blanket lexigram and point
toward the laundry room where blankets (for nest making) are kept,
then only on the next morning touch the orange lexigram, point to-
ward the woods, and recruit a caregiver’s assistance to recover the
orange outdoors. To say that Panzee appears to be capable of recalling
multiple features of her environment quickly and accurately, that she
appears to use this recalled information strategically, that she appears
throughout the day to be attempting to manipulate the outcome of

anticipated events, that she controls her environment indirectly by
controlling her caregivers, and that she uses a large range of behaviors
flexibly and interchangeably to exert this influence is not simply a
poetic metaphor. It is close to the direct impression gained by observ-
ing the interactions between Panzee and her caregivers in the daily
situations and tests described earlier.
As to what exactly makes some people believe that Panzee has the
same kind of autonoetic, self-knowing consciousness that humans do,
and why others will remain skeptical or even scornful of that possibil-
ity, one prime variable is assuredly firsthand personal experience:
with animals, with Panzee herself, with one or more live test sessions,
and with the investigators themselves. The pictures, tables, figures,
and written descriptions provided in this chapter are a pale represen-
tation of the real thing. If they provided all the information that any-
one should ever need, then funding agencies surely would not invest
so much in site visits and conferences. A person does not have to
work for years as a chimpanzee researcher or caregiver to experience
these things. First-time visitors, including neurobiologists and stu-
dents, who see a single live test session are typically impressed with
the forcefulness and clarity of Panzee’s behavior and with the signifi-
cance of her behavior for memory.
There are at least two approaches to the problem of self-knowing
consciousness in animals. One approach is to attack the problem head
on, by presenting animals with arbitrarily designed tasks in the tradi-
tion of Thorndike (1898). The other approach is indirect, by examining
what the animal already does. My study of Panzee is an example of
the second approach. To paraphrase Wehner (1996), I did not begin
with an attempt to test a preconceived idea about how a memory
system or a symbol system might work in principle. Rather, I have
begun by taking a detailed look at what this particular chimpanzee
does in her particular ecological and social context.
ACKNOWLEDGMENTS Thanks are due to John Kelley, Michael Beran, Stephanie
Berger, Christopher Elder, Cameron Hastie, John Mustaleski, Isabel Sanchez,
and Shelly Williams for their help conducting the experiments with Panzee.
This research was supported by the L.S.B. Leakey Foundation, the Wenner-
Gren Foundation, National Science Foundation grant SBR-9729485, and Na-
tional Institute of Health grants MH-58855, HD-38051, and HD-06016.
REFERENCES
Aristotle. (1990). Memory and reminiscence. In Great books of the western world:
The works of Aristotle, Vol. 1 (pp. 690–695, J. I. Beare, Trans.). Chicago: Ency-
clopaedia Britannica.
Balda, R. P., & Kamil, A. C. (1992). Long-term spatial memory in Clarks’s
nutcracker, Nucifraga columbiana. Animal Behaviour, 44, 761–769.
Bayley, P. J., & Squire, L. R. (2002). Medial temporal lobe amnesia: Gradual
acquisition of factual information by nondeclarative memory. Journal of
Neuroscience, 22, 5741–5748.
Beran, M. J., & Beran, M. M. (2004). Chimpanzees remember the results of
one-by-one addition of food items to sets over extended time periods. Psy-
chological Science, 15, 94–99.
Beran, M. J., Savage-Rumbaugh, E. S., Brakke, K. E., Kelley, J. W., & Rum-
baugh, D. M. (1998). Symbol comprehension and learning: A “vocabulary”
test of three chimpanzees (Pan troglodytes). Evolution of Communication, 2,
171–188.
Bickerton, D. (1995). Language and human behavior. Seattle: University of Wash-
ington Press.
Bischof, N. (1978). On the phylogeny of human morality. In G. S. Stent (Ed.),
Morality as a biological phenomenon (pp. 48–66). Berkeley: University of Cali-
fornia Press.
Boesch, C., & Boesch-Achermann, H. (2000). The chimpanzees of the Täi forest.
Brakke, K. E., & Savage-Rumbaugh, E. S. (1995). The development of lan-
guage skills in bonobo and chimpanzee: I. Comprehension. Language and
Communication, 15, 121–148.
Brakke, K. E., & Savage-Rumbaugh, E. S. (1996). The development of lan-
guage skills in Pan. II. Production. Language and Communication, 16, 361–
380.
Clayton, N. S., & Dickinson, A. (1999a). Memory for the content of caches by
scrub jays (Aphelocoma coerulescens). Journal of Experimental Psychology: Ani-
mal Behavior Processes, 25, 82–91.
Clayton, N. S., & Dickinson, A. (1999b). Scrub jays (Aphelocoma coerulescens)
remember the relative time of caching as well as the location and content
of their caches. Journal of Comparative Psychology, 113, 403–416.
Clayton, N. S., Griffiths, D. P., Emery, N. J., & Dickinson, A. (2002). Elements
of episodic-like memory in animals. In A. Baddeley, J. P. Aggleton, &
M. A. Conway (Eds.), Episodic memory: New directions in research. New York:
Oxford University Press.
Donald, M. (1993). Precis of origins of the modern mind. Behavioral and Brain
Sciences, 16, 737–791.
Einstein, G. O., McDaniel, M. A., Manzi, M., Cochran, B., & Baker, M. (2000).
Prospective memory and aging: Forgetting intentions over short delays.
Psychology and Aging, 15, 671–683.
Gaffan, D. (1994). Scene-specific memory for objects: A model of episodic
memory impairment in monkeys with fornix transection. Journal of Cogni-
tive Neuroscience, 6, 305–320.
Gallistel, C. R. (1990). The organization of learning. Cambridge, MA: MIT Press.
Gallup, G. G., Jr. (1970). Chimpanzees: Self-recognition. Science, 167, 86–87.
Garber, P. (1989). Role of spatial memory in primate foraging patterns: Saguinus
mystax and Saguinus fuscicollis. American Journal of Primatology, 19, 203–216.
Hashiya, K., & Kojima, S. (2001a). Acquisition of auditory-visual intermodal

matching-to-sample by a chimpanzee (Pan troglodytes): Comparison with
visual-visual intramodal matching. Animal Cognition, 4, 231–239.
Hashiya, K., & Kojima, S. (2001b). Hearing and auditory-visual intermodal
recognition in the chimpanzee. In T. Matsuzawa (Ed.), Primate origins of
human cognition and behavior (pp. 155–189). New York: Springer.
Herman, L. M., & Forestell, P. H. (1985). Reporting presence or absence of
named objects by a language-trained dolphin. Neuroscience and Biobehav-
ioral Reviews, 9, 667–681.
Herman, L. M., Pack, A. A., & Palmer, M. (1993). Representational and con-
ceptual skills of dolphins. In H. L. Roitblat, L. M. Herman, & P. E. Nachti-
gall (Eds.), Language and communication: Comparative perspectives (pp. 403–
442). Hillsdale, NJ: Lawrence Erlbaum.
Hunter, W. (1913). The delayed reaction in animals and children. Behavior
Monographs, 2(1), Serial No. 6.
James, W. (1890). Principles of psychology. New York: Holt.
Janson, C. H. (1998). Experimental evidence for spatial memory in foraging
wild capuchin monkeys, Cebus apella. Animal Behaviour, 55, 1229–1243.
Johnson, M. K., DeLeonardis, D. M., Hashtroudi, S., & Ferguson, S. A. (1995).
Aging and single versus multiple cues in source monitoring. Psychology
and Aging, 10, 507–517.
Kawai, N., & Matsuzawa, T. (2001). Reproductive memory processes in chim-
panzees: Homologous approaches to research on human working memory.
In T. Matsuzawa (Ed.), Primate origins of human cognition and behavior (pp.
226–234). Tokyo: Springer-Verlag.
Kellogg, W. N., & Kellogg, L. A. (1933). The ape and the child. New York: Mc-
Graw-Hill.
Köhler, W. (1925). The mentality of apes (E. Winter, Trans.). New York: Live-
right.
Kuhlmeier, V. A., Boysen, S. T., & Mukobi, K. L. (1999). Scale model compre-
hension by chimpanzees (Pan troglodytes). Journal of Comparative Psychology,
113, 396–402.
Ladygina-Kohts, N. N. (2002). Infant chimpanzee and human child: A classic 1935
comparative study of ape emotions and intelligence (F. B. M. de Waal, Ed.).
New York: Oxford University Press. (Original work published 1935)
Menzel, C. R. (1991). Cognitive aspects of foraging in Japanese monkeys. Ani-
mal Behaviour, 41, 397–402.
Menzel, C. R., Savage-Rumbaugh, E. S., & Menzel, E. W., Jr. (2002). Bonobo
(Pan paniscus) spatial memory and communication in a 20-hectare forest.
International Journal of Primatology, 23, 601–619.
Menzel, E. W., Jr. (1973a). Chimpanzee spatial memory organization. Science,
182, 943–945.
Menzel, E. W., Jr. (1973b). Leadership and communication in young chimpan-
zees. In E. Menzel, Jr. (Ed.), Symposium of the Fourth International Congress of
Primatology: Vol. 1. Precultural primate behavior (pp. 192–225). Basel: Karger.
Menzel, E. W., Jr. (1978). Cognitive mapping in chimpanzees. In S. H. Hulse,
H. Fowler, & W. K. Honig (Eds.), Cognitive processes in animal behavior (pp.

375–422). Hillsdale, NJ: Lawrence Erlbaum.
Menzel, E. W., Jr. (1984). Spatial cognition and memory in captive chimpan-
zees. In P. Marler & H. S. Terrace (Eds.), The biology of learning (pp. 509–
531). New York: Springer-Verlag.
Menzel, E. W., Jr., & Juno, C. (1985). Social foraging in marmoset monkeys
and the question of intelligence. Philosophical Transactions of the Royal Soci-
ety of London, Series B, 308, 145–158.
Menzel, E. W., Jr., Premack, D., & Woodruff, G. (1978). Map reading by chim-
panzees. Folia Primatologica, 29, 241–249.
Menzel, E. W., Jr., Savage-Rumbaugh, E. S., & Lawson, J. (1985). Chimpanzee
(Pan troglodytes) spatial problem solving with the use of mirrors and tele-
vised equivalents of mirrors. Journal of Comparative Psychology, 99, 211–217.
Menzel, R., Geiger, K., Chittka, L., Joerges, J., Kunze, J., & Muller, U. (1996).
The knowledge base of bee navigation. Journal of Experimental Biology, 199,
141–146.
Mishkin, M., Suzuki, W. A., Gadian, D. G., & Vargha-Khadem, F. (1997). Hier-
archical organization of cognitive memory. Philosophical Transactions of the
Royal Society of London, Series B, 352, 1461–1467.
Nissen, H. (1951). Phylogenetic comparison. In S. S. Stevens (Ed.), Handbook
of experimental psychology (pp. 347–386). New York: Wiley.
Olton, D. (1978). Characteristics of spatial memory. In S. H. Hulse, H. Fowler,
& W. K. Honig (Eds.), Cognitive processes in animal behavior (pp. 341–373).
Hillsdale, NJ: Lawrence Erlbaum.
Olton, D. S. (1984). Comparative analysis of episodic memory. Behavioral and
Parkinson, J. K., Murray, E. A., & Mishkin, M. (1988). A selective role for the
hippocampus in monkeys: Memory for the location of objects. Journal of
Neuroscience, 8, 4159–4167.
Pavlov, I. (1904). Physiology of digestion. Nobel lecture, December 12, Stock-
holm.
Roberts, A. C. (1996). Comparison of cognitive function in human and non-
human primates. Cognitive Brain Research, 3, 319–327.
473–489.
Ruggiero, F. T., & Flagg, S. F. (1976). Do animals have memory? In D. L.
Medin, W. A. Roberts, & R. T. Davis (Eds.), Processes of animal memory (pp.
1–19). Hillsdale, NJ: Lawrence Erlbaum.
Rumbaugh, D. M., & Washburn, D. A. (2003). Intelligence of apes and other
rational beings. New Haven, CT: Yale University Press.
Savage-Rumbaugh, E. S., McDonald, K., Sevcik, R. A., Hopkins, W. D., & Rub-
ert, E. (1986). Spontaneous symbol acquisition and communicative use by
pygmy chimpanzees (Pan paniscus). Journal of Experimental Psychology: Gen-
eral, 115, 211–235.
Savage-Rumbaugh, E. S., Murphy, J., Sevcik, R., Brakke, K. E., Williams, S. L., &
Rumbaugh, D. M. (1993). Language comprehension in ape and child. Mono-
graphs of the Society for Research on Child Development, 58(3–4, Serial No. 233).
Savage-Rumbaugh, E. S., Pate, J. L., Lawson, J., Smith, S. T., & Rosenbaum, S.
(1983). Can a chimpanzee make a statement? Journal of Experimental Psy-

Savage-Rumbaugh, E. S., Rumbaugh, D. M., & McDonald, K. (1985). Lan-
guage learning in two species of apes. Neuroscience and Biobehavioral Re-
views, 9, 653–665.
Schacter, D. L., Moscovitch, M., Tulving, E., McLachlan, D. R., & Freedman,
M. (1986). Mnemonic precedence in amnesic patients: An analogue of the
AB error in infants? Child Development, 57, 816–823.
Schacter, D. L., & Tulving, E. (1994). What are the memory systems of 1994?
In D. L. Schacter & E. Tulving (Eds.), Memory systems 1994 (pp. 1–38). Cam-
(Gorilla gorilla gorilla): Implications for episodic memory. Animal Cognition,
5, 85–90.
Sherry, D. (1984). Food storage by black-capped chickadees: Memory for the
location and contents of caches. Animal Behaviour, 32, 451–464.
Shettleworth, S. J. (1995). Memory in food-storing birds: From the field to the
Skinner box. In E. Alleva, A. Fasolo, H.-P. Lipp, & L. Nadel (Eds.), Behav-
ioural brain research in naturalistic and semi-naturalistic settings (pp. 158–179).
The Hague: Kluwer Academic.
Shettleworth, S. J. (1998). Cognition, evolution and behavior. New York: Oxford
University Press.
Squire, L. R. (1992). Memory and the hippocampus: A synthesis from findings
with rats, monkeys, and humans. Psychological Review, 99, 195–231.
123, 133–167.
Thorndike, E. L. (1898). Animal intelligence: An experimental study of the
associative processes in animals. Psychological Monographs, 2(8).
Tinklepaugh, O. L. (1932). Multiple delayed reaction with chimpanzees and
monkeys. Journal of Comparative Psychology, 13, 207–243.
Tomasello, M., & Call, J. (1997). Primate cognition. Oxford, UK: Oxford Univer-
sity Press.
Tulving, E. (1983). Elements of episodic memory. Oxford, UK: Clarendon Press.
Tulving, E. (1984). Precis of elements of episodic memory. Behavioral and Brain
Sciences, 7, 223–268.
Vander Wall, S. B. (1990). Food hoarding in animals. Chicago: University of
Chicago Press.
Vander Wall, S. B. (1991). Mechanisms of cache recovery by yellow pine chip-
munks. Animal Behaviour, 41, 851–863.
Vauclair, J. (1990). [Complex cognitive processes: Study of mental perfor-
mance in baboons.] In J.-J. Roeder & J. R. Anderson (Eds.), Primates: Recher-
ches actuelles (pp. 170–180). Paris: Masson.
Wehner, R. (1996). Preface. Journal of Experimental Biology, 199, i.

Woodworth, R. S., & Schlosberg, H. (1954). Experimental psychology (Rev. ed.).
New York: Henry Holt.
Yerkes, R. M. (1943). Chimpanzees: A laboratory colony. New Haven, CT: Yale
University Press.
Yerkes, R. M., & Yerkes, D. N. (1928). Concerning memory in the chimpanzee.
Journal of Comparative Psychology, 8, 237–271.
9
Do Nonhuman Primates
Have Episodic Memory?
Bennett L. Schwartz
Consider the familiar image of a deposed alpha male ape, much be-
loved by the females of a troop, who has been overthrown by a bunch
of rowdy young males eager to make their own mark. The deposed
male, now an outcast, looks back at the troop he has known for years.
Does he reflect on his glory days as an alpha male and recall specific
episodes of his former life? Does he worry about his diminished status
in times to come? If he did, he would have abilities that cognitive
scientists claim are uniquely human: a sense of his self, his past, and
his future.
Consider another scenario. A murder takes place at a captive pri-
mate facility. The only witness to the crime is a language-trained
chimpanzee. The police have no clues, save for the agitated chatter
of the chimpanzee. Could the chimpanzee recall the murderer and
communicate his identity to the police? This very plot was actually
used in a mystery novel (Dickinson, 1974).
Empirical questions suggested by these scenarios have only re-
cently been posed in the scientific literature (see Griffiths, Dickinson,
& Clayton, 1999; Roberts, 2002; Schwartz & Evans, 2001, for reviews).
Based on the assumption that memory is a unitary phenomenon, re-
search on animal memory addressed more tractable questions. There
is, for example, a large literature on delayed matching to sample (see
Dewsbury, 2000; Harper & Garry, 2000; Morris, 2002), spatial memory
(e.g., Burke, Cieplucha, Cass, Russell, & Fry, 2002; MacDonald, 1994;
MacDonald & Agnes, 1999; Sherry, 1984), and memory of species-typi-
cal behavior (e.g., Platt, Brannon, Briese, & French, 1996). However,
none of these studies addressed the question whether animals have
episodic memory.
Most cognitive psychologists consider episodic memory to be the
sole domain of human beings and therefore do not consider it an ap-
225
propriate topic for study in nonhuman animals (e.g., Tulving, 1983).

That situation is rapidly changing. During the past few years, a num-
ber of investigators have begun to study episodic memory in a variety
of species: pigeons (Zentall, Clement, Bhatt, & Allen, 2001), rats (Eichen-
baum & Fortin, 2003; Morris, 2002), scrub jays (Clayton & Dickinson,
1998; Clayton, Yu, & Dickinson, 2001), chimpanzees (Menzel, 1999),
and gorillas (Schwartz, Colon, Sanchez, Rodriguez, & Evans, 2002).
Although the results of those studies have been suggestive, they have
not provided clear evidence that episodic memory exists in nonhuman
animals.
In this chapter, I evaluate evidence of episodic memory in nonhu-
man species, in primates in particular. I begin by discussing opera-
tional definitions of episodic memory in humans and animals and
indicate how they differ. I then describe research I have conducted on
the episodic memory of a western lowland gorilla and discuss the
implications of that research for narrowing the gap in the study of
episodic memory between humans and animals.
EPISODIC MEMORY IN THE HUMAN LITERATURE

Tulving (1972, 1983, 1993, 2002) introduced the term “episodic mem-
ory” to refer to the human capacity to recollect individual events from
their personal lives. Definitions of episodic memory have varied, but
all center on one idea, which is that episodic memory is a memory
system that involves the conscious retrieval of events from one’s per-
sonal past. Tulving and Markowitsch (1998) defined episodic memory
as having to do “with the conscious recollection of previous experi-
ences of events, happenings, and situations” (p. 202). They argued fur-
ther that episodic memory is directed toward the past, whereas the
focus of other memory systems is the present. Wheeler (2000) defined
episodic memory as the “the type of memory that allows people to
reflect upon personal experience” (p. 597). Cabeza (1999) observed
that “episodic memory retrieval allows one to travel back in time and
re-experience events that happened minutes ago or decades ago” (p. 76).
I consider three aspects of episodic memory to be important. First,
the memory refers to a specific event from personal experience. Sec-
ond, retrieval involves some form of “mental time travel” or reexperi-
encing of the past event (Roberts, 2002; Suddendorf & Corballis, 1997;
Tulving, 1983). Thus, episodic memory concerns past states of the
world and not the present. Third, episodes are retrieved from long-
term memory rather than from working memory (Schwartz & Evans,
2001).
A wealth of data and theory support these definitions. Indeed, the
past three decades of research have demonstrated that episodic mem-
ory is cognitively, neuropsychologically, and developmentally disso-
Do Nonhuman Primates Have Episodic Memory? 227
ciable from its close cousin, semantic memory. That literature has
been reviewed by Tulving (2002) and is not discussed in this chapter.
Instead, I focus on a functional analysis of episodic memory.
In humans, episodic and semantic memory differ functionally. To
characterize that difference, Tulving and Lepage (2000) distinguished
between proscopic (for semantic memory) and palinscopic memory (for
episodic memory). The function of proscopic memory is to update
current knowledge. When we remember proscopically, there is no ref-
erence to particular past events. To refer to proscopic memories, we
use the verb “know,” as in “I know where my keys are,” or “I know
the scientific name of the orangutan.” I need not remember the event
of placing my keys on the table by the door; I merely need to know
that they are there now. Similarly, I do not need to remember when
and how I first learned the phrase “Pongo pygmaeus”; I merely need
to access the term now. Proscopic memory can be broken down into
procedural memory (knowledge of motor routines), lexical memory
(knowledge of words), and semantic memory (knowledge of mean-
ing) (Tulving & Lepage, 2000).
Palinscopic memory refers to memory of specific events. The men-
tal focus of palinscopic memory is the past, not the present. Palins-
copic memories are retrieved by mental time travel. To refer to palins-
copic memories, we use the verb “remember,” as in “I remember
eating lunch with my colleagues at a Korean restaurant on Broadway”
or “I remember testing an orangutan in my lab yesterday.” It is impor-
tant to keep in mind that the retrieval of information based on unique
events does not implicate palinscopic memory. It could be proscopic
memory if that memory refers to a general fact (e.g., “My keys are in
the bowl by the door”). It is palinscopic memory if it refers to past
events (e.g., “I remember leaving my keys in the bowl when I came
home last night”).
The issue of palinscopy has been examined by researchers investi-
gating the remember/know distinction in recognition memory of
words. In one experiment, for example, Gardiner and Java (1991) pre-
sented human participants with a list of words, which they were
asked to study for a future memory test. Each word was presented
only once. Later, they were given an “old-new” recognition test. In
this test, they were presented with some words that had been pre-
sented on the earlier list and some words that were not presented on
the earlier list. The participants were asked to decide “old” if the word
had appeared in the list, and “new” if it had not. Most important
for the current discussion, the participants were asked to distinguish
whether they “remembered” the word from the list (that is, whether
an experience of recollection accompanied the memory decision) or if
they simply “knew” the word had been on the list. Gardiner and Java
showed that human participants were accurate at deciding which
words had been presented earlier and assigned many of them to the
“remember” category. Subsequent studies have shown that items as-
signed to the “remember” category correlate with variables that affect
human episodic memory (e.g., Wheeler, 2000). For example, amnesics
who show selective deficits in episodic memory also show a lower
proportion of remember judgments (Schacter, Verfaillie, & Anes, 1997).
DEFINING EPISODIC MEMORY IN NONHUMAN SPECIES

How do we apply the concept of episodic memory from humans to
nonhumans? This is a thorny issue because of the way in which epi-
sodic memory is conceptualized in humans. Episodic memory is
thought of as a distinct neurocognitive system that stores events and
provides for their retrieval, but its signature is the mental experiences
that accompany that retrieval, namely feelings of pastness. Moreover,
the usual way of assessing episodic memory in humans is with verbal
tests. In the human literature, it is easy to ask a person, “What do you
recall?” or “What happened to you?” We can gather data on what
kinds of information people can recall, the context and source of those
memories, and qualitative aspects of memory. Obviously, these kinds
of verbal tests cannot be conducted with animals, except perhaps with
a small number of language-trained apes. Thus, when transferring the
concept of episodic memory from humans to nonhumans, there is an
immediate methodological impediment. How do you test for episodic
memory functions in nonlinguistic animals?
Four operational definitions of episodic-like memory have been
used in the recent animal cognition literature. Two come from work
on birds (i.e., Clayton & Dickinson, 1998; Zentall et al., 2001) and two
from work on apes (Menzel, 1999; Schwartz et al., 2002; also see Eich-
enbaum & Fortin, 2003; Morris, 2002). These definitions focus on be-
havioral rather than phenomenological criteria because, for obvious
reasons, the issue of the mental experience of a nonhuman is not a
tractable issue. Each definition focuses on a different aspect of how
the human episodic memory system operates. I describe each ap-
proach, as studied in different species, and then assess how well it
addresses the issue of episodic memory.
Birds
What, Where, and When Clayton and Dickinson developed criteria,

based on Tulving’s (1972) original conception of episodic memory, to
examine episodic memory in nonhumans (Clayton & Dickinson, 1998,
1999, 2000; Clayton & Griffiths, 2002; Clayton et al., 2001; Griffiths et
al., 1999). Although their work involves scrub jays, they intended their
operational definition to extend to other animals as well (Griffiths et
al., 1999). According to their view, episodic memory stores informa-

tion about temporally dated events and the spatial-temporal relation
inherent in the event. Episodic memories are not of single bits of infor-
mation; they involve multiple components of an event linked (or
bound) together. They sought to examine evidence of accurate mem-
ory of “what,” “where,” and “when” information, and the binding of
this information, based on trial-unique learning events.
Clayton and Dickinson (1998) focused their research on food-stor-
ing birds, in particular, the scrub jay (Aphelocoma coerulescens). These
birds cache extra food in the wild and return to the cache sites later
when food is in short supply. Clayton and Dickinson devised an inge-
nious experiment to determine if jays would act similarly in the labo-
ratory, specifically, if they would remember the location of two items
of food that were cached under trial-unique conditions. Their para-
digm was based on a jay’s preference for fresh crickets over peanuts
during short intervals of time and a shift in their preferences over
time because crickets degrade more quickly than peanuts. At the start
of each trial, the jays were shown an ice cube tray in which crickets
were stored on one side and peanuts on the other. After varying inter-
vals of time, the jays were allowed to choose one side of the tray.
Clayton and Dickinson showed that jays preferred crickets at short-
retention intervals and peanuts at long-retention intervals. That is, de-
pending on how long ago the food was stored before testing, the
birds’ preference switched from crickets to peanuts. These results
show that jays not only remembered what type of food was hidden,
but when and where it was stored.
This methodology has proved a reliable method for exploring the
remarkable memory capabilities of scrub jays exposed to unique
events. The birds show an amazing ability to recognize the site in
which particular foods were stored at long temporal intervals, al-
though so far the birds have only been tested with a binary choice.
However, the methodology falls short of addressing the issue of palin-
scopy. The birds merely have to continually update their knowledge
of the world at the current moment, rather than reflect on the past. In
this way, it is not dissimilar to the memory of where one’s keys are.
In essence, it would be perfectly reasonable to hypothesize that the
jays accomplish this task with something much closer to semantic
memory than to episodic memory. Thus, however compelling this re-
search may be, the methodology and the operational definition of epi-
sodic-like memory behind it do not accord with a key aspect of human
episodic memory, palinscopy.
Surprise Test After Unique Events The next operational definition

comes from the work of Zentall and his colleagues (2001). Their defi-
nition is based on tests of an animal’s ability to recall a unique event
when it is not expecting a test. Zentall et al. argue that an animal

would not attempt to update information when trained on a paradigm
on which it did not anticipate a test.
In order to apply this definition, Zentall et al. (2001) trained pi-
geons (Columbia livia) to respond to a red light if they had just pecked
and to respond to a green light if they had not. This would later serve
as the method by which the pigeons could demonstrate memory of an
event. The pigeons were then trained to peck when presented with a
yellow stimulus and not to peck when presented with a blue stimulus.
The two kinds of training were not combined during the pretest period.
After these responses had been trained, Zentall et al. introduced a
novel manipulation. The pigeons were given the opportunity to
choose red or green after being presented with blue or yellow. Zentall
et al. asked whether the pigeons would “comment” on their memory
by choosing the appropriate color to indicate if they had pecked or
not. The results clearly indicated that the pigeons could reliably indi-
cate whether or not they had pecked by choosing green and red ap-
propriately. The data from eight pigeons also indicated that they were
above chance on the very first few trials, when the element of “sur-
prise” was still present.
Zentall et al.’s study has one major drawback. The retention inter-
vals were short enough to fall within the domain of short-term mem-
ory or working memory (Bjork & Bjork, 1992; Washburn & Astur,
1998). Thus, it is unclear if the pigeons’ performance is being guided
by retrieval from working memory or retrieval from long-term mem-
ory, let alone long-term episodic memory. In addition, like all studies
conducted with animals to date, there is no evidence that the pigeon’s
choice was based on self-reflection. For that reason, the Zentall et al.
(2001) study provides no conclusive evidence for an episodic-like sys-
tem in pigeons.
Apes
Spontaneous Recall Menzel (1999) investigated memory of unique

events in a language-trained chimpanzee (Pan trogolodytes) named
Panzee at the Language Research Center. The unique symbolic ability
of this chimpanzee allowed for the kind of experiment that cannot be
done with most animals. Panzee is capable of communicating about a
vast number of objects, places, and other information by using a lexi-
gram keyboard (see Beran, Savage-Rumbaugh, Brakke, Kelley, &
Rumbaugh, 1998). As such, Menzel evaluated Panzee’s ability to spon-
taneously recall unique events by using particular symbols after long
retention intervals.
In this study, an experimenter hid foods and assorted objects just
beyond the fence of Panzee’s outdoor enclosure. Panzee could see the
food being placed outside but was moved back inside before she
could get the food. The experimenter then left the area and Panzee’s
caretakers noted what spontaneous communications Panzee used
with them. The caretakers had no knowledge of what was hidden and
when it was hidden. They were also blind as to the location of the
foods or objects, ruling out any form of experimenter bias. The care-
takers noted the lexigrams that Panzee chose to communicate and the
physical gestures that Panzee used to capture the caretaker’s attention
and to direct the caretakers to the hidden objects.
At retention interval as long as 16 hours, Panzee indicated both
memory for the food type (by selecting the appropriate lexigram) and
its location (by pointing in the general direction of where it was hid-
den). She was more likely to use a lexigram for a particular object
during the period in which it was hidden than during control periods
during which there were no hidden objects. Panzee was also more
likely to elicit help from the caretakers when an object was hidden
than when it was not.
Importantly, Panzee remembered both the what and where compo-
nents of events that were based on unique events. Furthermore, Pan-
zee’s retrieval of this information was spontaneous, that is, un-
prompted by any of her caretakers. Panzee’s working vocabulary was
about 120 lexigrams, most of which refer to food, and therefore, her
use of the keyboard to indicate the hidden item approximates a recall
measure. Buchanan, Gill, and Braggio (1981) also used a recall-like
measure with a language-trained chimpanzee in a list-learning para-
digm. However, the retention interval between the beginning of the
learning phase and the beginning of the retrieval phase lasted for a
maximum of around 30 seconds. Therefore, the Buchanan et al. data
are likely to have been recalled from short-term or working memory,
whereas Menzel’s data were retrieved from long-term memory. Fi-
nally, pointing to food hidden outside Panzee’s enclosure was not a
common task for Panzee. Therefore, the trials used also satisfy the
trial-unique constraint of episodic memory.
Menzel’s study represents a significant advance in the study of
nonhuman episodic memory, but there is a problem when interpret-
ing Menzel’s results as evidence of episodic memory. The chimpan-
zee’s memory need not have been palinscopic. It is possible that Pan-
zee updated her memory about spatial landmarks and their contents
and did not need to mentally refer to the event of watching Menzel
hiding the food. In this view, retrieval would be based on the present
state of the world and not on the occurrence of a past event.
Reporting on Past Events A fourth approach comes from my own lab.

I have argued that the three main features of episodic memory in
humans are long-term memory based on unique and nonrepeatable
events, feelings of pastness when those memories are retrieved, and a

feeling of veracity and confidence about those memories (Schwartz &
Evans, 2001). The last two features are extremely difficult to demon-
strate in nonhuman animals. Accordingly, we have opted for an oper-
ational definition of episodic-like memory in our work with apes that
has the following features. The ape’s response must be based on re-
trieval from long-term memory of a unique event that was learned
in a single trial. In addition, the animal’s response should provide
information about its past rather than about the current state of its
knowledge. This allows us to address the issue of palinscopy without
concern for the animal’s ability to describe its internal mental states.
Note that in the prior research described, only Zentall et al. (2001)
asked their animals to report on the past, albeit at very short retention
intervals. Clayton et al. (2001) and Menzel’s (1999) animals were re-
quired to report the location of food that was hidden earlier on a
single trial (also see MacDonald, 1994, for single-trial spatial learning
in gorillas). We now turn to data collected using the “reporting on
past events” definition.
KING, THE GORILLA

I have had the opportunity to study King, a lowland gorilla (Gorilla
gorilla gorilla) who has lived at Monkey Jungle in Miami, Florida, for
the past 20 years (figure 9.1). At the time I began testing him, King
was 30 years old and had a long history of contact and social interac-
tions with his human caretakers. King was wild-caught around the
age of 2 in Cameroon, in western Africa. He was then sold to a Las
Vegas circus, where he performed on a daily basis for nearly 8 years.
Little is known now about his life in the circus, other than that he was
trained to perform and his front teeth were removed (making social
contact with other gorillas problematic). By the age of 10, King was
too large, strong, and unmanageable for the circus, and he was sold
to Monkey Jungle in 1980. For many years, he lived alongside a female
gorilla, who died in 1989. Since then, King has not been in the com-
pany of other gorillas. King has received daily enrichment sessions to
keep his mind active, to educate the public, and to facilitate medical
exams.
Perhaps because of his extensive human contact and his limited
contact with his own species, King shows behaviors uncommon to his
species. Few of these abilities have ever been studied formally. They
include his language comprehension of both spoken English and
Spanish speech, his response to eye contact as a social bond rather
than as a threat, and, indeed, his schadenfreude (we have noted but
not attempted to measure his pleasure vocalizations when he wit-
nesses someone tripping or falling). Some prior formal testing had
Figure 9.1 King knuckle-walking in his outdoor area. He occupies .6 hectares of Florida
hammock forest. See color insert.
been done with King. He is the only gorilla in the scientific literature
to have passed the mirror self-recognition test (see Swartz & Evans,
1994, for details). After observing the mark on his brow in the mirror,
he immediately engaged in touching it and then tried to rub off the
mark on the bars of his cage. For all these reasons, we considered
King an excellent animal to test for evidence of episodic memory.
Before formal testing began, King was trained to associate specific
cards with specific items of food. Thus, a card with a drawing of an
apple on it was associated with both apples and the spoken word
“apple.” Five such foods were trained. Because this was done as en-
richment prior to my involvement with King, there is no documenta-
tion of how many trials and how much time it took King to learn
these card/food associations. However, at the start of my research,
King was trained to use a particular card in the presence of three
different experimenters or when the experimenter’s name was spo-
ken. Training continued until King was 90% accurate at presenting
the card when either the name was mentioned or the person was pres-
ent. This 90% accuracy was achieved after three sessions (30 trials).
In Experiment 2 of Schwartz et al. (2002), King was expected to
make two responses, one about the “what” component (food) and one
about the “who” component, that is, which experimenter gave King
the food. One of the experimenters gave King a specific food. Reten-
tion intervals were either short (5 minutes) or long (24 hours). In order
to be reinforced (with an unrelated food), King had to respond with
a card representing both the correct food (the what component) and
the correct person (the who component). King was accurate at both
the short and long retention intervals at identifying both the food
(what) and the person (who) (see table 9.1).
The Schwartz et al. (2002) study has one important advantage over
the earlier work of Menzel (1999). In our study, the gorilla’s responses
were not based on the current state of the world—that is, the food
had been distributed and consumed. The response that King made is
therefore a reflection of what happened in the past. To be concrete,
King ate an apple at Time 1. After a retention interval, the apple was
eaten and gone, but remembering that action garnered a reward. We
rewarded King for the correct answer, which required King to retrieve
information about the past. Thus, like Zentall’s study, Schwartz et al.
showed that the gorilla was referring to a past event. However, unlike
Zentall et al., we showed retention at long intervals (up to 24 hours).
Nonetheless, this study is not without nonepisodic interpretations.
First, and most problematic, is that King was tested repeatedly in sim-
ilar trials. It is therefore possible that he expected the test and that he
planned the correct response by encoding the correct food and person
“semantically.” He could then rehearse that association (see Wash-
burn & Astur, 1998, for a discussion of rehearsal in nonhuman pri-
mates). Like Panzee, King’s responses are potentially guided by se-
mantic as well as episodic memory. Thus, we lack the element of
surprise that Zentall et al. (2001) consider so important. Second, be-
cause the tests were five-alternative recognition tests (for food) and
two-alternative recognition tests, it is also possible that some re-
sponses were guided by the familiarity of the targets rather than re-
trieval of past events (i.e., Jacoby, 1991; Kelley & Jacoby, 2000). How-
ever, the 24-hour retention intervals suggest that it was not simply
familiarity. Third, like all previous experiments with nonhumans, we
Table 9.1 King’s Percentage Correct
5 min. RI 24 hr. RI
“What” “Who” “What” “Who”
Experiment 1 70 82
Experiment 2 55 82 73 87
Note: 20% is chance baseline for “what” questions; 50%

is chance baseline for “who” quetions. RI, retention in-
terval. From Schwartz et al., 2002.
made no headway in examining the mnemonic experience of the go-

rilla when he was retrieving the target memory.
To address some of those issues, we have completed a new series
of studies with King (Schwartz, Meissner, Hoffman, Evans, & Frazier,
2004) that was based on the human eyewitness memory paradigm. In
the study, King witnessed unique events. These events involved either
eating novel foods (e.g., sapodilla, cactus fruit, passion fruit), seeing
unfamiliar people or familiar people doing odd things (e.g., stealing
a cell phone, doing jumping jacks, swinging a tennis racket), or seeing
unfamiliar objects (e.g., a plastic frog, a baby’s shoes, a guitar). After
eating a novel food or witnessing a novel event, King was given a
5- to 15-minute retention interval. After the retention interval, three
photographs were shown and then given to King. One photograph
depicted an aspect of the event (e.g., the object witnessed) and the
other two photographs were distractors (e.g., other objects). King’s
task was to select the correct photograph and pass that card through
the bars of his cage to a blind tester. A trainer outside of King’s view
would then verify if the photograph was the correct one. If it was,
King was given food reinforcement.
King’s performance in this task was less than stellar (55% accurate
overall). However, it was significantly higher than chance (33%). We
suspect his performance was weak because some of the photographs
might have been hard for him to see and some of the discriminations
themselves were difficult (e.g., choosing honeydew over cantaloupe
melon). The study does show that King could retrieve unique events
from the past based on unique events even when he did not know
what test to expect. Because 24-hour retention intervals were not
tested, responses based on familiarity could not be ruled out. Thus,
we cannot rule out a nonautonoetic response based on familiarity in
this study. On the other hand, King’s responses are based on past
events, not current knowledge, and, in that sense, they satisfy the
palinscopy constraint.
In another experiment in this series, we asked whether an ape’s
episodic memory could be manipulated in a manner similar to that
observed in experiments on human episodic memory. An eyewitness
to a crime sees a perpetrator engage in an illegal act. This is a unique
and often exceedingly quick event involving the observation of a com-
plete stranger doing something that may arouse high levels of stress
in the witness (e.g., Loftus, 1979; Loftus & Hoffman, 1989). Think of
witnessing a burglar climb out your window and descend the fire
escape. At some point later, either to police, lawyers, or under oath in
the courtroom, the eyewitness must recount the event in question.
The desired memory response is a palinscopic reflection on the events
witnessed. The witness must mentally time travel and recount the
event from the past. We know that people are generally accurate, but
that under many circumstances people are prone to errors (Dunning

& Stern, 1992). For example, in the famous misinformation paradigm,
witnesses often misremember events in systematic ways (see Loftus,
1979; Loftus & Hoffman, 1989).
Would a nonhuman primate be able to report who committed a
particular act? This is a potentially revealing line of inquiry because
the memory would necessarily be palinscopic, that is, about the past
and not about the present state of the world. Thus, even if the research
fails to show mechanisms common to human episodic memory, the
studies would be important because of the palinscopic nature of the
task. Another advantage of this paradigm is that it only requires an
animal to point or otherwise select a photograph. Thus, it can be done
with a range of animals in addition to primates and it does not involve
language-like communication. Indeed, Harper and Garry (2000) have
found that pigeons show some “misinformation” effects at short-term
memory retention intervals in a delayed matched-to-sample task.
As we discussed earlier, King is able to reliably choose the correct
person, object, or food from among three photographs following a
unique event (Schwartz et al., 2004). We recently completed a misin-
formation study with King. Similar to the paradigm described earlier,
King witnessed a unique event, such as a familiar person jumping
rope. Shortly after the event, King either was shown a photograph of
the person, the object, or food involved in the event (correct informa-
tion condition), was shown another person, object, or food item (mis-
information condition), or was not shown any new photograph (con-
trol condition). Thus, like misinformation experiments in humans, the
misinformation condition here was intended to create the possibility
of false memories in King.
The results were quite clear; whereas King was above chance at
retrieving the correct photograph in the control and correct informa-
tion condition (53%; chance was 33%), he was no different than chance
in the misinformation condition (40%). However, his percentage cor-
rect in the control and correct conditions was not significantly above
the 40% rate of the misinformation condition. Nonetheless, in the mis-
information condition, King was just as likely to choose the additional
distractor as he was the misinformation item. Therefore, like the eye-
witness relying on episodic memory, King’s memory was impaired
by the presentation of misinformation.
CONCLUSION
One of the most perplexing issues of science is the evolution of human
cognition and intelligence. Comparing humans and apes provides an
imperfect measurement of this evolution because of the 6 to 8 million
years of divergent paths. Depending on one’s orientation and prefer-
ences, one can marvel at the similarity between humans and apes or
skeptically nod at the profound differences. It would not be difficult
to demonstrate any number of common episodic memory tasks that
are far beyond the capability of a gorilla (e.g., flashbulb memories; see
Conway, 1995). I think, however, that I have demonstrated some strik-
ing similarities between human episodic memory and the trial-unique
learning and memory for individual events we have tested in King.
Because King’s memory attains the criterion of palinscopy or referen-
tial to the past, I think we may claim to have supported the contention
that King has the rudiments of an episodic memory system.
QUESTIONS AND ANSWERS
Dr. Josep Call: I’m curious how you ask King the key question. Do you
just present symbols?
Dr. Bennett Schwartz: The tester is experimentally blind to the food pre-
sented to King. In the initial studies, the tester asked, “What did you eat?”
and “Who gave it to you?” We know that King responds appropriately to
a wide variety of English words and commands, but what interests us is
not his behavior in response to language, but if he can demonstrate mem-
ory. Nonetheless, at some point, we do wish to test the extent of his com-
prehension of English words and grammar. We have no idea to what ex-
tent he is responding to words or other cues.
King responds to us by using the cards. He has a set of cards with
pictures of food on them and he has a set of cards with names of familiar
people written on them. The cards are put into his cage by the trainer, and
he responds by pushing his choice back out through the cage to the trainer.
Please see Schwartz et al. (2002) for complete details.
Dr. David Smith: At what time point do you go from short term to long
term? For example, how long would an animal have to hold on to a delay
match-to-sample trial for it to move from short to long term?
Dr. Schwartz: Because my main focus is comparing this to human mem-
ory, I use human memory differences between long- and short-term mem-
ory as an initial guide. In theory, apes could have working memories that
are either longer or shorter than ours (but see Washburn & Astur, 1998).
We thought 5 minutes would put us safely beyond what anyone would
expect possible of a short-term memory system.
Dr. Leonard Rosenblum: There is a study, I believe with baboons, which
shows that after some preliminary training that animals outside of the
group setting can correctly identify parent-offspring pairs as compared to
adult female nonpaired similarly aged animal sitting next to them in pho-
tographs. Would you comment on where that kind of ability would fit into
this framework?
Dr. Schwartz: That is what we would call semantic memory in humans,
our knowledge of the world. Countless studies demonstrate that primates
are expert learners and represent spatial and conspecific information quite
well (see MacDonald, 1994, for an example). We also know they learn sym-
bols in language-learning studies (e.g., Beran, Pate, Richardson, & Rum-
baugh, 2000).
Dr. Herb Terrace: Can you expand on the uniqueness of the experience?
Presumably there is a small and finite range of answers. Has the ape been
trained on that block so that the uniqueness may be with respect to that
particular block? How does the ape get the flavor of the task without pre-
vious experience?
Dr. Schwartz: When I started to test King, he had already been at Mon-
key Jungle for 20 years. He had been doing “enrichment” activities for
nearly that long. One of his enrichment activities was learning to associate
the wooden blocks with food items and people’s names. Nobody was able
to tell me how long he had been doing this form of enrichment, and there
are no records. So much of his learning history is undocumented. We did
document the rapidity with which he learned the names of new blocks,
such as the block bearing my name. By the second session of learning, he
was at 90% correct in providing the block with my name in response to
my spoken name.
To answer the first part of your question, his ability to report on unique
experiences is limited by the number of cards he can handle. Thus, in Ex-
periment 2 of Schwartz et al. (2002) he is limited to one of two people and
one of five foods. For this reason, the second set of studies described in
this chapter use photographs in which we were able to test him each time
with a unique event and not have to repeat stimuli.
Dr. Endel Tulving: I’d like to congratulate you on a fascinating and very
interesting presentation. You reminded us all that we may be witnessing
a new research area that has a potential at being successful. You under-
stand what the issues are perfectly. In all of the history of science many
people just talk past each other. A very important requirement for these
kinds of experiments is that you don’t let your animals pull a clever Hans
trick. Earlier I told you I spent the whole day with John, who has no epi-
sodic memory because of hippocampal damage. He has become very sensi-
tive to tests of science. Some of the things that amnesiacs can accomplish
in test situations are due to very subtle cues. It is very important that when
you do these tests with King, the experimenter is experimentally blind.
Dr. Schwartz: The trainer who did most of testing has little understand-
ing of the scientific process. The biggest trick is making sure the trainer
has no idea what the food is. If she did, she would try to cue him every
which way. In fact, we threw out several trials in which she came by while
we were giving food to King that she wasn’t supposed to see.
Dr. Tulving: If so, I think you should test her as well and correlate the
results.
Dr. Schwartz: In the second set of studies, the tester always guessed at
the correct response. Testers did not differ from chance, nor was there a
correlation between their choices and those of King.
ACKNOWLEDGMENTS I am grateful to Monkey Jungle for access to King and
for employees’ time. In particular, I thank Sharon Du Mond, Sian Evans, and
Steve Jacques for their continued support of this project. I thank Leslie Fra-
zier, Chris Meissner, Lisa Son, and Herb Terrace for comments on earlier
drafts of this chapter. I thank Herb Terrace, Janet Metcalfe, and the partici-
pants of the conference on the origins of self-knowing consciousness for their
insight and comments.
REFERENCES
Beran, M. J., Pate, J. L., Richardson, W. K., & Rumbaugh, D. M. (2000). A
chimpanzee’s (Pan troglodytes) long-term retention of lexigrams. Animal
Learning and Behavior, 28, 201–207.
Beran, M. J., Savage-Rumbaugh, E. S., Brakke, K. E., Kelley, J. W., & Rum-
baugh, D. M. (1998). Symbol comprehension and learning: A “vocabulary”
test of three chimpanzees (Pan troglodytes). Evolution of Communication, 2,
171–188.
Bjork, R. A., & Bjork, E. L. (1992). A new theory of disuse and an old theory
of stimulation fluctuation. In A. F. Healy, S. M. Kosslyn, & R. M. Shiffrin
(Eds.), From learning processes to cognitive processes: Essays in honor of William
K. Estes (Vol. 2, pp. 35–67). Hillsdale, NJ: Lawrence Erlbaum.
Buchanan, J. P., Gill, T. G., & Braggio, J. T. (1981). Serial position and the
clustering effect in a chimpanzee’s free recall. Memory and Cognition, 9,
651–660.
Burke, D., Cieplucha, C., Cass, J., Russell, F., & Fry, G. (2002). Win-shift and
win-stay learning in the short-beaked echidna (Tachyglossus aculeatus). Ani-
mal Cognition, 5, 79–84.
Cabeza, R. (1999). Functional neuroimaging of episodic memory retrieval. In
E. Tulving (Ed.), Memory, consciousness, and the brain (pp. 76–90). Philadel-
phia: Psychology Press.
Clayton, N. S., & Dickinson, A. (1999). Memory for the content of caches by
scrub jays (Aphelocoma coerulescens). Journal of Experimental Psychology: Ani-
mal Behavior Processes, 25, 82–91.
Clayton, N. S., & Dickinson, A. (2000). What-where-when memory in food-
storing scrub jays. Abstracts of the Psychonomic Society, 5, 59.
Clayton, N. S., & Griffiths, D. P. (2002). Testing episodic-like memory in ani-
mals. In L. Squire & D. L. Schacter (Eds.), The neuropsychology of memory
(pp. 492–507). New York: Guilford.
Clayton, N. S., Yu, K. S., & Dickinson, A. (2001). Scrub jays (Aphelocoma coer-
ulescens) form integrated memories of the multiple features of caching epi-
sodes. Journal of Experimental Psychology: Animal Behavior Processes, 27, 17–29.
Conway, M. A. (1995). Flashbulb memories. Hove, UK: Lawrence Erlbaum.
Dewsbury, D. A. (2000). Comparative cognition in the 1930’s. Psychonomic Bul-
letin and Review, 7, 267–283.
Dickinson, P. (1974). The poison oracle. New York: Pantheon.
Dunning, D., & Stern, L. B. (1992). Examining the generality of eyewitness
hypermnesia: A close look at time delay and question type. Applied Cogni-
tive Psychology, 6, 643–657.
Eichenbaum, H., & Fortin, N. J. (2003). Episodic memory and the hippocam-
pus: It’s about time. Current Directions, 12, 53–57.
Gardiner, J. M., & Java, R. I. (1991). Forgetting in recognition memory with
and without recollective experience. Memory and Cognition, 19, 617–623.
Griffiths, D., Dickinson, A., & Clayton, N. (1999). Episodic memory: What can
animals remember about their past? Trends in Cognitive Science, 3, 74–80.
Harper, D. N., & Garry, M. (2000). Postevent cues bias recognition perfor-
mance in pigeons. Animal Learning and Behavior, 28, 59–67.
Jacoby, L. L. (1991). A process dissociation framework: Separating automatic
from intentional uses of memory. Journal of Memory and Language, 30, 513–
541.
Kelley, C. M., & Jacoby, L. L. (2000). Recollection and familiarity. In E. Tulving
& F. I. M. Craik (Eds.), The Oxford handbook of memory (pp. 215–228). New
York: Oxford University Press.
Loftus, E. F. (1979). Eyewitness testimony. Cambridge, MA: Harvard University
Press.
Loftus, E. F., & Hoffman, H. G. (1989). Misinformation and memory: The cre-
ation of new memories. Journal of Experimental Psychology: General, 118,
100–114.
MacDonald, S. E. (1994). Gorilla’s (Gorilla gorilla gorilla) spatial memory in a
foraging task. Journal of Comparative Psychology, 108, 107–113.
MacDonald, S. E., & Agnes, M. M. (1999). Orangutan (Pongo pygmaeus abelii)
spatial memory and behavior in a foraging task. Journal of Comparative Psy-
chology, 113, 213–217.
chimpanzee (Pan trogolodytes) after extended delays. Journal of Comparative
Morris, R. G. M. (2002). Episodic-like memories in animals: Psychological cri-
teria, neural mechanisms and the value of episodic-like tasks to investigate
animal models of neurodegenerative disease. In A. Baddeley, M. Conway,
& J. Aggleton (Eds.), Episodic memory: New directions in research (pp. 181–
203). New York: Oxford University Press.
Platt, M. L., Brannon, E., Briese, T. L., & French, J. A. (1996). Differences in
feeding ecology predict differences in performance between golden lion
tamarins (Leontopithicus rosalia) and Wied’s marmoset (Callithrix kuhli) on
spatial and visual memory tasks. Animal Learning and Behavior, 24, 384–393.
473–489.
Schacter, D. L., Verfaillie, M., & Anes, M. D. (1997). Illusory memories in
amnesic patients: Conceptual and perceptual false recognition. Neuropsy-
chology, 11, 331–342.
(Gorilla gorilla gorilla): Implications for episodic memory. Animal Cognition,
5, 85–90.
Schwartz, B. L., Meissner, C. A., Hoffman, M., Evans, S., & Frazier, L. D.
(2004). Event memory and misinformation effects in a gorilla (Gorilla gorilla

gorilla). Animal Cognition, 7, 93–100.
Sherry, D. (1984). Food storage by black-capped chickadees: Memory for the
location and contents of caches. Animal Behavior, 32, 451–464.
Swartz, K. B., & Evans, S. (1994). Social and cognitive factors in chimpanzee
and gorilla mirror behavior and self-recognition. In S. T. Parker, R. W.
Mitchell, & M. L. Boccia (Eds.), Self-awareness in animals and humans: Devel-
opmental perspectives (pp. 189–206). Cambridge, UK: Cambridge University
Press.
Suddendorf, T. S., & Corballis, M. C. (1997). Mental time travel and the evolu-
123, 133–167.
Tulving, E. (1972). Episodic and semantic memory. In E. Tulving & W. Don-
aldson (Eds.), Organization of memory (pp. 382–403). New York: Academic
Press.
Press.
Tulving, E. (1993). What is episodic memory? Current Directions in Psychology,
3, 67–70.
Tulving, E. (2002). Episodic memory and common sense: How far apart? In
A. Baddeley, M. Conway, & J. Aggleton (Eds.), Episodic memory: New direc-
tions in research (pp. 269–288). New York: Oxford University Press.
Tulving, E., & Lepage, M. (2000). Where in the brain is the awareness of one’s
past. In D. Schacter & E. Scarry (Eds.), Memory, brain, and belief (pp. 208–
230). Cambridge, MA: Harvard University Press.
Washburn, D. A., & Astur, R. S. (1998). Nonverbal working memory of hu-
mans and monkeys: Rehearsal in the sketchpad. Memory and Cognition, 26,
277–286.
Zentall, T. R., Clement, T. S., Bhatt, R. S., & Allen, J. (2001). Episodic-like
memory in pigeons. Psychonomic Bulletin and Review, 8, 685–690.
10
Studies of Uncertainty Monitoring

and Metacognition in Animals
and Humans
J. David Smith
HUMAN METACOGNITION
Humans feel uncertain—they know when they do not know. More-
over, they often respond intelligently in difficult situations by paus-
ing, reflecting, and seeking help, hints, or information. These states of
feeling and knowing and these adaptive coping responses anchor the
extensive literatures on metacognition and the related phenomena of
metamemory and uncertainty monitoring (Brown, 1991; Brown, Brans-
ford, Ferrara, & Campione, 1983; Dunlosky & Nelson, 1992; Flavell,
1979; Hart, 1965; Koriat, 1993; Metcalfe & Shimamura, 1994; Nelson,
1992; Schwartz, 1994; Smith, Brown, & Balfour, 1991).
Metacognition can be defined as thinking about thinking, or cogni-
tion about cognition. The idea in this field is that in some minds men-
tal activities occur at a higher “metalevel” and at a lower “object level”
during cognitive processing. In these minds, there is a cognitive exec-
utive that supervises (i.e., oversees and facilitates) thought or problem
solving. Figure 10.1 summarizes the functions of the cognitive execu-
tive using a framework provided by Nelson and Narens (1990). The
overseeing or monitoring functions are shown at the top, and they
are studied by having human participants make basic metacognitive
judgments. There is an ease-of-learning judgment about whether ma-
terial will be easy or hard to learn, a judgment of learning about how
much has been learned, a feeling-of-knowing judgment about whether
information is potentially available in memory, and a confidence judg-
ment about a potential answer.
The cognitive executive also controls cognition, directing informa-
tion processing along more productive paths. These metacognitive
control processes are shown at the bottom of the figure. The executive
242
Uncertainty Monitoring and Metacognition 243
MONITORING
JUDGMENTS FEELING-OF-KNOWING
OF LEARNING JUDGMENTS
CONFIDENCE IN
EASE-OF-LEARNING RETRIEVED
JUDGMENTS ANSWERS
ACQUISITION RETENTION RETRIEVAL

IN ADVANCE MAINTENANCE OUTPUT
ON-GOING SELF-DIRECTED
OF OF OF
LEARNING SEARCH
LEARNING KNOWLEDGE RESPONSE
TERMINATION
OF STUDY
ALLOCATION OF SELECTION
STUDY TIME OF SEARCH
STRATEGY TERMINATION
SELECTION OF OF SEARCH
KIND OF
PROCESSING
CONTROL
Figure 10.1 A theoretical framework for research on metacognition, showing examples

of process-monitoring capacities above and process-control capacities below (after Nel-
son & Narens, 1990).
may devote disproportionate study time to difficult items, terminate

studying when sufficient learning has been achieved, select new re-
trieval strategies when the present ones are failing, or abandon re-
trieval efforts if success seems improbable.
When humans behave metacognitively, we naturally infer some in-
triguing aspects of mind. We take metacognition to be about hierar-
chical tiers of oversight in mind (i.e., the meta- and object levels of
cognition). We take it to demonstrate our awareness of the processes
of mind. We link metacognitive states to our self-awareness in a re-
flexive sense, because uncertainty and doubts are personalized (i.e., I
know that I am uncertain). We even link metacognitive states to de-
clarative consciousness (Nelson, 1996) because we so naturally intro-
spect about those states and speak about them to one another. It is
no wonder that we take metacognition to be one of humans’ most
sophisticated cognitive capacities and one of those most likely to be
humanly unique. For all these reasons, it is an intriguing question
whether animals have analogous or homologous metacognitive capac-
ities. The purpose of this chapter is to review some research that has
attempted to answer this question and to evaluate the progress and
prospects of the new research area that explores metacognition com-
paratively.
THE PROBLEM OF COMPARATIVE METACOGNITION RESEARCH

In beginning this exploration, comparative metacognition researchers
faced obstacles that slowed initial progress. For one thing, human
metacognition paradigms were not appropriate for animal partici-
pants. Their phenomena relied too heavily on introspection and on
verbal reports about feelings and judgments of knowing and learning.
Instead, researchers faced the problem of constructing a behavioral
(i.e., nonverbal) metacognition or uncertainty-monitoring paradigm
that could be used comparatively.
One key element of such a paradigm is that it must systematically
cause difficulty for animals. The difficulty is what will create some-
thing like an uncertainty state that the animal may monitor or respond
to adaptively. For example, psychophysical procedures accomplish
this because they let one reduce the contrast between two stimulus
input classes (e.g., faint sounds vs. silence) until the threshold is
reached at which perceptual discriminations become very difficult
(Au & Moore, 1990; Schusterman & Barrett, 1975; Yunker & Herman,
1974).
However, causing animals difficulty is not sufficient for studying
their uncertainty. Illustrating this point, figure 10.2A shows a bottle-
nosed dolphin’s performance near threshold in a difficult auditory
discrimination task (Smith et al., 1995). Pressing the high paddle was
correct when the animal heard a 2100-Hz tone. Pressing the low pad-
dle was correct when the animal heard any tone that was lower in
pitch. The dolphin was often able to make the response high for 2100-
Hz tones and to make the response low for tones below about 2075
Hz. But the trials surrounding 2085 Hz, his known psychophysical
threshold relative to 2100 Hz (Herman & Arbeit, 1972), produced
chance performance and many errors. This task caused the animal the
difficulty it was meant to; however, it did not show whether the ani-
FACING PAGE
Figure 10.2 A. Frequency discrimination performance by a bottlenosed dolphin (Tursiops
truncatus) in the procedure of Smith et al. (1995). The horizontal axis indicates the
frequency (Hz) of the tone. The response high was correct for tones at exactly 2100 Hz,
and these trials are represented by the rightmost data point for each curve. All lower
tones deserved the response low. The percentages of trials ending with the high response
(dashed line) or low response (dotted line) are shown. B. The dolphin’s performance in
the same auditory discrimination when he was also given the response uncertain. The
solid line represents the percentage of trials receiving the uncertainty response at each
difficulty level. The error bars show the lower 95% confidence limits. C. The intensity of
the dolphin’s incidental hesitation and wavering behaviors for tones of different fre-
quencies (Hz). See color insert.
A. 100
Low
90
80 High
Response Percentage
70
60
50
40
30
20
10
0
1950 1975 2000 2025 2050 2075 2100
Pitch Height (Hz)
B. 100
90 Low
80
Response Percentage
70
High
60
50
40
30
Uncertain
20
10
0
1950 1975 2000 2025 2050 2075 2100
Pitch Height (Hz)
C. 0.8
0.6
Hesitation and Wavering
0.4
0.2
-0.2
-0.4
-0.6
-0.8
1950 1975 2000 2025 2050 2075 2100
Pitch Height (Hz)
mal sensed this difficulty. In fact, it was not suited to show the animal
analogues of confidence and doubt or anything about the animal’s
system for coping with uncertainty. There were only the two primary
perceptual responses, high and low, so this task denied the animal any
way to demonstrate something like uncertainty by coping with it.
Therefore, a second key element of a comparative uncertainty para-
digm is to give animals a third, uncertainty response that lets them
cope with difficulty by declining the trials they do not choose to com-
plete. Then one can ask if animals sense the difficulty caused by the
task and respond adaptively to it.
THE UNCERTAINTY RESPONSE IN HUMAN PSYCHOPHYSICS

One sees from this discussion that one potential paradigm for study-
ing animals’ uncertainty monitoring would combine the perceptual
difficulty caused by psychophysical procedures with an uncertainty
response that animals could use to report on and cope with their un-
certainty. I soon realized that this paradigm had a long history in
human experimental psychology that was friendly to the possibility
of studying metacognition comparatively.
Humans in the founding psychophysical studies were often al-
lowed to respond Uncertain when they felt unable to classify an event
into one of the two primary input classes (sound-silence, lighter than–
heavier than, etc.; Angell, 1907; Fernberger, 1914, 1930; George, 1917;
Watson, Kellogg, Kawanishi, & Lucas, 1973; Woodworth, 1938). How-
ever, some came to question this approach (Fernberger, 1914; Wood-
worth, 1938, pp. 419–427) after noting what they took to be the special
psychological status of the response uncertain. That response had
longer latencies (Angell, 1907; George, 1917; Woodworth, 1938). It was
strongly affected by instructional set (Brown, 1910; Fernberger, 1914,
1930; Woodworth, 1938). It seemed to be linked to participants’ temper-
amental tentativeness (Angell, 1907; Fernberger, 1930; Thomson, 1920).
It seemed to be more reflective and cognitive than the two primary
discrimination responses (Angell, 1907; Fernberger, 1930; George, 1917).
Theorists also came to believe that uncertainty responses were meta
to the primary discrimination and were a comment on the participant’s
failure to assign a stimulus to one of the primary input classes. For
example, Boring (1920) suggested that doubt was an attitudinal seducer
that distracted observers from the series of mental states that are a con-
tinuous function of the series of stimuli. Similarly, George (1917) con-
cluded that doubtful responses disrupted the constant attitude required
for psychophysical judgments because they introduced “extra-serial”
attitudes into a task that depended on intraserial, sensory attitudes.
The concerns the early theorists had about the uncertainty re-
sponse—its special psychological status; its metarole in the perfor-
mance of the psychophysical task—make this response interesting in

the context of exploring metacognition comparatively. For we shall
soon see that animals can respond uncertain in psychophysical tasks.
Given this fact, it is possible that the uncertainty response plays a
metarole in their performance, too. It is possible that it is for them,
too, a comment on or a response to indeterminacy and difficulty in
the primary discrimination.
Weiskrantz (1986, 1997) saw the same possibility. In a provocative
thought experiment, he considered giving blindsight animals a com-
mentary key with which they could report on knowing or not know-
ing, or on seeing or not seeing. Cowey and Stoerig (1992) considered
a similar gedanken procedure. They proposed to use psychophysical
procedures to bring animals to threshold (where they could be only
50% correct) in a light–no light discrimination, but also to give the
animal a third lever reinforced on a 75% schedule. Then adaptive ani-
mals could report on their state of not knowing whether a light was
seen by choosing the third lever selectively on threshold trials.
These converging ideas suggested that a psychophysical task com-
bined with an uncertainty response might constructively initiate the
comparative study of metacognition and uncertainty monitoring. In
the next section, I discuss the results when a dolphin, monkeys, and
humans were placed into uncertain perceptual situations and asked
to make difficult threshold judgments while also being allowed to re-
spond uncertain.
As I discuss these results, readers will naturally consider both met-
acognitive and lower level interpretations of performance. For exam-
ple, it is possible that a dolphin trainer might cue the animal to re-
spond uncertain on difficult trials. For another example, the animal
might respond uncertain not to cope adaptively with uncertainty but
simply to avoid those stimuli that are associated with errors and lean
reinforcement. Additional research in the domain of memory monitor-
ing is discussed below, and it addresses some alternative accounts.
Moreover, in the last sections of this chapter, I try to address carefully
the appropriate psychological interpretation of animals’ performances
in these procedures. Meanwhile, I invite readers to read on with both
healthy skepticism and an open mind.
ANIMALS’ USE OF THE PSYCHOPHYSICAL UNCERTAINTY RESPONSE
Dolphin Uncertainty Responses

To evaluate animals’ capacity to respond uncertain in a psychophysical
task, Smith et al. (1995) gave a bottlenosed dolphin (Tursiops truncatus)
the auditory discrimination task that was described above. Pressing
the paddles high or low, respectively, was correct for a 2100-Hz tone
or for any lower-pitched tone. The third paddle, uncertain, advanced

the animal into an easy, low-pitched trial that was rewarded when
answered correctly. Initially, the animal was stabilized on an easy dis-
crimination between 2100-Hz and 1200-Hz trials. Then the difficulty
of the discrimination was gradually increased by raising the pitch of
the lower tones until the dolphin was trying to distinguish trials of
2100 Hz and 2085 Hz. At mature performance, trial difficulty was
adjusted dynamically based on performance within a session to main-
tain difficulty at a high level.
Figure 10.2A shows the dolphin’s performance with the response
uncertain disallowed. High and low responses mapped generally to
2100-Hz and lower-pitched tones, respectively, with these response
curves crossing (signifying chance performance) at the dolphin’s
threshold. However, this two-response procedure denied the animal
any way to cope with his uncertainty on threshold trials. To study
uncertainty processes, one must map the confidence landscape more
finely by providing a third response alternative.
Figure 10.2B shows the dolphin’s use of this uncertainty response.
Against the backdrop of the same primary discrimination perfor-
mance as before, he used the uncertainty response selectively for the
difficult trials surrounding his discrimination threshold.
The dolphin amplified his primary evidence of uncertainty re-
sponding through his incidental confidence and uncertainty behav-
iors. When the trial was easy, he swam toward the response paddles
so fast that he created a bow wave that required us to cover our
electronics in condoms to prevent their being soaked. When the trial
was uncertain, we observed him to slow his approach to the re-
sponse paddles or to waver among them. To formalize these ob-
servations, raters judged the intensity of these hesitation and waver-
ing behaviors during the trials seen in videotaped sessions. Figure
10.2C shows the overall intensity of these behaviors for trials at dif-
ferent pitch levels. These behaviors were most intense at 2087 Hz
and were distributed like the uncertainty response was (figure
10.2B).
Behaviors like these—which Tolman (1938, p. 27) called “lookings
or runnings back and forth”—were always challenging to behaviorists
because they hinted at mental turmoil in animals and because they
were intuitive symptoms of uncertainty states. Indeed, Tolman (1927)
offered these behaviors as a “behaviorist’s definition of animal con-
sciousness.” Given the time and theoretical climate, this was an ex-
traordinary statement. In a limited sense, therefore, the dolphin’s an-
cillary behaviors reinforced an uncertainty interpretation of the
animal’s uncertainty responses. However, the concrete uncertainty re-
sponse is more easily measured and compared across situations and
species than are incidental waverings.
Monkey Uncertainty Responses

In a companion study, Smith, Shields, Schull, and Washburn (1997)
placed two rhesus monkeys (Macaca mulatta) in a visual density-
discrimination task. The animals (tested using the Language Research
Center’s Computerized Test System, LRC-CTS; Washburn & Rum-
baugh, 1992) used a joystick to move a cursor to one of three objects
on a computer screen (figure 10.3A). Moving the cursor to the box
was correct if it contained exactly 2,950 illuminated pixels. Choosing
the S was correct if the box contained any fewer pixels. Choosing the
star allowed participants to decline the trial and move into a new,
guaranteed-win trial. Initially, animals were stabilized on a dense-
sparse discrimination involving 2,950-pixel and 450-pixel boxes. As
training progressed, the difficulty of the discrimination was increased
by increasing the density of the sparse boxes until participants’ perfor-
mance faltered (i.e., at about 2,950 vs. 2,600 pixels). At mature perfor-
mance, trial difficulty was adjusted based on animals’ performance
within a session to maintain difficulty at a high level.
Figures 10.3B and 10.3C show the monkeys’ performance. Box re-
sponses predominated on dense trials and the most difficult sparse
trials. S responses predominated on the sparser trials. The primary
discrimination was performed at chance where these two response
curves cross, and the star was used most in this region of maximum
uncertainty. Both monkeys assessed accurately when they were liable
to make an error in the primary discrimination, and they bailed out
of those trials selectively and adaptively. Figure 10.3D shows that hu-
mans did, too. In fact, the graphs in figure 10.3 show one of the stron-
gest performance similarities between humans and animals in the
comparative literature.
Humans’ descriptions of performance were revealing about the
cognitive and decisional organization of the task. They said that their
responses sparse and dense were cued by the objective stimulus condi-
tions (i.e., sparseness or denseness) on a trial. Notice that they
mapped the task’s two primary responses to the task’s two stimulus
input classes. In contrast, they said that their uncertainty responses
were prompted by feelings of uncertainty and doubt about the correct
answer on a trial. Notice that they mapped the uncertainty response
to monitored uncertainty. It is also interesting that humans stated the
uncertainty and doubt personally (I was uncertain; I didn’t know).
They understood that the trial was still either sparse or dense—they
just felt they could not tell. Thus humans’ uncertainty responses may
also be showing reflexive self-awareness in the sense that humans are
aware of themselves as cognitive monitors.
Given this interpretation of the uncertainty response, many hu-
mans underuse it because they are overconfident or because they feel
A. Trial Screen B. Monkey Abel

100
90
Response Percentage
Sparse
80 Dense
70
60
50
40
30
20
10 Uncertain
0
1250 1650 2050 2450 2850
Box Density (pixels)
C. Monkey Baker D. Humans

100 100
90 Sparse Sparse
90
Response Percentage
Response Percentage
80 80
Dense Dense
70 70
60 60
50 50
40 40
30 30
20 Uncertain 20 Uncertain
10 10
0 0
1250 1650 2050 2450 2850 1250 1650 2050 2450 2850
Box Density (pixels) Box Density (normalized pixels)
Figure 10.3 A. The screen from a trial in the dense-sparse discrimination of Smith et al.
(1997). B. The performance of monkey Abel in the dense-sparse task. The response dense
was correct for boxes with exactly 2,950 pixels—these trials are represented by the
rightmost data point for each curve. All other boxes deserved the response sparse. The
horizontal axis indicates the pixel density of the box. The solid line represents the per-
centage of trials receiving the uncertainty response at each density level. The error bars
show the lower 95% confidence limits. The percentages of trials ending with the dense
response (dashed line) or sparse response (dotted line) are also shown. C. The perfor-
mance of monkey Baker in the dense-sparse discrimination depicted in the same way.
D. The performance of seven humans in the dense-sparse discrimination. To equate
discrimination performance across participants, the data have been normalized to place
each participant’s discrimination crossover at a pixel density of about 2,700. The hori-
zontal axis indicates the normalized pixel density of the box. See color insert.
they should try or because they want to know if they can be right.
They think escaping from the trial is lame or a copout. Humans have
underused the response uncertain for these reasons for nearly 100
years. Psychophysicists have sometimes agreed with them that re-
sponding uncertain amounts to shirking and mental inertia (Brown,
1910, p. 32) or an admission of weakness (Fernberger, 1930, p. 110).
Recently, Washburn, Smith, Baker, and Raby (2001) found that there
are sex differences in this area—males are especially overconfident

and tend to think they know even when they do not. Apparently,
males are unlikely to stop and ask for directions even in a psycho-
physical discrimination.
Given the strong analogy in the data patterns produced by humans
and monkeys, one wonders whether monkeys’ uncertainty responses
are also indicative of some form of cognitive monitoring. If so, the
response uncertain would be a very important behavioral ambassador.
Individual Differences in Uncertainty Responding

The cross-species behavioral analogy is made stronger by the fact that
there are parallel individual differences in how individual members
of the two species use the uncertainty response. In a related experi-
ment, Smith et al. (1997) gave humans and monkeys another common
type of psychophysical task (Woodworth, 1938). Instead of dynami-
cally adjusting the level of sparse trials to hold perceivers at their
sparse-dense threshold, we created a sparse-dense break point at the
density continuum’s center and then continuously tested perceivers
on a wide range of densities from sparse to dense including the diffi-
cult and uncertain region near the discrimination’s break point.
Figures 10.4A and 10.4B, respectively, show the results from eight
humans and one particular human. In both graphs, the discrimination
responses sparse and dense were used in the same way. However,
whereas humans generally responded uncertain most often to the diffi-
cult trials near the discrimination’s break point, the one human did
not. Figures 10.4C and 10.4D, respectively, show the performance of
monkeys Baker and Abel. Both animals used the sparse and dense dis-
crimination responses in the same way as each other and as humans
did. However, whereas Baker selectively declined the indeterminate
trials near the discrimination’s break point, Abel did not. The vanish-
ing error bars in Abel’s graph make plain how long and hard we tried
to get him to behave like Baker.
Everyone (humans and monkeys) in both dense-sparse tasks used
the dense and sparse discrimination responses in the same way. This
suggests that the two primary responses in a discrimination task have
a transparent and highly stable function. Probably this is because
these two responses are mapped to the two stimulus input classes. In
contrast, the uncertainty response showed strong differences across
individuals and tasks. The pronounced changeability of the uncer-
tainty response persuaded early psychophysicists that it was not
mapped to a stimulus input class, that it was a report on the failure to
assign a stimulus to an input class, that it was related to uncertainty,
extraserial attitudes, and decisional temperaments (Angell, 1907; Fern-
berger, 1930; Thomson, 1920). It is interesting that monkeys’ use of
the response uncertain shows this changeability, too.
A. Humans B. Humans
100 100
90 90 Dense
Response Percentage
Response Percentage
80 Dense 80
70 70
60 60 Sparse
50 Sparse 50
40 40
30 30
20 20
Uncertain
10 10
0 0 Uncertain
450 550 650 750 450 550 650 750
Box Density (pixels) Box Density (pixels)
C. Baker D. Abel
100 100
90 90 Dense
Response Percentage
Response Percentage
80 80
70 Dense Sparse
70
60 Sparse 60
50 50
40 40
30 30
20 Uncertain
20
10 10
0 0 Uncertain
450 550 650 750 450 550 650 750
Figure 10.4 A. The performance of eight humans in the second kind of dense-sparse
discrimination used by Smith et al. (1997). The lower and upper half of a range of
densities deserved the sparse and dense response, respectively. The percentage of trials
receiving the uncertainty response at each density are shown by the solid line. The error
bars show the lower 95% confidence limits. The percentages of trials ending with the
dense response (dashed line) or sparse response (dotted line) are also shown. B. The
performance of one human in the same dense-sparse discrimination. C. The perfor-
mance of monkey Baker in this discrimination. D. The performance of monkey Abel in
this discrimination. See color insert.
COMPARATIVE STUDIES OF MEMORY MONITORING IN A SERIAL PROBE

RECOGNITION TASK
There are other kinds of difficulty than perceptual difficulty. In fact,
memory tasks that cause memory difficulty are a sharp focus in stud-
ies of human metacognition. Accordingly, comparative researchers
have also asked whether animals can monitor their memory and re-
spond adaptively when the state of their memory does not justify
completing a memory test—that is, whether animals have cognitive
capacities that are analogous to humans’ memory monitoring.
For example, Smith, Shields, Allendoerfer, and Washburn’s (1998)

exploration of these capacities took advantage of the fact that memory
performance changes in predictable ways across the serial positions
of a memory list. The predictability tells the experimenter which
memory trials are difficult and uncertain and lets him or her ask
whether animals decline those trials selectively. Smith et al. adopted
the serial probe recognition (SPR) task that is ubiquitous in compara-
tive memory research (Castro & Larsen, 1992; Roberts & Kraemer,
1981; Sands & Wright, 1980; Wright, Santiago, Sands, Kendrick, &
Cook, 1985). In their procedure, the animal saw a “list” of four pic-
tures and then was asked whether a fifth probe picture had been in
the list or not. Working from left to right at about 1 second each, scan
the four shapes in figure 10.5 to get an idea of what the presentation
of a memory trial was like. If you think this kind of trial is too easy,
go pick on someone your own brain size—the participants were mon-
keys. Monkeys made a there or not there response as the probe was
judged to have been in the list or not. But we also gave animals the
response uncertain that let them decline any memory tests of their
choosing. (Robert Hampton’s closely related research on memory
monitoring by monkeys is discussed in chapter 11, this volume.)
Figure 10.6A shows a monkey’s performance in this task. When he
chose to complete the memory tests, he showed classic primacy and
recency effects. Most important, though, he used the uncertainty re-
sponse in the mirror image of those effects. That is, he declined trials
selectively when his worst serial positions were probed. Figure 10.6B
shows that the same was true for 10 humans under similar conditions
(though these conditions were not the right ones for producing a ro-
bust human primacy effect). The similarity between the human and
animal graphs is made more compelling by the fact that humans were
told explicitly to use the uncertainty response to report on and cope
with memory indeterminacy. Thus, humans declined memory tests
when they felt uncertain. Monkeys behaved like humans.
Figure 10.5 An example of the memory lists Smith et al. (1998) presented to monkeys
in a serial probe recognition task. The pictures would have been presented to monkeys
successively for about 1 second each, by computer and in different colors.
A. Monkey B. Humans
100 100
90 90
Response Percentage
Response Percentage
80 80
70 Correct 70 Correct
60 60
40 40
30 30
20 20
10 10
0 0
1 2 3 4 NT 1 2 3 4 NT
Serial Position Serial Position
C. SDT Model D. Simulant

0.4 100
NT T 90
0.35
Response Percentage
Response Percentage
80
0.3
70 Correct
0.25 60
0.2 50 Uncertain
0.15 40
30
0.1
20
0.05 10
NT U T 0
0
-4 -3 -2 -1 0 1 2 3 4 5 1 2 3 4 NT
Trace Strength Impression Serial Position
Figure 10.6 A. Serial probe recognition (SPR) performance by monkey Baker in the task
of Smith et al. (1998). NT denotes not there trials. The serial position (1–4) of the probe
in the list of pictures is also given along the x-axis for the probes on there trials. The
percentage of trials of each type that received the uncertainty response is shown (bold
line). The percentage correct (of the trials on which the memory test was attempted) is
also shown (dotted line). B. Performance by 10 humans in a similar SPR task used by
Smith et al. C. A signal detection theory (SDT) portrayal of monkey Baker’s decisional
strategy in the SPR task of Smith et al. (1998). Unit-normal trace-impression distributions
are centered at the locations along the trace-strength continuum corresponding to the
animals d for probes of the four serial positions in the memory lists (there, T), and at 0.0
for the NT probes. These normal curves are overlain by the decision criteria that define
the animal’s three response regions (from left to right, NT, uncertain [U], and T). D. Perfor-
mance by the SDT simulant that fit best monkey Baker’s performance. See color insert.
On some occasions, Smith et al. ran monkey Baker without the un-
certainty response available. This condition let us test an additional
prediction that arises from a memory-monitoring interpretation of
monkeys’ performance. The prediction is that animals should perform
better when they choose to complete the memory test than when they
are forced to complete it. This prediction follows because the monkey
should accept memory tests when the probe picture encounters either
quite strong or quite weak traces, and he will often be able to give
these probes a correct there or not there response, respectively. In con-
trast, if the animal is forced to complete all memory tests, his perfor-
mance will include those memory tests he would have declined be-
cause he monitored indeterminate traces. He will often be wrong on
these trials and so his overall percentage correct will fall. Smith et al.
found that Baker’s performance did deteriorate substantially when he
was denied the uncertainty response. Baker was clearly sensing some-
thing real about his memory that was rationally attended to in decid-
ing whether to accept or decline memory tests. In fact, evidently Baker
was using the uncertainty response adaptively to avoid errors when
he monitored an indeterminately available memory.
A Formal Model of Uncertainty Monitoring

Signal detection theory (SDT) provides a constructive framework for
thinking about the psychological organization of the animal’s perfor-
mance in this and the perceptual threshold tasks considered earlier.
Regarding the memory-monitoring task, one can assume that the list
items create memory impressions that lie along a continuum of trace
strength (the x-axis in figure 10.6C). Then the probe picture queries
the strength of one trace. Probes on not there trials will generally point
to weak traces, perhaps averaging 0.0 plus or minus the scatter of
memory variability (the normal distribution NT in the figure). Probes
on there trials will point to stronger traces on average (though still
with memory variability), especially for the primacy and recency list
items (the four T normal distributions in the figure). These four distri-
butions would be centered at the memory sensitivity (d) appropriate
to the performance that the animal showed at each serial position
(MacMillan & Creelman, 1991, pp. 209–230). The overlap between the
not there and there distributions is what makes the SPR task difficult
and uncertain, because it means that probes on there and not there
trials often feel equally strongly remembered to the animal. Facing
this difficult memory situation, the animal needs to find a way to
divide the memory continuum into three response regions using re-
sponse criteria, hopefully being able to respond not there, there, or un-
certain when the probe contacts a memory trace that is weak, strong,
or ambiguous/indeterminate. These criterion lines are also shown in
the figure.
To feel this model in your mind, think which of the two probe
pictures in figure 10.7 contacts a stronger trace in memory. Many will
say that the probe on the left does. It was the fourth item in the sample
trial given previously and so received a recency boost of activation
from being in that position in the list. The animals would likely re-
spond there to this probe item. The probe on the right may have made
a dimmer impression on memory because it was presented in the sec-
Figure 10.7 Two probes that might have followed the

memory list given in figure 10.5. Both would be there
probes. The item on the left reprises the item in the
list that occurred at the final, strongest, recency-advan-
taged serial position. The item on the right reprises the
item in the list that occurred at the weakest, second
serial position. The monkeys’ memory probes were
computer-presented in color.
ond, weakest serial position. The animals would often respond uncer-
tain to this probe item.
I conducted simulations to evaluate the SDT model of monkeys’
memory-monitoring performances and to find out what decisional
strategy (i.e., what placement for the two criterion lines) they were
probably using. To do so, I evaluated the data patterns produced by
many thousands of simulated creatures (simulants), who placed their
decision criteria at different points along the trace-strength contin-
uum. Each simulant completed 8,000 trials in a virtual version of the
SPR task. On each trial, the simulant received one of five trial types
(not there or a there probe of one of four serial positions), assessed
(with memory variability) the trace strength this probe item contacted
(following the five probability-density functions shown in figure
10.6C), and responded according to its criterion placements. I summa-
rized the simulant’s performance over the 8,000 trials and compared
its performance pattern mathematically to monkey Baker’s observed
performance pattern. The criterion of best fit was the sum of the
squared deviations (SSD) between the observed and simulated per-
centages. On average, the best-fitting response percentages were
within about 3% of their observed targets. Thus the performance of
this model was competitive with the performance of other formal
models in the experimental literature (Smith & Minda, 1998, 2000).
Figure 10.6D shows the performance of the simulant that most closely
reproduced Baker’s performance (compare figure 10.6A). Figure 10.6C
shows the criterion placements of this simulant. One sees intuitively
that Baker found an adaptive decisional strategy. He took the memory
tests (i.e., he responded not there or there) that presented with weak or
strong traces. He responded uncertain for indeterminate, ambiguous

trace strengths that could easily have been caused by either a there or
a not there probe. In fact, in additional simulations I was able to estab-
lish that Baker’s decisional strategy in the task was essentially optimal
in terms of its reward efficiency.
CONSIDERING THE PSYCHOLOGY OF UNCERTAINTY RESPONSES

AND UNCERTAINTY-MONITORING PERFORMANCES
The data patterns produced by the dolphin and monkeys in the per-
ceptual and memory tasks are noncontroversial matters of fact. How-
ever, there remains the problem of interpreting these performances
psychologically. What representations underlie performance? What
level do animals’ decisional processes occupy in their cognitive sys-
tem? How close to consciousness do they lie? How close is the perfor-
mance of animals to those performances of humans that we comfort-
ably call metacognitive? Difficult questions like these have sparked
theoretical tensions throughout a century of scholarship within com-
parative psychology. In the present discussion, I raise six points that
help me think psychologically about animals’ use of the response un-
certain and that may help readers of this volume, too.
1. The Parsimony Embodied in Morgan’s Canon Is Sometimes False

Actually, it is easy to get theoretical sparks to fly regarding this re-
search. For example, consider the dolphin’s and monkeys’ perceptual
threshold tasks that began the exploration of animals’ metacognitive
capacities and that were discussed earlier. In these cases, one could
explain performance using behaviorist constructs like stimulus control
or reinforcement history. One might say that stimuli of intermediate
(threshold) pitch or density were mildly aversive for being associated
with errors and timeouts and that uncertainty responses were condi-
tioned in these stimulus contexts. In this case, the uncertainty re-
sponse would be poorly named and would have nothing to do with
metacognition. This low-level description of animals’ responses in
perceptual threshold tasks would have a distinguished pedigree. It
would defend the principle of parsimony embodied in Morgan’s
Canon (Morgan, 1906, p. 53) that one should always interpret an or-
ganism’s behavior at the lowest possible psychological level. Thus,
given performances by animals that might be about uncertainty moni-
toring, there is a 100-year-old urge to dismiss them as low-level asso-
ciative phenomena. I suspect that many readers felt this urge as they
read about those results. The first point in my discussion is to caution
readers sharply that this urge is inappropriate in the present instance.
The parsimony embodied in Morgan’s Canon is false in this case.
To see why, consider the graphs in figure 10.3 that show humans
and monkeys using the uncertainty response identically in the dense-
parse threshold task and that show one of the strongest existing paral-
lels between human and animal performance. These graphs make it
clear that one cannot interpret the monkeys’ performance in a vac-
uum. Humans perform the same way. Moreover, humans report that
they are consciously uncertain and reflexively (i.e., personally) self-
aware as they produce these graphs. Humans and monkeys have
shared much of their evolutionary histories, and probably shared the
set of adaptive pressures that could have led to the emergence of an
adaptive uncertainty-resolving cognitive system. Humans and mon-
keys even share homologous brain structures. Thus it is unparsimoni-
ous to interpret the same graph produced by humans and monkeys in
qualitatively different ways—consciously metacognitive versus low-
level associative. It uses two opposed behavioral systems to produce
the same phenomenon when one might do.
In fact, this duality of interpretation would even be an inappropri-
ate scientific stance in this case. Imagine if, in any other domain, a
researcher showed identical graphs by two populations (older and
younger children, nondepressed and depressed individuals, etc.) and
then nonetheless offered qualitatively different high-level and low-
level interpretations to explain them. This would cause a peer-review
massacre. If it were older and younger children, or individuals with
and without depression, the researcher would have no warrant to
make this dual interpretation. Likewise, in the case of humans and
animals, there is no warrant to do so, either. It is extraordinary that
this duality of interpretation is given momentary credence when the
two populations are humans and animals. However, this is an acci-
dent of the history of comparative psychology that serves the field
poorly. To the contrary, scientific inference in this case would require
that the two graphs be provisionally interpreted as instances of the
same phenomenon—a phenomenon that humans describe very
clearly—until other evidence required the monkey graph to be given
a qualitatively different and lower interpretation. Thus, given identi-
cal graphs by humans and animals, a significant part of the burden
of proof falls on the behaviorist who would interpret the monkeys’
performances in a dismissive way. The dismissive interpretation is not
the scientifically appropriate default assumption, though it may be
the historically preconceived notion.
2. Associative Interpretations Are Not Sufficient for Explaining Animals’

Performances in Uncertainty-Monitoring Tasks
Given this historical context, it is a helpful consideration that a variety
of low-level interpretations that concern the control of behavior by
stimulus cues are insufficient to explain the performances described
in this chapter. For example, in the memory-monitoring task of Smith

et al. (1998), all stimuli, across trials, became targets and foils and
were rewarded and nonrewarded following both primary responses.
No stimulus cue indicated any response. Only the presence or absence
of the probe in the preceding list was ever relevant. The psychological
action in this experiment occurred along an internal continuum of
subjective trace strength, with the animals responding uncertain when
probes met memory traces that were indeterminately strong or weak.
Thus, the signals for behavior in this study were cognitively derived,
abstract, and subjective. These memory-based assessments are pro-
foundly different from the signals available in many traditional oper-
ant situations, wherein a tone or a keylight comes over many trials to
stably, predictably signal rich or poor reinforcement. Accordingly, one
sees that the animals’ behavior in the memory task lies far from tradi-
tional senses of stimulus control.
This is an important theoretical and psychological interpretation.
In essence, the animals are monitoring the contents of memory on
each trial. In essence, they are realizing a capacity for metamemory,
though further research might be able to draw distinctions between
what humans and monkeys are capable of in this area, and between
what the two species know, feel, and experience consciously in the
tasks.
A similar conclusion would apply to the uncertainty-monitoring
task used by Shields, Smith, and Washburn (1997). Shields et al. asked
whether monkeys could recruit adaptive uncertainty responses when
pushed to their psychophysical limit in a same-different task. The
same-different task requires one to go beyond the absolute stimulus
qualities in order to make a judgment about the relation between two
stimuli. The abstractness of this task explains why same-different per-
formance is so hard to train, why it is so fragile when placed into
opposition with absolute stimulus qualities, and why true same-differ-
ent performance is so phylogenetically restricted. Shields et al. found
that monkeys did respond uncertain adaptively when they were pre-
sented with stimulus relations that were near their threshold of dis-
criminating same from different. The monkeys’ data pattern was once
again essentially identical to that of humans. Yet behavior in this task
was not controllable by absolute stimuli, whether through the general-
ization gradients surrounding them or the reinforcement histories as-
sociated with them. As within the memory task, the cues were more
cognitively derived. In the same-different task, they represented a de-
cision about the (indeterminate) status of the relation assessed on a
trial.
The principal conclusion of this section—that associative interpre-
tations do not suffice for the memory-monitoring task or for Shields
et al.’s (1997) same-different task—feeds back to provide an additional
reason why an associative interpretation of animals’ perceptual

threshold performances is unparsimonious. One cannot consider the
perceptual threshold results in a vacuum because they do not exist
alone. That animals can base uncertainty responses on derived cogni-
tive states (e.g., indeterminate same-different relations, indetermi-
nately available memory traces) already grants them a sophisticated
form of cognitive monitoring. Now, if one explains their threshold
performances using a low-level associative mechanism, one grants
them (unparsimoniously) two different indeterminacy-resolution sys-
tems, one of which is already fairly high level. However, one could
explain the data more simply just by invoking one uncertainty-resolu-
tion system that applies to indeterminate memories, same-different
relations, and threshold events. Readers might want to consider, if
they find themselves resisting the idea that animals would have a
generalized indeterminacy-resolution system of this kind, why they
think it is implausible. Later I discuss the possible benefits animals
might derive from having such a system.
3. A Failure of Uncertainty Monitoring by Rats

Additional data suggest indirectly that monkeys’ and the dolphin’s
performances at threshold were not ascribable to associative mecha-
nisms. Smith and Schull (1989) gave a perceptual threshold uncer-
tainty-monitoring task to rats. A repeating 400-Hz tone deserved one
response. A 400-Hz tone alternating with any other deserved another
response. The third, uncertainty response let the animal wave off any
trial and begin a new one. Animals knew the effect of this response
because they also used it to initiate trials. Once rats had mastered an
easy discrimination between 400–400 Hz tone pairs and 400–700 Hz
tone pairs, the discrimination’s difficulty was increased until animals
were struggling to distinguish 400–400 Hz tone pairs from tone pairs
of about 400–410 Hz.
Figure 10.8 shows that rats mastered well the two primary discrimi-
nation responses. They made the alternating and repeating responses
mainly on alternating and repeating trials, respectively, though of
course the discrimination was performed at chance where these two
response curves cross at threshold. However, unlike humans, mon-
keys, and a dolphin, the rats did not selectively make the uncertainty
response at threshold. They could not tell when they were at risk for
error in the primary discrimination and could not decline those trials
selectively, even though Smith and Schull tried for months to coax
this monitoring capacity from them.
Yet rats could certainly learn a reinforcement contingency attached
to a middle stimulus class between two others and have a response
brought under the control of those middle stimuli. If this were all
there is to the threshold task, rats would escape from threshold trials
Rats
100
90
Response Percentage
80
Alternating
70 Repeating
60
50
40
30
20
Uncertain
10
0
40 35 30 25 20 15 10 5 0
Pitch Disparity (Hz)
Figure 10.8 Performance by six rats in the frequency discrimi-

nation used by Smith and Schull (1989). The horizontal axis
indicates the frequency difference between the alternating
pitches on a trial. The repeating response was correct for trials
with a frequency difference of 0, and these trials are repre-
sented by the rightmost data point for each curve. All other
trials deserved the alternating response. The solid line repre-
sents the percentage of trials receiving the uncertainty re-
sponse at each pitch disparity. The percentages of trials ending
with the repeating response (dashed line) or alternating re-
sponse (dotted line) are also shown. See color insert.
naturally and easily. They do not. This suggests that uncertainty re-
sponses in perceptual threshold tasks are not just about middle-stimu-
lus avoidance. The threshold tasks seem to be psychologically struc-
tured in some way that leaves rats out (insofar as the methods were
sufficient to elicit the crucial capacities from them) but leaves humans,
monkeys, and dolphins in.
An additional manipulation supported this interpretation. We were
concerned that the rats’ null result arose because, for the rats, the un-
certainty response doubled as a trial-initiation response. To evaluate
this possibility, we examined how rats responded to an occasional
higher, faster repeating tone that the computer rewarded randomly
so that rats could only be 50% correct (just as they are at threshold).
The rats declined these trials three times as often as they declined
threshold trials. So rats did bail out of the objective, stimulus-borne
50:50 contingency, but not the subjective, threshold-borne 50:50 con-
tingency. Once again, this suggests that the two kinds of behavior
have a different underlying psychological organization. Moreover, it
reinforces the idea that the capacity to sense threshold is relatively
inaccessible in the rat’s cognitive system (inaccessible to the rat and
to the experimenter), whereas we know from earlier sections of this

chapter that this capacity is easily accessible in the cognitive systems
of humans, monkeys, and dolphins.
4. Tests of Uncertainty Monitoring Foster Controlled Cognitive Processes

Because They Ensure Inconsistent Mappings Between Stimulus Inputs
and Behavioral Outputs
One can make two principled statements about the psychological or-
ganization of the capacity for monitoring threshold perceptions and
memories and for responding adaptively to them. This section consid-
ers the first of these. All the tasks described in this chapter trade on
the scatter and variability that accompany all acts of perceiving and
remembering by humans and animals. For example, the perceptual
threshold tasks deliberately challenge the observer’s discrimination
ability. Given the variable impression made by the same objective per-
ceptual event on different occasions, true dense trials and threshold
sparse trials (for example) will often produce the same subjective psy-
chological impression. Likewise, the same memory event on different
occasions (e.g., a second-serial-position presentation) will create a dis-
tribution of trace strengths (this is why the T and NT distributions in
figure 10.6C have widths). Consequently, the trace strength monitored
on a trial could be caused by either a there probe or a recently seen
not there probe (this is why the T and NT distributions in figure 10.6C
overlap).
Therefore, neither the impression of density nor the monitored
trace strength can reliably tell the animal what to do. These percep-
tual/memorial impressions are inherently ambiguous. Shiffrin and
Schneider (1977, pp. 167–168) discussed the information-processing
consequences of this ambiguity. In their description, indeterminate
mental representations of stimuli necessarily mapped inconsistently
onto behavioral responses, making those representations poor indica-
tors of what the organism should do, and ruling out that an adaptive
response could safely be automatically associated to them. Instead,
higher levels of controlled cognitive processing would be needed to
resolve the indeterminacy and produce a decision about behavior. All
the uncertainty-monitoring tasks described in this chapter are incon-
sistently mapped in Shiffrin and Schneider’s sense. All would require
this higher-level form of controlled cognitive processing. From this
perspective, the uncertainty response would be one manifestation of
the controlled indeterminacy-resolution processes that are necessary
near threshold. It might represent a decision that the trial should be
declined because the primary perceptual or memory process had
failed. Note that this description applies to both human and animal
cognition. Psychophysical procedures and uncertainty-monitoring
tasks ensure indeterminate stimulus-response mappings and encour-
age controlled decision-making processes no matter the participant

species. This realization provides a theoretically gentle but principled
way to grant animals’ uncertainty responses some of the cognitive
sophistication that is due them.
An everyday example can clarify the contrast between stimulus-
controlled and decisionally controlled processes. Many humans per-
form the following conditional discrimination daily—green-go; red-
no go. This task is consistently mapped. The underlying perceptual
representations (green and red) are not confuseable and they point
absolutely consistently and reliably to the appropriate behavior. This
task needs no fine discriminations, no decision criteria, and no deci-
sion making. In fact, behavior in the stoplight task may be triggered
automatically, reflexively, fast, and effortlessly simply because the
task’s consistent mapping allows stimulus and response to associate
very strongly. This is the reason that stoplights worldwide present the
same consistent mapping.
In contrast, imagine the decisional nightmare we would all face if
traffic lights gradually morphed between red and green, so that driv-
ers had to decide whether their light was green enough to go. This
situation would require decision making and decision criteria. It
would involve controlled processing—slow, attentional, and capacity
intensive. The situation would also be dangerous, as behind-schedule
businesswomen and soccer dads cheated on their green criterion at
the same crossing. It is an important fact that all of the uncertainty-
monitoring paradigms described in this chapter represent this latter
kind of controlled, decisional task. Once again, this claim applies to
both humans and animals. Humans and animals must decide if the
box is dense or sparse enough to try, if the tone’s pitch is high or low
enough, if the memory trace is weak or strong enough.
I point out that this kind of information-processing analysis, done
deliberately to include both species, can serve a constructive purpose
in comparative psychology. Too often our comparisons of human and
animal performance are polarizing. We describe humans’ conscious,
verbal metacognition in a way that excludes animals. We try to de-
scribe animals’ similar performances using low-level, behaviorist de-
scriptions that do not apply to humans. (The same thing happens in
discussions about tool use, language, and so forth.) But these issues do
not have to reduce to either elevating humans or denigrating animals.
Rather, given true comparative data like those shown in figure 10.3
and the strong cross-species analogy between the performance of hu-
mans and animals, it is possible to pursue a reasoned middle ground
that in my view is preferable to a clash between divergent explanatory
frameworks at different psychological levels. The real parsimony in
our explanation of the situation may lie in being able to explain sim-
ply the performance of several species using a single, structural infor-
mation-processing description. Such multiple-species descriptions are

still a rarity in comparative psychology, because so many laboratories,
research programs, and articles are single-species driven. But they
may become an increasingly important feature of comparative psy-
chology as it gains in maturity and in information-processing comfort
and sophistication. To me, an important possibility for the next de-
cade of comparative research is that we may come to treat cognition
and behavior more integratively and more truly comparatively by at-
tending systematically to strong cross-species similarities in perfor-
mance and by seeking the information-processing middle ground. Of
course, this reasoned middle ground could consider both common
processing principles across species while also acknowledging possi-
ble experiential differences across species in whatever way the whole
empirical picture warranted.
5. Animals May Share With Humans a Theoretically Important Construal

of the Threshold and Memory-Monitoring Tasks
One can make another principled statement about the psychological
organization of the capacity for monitoring threshold perceptions and
memories and for responding adaptively to them. Humans under-
stand the grammar of uncertainty-monitoring tasks in a distinctive
way that explains their data patterns and their self-descriptions of per-
formance. Humans know that these tasks have two primary input
classes (dense-sparse, there-not there). They know that one of these
two input classes is instantiated on every trial, so that every trial has
a correct answer if they could only discover it. The consequence is
that humans map the two stimulus input classes to the two primary
responses, use these responses when they think they know which
event occurred, and use the uncertainty response when they cannot
tell.
Given this understanding of this task, the uncertainty response is
not associated with a stimulus input class. Instead, it is a judgment
about the probable failure of the primary discriminatory process on a
trial. In this sense, it is structurally outside the discrimination and
meta to it. For humans, it seems clear that the uncertainty response is
also linked to conscious uncertainty states that can be verbalized and
that reveal personal self-awareness.
This understanding of the uncertainty-monitoring task sheds a
clear light on why the uncertainty response feels less stimulus-based
(it is!), and why it can feel like mental shirking or like an admission
of weakness (I should be able to discover the answer and should try!).
This understanding explains why the uncertainty response can be nat-
urally excluded from the task’s structure, and why some humans
choose to do so because of their scruples against it or because of their
unwarranted bravado in the face of the difficult threshold task. The
other two responses cannot be omitted, and humans would not do so.
This understanding also justifies the special psychological status that
psychophysicists always accorded the uncertainty response.
There are strong reasons for suggesting that animals share with
humans their understanding of the perceptual and memory uncer-
tainty-monitoring tasks, though animals’ understanding of the task
will of course be more tacit and less verbal. First, we generally train
animals extensively in the primary discrimination before giving them
an uncertainty response in the task. The uncertainty response joins as
an extra, optional response a mature discrimination that has the two
input classes and the two primary responses already established,
mapped to one another, and sufficient for performance. Second, we
generally provide animals a daily warm-up in which they receive easy
discrimination trials that then gradually increase in difficulty. This
warm-up, during which animals do not choose to use the uncertainty
response, reestablishes the idea that the task has two input classes and
two primary response. Third, we sometimes give animals sessions in
which they perform the mature discrimination without the uncer-
tainty response available. This reinforces again the understanding of
the task I am discussing here. Fourth, in our tasks the uncertainty
response never earns a direct reward, never earns a timeout penalty,
and always has the same neutral function and result in every stimulus
context. These things are not true of the two primary responses in the
task. This also helps the animal differentiate the uncertainty response
from the two discrimination responses. All in all, I believe that my
animals are highly trained to have tacitly just the task understanding
that humans have explicitly.
To summarize Points 1 to 5, the identity between the data patterns
of humans and animals in figure 10.3 suggests a strong homology
between the behavior of the two species. Given this, one does not
want to interpret the human performance in a qualitatively high-level
metacognitive way, while interpreting the monkey performance in a
qualitatively low-level associative way. There is no warrant for this
duality of interpretation (indeed, the graphs provide a negative war-
rant), and the duality is unparsimonious because a common informa-
tion-processing description might explain both data patterns and their
identity (Point 1).
Fortunately, one can show that low-level associative interpretations
are not sufficient to explain the animals’ performances. They certainly
do not work for the serial probe recognition task. They would not
work for Shields et al.’s same-different paradigm either. Though they
might work for the perceptual threshold tasks, it is unparsimonious
again to explain the animals’ memory data in a high-level way but
then insist on an associative interpretation of the same animals’
threshold data. The two performances could be explained in just the
same way if only the animal were granted a generalizable uncertainty-

resolution system (Point 2). Moreover, it is interesting that rats failed
to use the uncertainty response at threshold as humans, monkeys, and
a dolphin did, though probably they would have succeeded if low-
level associative cues had been available at threshold (Point 3).
Given the failure of an associative account, it is constructive to con-
sider a reasoned, information-processing middle ground that ad-
dresses aspects of the task and its decisional processes that apply to
performance across species. One element of this middle ground is that
the uncertainty-monitoring tasks are inherently inconsistently mapped
and will require of humans and animals a more decisional, controlled
brand of cognitive processing to disambiguate perception or memory
and to adaptively choose behavioral options (Point 4). A second ele-
ment of this common ground is that humans and animals probably
share an abstract understanding of the uncertainty-monitoring task in
which the uncertainty response is sharply differentiated from the two
primary discrimination responses so that it serves the function of a
negative judgment about the outcome of the primary discriminatory
process (Point 5).
It seems to us that these five points lay the groundwork for a psy-
chological understanding of humans’, monkeys’, and a dolphin’s suc-
cessful performances in uncertainty-monitoring tasks. They offer a
balanced psychological interpretation of uncertainty responses, grant-
ing them the cognitive sophistication they deserve without burdening
them with consciousness or with equally heavy behaviorist assump-
tions. These points also suggest the possibility of integrating findings
from humans and animals into an overall information-processing
framework instead of the polarizing descriptions that are more common.
6. Do Controlled Decisional Processes Set the Stage for Conscious Cognition?

Smith et al. (1998), in considering monkeys’ memory-monitoring per-
formance in an SPR task, noted that “one could scale back the claims
of consciousness while preserving something of the sophisticated,
memory-based, flexible, controlled mediational processes that do rep-
resent a higher level of choice and decision making and that are
needed to explain how monkeys cope with (and escape) indetermi-
nate memory events” (p. 245). This statement let Smith et al. give
animals’ uncertainty responses considerable cognitive credit without
prematurely engaging the nettlesome question of the role that con-
sciousness plays in memory monitoring and uncertainty responding.
I close by engaging this nettlesome question briefly, first regarding
human cognition, then regarding animal cognition. Humans encoun-
ter situations daily in which their automatic and habitual behavior
patterns are thwarted and they enter difficult and uncertain straits. It
is worth reflecting on how cognition meets these challenging situa-
tions. Simply put, when the going gets tough, the tough get conscious.
Times of difficulty and uncertainty seem to shift the mind into con-
scious overdrive as humans ask themselves that all-important ques-
tion—so what do I do now?
This is not my original intuition. In fact, it is a beloved idea in
cognitive science that cognitive indeterminacy and difficulty inher-
ently elicit higher level and even conscious modes of cognition and
decision making in the organism. James (1890/1952, p. 93) noted that
consciousness provides extraneous help to cognition when nerve pro-
cesses are hesitant. “In rapid, automatic, habitual action it sinks to a
minimum.” In contrast, he added, “Where indecision is great, as be-
fore a dangerous leap, consciousness is agonizingly intense” (p. 93).
Dewey (1934/1980, p. 59) also argued that in habitual, well-learned
behaviors, the behavioral impulses are “too smooth and well-oiled
to admit of consciousness.” Tolman (1932/1967, p. 217) noted that
“conscious awareness and ideation tend to arise primarily at moments
of conflicting sign-gestalts, conflicting practical differentiations and
predictions.” Karoly (1993, p. 25) emphasized that uncertain, con-
flicted conditions are the ones that initiate self-regulation. Gray (1995)
described the special neural circuits that may arrest behavior, increase
arousal, and redirect attention and mental resources toward the causes
of difficulty (see also Smith, 1995).
There is a structural, almost logical, reason why mind needs to be
so composed that it responds to difficulty and uncertainty in this way.
Difficulty and uncertainty imply that the well-learned behavior, the
well-oiled habit, will not work and could be dangerous. Instead, a
close cognitive call has to be made and a behavioral solution chosen
online. To do so, the human has to gather together within some imme-
diate/current information-processing resource—that is, within work-
ing memory or working consciousness—his or her situation, goal,
possible approaches, and their risks, benefits, costs, and prospects.
The elements that enter cognition as a human resolves indeterminacy
and chooses a behavioral solution I call a decisional assemblage. This
is the multidimensional pool of knowledge, situational perception,
reasoning, and evaluation that lets him or her maximize within the
difficult problem space or at least satifice safely.
I hope that I have communicated successfully a theme in this chap-
ter that there are sometimes structural similarities in the information-
processing situations that humans and animals face, in the task con-
struals they make, and in the character of their information-processing
approaches. The idea of the decisional assemblage provides a final
case in point. As with humans, animals will encounter difficult and
uncertain situations in which their well-learned associations and auto-
matic behavioral patterns cannot help them because a finely balanced
judgment call is required or because there is no existing template for
an appropriate behavior. Then, in the immediate moment, animals too

will have to rethink the problem, to assemble together what they
know, what they remember, what the goal is, what the choices are,
and their prospects and risks. Interestingly, animals have faced this
same immediate need to rethink their situation in many classic com-
parative paradigms. For example, Rumbaugh (see Rumbaugh & Pate,
1984) would suddenly reverse animals’ discrimination contingencies
to see if they could use an environmental signal mediationally and
change behavior quickly. For another example, Harlow (1949) would
suddenly instantiate a qualitatively new contingency to see if animals
had a learning set that would let them respond to a single error in-
sightfully.
As with humans, something like a working consciousness could be
especially appropriate for holding the cognitive and decisional assem-
blage that allows the momentary rethinking in which the animal deals
in its own tacit way with the question of what to do now. If a working
consciousness is especially useful or appropriate, it would have sub-
stantial fitness utility and phylogenetic breadth. In that case, one cog-
nitive utility serving difficult and uncertain decisions by animals
would be a working consciousness that is the functional parallel of
human declarative consciousness.
Therefore, I hope that readers will reflect on the long-lived intuition
in cognitive science that the highest levels of information processing
and especially consciousness present themselves when difficulty and
indeterminacy are encountered. I would suggest, though, that this
may not be just a human phenomenon but rather a functional prop-
erty of adaptive cognitive systems more generally. Thus it is poten-
tially important that all the tasks considered in this chapter create
decisional difficulty and indeterminacy. This makes it possible that
these tasks are well suited to elicit higher-level and possibly conscious
regulatory processes in animals. If so, then they and additional tasks
that others will contrive may eventually ground the systematic study
of animal consciousness. This would be what Tolman hoped for so
long ago when he saw animals poised in decisional conflict at the
choice point of a maze and he wondered at their lookings and run-
nings back and forth.
ACKNOWLEDGMENTS Correspondence concerning this chapter should be ad-

dressed to J. David Smith, Department of Psychology, Park Hall, State Univer-
sity of New York at Buffalo, Buffalo, NY, 14260, or to psysmith@buffalo.edu.
The preparation of this chapter was supported by NIH grant HD-38051,
which also funds ongoing research in this area. I thank my collaborators,
Wendy E. Shields and David A. Washburn, for their important contributions
to all phases of this research. I also thank Kenneth Allendoerfer, Roian Egnor,
Linda Erb, Katerina Guttmanova, Kelli Mcgee, Joshua Redford, Jonathan
Schull, Jared Strote, the staff and trainers of the Dolphin Research Center,
colleagues at the Language Research Center, and many laboratory RAs at the
University at Buffalo for their contributions to this research.
REFERENCES
Angell, F. (1907). On judgments of “like” in discrimination experiments.
American Journal of Psychology, 18, 253.
Au, W. W., & Moore, P. W. (1990). Critical ratio and critical bandwidth for
the Atlantic bottlenose dolphin. Journal of the Acoustical Society of America,
88, 1635–1638.
Boring, E. G. (1920). The control of attitude in psychophysical experiments.
Psychological Review, 27, 440–452.
Brown, A. L., Bransford, J. D., Ferrara, R. A., & Campione, J. C. (1983). Learn-
ing, remembering, and understanding. In J. H. Flavell & E. M. Markman
(Eds.), Handbook of child psychology (Vol. 3, pp. 77–164). New York: Wiley.
Brown, A. S. (1991). A review of the tip-of-the-tongue experience. Psychological
Bulletin, 109, 204–223.
Brown, W. (1910). The judgment of difference. University of California Publica-
tions in Psychology, 1, 1–71.
Castro, C. A., & Larsen, T. (1992). Primacy and recency effects in nonhuman
primates. Journal of Experimental Psychology: Animal Behavior Processes, 18,
335–340.
Cowey, A., & Stoerig, P. (1992). Reflections on blindsight. In D. A. Milner
& M. D. Rugg (Eds.), The neuropsychology of consciousness: Foundations of
neuropsychology (pp. 11–37). London: Academic Press.
Dewey, J. (1980). Art as experience. New York: Perigee Books. (Original work
published 1934)
Dunlosky, J., & Nelson, T. O. (1992). Importance of the kind of cue for judg-
ments of learning (JOL) and the delayed JOL effect. Memory and Cognition,
20, 374–380.
Fernberger, S. W. (1914). The effect of the attitude of the subject upon the
measure of sensitivity. American Journal of Psychology, 25, 538–543.
Fernberger, S. W. (1930). The use of equality judgments in psychophysical
procedures. Psychological Review, 37, 107–112.
Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of
cognitive-developmental inquiry. American Psychologist, 34, 906–911.
George, S. S. (1917). Attitude in relation to the psychophysical judgment.
American Journal of Psychology, 28, 1–38.
Gray, J. A. (1995). The contents of consciousness: A neuropsychological con-
jecture. The Behavioral and Brain Sciences, 18, 659–722.
Harlow, H. F. (1949). The formation of learning sets. Psychological Review, 56,
51–65.
Hart, J. T. (1965). Memory and the feeling-of-knowing experiments. Journal of
Herman, L. M., & Arbeit, W. R. (1972). Frequency difference limens in the
bottlenose dolphin: 1–70 kc/s. Journal of Auditory Research, 2, 109–120.
James, W. (1952). Great books of the western world: Vol. 53. The principles of psy-
chology. Chicago: University of Chicago Press. (Original work published
1890)
Karoly, P. (1993). Mechanisms of self-regulation: A systems view. Annual Re-

view of Psychology, 44, 23–52.
MacMillan, N. A., & Creelman, C. D. (1991). Detection theory: A user’s guide.
Cambridge, UK: Cambridge University Press.
Metcalfe, J., & Shimamura, A. (1994). Metacognition: Knowing about knowing.
Cambridge, MA: Bradford Books.
Morgan, C. L. (1906). An introduction to comparative psychology. London: Walter
Scott.
Nelson, T. O. (Ed.). (1992). Metacognition: Core readings. Toronto: Allyn and
Bacon.
Nelson, T. O. (1996, February). Consciousness and metacognition. American
Psychologist, 51, 102–116.
Nelson. T. O., & Narens, L. (1990). Metamemory: A theoretical framework
and new findings. The Psychology of Learning and Motivation, 26, 125–141.
Roberts, W. A., & Kraemer, P. J. (1981). Recognition memory for lists of visual
stimuli in monkeys and humans. Animal Learning and Behavior, 9, 587–594.
Rumbaugh, D. M., & Pate, J. L. (1984). The evolution of cognition in primates:
A comparative perspective. In H. L. Roitblat, T. G. Bever, & H. S. Terrace
(Eds.), Animal cognition (pp. 569–587). Hillsdale, NJ: Lawrence Erlbaum.
Sands, S. F., & Wright, A. A. (1980). Primate memory: Retention of serial list
items by a rhesus monkey. Science, 209, 938–939.
Schusterman, R. J., & Barrett, B. (1975). Detection of underwater signals by a
California sea lion and a bottlenose porpoise: Variation in the payoff ma-
trix. Journal of the Acoustical Society of America, 57, 1526–1537.
Schwartz, B. L. (1994). Sources of information in metamemory: Judgments
of learning and feelings of knowing. Psychonomic Bulletin and Review, 1,
357–375.
humans and rhesus monkeys (Macaca mulatta) in a psychophysical same-
Shiffrin, R. M., & Schneider, W. (1977). Controlled and automatic human in-
formation processing: II. Perceptual learning, automatic attending, and a
general theory. Psychological Review, 84, 127–190.
Smith, J. D. (1995). The homunculus at home: Commentary on J. A. Gray, The
contents of consciousness: A neuropsychological conjecture. The Behavioral
and Brain Sciences, 18, 697–698.
Smith, J. D., & Minda, J. P. (1998). Prototypes in the mist: The early epochs of
category learning. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 24, 1411–1436.
Smith, J. D., & Minda, J. P. (2000). Thirty categorization results in search of a
model. Journal of Experimental Psychology: Learning, Memory, and Cognition,
26, 3–27.
Smith, J. D., & Schull, J. (1989). A failure of uncertainty monitoring in the rat.
Unpublished data.
Smith, J. D., Schull, J., Strote, J., McGee, K., Egnor, R., & Erb, L. (1995). The
uncertain response in the bottlenosed dolphin (Tursiops truncatus). Journal
of Experimental Psychology: General, 124, 391–408.
Smith, J. D., Shields, W. E., Allendoerfer, K. R., & Washburn, W. A. (1998).

Smith, J. D., Shields, W. E., Schull, J., & Washburn, D. A. (1997). The uncertain
response in humans and animals. Cognition, 62, 75–97.
Smith, S. M., Brown, J. M., & Balfour, S. P. (1991). TOTimals: A controlled
experimental method for studying tip-of-the-tongue states. Bulletin of the
Psychonomic Society, 29, 445–447.
Thomson, G. H. (1920). A new point of view in the interpretation of threshold
measurements in psychophysics. Psychological Review, 27, 300–307.
Tolman, E. C. (1927). A behaviorist’s definition of consciousness. Psychological
Review, 34, 433–439.
Tolman, E. C. (1938). The determiners of behavior at a choice point. Psycholog-
ical Review, 45, 1–41.
Tolman, E. C. (1967). Purposive behavior in animals and men. New York: Mere-
dith. (Original work published 1932)
Washburn, D. A., & Rumbaugh, D. M. (1992). Testing primates with joystick-
based automated apparatus: Lessons from the Language Research Center’s
Computerized Test System. Behavior Research Methods, Instruments, and
Computers, 24, 157–164.
Washburn, D. A., Smith, J. D., Baker, L. A., & Raby, P. R. (2001). Responding
to uncertainty: Individual differences and training effects. Proceedings of the
45th annual meeting of the human factors and ergonomics society (pp. 911–915).
Santa Monica, CA: HFES.
Watson, C. S., Kellogg, S. C., Kawanishi, D. T. & Lucas, P. A. (1973). The
uncertain response in detection-oriented psychophysics. Journal of Experi-
Weiskrantz, L. (1986). Blindsight: A case study and implications. Oxford, UK:
Oxford University Press.
Weiskrantz, L. (1997). Consciousness lost and found: A neuropsychological explora-
tion. Oxford, UK: Oxford University Press.
Woodworth, R. S. (1938). Experimental psychology. New York: Holt.
Wright, A. A., Santiago, H. C., Sands, S. F., Kendrick, D. F., & Cook, R. G.
(1985). Memory processing of serial lists by pigeons, monkeys, and people.
Science, 229, 287–289.
Yunker, M. P., & Herman, L. M. (1974). Discrimination of auditory temporal
differences by the bottlenose dolphin and by the human. Journal of the
Acoustical Society of America, 56, 1870–1875.
11
Can Rhesus Monkeys Discriminate

Between Remembering and Forgetting?
Robert R. Hampton
As I entered the study, I flipped the light switch and was startled by a loud
“pop!” and an electric flash. There was a loose connection that was shorting.
I was headed for the fuse box to remove the fuse when the phone rang. An-
other annoying telemarketer! I went upstairs and got a screwdriver. As I
began to remove the switch cover, I thought, “Did I remove the fuse?” I
could not bring to mind a memory of having done it, so I put down my
screwdriver and checked the fuse box. Sure enough, the fuse for the study
was still in place. I removed it before proceeding to open the switch and
tighten the loose wire.
A form of self-reflective cognition saved me from getting a shock.

Like most humans, I can tell the difference between knowing and not
knowing, or between remembering and forgetting. While we take
such memory awareness for granted, there is surprisingly little com-
pelling evidence that it exists in nonhuman species. The difficulties in
demonstrating memory awareness in nonhuman species stem from
the existence of unconscious memory in humans. In the absence of
awareness, humans show classical conditioning (Clark & Squire,
1998), word priming (Hamman & Squire, 1997; Tulving & Schacter,
1990), and learn new skills (Cohen, Eichenbaum, Deacedo, & Corkin,
1985; Knowlton, Ramus, & Squire, 1992; Knowlton & Squire, 1993).
Given that these behaviors can occur without awareness in humans,
isn’t it possible that unconscious, inaccessible forms of memory and
cognition can account for all nonhuman behavior? Indeed, it is rightly
argued that we do not need to invoke concepts like consciousness or
memory awareness to account for most, possibly all, of nonhuman
behavior (e.g., Shettleworth, 1998, pp. 5–10). In this chapter, I outline
a framework that makes tests for memory awareness in nonhumans
possible, and describe some of the few observations of nonhuman be-
272
Discriminating Remembering and Forgetting 273
havior that are difficult to explain without reference to memory

awareness.
AN ASSUMPTION OF ADAPTATION
I assume that self-reflective cognition evolved by natural selection.
Most readers probably do not find this position radical. Some conclu-
sions that follow from this assumption may be more controversial, but
they suggest a way to make the problem of detecting self-reflective
cognition in nonhuman animals tractable. The central working hy-
pothesis of an evolutionary approach, and a point that distinguishes
it from many more philosophical approaches, is that self-reflection
is not epiphenomenal but is instead functional (Dennett, 1992). Self-
reflective cognition evolved because it enabled organisms to behave
overtly in adaptive ways that would not be possible without it. The
behaviors enabled by self-reflection were the basis on which natural
selection acted to increase the occurrence of self-reflection in subse-
quent generations. Self-reflection is therefore a mechanism that sup-
ports a set of potentially identifiable adaptive behaviors. Like other
cognitive mechanisms, it can be inferred from overt, nonverbal behav-
ior, without reference to private experience.
In studies of human cognition, it is convenient to ask subjects about
their experiences as they carry out cognitive tasks. To confirm the
occurrence of episodic memory, for example, subjects might be asked
whether memory was accompanied by the “experience of having been
there,” or whether they “remembered” or just “knew” the correct an-
swer (Tulving, 1985, 1993). But emphasizing these verbal reports of
private phenomenology is problematic for the wider effort of compar-
ative cognition. Defining cognitive systems in ways that depend on
verbal reports of private experience creates a rift between studies of
cognition in humans and nonhumans because such reports cannot be
obtained from nonhumans. Furthermore, the definition of phenomena
like episodic memory, declarative memory, and metacognition may
become so dependent on reported phenomenology that the most im-
portant feature of these cognitive mechanisms, their function, is ob-
scured. As a result, it can appear impossible, even in principle, to
demonstrate the existence of these cognitive phenomena in nonhu-
mans, simply for lack of verbal behavior.
I argue that verbal reports are not necessary to demonstrate self-
reflective cognition. It can be inferred instead from the nonverbal abil-
ities it confers on a human or nonhuman animal (see also Inman &
Shettleworth, 1999; Smith, Shields, Washburn, & Allendoerfer, 1998;
Weiskrantz, 2001). By treating self-reflection as a biological phenome-
non that resulted from the mechanistic action of natural selection, we
can define self-reflective cognition by what it accomplishes for an or-
ganism, rather than by how it is experienced. Investigations of self-

reflective cognition in nonhumans can then focus on the same material
that natural selection acts on—overt behavior. If self-reflective cogni-
tion was naturally selected on the basis of overt behavior, then the
occurrence of self-reflection can be identified on the basis of those
same, or related, overt behaviors.
MENTAL CONTENTS AND BEHAVIORAL CAPABILITY

In over a century of scientific study of animal cognition, well-designed
experiments have allowed investigators to infer some of the contents
of animal information processing systems. Pigeons demonstrate that
they mentally represent specific rewards when they peck. When peck-
ing a response key associated with water reward, pigeons peck with
a “nibbling” motion like that used when drinking. In contrast, when
pecking a key associated with grain, they peck with the beak open, as
is done when actually pecking at grain (Jenkins & Moore, 1973). Rats
represent the durations of sounds and will appropriately press levers
associated with long and short sounds (Meck, 1983). Scrub jays repre-
sent which foods they have hidden, where they hid them, and even
when they did the hiding, demonstrated by their avoidance of the
locations of perishable foods stored long enough to have spoiled
(Clayton & Dickinson, 1999). In each of these cases, the animals be-
haved in a way that would not be possible had they not mentally
encoded the types of reward, the durations of the tones, and the loca-
tions of the foods. The observed behavior requires the inference of
mental representations of these particular features of events or stim-
uli. In other words, these experiments show that nonhuman animals,
in some sense, remember. The success of this behavioral approach, in
which the unobservable features of memory are logically inferred
from the observable properties of behavior, encourages emulation.
However, while these clever experiments do show that animals re-
member, there is nothing in these findings to indicate that the animals
were aware of having their memories the way that humans are aware
of many memories. The experiments provide no basis for discriminat-
ing between memories that are like human conscious declarative
memory and those that are unconscious or implicit. That is, they do
not demonstrate memory awareness. However, that does not mean
that other experiments could not provide relevant data. I argue that it
is possible to treat memory awareness as another feature of a memory,
like the type of reward, duration of a tone, or location of a seed. It
follows that observable behavior, if measured under appropriately
controlled conditions as in the preceding examples, can be sufficient
to demonstrate memory awareness in a nonverbal animal.
A FUNCTIONAL DEFINITION OF MEMORY AWARENESS

The conceptual approach used to demonstrate memory in nonhuman
animals may be adapted to generate tests for memory awareness as
well. The presence of memory is inferred when a subject demonstrates
behavior that cannot be explained without reference to memory (see
Shettleworth, 1998). For example, a monkey might be presented with
an object overlaying a candy. It removes the object and eats the candy.
The object is then taken out of sight for a minute, after which the
monkey is presented with the object it just saw, plus another object.
In many trials, the monkey shows that it can select the rewarded ob-
ject and avoid the object that had not been paired with a candy. The
monkey can “match to sample,” correctly identifying a recently seen
object. In order for the monkey to do this, some representation, or
memory, of the object must persist in the brain after the object is re-
moved from view to later control behavior. Under similar circum-
stances, humans could not only respond correctly, but could also pre-
dict their ability to respond correctly. Thus, humans not only have
primary mental states of memory about events or objects in the world,
they also have secondary mental states about memories—for instance,
states indicating the presence or absence of memory. That is, we can
often know whether or not we know the correct answer. How could
a nonhuman animal demonstrate such a secondary mental state with-
out the use of language?
Conventionally, nonhuman memory is assessed in forced-choice
situations. In the case of our monkey looking for the candy, the re-
warded object and the distracter object are presented simultaneously,
and the monkey has to choose between them. If it is unsure, the mon-
key has to guess. A human confronting uncertainty in such a situation
might say he was uncertain, or state that he was guessing. But the
monkey is afforded no opportunity to do this in conventional tests of
memory (Inman & Shettleworth, 1999; Smith et al., 1998; Weiskrantz,
2001). We can accurately measure how well the monkey remembers,
but we have no measure of whether it knows it remembers, or is un-
certain. To get such a measure, we need to explicitly posit what mem-
ory awareness might afford an organism.
Memory awareness can make it possible to determine whether or
not the information required to successfully perform a particular task
is available. Consider the example of making a phone call to order
take-out food from a restaurant you have used several times. Before
starting, you would likely attempt to retrieve a memory of the num-
ber. If you successfully recall the number, you go ahead and dial. If
retrieval fails, you look up the number before dialing. Awareness of
the presence or absence of a memory of the number permits you to
choose appropriately between dialing immediately or looking up the

number. It should be noted, though, that awareness of the memory is
not the only way you could accurately determine whether or not to
look up the number. If you call every day to order food, you would
likely cease to check whether you know the number, and simply as-
sume that you do—not necessarily on the basis of memory awareness,
but rather on the basis of your knowledge that you have succeeded
many times in dialing correctly. Nonetheless, normal humans are in-
deed capable of making this sort of judgment on the basis of memory
awareness.
The alternatives to evaluating knowledge states through memory
awareness can be quite clumsy. Imagine being unable to introspect
about your memory of a phone number. You would either have to
look up the number every time you call, or use trial and error to
determine whether or not you remembered the correct number. In the
case of phone calls, this could be embarrassing. If you were unable to
determine whether or not you know how to rewire a fuse box, or need
to hire an electrician for the job, trial and error testing could have
more serious consequences. Humans can sometimes avoid putting
themselves in dangerous or futile situations through accurate self-
reflective assessments of knowledge.
One can imagine similar situations in which memory awareness
would be helpful to nonhuman animals. Consider a rabbit that detects
a fox in the field. It faces a choice between freezing, possibly avoiding
detection, or running for shelter. The ability to assess whether it
knows the location of a shelter would be helpful to the rabbit in mak-
ing an adaptive choice. If the rabbit knows of a shelter, nearby but
out of sight, running to it might be the best option. In contrast, making
itself conspicuous by dashing around the field at random searching
for shelter would be a bad choice. Thus the rabbit could adaptively
choose to freeze or run based on awareness, or lack of awareness, of
a memory for a nearby shelter. Of course, the fact that it would be
useful for a rabbit to have memory awareness does not mean that
rabbits have it. It would also be useful for the rabbit to sprout wings
and fly away from danger, yet rabbits do not fly. Nonetheless, think-
ing about how memory awareness would be useful suggests how to
design experiments to test for it. Speculation can then be replaced by
experimentation.
A SIMPLE MODEL
Awareness of the presence of memory can permit an animal to choose
adaptively between behavioral options when the success of one or
more of the options depends on remembered information, as when
the rabbit’s decision to run or freeze should be contingent on knowing
of a good place to run toward. Humans can choose between dialing a

number and looking it up, first by assessing whether or not we re-
member the number. We can either choose to undertake a home repair
ourselves, or hire a professional, based on an assessment of our own
knowledge relevant to completion of the repair. Of course, in initially
testing for this ability in nonhuman animals, a simple situation and a
simple model are desirable. Such a simple model of memory aware-
ness is diagrammed in figure 11.1. A prerequisite for a metacognitive
or self-reflective system is a cognitive system in which stimuli in the
world are mentally represented. Modern cognitive theory posits that
mental representations link events outside the organism to behavior
emitted by the organism (Terrace, 1984). For example, without mental
representations of objects, the monkey in our earlier example could
not correctly choose the rewarded object. In figure 11.1, seeing the
stimulus object (S) causes a mental representation of the object (M) to
be formed. This representation persists after the object is removed
from sight and later guides the response (R). We can therefore infer
the existence of M because it makes a logically necessary link between
S and R.
Accepting the existence of M in figure 11.1 admits the possibility
of another, secondary or meta mental representation (m), as shown in
the lower part of figure 11.1. While M is a cognitive representation of
an object in the world, m is a metacognitive representation about M.
Figure 11.1 Measuring memory awareness

behaviorally. Viewing an object (S) causes
the formation of a mental representation,
or memory, of the object (M) that persists in
the brain even when the object is removed
from view. The memory later guides the re-
sponse (R) to the remembered object in a
memory test. In the same way that M is
about the object in the world, m is a sec-
ondary mental representation that is about
the memory M. A secondary response (r) is
required to infer the presence of m. A pro-
cedure for doing so is shown in figure 11.2.
See color insert.
It is a thought about a thought (Nelson, 1996). To test for the existence

of m, an experimental situation must be created in which a second
response (r) allows us to infer the presence or absence of m, just as R
allows us to infer M.
THE EXPERIMENTS
A schematic of the experiment used to test for metacognitive repre-
sentation is shown in figure 11.2 (Hampton, 2001). Matching to sam-
ple was the primary task, which requires first-order mental represen-
tations, such as M in figure 11.1. At the beginning of each trial,
monkeys saw a centrally located image on a touch-sensitive computer
monitor (figure 11.2, top panel). They touched the image, demonstrat-
ing that they had viewed it. The screen was then blank for a delay
period over which monkeys should remember the image. At the end
of the delay, monkeys were presented with four images, one occupy-
ing each of the four corners of the computer monitor (lower left
panel). In order to receive a reward, monkeys had to touch the image
that was identical to the one seen at the beginning of the trial. The
task was made difficult by increasing the delay to a point at which
the monkeys often made mistakes. Thus, on some trials they remem-
bered while on other trials they had forgotten (see figure 11.2).
In the secondary, or metacognitive, task, the monkeys were given
a choice between taking the memory test and receiving a favored re-
ward for correct responses, or declining the test and receiving a less
desirable but guaranteed reward (middle panels). This choice phase
occurred at the end of the delay on each trial, and before the monkey
saw the test array. Facing these contingencies, a monkey could maxi-
mize reward by choosing to take the memory tests when memory for
the studied image was good, and avoiding the test (settling for the
lesser but guaranteed reward) when memory was relatively poor.
An important feature of this procedure is that it requires a prospec-
tive judgment of memory, that is, the monkeys had to decide whether
to take or decline the memory test before being presented with the
test. A prospective judgment addresses two concerns. First, because
the test stimuli were absent when the judgment was made, the choice
to decline tests could not be based on familiarity or perceptual fluency
of the correct test image (Verfaelli & Cermak, 1999; Wagner & Gabri-
eli, 1998). Second, a prospective judgment also precludes a decision to
decline tests based on direct experience with the difficulty of a given
test. Previous studies using pigeon subjects showed that the distinc-
tion between prospective and concurrent memory judgment can be
critical (Inman & Shettleworth, 1999; J. E. Sutton & S. J. Shettleworth,
personal communication, February 12, 2003). Pigeons offered the op-
tion of declining tests concurrently with presentation of the test dis-
Figure 11.2 Method for assessing whether monkeys know when they remember. Each
colored panel represents what monkeys saw on a touch-sensitive computer monitor at
a given stage in a trial. At the start of each trial, subjects studied a randomly selected
image. A delay period followed, during which monkeys often forgot the studied image.
On two thirds of trials, animals chose between taking a memory test (right panel, left-
hand stimulus) and declining the test (right panel, right-hand stimulus). On one third of
trials, monkeys were forced to take the test (left panel). Better accuracy on chosen
memory tests than on forced tests indicates that monkeys know when they remember
and decline tests when they have forgotten, if given the option. See color insert.
play behaved as if they knew when they remembered. However,

when these same birds were required to judge memory prior to pre-
sentation of the memory test, they were unable to discriminate trials
on which they remembered from those on which they had forgotten
the sample.
In earlier training, two monkeys, Shepard and Gagarin, were famil-
iarized with the significance of the choice phase stimuli (figure 11.2,
middle panels). Preference for peanut over pellet rewards was also
established earlier and confirmed throughout the experiments. The
monkeys were tested on matching to sample with four new images
used each day. On each trial, one of the four images was randomly
selected to serve as the sample for that trial; the other three images
served as distracters, yielding a chance rate of 25%. Initially, the mon-
keys were given only the primary matching task, without the second-
ary choice phase in which they could decline memory tests. The delay
over which the monkeys were required to remember the sample on
each trial was gradually increased until they made errors on approxi-
mately 35% of trials. Thus, the monkeys were sure to be experiencing
a substantial number of trials on which they forgot the identity of the
sample. Shepard was tested at a delay of 34 seconds and Gagarin at
38 seconds.
The choice phase stimuli were then incorporated into testing. At
the end of the delay period on each trial, one of two choice phase
arrays was presented. On one third of trials, the “choice” consisted of
only the stimulus associated with taking the memory test. On these
forced-test trials, monkeys had to touch this stimulus in order to pro-
ceed to the memory test. At test, correct responses were rewarded
with highly desirable peanuts, while a distinctively colored error
screen and no reward followed incorrect choices. These trials were
therefore almost identical to the normal matching-to-sample trials, ex-
cept the monkeys had to touch the stimulus associated with tests be-
fore proceeding to the test. On the other two thirds of trials, monkeys
were given a choice between taking the test and declining the test. If
they selected the stimulus associated with taking the memory test, the
trial proceeded as just described. If they chose the stimulus associated
with avoiding the memory test, an escape response screen appeared.
The monkeys were rewarded with less desirable primate pellets after
they touched the image on this screen, which was unrelated to the
other test images and presented in a location not used for other im-
ages.
If monkeys can accurately assess whether or not they remember
the studied image, they should choose to take the test when they re-
member and avoid the test when the studied image is forgotten. Reli-
able discrimination between trials on which memory was good and
trials on which memory was poor should result in higher accuracy on
chosen tests compared to forced tests. Both monkeys were more accu-
rate on trials in which they freely chose memory tests than they were
on forced-test trials in which they were not given the option of declin-
ing the test (figure 11.3, paired t tests: Shepard, t9 = 3.91, p < .01; Gaga-
rin, t9 = 4.51, p < .01). However, accuracy on these tests does not di-
rectly reflect the accuracy expected on trials in which the monkeys
declined the memory test.
Unfortunately, it is difficult to get a direct measure of accuracy on
trials on which the monkeys declined to take the test, for two reasons.
The first is that the contingencies of the experiment, by which the
monkeys learned the significance of taking and declining memory
tests, would be undermined by any reasonably large number of probe
tests in which the monkey was forced to take a memory test, even
though it had declined to do so. Second, performance on such trials
is unlikely to reflect memory accurately. This is because the monkeys
expect a small, reliable reward after declining the test. If they were
forced to take the test instead, they would likely be surprised and
possibly frustrated by the unexpected change. Surprise or frustration
could disrupt performance and impair accuracy. Thus, forced tests
following the decline response do not provide a good measure of
memory on declined trials. Instead, we have to rely on a less direct
measure, as follows.
Accuracy on forced tests is a weighted average of accuracy on tests
subjects would have declined, given the choice, and those they would
freely have chosen to take anyway. The expected accuracy on declined
trials is substantially lower than this weighted average, and avoiding
Figure 11.3 Accuracy on freely chosen tests and forced

tests. Dark bars represent accuracy on tests the mon-
keys chose to take. Striped bars represent performance
on trials in which the animals were not given the
choice of declining tests. Scores for the two monkeys
are the means of 10 daily sessions. Error bars are stan-
dard errors. Subjects would be correct 25% of the time
if guessing.
these tests is of considerably more value than it at first appears. Be-

cause accuracy on freely chosen tests is known, as is the proportion
of tests taken and declined on free-choice trials, the expected accuracy
in trials on which the monkeys declined the memory test can be deter-
mined. Using the proportions of tests monkeys declined when given
the opportunity (Shepard, 0.51; Gagarin, 0.36), and the accuracies
shown in figure 11.3, expected accuracies on the trials monkeys de-
clined are determined to be 58.1% and 46.8%, respectively. The accu-
racy of memory on trials the monkeys declined is thus substantially
lower than it is on other trials, and monkeys experience a substantial
increase in reward rate by avoiding these tests.
How Did the Monkeys Know That They Had Forgotten?

The relevance of these findings to memory awareness depends on de-
termination of the discriminative stimulus used by monkeys to guide
their choice to decline some of the memory tests. The working hypoth-
esis is that it is awareness of the memory that serves as the discrimina-
tive stimulus. Therefore, the critical concern is that some factor other
than the absence of memory per se might control the choice to decline
memory tests. There are many candidate alternatives. Events that
might occur during the delay interval, such as noises, bouts of groom-
ing, or changes in motivation, might result in forgetting by distracting
the animal or interrupting a memory process. These events them-
selves could therefore cue animals to decline tests and yield a false
impression that the animals attended to their own memory states. To
rule out the possibility that monkeys’ decisions to decline tests were
controlled by such external cues, monkeys were presented with infre-
quent randomly distributed probe trials. These unpredictable probe
trials were identical to normal trials in every way except that no image
was presented for study (figure 11.4). When the absence of memory
was experimentally controlled in this way, a priori prediction of
which tests the monkeys would decline was possible. After an inter-
trial interval, and a delay period equivalent to that on normal trials,
animals were given the choice of declining or taking a memory test,
just as would occur on normal trials.
If the absence of memory causes the monkeys to decline tests, they
should decline tests on these no-sample probe trials, treating them like
trials on which they have forgotten the studied image. If, however,
the decision to decline tests is controlled by some environmental or
behavioral event, subjects should decline normal and probe trials with
equal probability, because such events are evenly distributed among
the randomly intermixed normal trials and the no-sample probe trials.
In six test sessions with no-sample probe trials, both monkeys were
much more likely to decline tests if no image was presented for study
than they were to decline tests on normal trials (figure 11.5; Shepard:
Figure 11.4 Schematic of a normal trial (upper panel) and a no-sample

probe trial (lower panel). The white lines represent the passage of time.
During the intertrial interval (ITI), the screen was black. On normal trials,
a sample appeared and the monkeys had to touch it to advance the trial.
After a delay, the monkeys could choose to either take or decline the
memory test. Probe trials were identical, with the exception that no sam-
ple appeared, and the monkey did not have to touch the screen to advance
the trial. See color insert.
t5 = 24.34, p < .01; Gagarin: t5 = 10.19, p <.01). Shepard declined the

memory test on 49% of normal trials, and 100% of the probe trials
where no sample was presented. Gagarin declined 18% of normal
trials and 63% of probe trials. These results, following the experimen-
tal manipulation of memory, provide compelling support for the hy-
pothesis that the choice to decline tests was based on the absence of
memory per se, rather than due to some external event correlated with
forgetting.
Monkeys might have gradually learned to base the decision to de-
cline tests on some distinguishing feature of probe trials, rather than
on the absence of a memory per se. The first session of probe trials
was therefore analyzed separately. Both monkeys declined probe tests
from the first session of testing (Shepard, χ1,n=66
2
= 7.66, p < .01; Gagarin,
χ1,n=69 = 7.88, p < .01), indicating that learning did not take place after
2
the onset of probe trials. Additionally, Shepard declined every probe

trial presented and could never have learned the negative conse-
quences of choosing a memory test on such trials. The high probabil-
ity with which monkeys declined the no-sample probe trials therefore
Figure 11.5 Probability of declining tests on normal

trials and on probe trials lacking an image to study.
Dark bars show the proportion of normal trials on
which monkeys declined tests; striped bars represent
this proportion on probe trials. Error bars are standard
errors. Inset in each bar is the percentage of each test
type declined in the first session of testing only. These
results indicate that it is the absence of a memory that
causes the monkeys to decline tests. If some factor
other than the absence of memory per se, such as dis-
tracting noises, variation in motivation, or fatigue, con-
trolled the decision to decline tests, normal and probe
trials would be affected equally.
reflects spontaneous generalization to these probes as equivalent with

trials on which the sample was forgotten.
Monkeys might have used delay, or a correlate of delay, as a cue
to decline tests. In these experiments, the computer controlled when
the monkey could first respond, but the monkey could respond at any
time after that point. Thus, on some trials the monkey might complete
the choice phase promptly, and other times slowly. Because perfor-
mance deteriorates as the delay before the memory test increases, the
monkeys could theoretically have produced the pattern of results ob-
served by choosing to take the memory test when they completed the
choice phase promptly, and choosing to avoid the test when this
phase was completed more slowly. Indeed, the monkeys responded
slightly more quickly when choosing to take the test than they did
when choosing to decline the test (Shepard, mean difference 0.29 sec-
onds, t9 = 2.09, p < .10; Gagarin, mean difference = 0.46 seconds, t9 =
3.7, p < .01). However, these differences of a fraction of a second in
response latency, added to mean delays of about 36 and 40 seconds
respectively, are hardly enough to produce the difference in accuracy
between freely chosen and forced memory tests. It is rather more
likely that this small difference in latency to complete the choice phase
reflects a slightly longer decision process preceding the decision to
avoid the memory test. Apparently, when the monkeys are sure they
remember, they decide quickly and surely to take the test. But when
they are less certain, it takes slightly longer to make a decision. This
difference may reflect a search of memory conducted at the time the
monkey is choosing to take or decline the memory test. On trials in
which a memory is located, the search is terminated and the monkey
chooses to take the test. On trials on which the sample is forgotten,
the search continues for some time without a memory being located.
Eventually the search of memory is terminated and the monkey de-
clines the test. Thus longer latencies are associated with failure to find
the item in memory (Briggs & Blaha, 1969; for review, see Van Zandt
& Townsend, 1993).
Forgetting as Time Passes

Varying the delay over which the monkeys were required to remem-
ber the studied image provides a second way of directly manipulating
memory in these experiments. Because memory declines as time pas-
ses, monkeys should decline tests more often after long than after
short delay intervals. Each monkey was tested with five delay inter-
vals, which were randomly intermixed in test sessions so that the
monkeys could not anticipate the delay on a given trial. For Shepard,
the delay intervals ranged from 12.5 to 200 seconds; for Gagarin the
range was 15 to 240 seconds. Accuracy on forced tests was high for
both monkeys at short delays, but declined as the delay interval in-
creased (figure 11.6, filled circles; Shepard: F4,116 = 45.66, p < .01; Gaga-
rin: F4,116 = 44.72, p < .01). As would be expected if monkeys declined
tests when memory was poor, both monkeys declined tests more often
after long delays than after short delays (figure 11.6, filled squares;
Shepard: F4,116 = 196.02, p < .01; Gagarin: F4,116 = 30.76, p < .01). The
first 100 free-choice trials with these variable delays were analyzed
separately to test whether monkeys generalized immediately to the
delays or learned gradually with experience. Trials were divided into
two groups: those with delays shorter than the mean and those with
delays longer than the mean. Shepard declined 85% of these long-
delay trials, compared with 43% of short-delay trials (χ1,n=100 2
= 17.34,
p < .01), while Gagarin declined 32% and 16% of these trials, respec-
2
tively (χ1,n=100 = 3.72; p = .05).
By selectively declining tests when they had forgotten the studied
image, both monkeys were significantly more accurate overall on cho-
sen tests than they were on forced tests (Shepard: 88% vs. 67%, F1,29 =
152.62, p < .01; Gagarin: 78% vs. 69%, F1,29 = 21.20, p < .01). But in
addition to improving overall accuracy, monkeys able to detect the
absence of memory with high sensitivity should also be able to im-
prove memory performance at each delay at which substantial forget-
ting occurs. No difference between free- and forced-test accuracy is
0.8
0.6 Shepard
0.4
Proportion Correct or Declined
0.2
0
12.5 25 50 100 200
0.8
0.6 Gagarin
0.4
0.2
0
15 30 60 120 240
Delay (s)
Figure 11.6 Accuracy and the probability of declining tests after ran-
domly intermixed variable delay intervals. Filled squares depict the
probability of declining tests. Shaded circles represent accuracy on
freely chosen tests, and filled circles accuracy on forced tests. Error
bars are standard errors.
expected at delays at which animals rarely forget, and accuracy is

high, because there is little room for improvement. However, as the
delay interval increases, and forced-choice accuracy declines, a differ-
ence between accuracy on chosen tests and forced tests should
emerge. Shepard clearly showed this pattern (figure 11.6, shaded cir-
cles; F1,23 = 53.66, p < .01; only the first four delays were analyzed,
because at the longest delay Shepard declined nearly 100% of tests
and did not choose any of these trials in over half of test sessions,
thereby precluding an accurate estimate of accuracy). Gagarin was
also more accurate on freely chosen tests at longer delays, but this
difference was not statistically significant (F1,25 = 3.17, p < .09).
Together, these findings indicate that rhesus monkeys know when
they know. In addition to often retaining a representation of a recently
seen image, the monkeys were able to discriminate between the pres-
ence and absence of such a representation in memory. Competence in
this discrimination allowed the monkeys to choose appropriately to

take memory tests when memory for the studied image was strong
and to decline the test when memory was weak, thus increasing the
total number of rewards received in a session. The ability of the mon-
keys to generalize performance to the no-sample probe trials and to
the sessions with variable delays rules out many “simpler” accounts
of this behavior, such as a discrimination based on some external
event rather than attention to the presence or absence of memory per
se. Thus, monkeys are aware of at least some of their memories. Mem-
ory awareness is a form of metacognition, or thinking about thinking
(Nelson, 1996). For metacognition to occur, two types of mental state
are required, an object-level state and a metalevel state. In the present
case, memory for the recently seen image constitutes the object-level
state, M in figure 11.1. The metalevel is in some way contingent on
the object-level memory, and is represented by m. This contingency
relationship between the object level and the metalevel is often re-
ferred to as monitoring, reflecting the fact that the meta-level state is
indicative of the condition of the object level, for example, indicating
the quality, vividness, or, in the most basic case, presence or absence
of the object-level state. The metalevel state can thereby provide the
basis for a discrimination between remembering and forgetting, by
reflecting the presence or absence of memory.
CANDIDATE PSYCHOLOGICAL MECHANISMS

I have argued that monkeys are aware of the presence of memory
and have bolstered this position by eliminating many other possible
explanations for the behavior of declining memory tests. But narrow-
ing the field to some sort of memory awareness leaves a great deal of
explanatory terrain to cover. Of what aspects of a memory can mon-
keys be aware? Can we determine any more about the psychological
mechanism underlying memory awareness? Future work might dis-
criminate among several candidate mechanisms (figure 11.7).
Before describing these candidate mechanisms, it should be said
that it is not certain whether monkeys were using short-term memory
or long-term memory in these tests, and so it may be premature to
analyze performance in terms specific to either type of memory. How-
ever, several facts suggest that the monkeys were using a long-term
memory mechanism. Monkeys were completing the memory tests
about 40 seconds after the offset of the sample, and in subsequent
experiments, delays of up to 240 seconds were used. These delays far
exceed what are often considered the temporal limits of short-term
memory for monkeys (e.g., Alvarez-Royo, Zola-Morgan, & Squire,
1992). While neurobiological evidence was not collected in these ex-
periments, many experiments have shown that monkeys performing
Figure 11.7 Cartoons of candidate processes underlying memory aware-

ness. The left column represents the contents of cognitive processing dur-
ing trials on which monkeys choose to take the memory test. The right
column depicts the same on trials on which the test is declined. The mem-
ory flag hypothesis posits an indicator for the presence of memory. Mon-
keys can respond on the basis of the indicator but are not aware of the
memory itself. In the case of the other two hypothetical mechanisms, the
decision to take the memory test is based on the richness of the memory
retrieved. See text for more details. See color insert.
very similar memory tasks, at delays shorter than 40 seconds, are im-
paired following lesions of the medial temporal lobe (Mishkin & Mur-
ray, 1994; Squire & Zola-Morgan, 1991). Thus, performance on the
tests described here almost certainly depends on the medial temporal
lobe memory system, and falls in the category of long-term memory.
Nonetheless, depending on the criteria adopted for distinguishing be-
tween working and long-term memory, the data described could fall
in either category.
Here I suggest three candidate mechanisms to account for the per-
formance of monkeys in these experiments. The first account, the
“memory flag” hypothesis, posits a mental representation of memory
strength. Such a representation might be as simple as a binary signal
representing only the presence or absence of memory. This represen-
tation would encode nothing about the context of the study episode
or the features of the image being remembered. The activity of the
memory flag could merely reflect the presence or absence of something
in memory, and monkeys would choose to take the test when the flag
was active. Because activation of the flag is contingent on the presence
of a memory, monkeys would be more accurate at test if the flag was
active than when it was inactive.
A more elaborate decision process could be based on the richness
of the recalled memory. If a retrieval attempt results in a very detailed
representation of a studied image, including its shape and color for
example, this could guide the monkey to choose to take the memory
test on that trial. In contrast, if few or no features of the item could be
recovered, the monkey would choose to decline the memory test. At
the extreme of the richness of recall account would be episodic re-
trieval, in which the monkey recalled a sufficient amount of detail
from the study episode to reexperience it (Tulving, 1985). In any case,
the monkeys could base their decision on the number of features of
the study episode or the studied image that could be successfully re-
trieved. When retrieval is rich, the monkey chooses to take the mem-
ory test. When the retrieved memory is impoverished, the test is de-
clined.
To determine which mechanism underlies memory awareness in
this case, we need to know more about what the monkey is thinking
when making the decision to take or decline the memory test. What
are the contents of the representation guiding the monkey’s choice? Is
the representation binary, encoding only two possible states—mem-
ory present and memory absent—or does it include many features of
the remembered item? Experiments in which the monkeys are re-
quired to make delayed categorization decisions about the studied
image might provide the answer. For example, after choosing to take
or decline the memory test, the monkey might be required instead to
categorize the image on the basis of color or shape (e.g., red vs. blue,
animal vs. plant). If the choice to take the test were based entirely
on a memory flag, there should be no basis on which to make the
categorization, because the activity of the flag does not reflect the spe-
cific contents of the memory. In contrast, if the monkey calls the image
to mind and assesses the richness of the memory before choosing to
take or decline tests, it will have information about the properties of
the image available to guide categorization following “remember” re-
sponses but not following “forget” responses. Recall that the monkey
has to choose to take or decline the memory test before seeing the test.
Therefore the tests just described could be used to determine which
features of the image the monkey can freely recall (see Menzel, 1999;
chapter 8, this volume).
Less elaborate mechanisms may be sufficient to account for similar
data collected under different conditions than those described here.
Pigeons declined tests selectively when memory was poor only if they
viewed the memory test at the time the choice to take or avoid the
test was made (Inman & Shettleworth, 1999; J. E. Sutton & S. J. Shet-
tleworth, personal communication, February 12, 2003). Given the suc-
cess of pigeons under conditions of simultaneous presentation and
the failure of the same birds when the judgment to decline memory
tests preceded the presentation of the memory test, it is likely that the
two types of test differ in the mechanisms required to solve them.
Pigeons apparently cannot judge the strength of a memory for an ab-
sent image by “calling it to mind,” although some other mechanism
does allow them to choose to decline tests when memory is weak
when the test is present (see Inman & Shettleworth, 1999). In related
work with monkeys, subjects similarly made the choice to take or de-
cline tests in the presence of the test discriminanda (Smith et al., 1998;
chapter 10, this volume). In the innovative work of Call and Carpenter
(2001; chapter 13, this volume), apes and children collected more in-
formation when confronted with a problem they lacked the necessary
information to solve. Because these primates were not required to
judge their memory before seeing the memory test, they could have
been using a mechanism similar to that used by Shettleworth and
Inman’s pigeons. While it remains to be determined what this mecha-
nism is, apparently it does not require free recall (Menzel, 1999) of the
memory. This does not mean that the subjects were not engaged in
metacognition. The metacognition may not have been based on direct
access to a mental representation, but rather on indirect evidence (Fla-
vell, 1979). In constrast, the prospective memory judgment used in the
experiments described in this chapter, and other control procedures,
made it very unlikely that monkeys based their judgments of memory
strength on indirect evidence. The monkeys most likely based their
decision to take or decline tests on a direct assessment of a recalled
memory.
MEMORY AWARENESS
Neurobiology of Memory Awareness

Our understanding of the neurobiology of memory in rhesus mon-
keys has advanced dramatically in the last 15 years (Murray, 1996).
Studies continue to focus on the medial temporal lobes, but new em-
phasis has been placed on areas adjacent to the hippocampus, includ-
ing the perirhinal, entorhinal, and parahippocampal cortices. It has
also become clear that the function of these medial temporal lobe
structures is often dependent on interaction with other areas, includ-
ing the frontal cortex (Easton, Parker, & Gaffan, 2001; Gaffan, Parker,
& Easton, 2001; Parker, Wilding, & Akerman, 1998). Despite these ad-
vances in the localization of memory function, the animal models
used to date do not capture a conspicuous feature of human declara-
tive memory, memory awareness. As described in the introduction to

this chapter, this failure to capture memory awareness stems from the
exclusive use of forced-choice memory tests, which do not provide
any opportunity for subjects to demonstrate awareness of memory or
metacognitive ability. Instead, investigators have necessarily relied on
a largely circular argument to justify animal models of declarative
memory. This argument posits that because human declarative mem-
ory depends on medial temporal lobe structures, and memory deficits
have been observed in monkeys following damage to these same
structures, the tests used to detect these deficits measure declarative
memory or amnesia (e.g., Gaffan et al., 2001; Squire & Zola-Morgan,
1991). While this logic has permitted work to move forward on the
localization of structures necessary for memory in monkeys, it does
not provide much help in discriminating between declarative memory
and implicit memory in nonhuman animals. With the new techniques
described here and elsewhere (Call & Carpenter, 2001; Hampton,
2001; Smith et al., 1998) comes the promise of directly studying the
neurobiology of memory awareness and declarative memory in non-
human animals for the first time.
Evolution of Memory Awareness

The ability of these monkeys to appropriately decline memory tests
when they were unlikely to choose the correct image indicates that
they know when they remember. The monkeys demonstrated a form
of self-reflective cognition that I have referred to as memory aware-
ness in this chapter. Awareness is one of the properties of declarative
memory, strongly suggesting that the monkeys used declarative mem-
ory, at least in part, to identify the correct images in these experi-
ments. This work reinforces other experiments with nonhuman pri-
mates (Call & Carpenter, 2001; Smith et al., 1998; chapters 10 and 13,
this volume). In each of these cases, nonhuman primates appear to
demonstrate the ability to make judgments about their own knowl-
edge states. In striking contrast to these observations of primates, pi-
geons tested under similar conditions appear to lack memory aware-
ness, suggesting that it may not be widely distributed among species
(Inman & Shettleworth, 1999; J. E. Sutton & S. J. Shettleworth, personal
communication, February 12, 2003). However, considerably more
comparative work with pigeons and other species will be required to
substantiate the presence of memory awareness in some, but not
other, animals. For one thing, further work will be required to deter-
mine the importance of methodological differences, such as whether
the subject makes a memory judgment before being presented with
the memory test, or in the presence of the test.
One central question to be answered about the evolution of self-
reflective cognition is phylogenetic. When in our evolutionary history
did it originate? The existence of memory awareness in many species

would suggest an early emergence in our ancient pre-primate ances-
tors. In contrast, a more restricted distribution among our closer rela-
tives would indicate a more recent emergence of memory awareness
(Riley & Langley, 1993). Answering the phylogenetic question thus
requires testing many species under similar conditions. To start, it
would be interesting to test more species of nonhuman primates. Cur-
rently there is evidence for memory awareness in apes (Call & Car-
penter, 2001) and in Old World monkeys (Hampton 2001; Smith et al.,
1998). Old World monkeys and apes shared a common ancestor 20 to
25 million years ago (Tomasello & Call, 1997, p. 15). This suggests that
memory awareness likely first evolved at least 20 to 25 million years
ago in an ancestor apes share with the Old World monkeys. Should
memory awareness be shown to occur in New World monkeys (capu-
chins or marmosets, for example), or in the prosimians (such as le-
murs), that would suggest pushing back the date for the emergence
of memory awareness even farther.
A related question about the emergence of self-reflective cognition
is ecological. What selection pressures promote the evolution of self-
reflective cognition (e.g., Povinelli & Cant, 1995)? One approach to
this problem is to think of self-reflective cognition as a set of compo-
nent functions, such as memory awareness, body awareness, emo-
tional awareness, and so on, rather than as a unified capacity. Because
an animal demonstrates self-reflective cognition in one domain does
not necessarily mean that it possesses it in other domains. Types of
information processing that are particularly crucial to a given species,
or which have to be carried out in a way only possible with self-
reflective processing, may be selectively subject to awareness (see
Kamil, 1988, for a discussion of this view on the evolution of cogni-
tion). In this scenario, self-reflective cognition might be widespread
among species, but restricted to particular ecologically relevant infor-
mation processing domains in each species. Alternatively, a domain-
general form of self-reflective cognition may have arisen more than
once, independently, for example, in primates and cetaceans (Reiss &
Marino, 2001). Comparisons between these groups could identify life
history variables that are common to them but are not shared by
groups lacking self-reflective cognition.
Critique of Memory Awareness in Monkeys

There are both experimental and conceptual issues regarding the va-
lidity of the concept of memory awareness in nonhuman animals. The
experimental difficulties consist of the possibility that the experimen-
tal procedures used are not adequate to discriminate memory aware-
ness per se from other explanations of the behavior observed in these
and other experiments. These threats are addressed using experimen-
tal manipulations, as described in the preceding sections. New control

procedures will likely be required as new critiques of these results
emerge. In contrast, conceptual criticisms are much more resistant to
solution by technical innovation. These have to do with the definition
of memory awareness and whether we can agree on a definition that
allows for the inclusion of nonverbal animals. I have adhered to a
functional definition of memory awareness in this chapter and have
argued that a functional account can capture the essential features of
memory awareness (see Dennett, 1992, for related views). For others,
a functional account of self-reflective cognition, one that strictly es-
chews phenomenology in the sense of private experience, will be un-
satisfying (Chalmers, 1996; see Zeman, 2001, for review of these is-
sues).
Although we are sure to learn more about what nonhuman species
know, it is unlikely that procedures can be developed to directly mea-
sure the experience of memory in a nonhuman animal. It will there-
fore always remain a matter of conjecture whether monkeys experi-
ence conscious states of knowing like our own. No amount of functional
convergence between rhesus monkey and human will convince the
hardened skeptic of similar experience in the two groups of animals.
Such a skeptic holds that conscious experience, Descartes’s one cer-
tainty, is the essence of self-reflection. However, I argue that empha-
sizing experience over function relegates memory awareness to an
epiphenomenal domain and undermines its significance. Furthermore,
the phenomenological position may be irrelevant to the endeavor of
understanding the evolution of self-reflective cognition. For this, we
need to put aside introspection, treat self-reflection as a biological
mechanism, and focus on what self-reflection does for us and may
have done for our ancestors. The evolution of self-reflective cognition
cannot be understood by delving further into our own thoughts, or
by wondering what it is like to be a monkey. Rather we need to adopt
the performance-based view of natural selection, a vantage from
which cognition is seen through what it does, not through how it
is experienced. Paradoxically, by avoiding the allure of our private
experience, we might learn something new about ourselves.
ACKNOWLEDGMENTS The writing of this chapter was supported by the NIMH
Intramural Research Program. I thank Cindy Buckmaster and Kristen Hamp-
ton for helpful discussions and comments on a draft.
REFERENCES
Alvarez-Royo, P., Zola-Morgan, S., & Squire, L. R. (1992). Impairment of long-
term memory and sparing of short-term memory in monkeys with medial
temporal lobe lesions: A response to Ringo. Behavioural Brain Research, 52,
1–5.
Briggs, G. E., & Blaha, J. (1969). Memory retrieval and central comparison
times in information processing. Journal of Experimental Psychology, 79, 395–
402.
Call, J., & Carpenter, M. (2001). Do apes and children know what they have
seen? Animal Cognition, 4, 207–220.
Chalmers, D. J. (1996). The conscious mind. Oxford, UK: Oxford University
Press.
Clark, R. E., & Squire, L. R. (1998). Classical condition and brain systems: The
role of awareness. Science, 280, 77–81.
Clayton, N. S., & Dickinson, A. (1999). Scrub jays (Aphelocoma coerulescens)
remember the relative time of caching as well as the location and content
of their caches. Journal of Comparative Psychology, 113, 403–416.
Cohen, N. J., Eichenbaum, H., Deacedo, B. S., & Corkin, S. (1985). Different
memory systems underlying acquisition of procedural and declarative
knowledge. In D. S. Olton, E. Gamzu, & S. Corkin (Eds.), Memory dysfunc-
tion: An integration of animal and human research from preclinical and clinical
perspectives (pp. 54–71). New York: New York Academy of Sciences.
Dennett, D. C. (1992). Consciousness explained. Boston: Little, Brown.
Easton, A., Parker, A., & Gaffan, D. (2001). Crossed unilateral lesion of medial
forebrain bundle and either inferior temporal or frontal cortex impair ob-
ject recognition memory in rhesus monkeys. Behavioural Brain Research, 121,
1–10.
Flavell, J. H. (1979). Metacognition and cognitive monitoring: A new area of
cognitive-developmental inquiry. American Psychologist, 34, 906–911.
Gaffan, D., Parker, A., & Easton, A. (2001). Dense amnesia in the monkey
after transection of fornix, amygdala, and anterior temporal stem. Neuro-
psychologia, 39, 51–70.
Hamman, S. B., & Squire, L. R. (1997). Intact perceptual memory in the ab-
sence of conscious memory. Behavioral Neuroscience, 111, 850–854.
Inman, A., & Shettleworth, S. J. (1999). Detecting metamemory in nonverbal
subjects: A test with pigeons. Journal of Experimental Psychology: Animal Be-
havior Processes, 25, 389–395.
Jenkins, H. M., & Moore, B. R. (1973). The form of the autoshaped response
with food or water reinforcers. Journal of the Experimental Analysis of Behav-
ior, 20, 163–181.
Kamil, A. C. (1988). A synthetic approach to the study of animal intelligence.
In D. W. Leger (Ed.), Nebraska Symposium on Motivation: Vol. 35. Compara-
tive perspectives in modern psychology (pp. 257–308). Lincoln: University of
Nebraska Press.
Knowlton, B. J., Ramus, S., & Squire, L. R. (1992). Intact artificial grammar
learning in amnesia. Psychological Science, 3, 172–179.
Knowlton, B. J., & Squire, L. R. (1993). The learning of categories: Parallel
brain systems for item memory and category knowledge. Science, 262,
1747–1749.
Meck, W. H. (1983). Selective adjustment of the speed of internal clock and
memory processes. Journal of Experimental Psychology: Animal Behavior Pro-
cesses, 9, 171–201.

Mishkin, M., & Murray, E. A. (1994). Stimulus recognition. Current Opinion in
Neurobiology, 4, 200–206.
Murray, E. A. (1996). What have ablation studies told us about the neural
substrates of stimulus memory? Seminars in the Neurosciences, 8, 13–22.
Nelson, T. O. (1996). Consciousness and metacognition. American Psychologist,
51, 102–116.
Parker, A., Wilding, E., & Akerman, C. (1998). The von Restorff effect in visual
object recognition memory in humans and monkeys: The role of frontal/
perirhinal interaction. Journal of Cognitive Neuroscience, 10, 691–703.
Povinelli, D. J., & Cant, J. G. H. (1995). Arboreal clambering and the evolution
of self-conception. Quarterly Review of Biology, 70, 393–421.
Reiss, D., & Marino, L. (2001). Mirror self-recognition in the bottlenose dol-
phin: A case of cognitive convergence. Proceedings of the National Academy
of Sciences U.S.A., 98, 5937–5942.
Riley, D. A., & Langley, C. M. (1993). The logic of species comparisons. Psy-
chological Science, 4, 185–189.
Shettleworth, S. J. (1998). Cognition, evolution, and behavior. New York: Oxford
University Press.
Smith, J. D., Shields, W. E., Washburn, D. A., & Allendoerfer, K. R. (1998).
Squire, L. R., & Zola-Morgan, S. (1991). The medial temporal lobe memory
system. Science, 253, 1380–1386.
Terrace, H. S. (1984). Animal cognition. In H. L. Roitblat, T. G. Bever, & H. S.
Terrace (Eds.), Animal cognition (pp. 7–28). Hillsdale, NJ: Lawrence Erl-
baum.
sity Press.
Tulving, E. (1985). Memory and consciousness. Canadian Psychology, 26, 1–12.
Tulving, E. (1993). What is episodic memory? Current Directions in Psychologi-
cal Science, 2, 67–70.
Tulving, E., & Schacter, D. L. (1990). Priming and human memory systems.
Science, 247, 301–306.
Van Zandt, T., & Townsend, J. T. (1993). Self-terminating versus exhaustive
processes in rapid visual and memory search: An evaluative review. Per-
ception and Psychophysics, 53, 563–580.
Verfaelli, M., & Cermak, L. S. (1999). Perceptual fluency as a cue for recogni-
tion judgments in amnesia. Neuropsychology, 13, 198–205.
Wagner, A. D., & Gabrieli, J. D. E. (1998). On the relationship between recog-
nition familiarity and perceptual fluency: Evidence for distinct mnemonic
processes. Acta Psychologica, 98, 211–230.
Weiskrantz, L. (2001). Commentary responses and conscious awareness in hu-
mans: The implications for awareness in non-human animals. Animal Wel-
fare, 10, S41–S46.
Zeman, A. (2001). Consciousness. Brain, 124, 1263–1289.
12
Metaconfidence Judgments
in Rhesus Macaques:
Explicit Versus Implicit Mechanisms
Lisa K. Son & Nate Kornell
To date, psychological research on human metacognition centers on

whether and to what extent people know what they know. For exam-
ple, feeling-of-knowing judgments ask people how certain they are
that they know an answer that they are unable to retrieve. Judgments
of learning ask how certain they are that they will be able to remem-
ber a recently learned item in the future. Confidence judgments ask
people how certain they are that their response to a question is correct.
In this chapter, we focus on confidence judgments, which in humans
are typically made verbally. Our chapter describes an experiment on
confidence judgments in rhesus macaque monkeys. Because of their
obvious lack of verbal ability, the monkeys were given the opportu-
nity to express their confidence by placing bets on the accuracy of
their responses in a cognitive task.
BACKGROUND ON HUMAN METACOGNITION

Until recently, the experimental study of metacognition had generally
been limited to the human species. We begin by summarizing the
major theories of human metacognition before discussing their exten-
sion to nonhuman species. The term “metacognition” has been associ-
ated with uniquely human qualities such as introspection, self-reflec-
tion, frontal lobe function, theory of mind, and even consciousness
(Flavell, 2000; James, 1890; Janowsky, Shimamura, & Squire, 1989; Shi-
mamura & Squire, 1986; Tulving, 1994; Tulving & Madigan, 1990).
For example, Tulving (1994) says that metacognitive researchers use
“behavioristically safe expressions, such as memory ‘monitoring,’
mnemonic ‘behavior,’ memory ‘search,’ tip-of-the-tongue ‘states,’ and
296
Metaconfidence Judgments in Rhesus Macaques 297
feeling of knowing ‘experience’ . . . possibly to avoid the big bad ‘C’

word” (p. ix), “C” being consciousness. Metcalfe and Kober (chapter
2, this volume) say that metacognition is associated with having a self-
reflective “inner eye” that can look at other cognitive functions and
content. Thus, the ability to self-reflect, or introspect on one’s internal
mental representations, has been considered a foundation for human
consciousness (see Metcalfe & Shimamura, 1994, preface).
Metacognitive processes, which have been more formally specified
by Nelson and Narens (1990, 1994), consist of a basic structure con-
taining two interrelated levels, an object-level and a meta-level. The
two levels are shown in figure 12.1A. The object-level may include an
Meta-level
Monitoring Flow of
Control
Information
Object-level
Bet Response Meta-level
Monitoring Flow of
Control
Information
Line Response Object-level
Figure 12.1 A. An overview of the two separate, but interact-

ing, processes during learning as illustrated by Nelson and
Narens (1990, 1994). During metacognitive monitoring, the
meta-level is informed by the object-level of the present
state, and, in turn, during metacognitive control, the meta-
level modifies the object-level. B. The current investigation:
A meta-level confidence judgment, as measured by risk re-
sponse, assessing an object-level line task.
individual’s memories, cognitions, and behaviors, and describes the

state of the present situation. The meta-level monitors (and sometimes
can control) the object-level. The interaction between the two levels
has generally been thought of as being one of self-reflection. Nelson
and Narens (1994) described people as “systems containing self-reflec-
tive mechanisms for evaluating (and re-evaluating) their progress and
for changing their ongoing processing” (p. 7). The results of process-
ing at the meta-level are judgments and feelings, such as confidence
judgments and other expressions of certainty.
One important line of research within the area of metacognition is
the issue of how the metacognitive judgments, or meta-judgments, are
made. Two categories of mechanisms have been proposed to underlie
meta-level judgments in the human literature: direct-access mecha-
nisms and inferential mechanisms (see Nelson, Gerler, & Narens,
1984; Schwartz, 1994). Direct-access mechanisms rely on the accessibil-
ity of the internal memory trace. An example is the metacognitive
judgment of being in a tip-of-the-tongue state (TOT)—in a TOT, peo-
ple can often retrieve part of the target memory (e.g., only the first
letter), causing a state of frustration. But it is the state of being “close”
to accessing the internal memory that causes the high feeling of know-
ing in a TOT, and the more features of the memory one is able to
retrieve, the stronger the meta-judgment of certainty.
Some of the earliest studies of metacognition were interpreted as
supporting the direct-access view. For example, in the first investiga-
tion of feeling-of-knowing judgments (FKJs), Hart (1965) asked partic-
ipants trivia questions, and then asked for FKJs on those items the
participants could not answer. Then a final recognition test was given.
Results showed that people’s judgments were positively correlated
with recognition on the final test—items given high judgments were
more likely to be recognized than items given low judgments. The
results indicated that people’s judgments were probably based on an
accurate assessment of what was in memory. Underwood (1966) con-
ducted a similar study, presenting participants with three-letter tri-
grams. The trigrams varied from common three-letter words to diffi-
cult consonant syllables. Participants had to judge the difficulty of
learning the items, after which they were all given a recall test. Results
showed that individuals predicted their own recall with high accu-
racy. Also, Arbuckle and Cuddy (1969) presented participants with
lists of paired associates, and asked them to predict whether they
would be able to recall each pair. After giving the judgment, the par-
ticipants were given a memory test. Results showed that the predic-
tions were accurate—those paired associates given lower ratings were
recalled less well than those given higher ratings. The main conclu-
sion from these early studies was that people’s meta-judgments were
based on the amount of information they had been able to retrieve, or
directly access, from an internal memory representation at the time of

judgment.
Unlike the direct-access view, the inferential view posits that meta-
judgments are based on cues that are external, rather than internal, to
the memory trace. For example, a judgment may be made based on
how familiar the topic of the question is (e.g., giving a high judgment
to the question, “Who is the tallest basketball player in the NBA?”
because basketball is a familiar topic). Or a judgment may be based
on how quickly the answer comes to mind (e.g., giving a higher judg-
ment following a quickly retrieved answer than a slowly retrieved
answer). Both of these can predict the accuracy of a memory, but nei-
ther relies on directly accessing information about the internal repre-
sentation.
Schwartz and Metcalfe (1992) compared the direct-access and infer-
ential accounts in an experiment by comparing the relative contribu-
tion of the memory strength of the cue and target in making FKJs.
They presented participants with paired associates and asked for FKJs
in presence of only the cue. Prior to presenting the list, they also ma-
nipulated the accessibility of some of the cues by preexposing them
in a pleasantness-rating task, and they did the same for a different set
of target items. Their results showed that priming cues influenced
FKJs, but priming targets did not. They concluded that the strength
of the memory trace did not influence meta-judgments, as a direct-
access account would predict; instead the judgment was inferred from
the familiarity of the cue (see also Glenberg, Sanocki, Epstein, & Mor-
ris, 1987; Metcalfe, 1993a, 1993b; Metcalfe, Schwartz, & Joaquim, 1993;
Miner & Reder, 1994; Reder, 1987; Reder & Ritter, 1992).
Benjamin, Bjork, and Schwartz (1998) found further evidence for
inferentially based meta-judgments. They had participants answer
general information questions, such as “What is the color of topaz?”
Participants were also told that the time it took them to answer each
question was of primary interest, and to press the Enter key as soon as
they knew the answer to the question. After answering each question,
participants gave a judgment predicting how well they would be able
to remember their answer on a later test on a scale from 0 (no chance
of later recall) to 100 (certain later recall). The data showed what was
called a retrieval fluency effect—participants’ judgments were nega-
tively correlated with response times for answering the questions. The
researchers concluded that the speed or ease with which one is able
to come up with an answer seems to be a source of information for
making meta-judgments—even though, in this case, it led to very in-
accurate judgments. Furthermore, the judgments could not have been
based on the amount of information directly accessed from the mem-
ory trace, because all of the answers were fully retrieved. Based on
these studies, then, there is ample evidence that meta-judgments can
be based on inferential or external cues—cues other than those based

on an assessment of an internal representation.
The results of these studies point out that meta-judgments need not
be based on introspection of the internal memory trace. Instead, a host
of inferential or external cues may be the basis of our judgments. An
interesting question, then, might be to ask whether awareness of such
cues is necessary in order to make the judgment—some think not
(see Cary & Reder, 2002, an article titled, “Metacognition . . . Giving
Consciousness Too Much Credit”). Below, we present data that sug-
gest that meta-judgments need not be based on decisions that one is
aware of, but instead may be based on implicit mechanisms.
IMPLICIT METACOGNITION
Outside of psychology laboratories, people are not asked to verbally
provide meta-judgments very often. However, we believe that they
may, nevertheless, be made frequently and, more important, without
our awareness. To illustrate, most people, if asked what Lyndon John-
son had for dinner on February 23, 1958, would probably feel uncer-
tain and say, “I don’t know.” When asked their own name, they would
feel very certain and would state their name immediately. It seems
rather absurd that meta-judgments such as “I’m quite sure that I know
my name” would or should reach awareness. Yet, merely responding
“I don’t know” might contain a tacit meta-judgment of uncertainty.
Here is another example: On the game show Jeopardy, a successful
contestant only rings her buzzer when she feels fairly certain that her
answer is correct (or in Jeopardy terms, she feels fairly sure that her
question is correct). Her decisions to press the buzzer must be made
very quickly, perhaps even more quickly than the time needed to be-
come aware of her decisions. Still, each of those decisions may consist
of several different tacit meta-judgments.
In several experiments using a game-show paradigm similar to
Jeopardy, Reder and colleagues (e.g., Reder, 1996) asked participants a
series of trivia questions. They were told to imagine that they were
competing against another contestant, and to say as quickly as possi-
ble whether they knew the answer to the question. Only then would
they be able to answer the question to earn points. Results showed
that these meta-judgments could be made more quickly than partici-
pants could retrieve the answers. Furthermore, the judgments were
usually accurate in predicting subsequent accuracy of the answer. Re-
der (1996) also found that priming words in the question led to in-
creased subjective estimates of knowing the answer, despite the fact
that this exposure did not improve actual rates of producing the cor-
rect answer. Reder and her colleagues concluded that some meta-
judgments operate at an implicit level (Reder, 1996; Reder & Schunn,
1996). We believe that certainty and uncertainty are feelings that can
occur to us even when we cannot explain them and can result from
implicit cues that are always present. For the most part, it is only
when we are asked about our level of certainty that we become aware
of that certainty. As we will see later in the chapter, the implications
of animals making judgments of uncertainty depend very much on
whether one views the judgments as being an implicit, nonverbal pro-
cess or a conscious experience.
In this chapter, we relate explicit mechanisms with that which per-
tains to an assessment of an internal representation, mainly because
this requires that one directly access retrieved pieces of information
from the memory trace. The notion that meta-judgments are based on
explicit mechanisms fits nicely with the original, more philosophical
definition of metacognition of labeling humans as self-reflective ma-
chines. On the other hand, we relate implicit mechanisms with meta-
judgments made on the basis of external cues that are currently present,
mainly because these judgments may be made without any internal
monitor. These judgments transform continuously as a result of the
constantly changing external cues and transient events that people
are typically unaware of. Based on the data in the human literature
summarized above, we would argue that people use both explicit and
implicit metacognitive processes. The question of whether any meta-
cognitive processes exist in nonhuman species remains.
ARE METACOGNITIONS SPECIAL TO HUMANS?

As mentioned earlier, metacognition is considered a very high-level
mental function unique to humans. “Machines without consciousness,
and animals whose consciousness is different from that of human be-
ings, could not perform many of the tasks that human subjects in
metacognitive experiments, and others of the same general kind, can
and do perform” (Tulving, 1994, p. ix). Tulving and Madigan (1970)
state that metacognition is “one of the truly unique characteristics of
human memory: knowing about knowing” (p. 477). Metcalfe and Shi-
mamura (1994) begin their book by saying, “The ability to reflect upon
our thoughts and behaviors is taken, by some, to be at the core of
what makes us distinctively human” (p. xi).
One obvious difference between humans and animals is the ability
to speak. A question, then, is whether it is the ability to verbalize
thoughts and feelings that grants humans the ability of self-reflection.
And if one did not—or more pertinently, could not—verbally express
one’s feelings, does that then rule out the possibility that meta-judg-
ments can occur? We think not. Thoughts or feelings, after all, are
not synonymous with verbal speech. Judgments that are based on the
familiarity of a question or cue, for instance, do not seem too cogni-
tively advanced for an animal. After all, animals can easily discrimi-
nate more familiar stimuli from less familiar ones. In addition, anyone
who has seen a dog hesitate before jumping into a high truck bed, or
a monkey waver between two choices in a psychological task, is likely
to imagine that the animal is experiencing some sort of uncertainty. Of
course, there is a difference between behaving in a way that appears
uncertain and being able to report feelings of uncertainty about cogni-
tions. The latter is analogous to answering the question, “Can you
give me a judgment about the certainty of your response?” and only
it qualifies as metacognitive.
A few empirical studies have shown that monkeys and dolphins
are able to make uncertainty responses (Hampton, 2001; Shields,
Smith, & Washburn, 1997; Smith et al., 1995; Smith, Shields, Allen-
doerfer, & Washburn, 1998; Smith, Shields, Schull, & Washburn, 1997;
also see chapters 10 and 11, this volume). In an early study of uncer-
tainty in animals conducted by Smith et al. (1995), human and dolphin
subjects were asked to respond high when a 2100-Hz tone was played,
and low for tones less than 2100 Hz. Then they were tested with tones
ranging from 1200 Hz to 2100 Hz, with the correct response being low
for any tone but the 2100 Hz tone. However, in addition to the high
and low choices, a third choice, escape, was offered simultaneously,
which, when pressed, returned a guaranteed reward, which was
smaller than the reward for a correct high or low answer. The results
showed that both humans and dolphins responded high for 2100-Hz
tones and low for low tones up to about 2080 Hz. Most interestingly,
they chose escape on trials with tones in between. Presumably these
were the most difficult trials, where the subjects would have felt the
most uncertain about their answers. Follow-up studies showed similar
evidence of uncertainty in monkeys.
No one would question that human participants chose to escape
difficult trials because they felt uncertain. But is this conviction based
on the fact that people can verbalize the feelings by saying, “I chose
to escape because I was uncertain”? Or is the behavior (pressing the
escape response) enough for us to assume feelings of uncertainty in
human participants? If so, then we could grant those same abilities
to animals that behave similarly. However, when animals do behave
similarly, the immediate reaction often is to ask whether the results
might be based on external cues, or whether the learning was based
on simple stimulus-response associations, or whether these behaviors
did not require any internal monitoring at all. For example, could the
animals have simply learned that to maximize reward, high tones
should be followed by high, low tones should be followed by low, and
middle-level tones should be followed by the escape response? In this
case, no meta-level process is necessary; object-level processes suffice.
Hampton’s procedure (see Hampton, 2001; chapter 11, this volume)

investigated prospective judgments in monkeys by using a modified
escape procedure. He presented two monkeys with a picture. Follow-
ing a variable delay, the monkeys were given a forced recognition test
in which they had to identify the picture that they had been shown
among three distractors. However, on some of the trials, rather than
being forced to take the test, the monkeys were free to either choose
to take the test (for a big reward only if correct) or escape the test (for
a guaranteed smaller reward). If the monkey chose to escape, the trial
ended without the test being taken. This decision was made prospec-
tively, before the test was presented. His results showed that both
monkeys performed better on tests that they freely chose to take than
on tests that they were forced to take, suggesting that the monkeys
chose to take the test when they were certain. The improvement in
this procedure over the original escape procedure is that the test was
not present simultaneously with the escape choice. However, a few
issues remain. For example, when the monkey chooses to escape the
trial, the trial ends, and no object-level response is made. Thus, on a
given trial, there is no way of knowing the monkey’s actual perfor-
mance on the test and his certainty of taking the test. In the rest of this
chapter, we present findings using a new paradigm in which meta-
judgments can be measured separately from object-level responses in
nonhuman animals.
META-CONFIDENCE JUDGMENTS IN ANIMALS

Is there a way to measure meta-level judgments, as in Nelson and
Narens’ (1990) framework in animals? We decided to investigate this
question using confidence judgments made about previous responses.
Using confidence judgments seemed like the obvious next step, partic-
ularly since others, mentioned above, had already shown that animals
could report uncertainty by using the escape response efficiently. In
our task, two responses were required, representing the two levels in
the Nelson and Narens (1990, 1994) framework. If an analogous task
were conducted on a person, the response at the object-level would
be to answer a question such as, “Who painted the Sistine Chapel?”
Then, after giving the response, he or she would make the meta-level
judgment about the correctness of the answer on a confidence rating
scale. Several studies have shown that humans can make accurate
confidence judgments (Koriat, Lichtenstein, & Fischhoff, 1980; Perfect
& Hollins, 1996; Shaughnessy, 1979). A schematic of our task in the
framework of Nelson and Narens is provided in figure 12.1B.
Our experiment was designed to test, in rhesus macaque monkeys,
the existence of true meta-judgments (explicit or implicit) about their
previous cognitive responses. The monkeys were first asked to touch

the longest of nine lines (analogous to a trivia question for a human).
After making their response, they were asked for their judgment of
certainty of their response by making a high or low bet. Unlike mak-
ing an escape response, where there is only one response, the diffi-
culty of our task was that the monkey had to remember the object-
level response he had just made, but he would also have to connect
that response to a subsequent meta-judgment. Furthermore, unlike the
escape procedures, the benefit of asking for both object-level and
meta-level responses is that we could tell on a given trial both whether
the monkey knew the answer and how sure he was. Of course, a mon-
key confidence judgment could not be a simple rating of 0 to 100 as
it would be for humans. Instead, the betting paradigm was used, in
which the monkey could wager a lot on a response (in terms of re-
ward), or play it safe and risk only a little.
EXPERIMENT 1: THE BETTING PARADIGM

Two male rhesus macaques (Macaca mulatta), named Ebbinghaus and
Lashley, were tested in a chamber containing a touch-sensitive com-
puter screen. On each trial, they were presented with nine lines posi-
tioned vertically on the screen, in a 3 × 3 array. Their task was a
psychophysical task, in which they had to press the longest of the
nine lines. Trials were either easy or difficult: On easy trials, one of
the nine lines was noticeably longer than the other eight, which were
all the same length. On difficult trials, all nine lines were the same
length, but one line was arbitrarily designated as correct, so the mon-
keys were correct on one out of nine trials by chance. For each touch,
a green border appeared around the item briefly. Then the screen
cleared and the monkeys were asked to make a bet by touching one
of two icons: a high-risk icon and a low-risk icon. Figure 12.2 displays
the four possible outcomes that could occur for the two risk responses.
If high risk was selected, the reinforcement was the gain of two tokens
given a correct response on the line task, or the loss of two tokens
given an incorrect response (the tokens were later exchanged for pel-
lets). If the low-risk icon was touched, the reinforcement was always
a gain of one token regardless of accuracy on the line task. The token
reservoir was located on the bottom right-hand side of the screen.
Tokens appeared to fly out of or into the reservoir with noticeable
noises. Once the number of tokens in the reservoir reached or ex-
ceeded 12, the monkeys received two real banana-flavored food pel-
lets and the token reservoir reset to 9 tokens. Using the tokens allowed
us to calibrate the amounts of reward for each level of risk, so as to
minimize bias toward one of the risk options. Toward the same end,
when a monkey developed a bias toward one of the risk choices, that
Correct Incorrect
"High Risk"
"Low Risk"
Figure 12.2 Contingency table showing the four possible outcomes of correct or
incorrect response to the line task and high- and low-risk bets, starting with two
tokens already present in the reservoir. When high risk was pressed, two tokens
fell into the reservoir after correct responses, but flew out of the reservoir after
incorrect responses. When low risk was pressed, one token fell into the reservoir
regardless of whether the line response was correct or incorrect. Note that in
Experiment 2, choosing high risk resulted in a gain or loss of three, rather than
two, tokens. See color insert.
risk icon began to appear only after a delay, which ensured that the
monkeys chose both levels of risk with some frequency. A sample
trial is presented in figure 12.3A.
The hypothesis was that on easy line-discrimination trials, the mon-
keys would get the answer correct, presumably feel certain about it,
and consequently bet high risk. On difficult (impossible) trials, though,
they would (usually) get the answer incorrect, presumably feel uncer-
tain, and bet low risk. Again, an advantage of this paradigm was that
two responses were required: the line discrimination at the object-level,
and a judgment of that response—the bet—at the meta-level.
Training
Typically, it is advantageous to reward animals immediately follow-
ing a response. This keeps the animal motivated. Furthermore, imme-
Figure 12.3 A. A sample trial for Experiment 1. In the first phase, nine lines appeared
on the screen. A token reservoir appeared in the bottom right-hand corner of the screen,
set at nine at the beginning of each session. The monkey’s task was to press the longest
line. For each touch, a green border appeared around the pressed item. Then the lines
disappeared, and two risk icons appeared. The monkey’s task was to report his confi-
dence by betting either low or high risk. Once he made his bet, the token reservoir
changed accordingly. In this sample trial, the monkey has bet high risk, and was correct
on the line task. Thus, two tokens were added to the reservoir. B. Sample numerical
stimuli for Experiment 2. The task was to press the stimulus with the most (Lashley) or
the least (Ebbinghaus) items in them. Otherwise, the general methods remain the same
as those in Experiment 1. See color insert.
diate feedback is most easily associated to the one and only most re-
cent response that has been made. Thus, an animal should have little
trouble understanding what response is being rewarded. However,
this was not the case in the current paradigm. Rather than immediate
feedback, the monkeys learned that feedback occurred only after mak-
ing two separate responses. Never before had these monkeys been
asked to perform two separate tasks in a single trial. Essentially, after
the monkey made his first response, the screen went blank and a new
question appeared. The monkey’s task was not only to make a second
response but also to understand that the feedback he received not
only depended on that second response but reflected two past re-
sponses—the bet response in addition to the earlier line response. We
believe this was perhaps the most challenging aspect of the experi-
ment. In addition, we required that the monkey remember four differ-
ent contingency combinations and respond accordingly. Finally, the
monkey needed to learn that accumulating a certain number of tokens
on the computer screen resulted in a food reward after a criterion had
been reached. Thus, because of the novelty of the paradigm for these
participants, a substantial amount of training was involved prior to
the actual initiation of the experiment. This included training of three
distinct aspects: the main task, the tokens (including reward and pun-
ishment), and the betting icons.
The monkeys were first trained on the psychophysical task alone. At
the start of training, Ebbinghaus and Lashley were presented with two
vertical lines of different lengths on the screen. Their task was to press
the longer of the two lines. A food pellet was earned only after correct
responses. Both participants learned the task very quickly, after only a
few sessions. Gradually, the number of lines presented for each ques-
tion was increased, to three, then four, until they were presented with
a total of nine vertical lines. All sessions were 20 minutes in length.
About a year after the beginning of training on the line task, a
token reservoir was added on the bottom right-hand side of the screen
during the entirety of each session. (During that year, we tried a simi-
lar betting procedure using food pellets as rewards and time-outs as
punishments. Specifically, a high-risk response resulted in either a big
food reward or a long time-out, whereas a low-risk response resulted
in a small food reward or a short time-out. However, using this para-
digm, both monkeys had a significant bias for pressing high risk. We
thought that this might have been because the aversion to the time-
outs was not nearly as salient as the attraction of the food rewards.
We attempted to fix the problem by tweaking the contingencies—in-
creasing the length of the high-risk time-out period—but there was
no evidence that the monkeys understood the connection between the
line task and the risk contingencies, and so we gave up.) Following
each correct response to the line task, a token flew into the reservoir.
After each incorrect response, the token reservoir remained empty. At

the start of token training, the criterion for receiving real food pellets
was set at one token, so that whenever the monkey responded cor-
rectly, a token would fly into the reservoir and then would disappear
at the bottom, simultaneously with the delivery of a food pellet. Grad-
ually, the criterion was increased. Thus, the monkeys learned that an
accumulation of these tokens was a positive feature that would even-
tually earn them food reward. At this time, the line trials were kept
at a fairly easy level, so that the monkeys would not get frustrated
and would continue to be motivated during each session. (We would
experience motivation difficulties several times over the course of the
research.) Then the number of starting tokens in the reservoir was
varied. For example, for some sessions, the reservoir started at a level
of three, with six as the criterion—meaning that in order to receive
food, the monkeys needed to get three trials correct. We had the start
number begin at a number greater than zero because the next step of
training would be to introduce punishment, or the loss of tokens. We
were also interested in having the monkeys connect the tokens to ac-
tual food pellets, which they did. In fact, throughout training, both
Ebbinghaus and Lashley spent some time licking the tokens on the
screen, and attempting to push the tokens down, as though they could
force the tokens to come out as real food pellets. In the meantime, the
trials were made gradually more difficult (the ratio of difficult to easy
trials increased).
The next major training step was to introduce the punishment—in
our paradigm, a loss of tokens—to the monkeys. We were not sure
how aversive such a punishment would be. (We only hoped that this
time, the punishment would be more effective than the time-out pun-
ishment that we had attempted to use prior to incorporating the token
economy.) Never had any food pellets, tokens, or other rewards been
“snatched away” from the monkeys prior to this experiment. Here,
when an incorrect response was made, a token (only if there were
already at least one in the reservoir) would fly up and out of the
reservoir, making a “negative” noise, and disappear. Of course, by
this time, the monkeys knew that fewer tokens in their reservoir
meant they were farther away from reaching criterion or winning a
food pellet. Thus, on the first session of punishment training, both
monkeys seemed almost stunned, staring at the reservoir on the
screen for a while and showing obvious frustration behaviors.
Lashley, after a couple minutes of frustration, soon settled in again,
performing some more trials and continuing on in training. Ebbing-
haus, on the other hand, stopped working, forcing us to put him on a
period of remedial training with easier trials without punishment for
a while. Slowly, Ebbinghaus got back on track and learned to work
consistently even with the punishment. Although we were happy to

know that the punishment using the token paradigm was adequately
aversive for the monkeys, one future interest of ours is to investigate
the differences in salience for rewards and punishments by manipu-
lating the contingencies.
Following punishment training, both monkeys were trained to re-
spond at a stable pace during 20-minute sessions, with a mixture of
both easy and difficult trials. For each session, the reservoir began
with 9 tokens, while the criterion was set at 12. Finally, the risk icons
were introduced. Both icons were presented on the screen after the
line response was made. One worry was that up until this point using
the token procedure, the monkeys had always received feedback im-
mediately after making their line response. This would be the first
time they would not receive immediate feedback. The risk icons that
we chose were two pictures, each symbolizing the number of tokens
that would be risked if pressed. For example, the high-risk icon was
a picture of a square with two tokens and the low-risk icon was a
picture of a square with one token. The contingencies, set up like a
logical gamble, were as follows: For betting high risk, they earned two
tokens after a correct line response and lost two tokens after an incor-
rect line response. For betting low risk, they earned one token after a
correct line response and lost one token after an incorrect line re-
sponse. The monkeys soon learned that each of the pictures resulted
in different contingencies. Whether they knew that it had anything to
do with their previous line response was not yet known. In fact, again,
both monkeys acquired an early bias for pressing high risk. This led
us to increase the number of difficult trials during the session. We
also changed the low-risk contingencies to always gaining a token
regardless of previous line accuracy (changed from the previous low-
risk contingencies of gaining a token when correct and losing a token
when incorrect). When we still had difficulties in encouraging the
monkeys to select the low-risk option more often, we changed the
low-risk icon to an arbitrary picture (see figure 12.3A, the middle pic-
ture of the row of five) because we were concerned that the original
low-risk icon had become unattractive. At the end of training, then,
which was approximately 7 months after the tokens had first been
introduced, the betting procedure was one in which a high bet meant
risking two tokens, while a low bet was gaining one token regardless
of previous line accuracy. Furthermore, due to the high-risk bias, a
time delay was added prior to the appearance of one—the more bi-
ased one—of the risk icons (i.e., the more biased they were toward
high risk, the longer they had to wait before the high-risk icon ap-
peared). Starting from this point, we were able to obtain 41 subse-
quent sessions of data for both Ebbingaus and Lashley.
Results
The data were analyzed over a total of 41 sessions for each monkey.
The average number of trials completed in each session was 87 and
60 for Ebbinghaus and Lashley, respectively (each session lasted 20
minutes). The mean percentage of easy trials answered correctly was
94% for Ebbinghaus and 88% for Lashley, and the mean percentage
of difficult trials answered correctly was 12% for both monkeys (11.1%
is chance with nine items to choose from). This is slightly higher than
chance, but the difference was not significant for either monkey.
For each session, a phi correlation was computed between accuracy
(correct vs. incorrect) and bet (high risk vs. low risk). We expected the
monkeys to bet high risk on correct trials and low risk on incorrect
trials, which would result in positive correlations. The correlations,
blocked in sessions of four, are presented in figure 12.4. As shown,
the correlations were generally positive.
In general, although Ebbinghaus took a little longer to achieve high
correlations, both monkeys were able to report stable positive correla-
tions by the end of training. For all 41 sessions, the mean correlations
0.7
Mean Phi Correlation Between Accuracy on
Ebbinghaus
0.6 Lashley
Line Task and Risk Choice
0.5
0.4
0.3
0.2
0.1
0
1 2 3 4 5 6 7 8 9 10
Blocked Sessions of 4
Figure 12.4 The mean phi correlations between accuracy of the object-level line task
and the meta-level risk choice for the final 41 sessions of training, in blocks of four.
A positive correlation indicates that on trials that were correct, the monkeys more
often bet high risk, whereas on trials that were incorrect, they more often bet low
risk.
were significantly above zero for both Ebbinghaus, M = 0.33, t(40) =

8.43, p < .0001, and Lashley, M = 0.37, t(40) = 13.37, p < .0001. We also
calculated the percentage of correct and incorrect trials on which the
monkeys chose high risk. As figure 12.5A indicates, both monkeys
chose high risk significantly more often after correct trials than after
incorrect trials—for Ebbinghaus, t(40) = 8.44, p < .0001; for Lashley,
t(40) = 13.31, p < .0001.
We also calculated reaction times (RT) for the difficult and easy
trials (for all RT analyses, we excluded trials over 10 seconds long).
The results showed that both monkeys took longer to respond to the
line-discrimination task on difficult trials than on easy trials. Ebbing-
haus took an average of 1.24 seconds to respond on difficult trials, but
only 1.07 seconds on easy trials, which significantly differed from each
other, t(40) = 3.19, p < .01. Lashley also took significantly longer to
respond on difficult trials, averaging 2.00 seconds as compared to 1.66
seconds on easy trials, t(40) = 2.55, p < .05. In general, a longer re-
sponse time could be interpreted as resulting from more uncertainty.
In summary, using a betting paradigm that included both an object-
level task and a meta-level task, it appeared that both Ebbinghaus and
Lashley were able make accurate confidence judgments. When they
responded correctly on the object-level task, they were able to report
that they were certain by tending to bet high risk. Similarly, after re-
sponding incorrectly, they bet low risk more often. If they were hu-
man, this chapter could now conclude without much controversy by
saying that “Ebbinghaus and Lashley were able to express feelings of
certainty and uncertainty about their cognitions, thus showing meta-
cognitive abilities.”
Discussion
However, the reader may be curious, as we were, to know whether
these responses were based on judgments resulting from direct or ex-
plicit access of internal representations, or from inferential mecha-
nisms. Regardless of what the basis was, both of these mechanisms,
according to the human literature, would indicate that the monkeys
possess metacognitive abilities. Given this, several questions arise. To
begin with, perhaps the monkeys relied on cues that did not require
any feelings of uncertainty. For example, the monkeys could have
been relying on their own response times to guide their bets. This
seems plausible, since easy and difficult trials resulted in a difference
in response latencies—they might have simply bet low risk when they
observed that they had taken a long time to make their response dur-
ing the line task, and high risk when they had been fast. Alternatively,
they could have relied on pattern recognition to make their bets. For
example, they might have responded low risk when they recognized
A
Figure 12.5 Mean percentage of correct and incorrect trials on which

high risk was chosen in (A) Experiment 1 and (B) Experiment 2. These
means were calculated for each monkey over the final 41 sessions for
the line task in Experiment 1 and over the first 4 sessions for the numero-
sity task in Experiment 2.
that all of the lines were the same length, without actually feeling
uncertain. If either of these was the case, then one hesitates to say that
the monkeys were responding based on explicit mechanisms.
It is interesting to note that these two possibilities are quite similar
to two inferential mechanisms in relation to human metacognition:
retrieval fluency and cue familiarity. As described at the beginning of
this chapter, in retrieval fluency, the faster one arrives at an answer,
the higher the judgment will be. The cue familiarity hypothesis states
that meta-judgments are based on how recognizable the cue is (Met-
calfe, 1993a). For example, a person may give a high judgment for the
question “What was the name of Batman’s butler?” because of high
familiarity in the subject of Batman and not because of any actual
memory of the answer. Cue familiarity as a metacognitive cue has
been supported by many experimental results (Glenberg et al., 1987;
Metcalfe, 1993a, 1993b; Metcalfe et al., 1993; Miner & Reder, 1994;
Reder, 1987; Reder & Ritter, 1992; Schwartz & Metcalfe, 1992). In both
retrieval fluency and cue familiarity, metacognition is inferred from
factors other than the explicit assessment of memory itself. Still, in the
above-mentioned literature, when these judgments were found to be
based on such inferential mechanisms, they were classified as meta-
cognitive judgments. Based on this, if Ebbinghaus and Lashley were
using cue-based pattern recognition or their reaction time as a basis
for their uncertainty judgments, one reaction is to say that they still
do possess metacognitive abilities. Furthermore, they would be a lot
like humans if they did either of these.
To investigate these possibilities, we first explored the question of
whether retrieval fluency predicted risk choice for our monkeys. To
do so, we calculated mean point-biserial correlations between re-
sponse time to the line-discrimination task and the bet choice. If the
monkey bet low risk when he was slow on the line task and bet high
risk when he was fast, the mean correlation across the 41 sessions
would be negative, and it was for both Ebbinghaus, M = .12, t(40) =
4.43, p < .0001, and Lashley, M = .17, t(40) = 5.64, p < .0001. Based on
this finding, it seems that response latencies could have helped guide
the monkeys confidence judgments. This opens up the possibility that
the monkeys behavior can be explained without positing that they
made judgments based on explicitly assessing internal representa-
tions. However, we had not yet investigated whether there was any
evidence that the monkeys might have, in fact, been making their risk
judgments based on something internal.
To explore the question of whether the monkeys were monitoring
an internal representation, we reanalyzed the data, factoring out the
effect of both trial difficulty and RT. First, we analyzed only the easy
trials, which were all equally difficult, meaning the monkeys could
not use difficulty as a cue, which even for a human makes meta-judg-
ments much more difficult. We then computed Pearson correlations

(although the Pearson correlations are standard for continuous vari-
ables, we used them here because there was no standard way of com-
puting partial correlations between binary and continuous data) be-
tween accuracy of the line task and risk choice, partialing out the
effect of RT, separately for each session. A positive correlation would
indicate that the judgments were attributable to mechanisms other
than those based on trial difficulty or RT. The mean correlations were
significantly positive for both Ebbinghaus, M = 0.07, t(35) = 2.67, p <
.05, and Lashley, M = 0.20, t(29) = 4.24, p <; .001, although more so for
Lashley than for Ebbinghaus. These results reveal that both monkeys
might be making meta-judgments based on something more than im-
plicit cues such as assessments of internal representations.
Another way to investigate whether the monkeys were not merely
basing their confidence judgments on external pattern recognition, re-
lying on features specific only to the line task, was to present the
monkeys with a novel question, a transfer task. Transfer tasks are
widely used, typically to ensure that the tested skill—in this case, the
reporting of accurate confidence judgments through bets—has not
been learned as an automated response to the external stimulus that
was presented, and can be generalized to other tasks. Thus, in Experi-
ment 2, we presented an experiment similar to Experiment 1, except
that the line task was replaced with a new task. First the monkeys
were trained on the new task until performance was stable, and then
the risk choices were added. If they could make appropriate risk
choices immediately, we reasoned that their ability to make confi-
dence judgments had generalized and was not based on features spe-
cific to the line task in Experiment 1.
EXPERIMENT 2: TRANSFER
The purpose of Experiment 2 was to test whether the ability to make
accurate meta-level confidence judgments would generalize to a num-
erosity task in which the monkeys had to differentially choose one
correct picture out of nine presented on the screen (see figure 12.3B
for a sample stimulus). Both Ebbinghaus and Lashley had previous
experience with a numerical task in which pictures containing varying
numbers of items (e.g., a yellow square containing 12 black dots) were
presented simultaneously. Ebbinghaus had learned to press the pic-
tures in an ascending order, and Lashley had learned to press the
pictures in a descending order. We modified our task so that the mon-
keys were presented with nine pictures, each containing some number
of items. The monkeys had to press only the one picture that con-
tained either the most items (in Lashley’s case) or the fewest items (in
Ebbinghaus’s case). For both monkeys, each of the eight incorrect
choices contained the same number of dots. As in Experiment 1, trials

varied in difficulty, but instead of only two levels of difficulty, there
were four levels (none of them being impossible as one was in Experi-
ment 1): On the very easy trials, one of the nine pictures consisted of
noticeably more (or fewer) items than the other eight choices. As the
trials became increasingly difficult, the correct answer became less
and less distinguishable from the distractors. All sessions were 20
minutes in length. Once the monkeys had been trained on this para-
digm (which took three sessions for Ebbinghaus and seven sessions
for Lashley), the risk choice was added. After the numerical response
was made, the computer screen cleared and the monkeys were asked
to make their bet. The reinforcement contingencies were slightly mod-
ified from Experiment 1, in which the monkeys gained or lost two
tokens when they bet high risk. Here, in order to increase the number
of pellets the monkeys received, the contingency was changed to a
gain or loss of three tokens following high-risk choices. Our hope was
that, starting with the very first sessions, both Ebbinghaus and
Lashley would behave metacognitively, by risking more when they
were correct than when they were incorrect. These results would sug-
gest that the monkeys had understood the meaning of the bet choices,
rather than choosing them in response to patterns of the line task.
Results and Discussion

In Experiment 1, we were interested in investigating whether the
monkeys acquired a metacognitive skill through training on the bet-
ting paradigm. Here, though, we were not interested in whether the
monkeys could acquire the metacognitive skill. Rather, our interest
lay in the very beginning of training, to examine whether immediate
transfer of the betting skill occurred for a new task. Thus, in Experi-
ment 2, we analyzed the first four sessions of training. To analyze
the relationship between accuracy and risk, we again calculated phi
coefficients, which allowed us to compute Fisher’s exact p-values for
a small number of observations. Both monkeys showed positive corre-
lations almost immediately. For Lashley’s first four sessions com-
bined, ϕ = 0.24, p < .0001. For Ebbinghaus’s first four sessions com-
bined, ϕ = 0.14, p < .05. We also calculated the percentage of trials on
which high risk was chosen separately for correct and incorrect trials,
again combining the first four sessions, as shown in figure 12.5B. As
in Experiment 1, high risk was chosen more often on correct trials
than on incorrect trials.
The transfer task showed that both monkeys began making good
risk decisions directly upon being introduced to their new task. They
did not need training to make confidence judgments on the new
task—unlike in the line/risk task, which took many months to learn.
Instead, they showed enough flexibility to generalize the risk response
to the new task. This makes sense if they were making their risk re-
sponses as meta-judgments. Such judgments about cognitions should
occur for virtually any cognition, and it appears that what the mon-
keys had learned was how to respond with accurate meta-cognitions.
We were aware of the fact that although the transfer results make
a stronger case for us, there are still similarities between the numerical
task and the line task, the most obvious being that both involve some
trials where all the stimuli look very similar, and other trials where
one stimulus is quite different than the others. Thus, the monkeys
could have succeeded by responding to the risk choices based solely
on whether the stimuli were the same or different, in both experi-
ments. We answered this criticism earlier by showing that risk choices
were appropriate even when only easy trials (which are all the same)
were included in the analysis. But it might be argued that on incor-
rectly answered easy trials, the monkeys simply did not notice the
correct answer, and therefore thought that all of the stimuli were the
same. In that case, even on easy trials, they would choose low risk and
get the answer incorrect without necessarily relying on uncertainty. In
short, one of the concerns of the present tasks is that they are both
perceptual tasks that may not have required internal representations.
So, although we think our findings are convincing, the best way to
answer criticisms about stimulus cues is to use a task where nothing
about the stimuli themselves makes it possible to make the risk choice
successfully. One way to address this concern would be by using the
betting paradigm with a memory task (e.g., Hampton, 2001; Smith et
al., 1998; chapter 11, this volume), not a perceptual task.
Tasks that rely on the monkeys’ memory make it difficult for a
meta-judgment to be made on the basis of external cues. For example,
Smith et al. (1998) used a serial probe recognition task, in which mon-
keys were shown four pictures in sequence and then a probe, and had
to indicate whether the probe had been in the preceding list. If the
probe had been presented in the list, the participants were to press
the probe. If the probe was new, they were to press a second option
that represented the idea of “new.” If they were unsure whether the
probe was old or new, they were to press a third option, the escape
response. In this paradigm, when the monkey is faced with the probe,
the only thing that indicates to him whether the trial is easy or diffi-
cult is the strength of his memory trace. The appearance of the stimuli
offers no hints. Combining the serial probe recognition task with the
current confidence-judgment task would have the advantage that on
each trial the monkeys would make a response before placing their
bet. As a result, we would know which trials were difficult for them.
Also, when using only the escape response, if the item is very familiar
or very unfamiliar, the monkeys will have a strong urge to respond
either old or new. On difficult trials, if they have no strong urge they
might simply respond randomly, or choose the escape response by

default, and as a result choose the escape response relatively fre-
quently. Confidence judgments do not allow such a strategy. Neither
does the match-to-sample task used by Hampton (2001; chapter 11,
this volume): By forcing the monkey to choose between high and low
risk and then perform the task, he ensured that his monkeys could
not choose low risk randomly or by default.
Memory tasks avoid certain criticisms, but there is a deeper reason
to favor them. Our monkeys’ estimations of size and number are cog-
nitions, and thus their performance fits the traditional definition of
metacognition. However, metacognitive research has been tradition-
ally associated with judgments and assessments about memory. Thus,
our next step is to test our monkeys’ ability to make confidence judg-
ments using a memory task, namely, one similar to the serial probe
recognition task.
CONCLUSION
The current definition of metacognition is associated with many dif-
ferent functions, some of which (e.g., theory of mind) go far beyond
making smart risk choices. For example, Nelson (1992) defines it as a
form of self-reflective consciousness, in which we are able to con-
sciously observe the workings of our own minds. This type of meta-
cognition is thought to be a high-level process that entails privileged
access into one’s own internal mind and may be impossible to observe
in nonverbal animals. However, another definition of metacognition
takes a different view—that many of the mechanisms involved in
making metacognitive judgments need not be open to conscious
awareness (Cary & Reder, 2002; Reder & Schunn, 1996). According to
this view, we make metacognitive judgments constantly and without
explicit knowledge of them. These implicit metacognitions can be
based on internal factors, but may also be based on ongoing external
factors that we are not aware of.
A link can be drawn between direct-access mechanisms and con-
scious metacognition, in that both involve inward reflection. Feelings
of uncertainty that are inferential and based on cue familiarity (for
example) do not necessarily require self-reflection any more than a
simple familiarity judgment does. Yet it is also clearly possible to feel
uncertain based on the familiarity of a cue and, at the same time, be
quite aware of that feeling. Likewise, it is also the case that an animal
(or human) might have a feeling of uncertainty based on a direct ac-
cess mechanism and not be aware of the process. In fact, some re-
searchers have claimed that people have little or no direct introspec-
tive access to mental processes such as those affecting judgments
(Koriat, 1993; Nisbett & Bellows, 1977; Nisbett & Wilson, 1977). For
example, Koriat (1993) provides evidence supporting an accessibility

mechanism of metacognition which states that, although people may
access internal representations and base their judgments on those rep-
resentations, they have no direct awareness of the accuracy of the in-
ternal representations. Instead, people’s meta-judgments are based on
both correct and incorrect pieces of information that were purportedly
retrieved from an internal memory trace.
Given these issues, although it may seem that there could be a cor-
respondence between direct access and consciously aware metacogni-
tion, it cannot be confirmed. If the process of accessing internal infor-
mation from memory is not necessarily open to conscious awareness,
then we see no reason why the result of the process has to be either.
Thus, neither explicit mechanisms nor implicit mechanisms are “more
metacognitive” than the other. Regardless of how humans, or our
monkeys, made their judgments, it does not take away from the fact
that the meta-judgments were made.
In this chapter, we have presented evidence that two rhesus ma-
caque monkeys were able to display metacognitive abilities by report-
ing high- and low-risk bets about their own cognitions. If metacogni-
tion is taken to mean self-reflective consciousness, then our conclusion
is, of course, endlessly arguable. However, we believe that the mon-
keys were making the type of metacognitive judgments that occur in
people all the time, with or without awareness. We readily concede
that because they cannot verbally express their judgments, we do not
have evidence of awareness of their judgments. If we did, we would
then have evidence of meta-metacognition (which people express
when they verbalize their metacognitions). The results from the pres-
ent experiments allow us to conclude that our monkeys are able to
reflect upon their cognitions, indicating that they possess at least one
level of metacognition. And that suffices for granting them metacogni-
tive abilities.
REFERENCES
Arbuckle, T. Y., & Cuddy, L. L. (1969). Discrimination of item strength at time
of presentation. Journal of Experimental Psychology, 81, 126–131.
Benjamin, A. S., Bjork, R. A., & Schwartz, B. L. (1998). The mismeasure of
memory: When retrieval fluency is misleading as a metamnemonic index.
Journal of Experimental Psychology: General, 127, 55–68.
Cary, M., & Reder, L. M. (2002). Metacognition in strategy selection: Giving
consciousness too much credit. In M. Izaute, P. Chambres, & P. J. Mares-
caux (Eds.), Metacognition: Process, function, and use (pp. 63–78). New York:
Kluwer.
Flavell, J. H. (2000). Development of children’s knowledge about the mental
world. International Journal of Behavioral Development, 24, 15–23.
Glenberg, A. M., Sanocki, T., Epstein, W., & Morris, C. (1987). Enhancing cali-
bration of comprehension. Journal of Experimental Psychology: General, 116,
119–136.
Hart, J. T. (1965). Memory and the feeling-of-knowing experience. Journal of
James, W. (1890). Principles of psychology. New York: Holt.
Janowsky, J. S., Shimamura, A. P., & Squire, L. R. (1989). Memory and meta-
memory: Comparisons between frontal lobe lesions and amnesic patients.
Psychobiology, 17, 3–11.
Koriat, A., Lichtenstein, S., & Fischhoff, B. (1980). Reasons for confidence.
Journal of Experimental Psychology: Human Learning and Memory, 6, 107–118.
Metcalfe, J. (1993a). Novelty monitoring, metacognition, and control in a com-
posite holographic associative recall model: Implications for Korsakoff am-
nesia. Psychological Review, 100, 3–22.
Metcalfe, J. (1993b). Monitoring and gain control in an episodic memory
model: Relation to the P300 event-related potential. In A. F. Collins & S. E.
Gathercole (Eds.), Theories of memory (pp. 327–353). Hove, UK: Lawrence
Erlbaum.
Metcalfe, J., Schwartz, B. L., & Joaquim, S. G. (1993). The cue-familiarity heu-
ristic in metacognition. Journal of Experimental Psychology: Learning, Memory,
Metcalfe, J., & Shimamura, A. P. (1994). Metacognition: Knowing about knowing.
Miner, A. C., & Reder, L. M. (1994). A new look at feeling of knowing: Its
metacognitive role in regulating question answering. In J. Metcalfe & A. P.
Shimamura (Eds.), Metacognition: Knowing about knowing (pp. 47–70). Cam-
Nelson, T. O. (1992). Metacognition: Core readings. Boston: Allyn and Bacon.
Nelson, T. O., Gerler, D., & Narens, L. (1984). Accuracy of feeling-of-knowing
judgments for predicting perceptual identification and relearning. Journal
Nelson, T. O., & Narens, L. (1990). Metamemory: A theoretical framework
and new findings. In G. H. Bower (Ed.), The psychology of learning and moti-
vation (Vol. 26, pp. 125–141). New York: Academic Press.
Nelson, T. O., & Narens, L. (1994). Why investigate metacognition? In J. Met-
calfe & A. P. Shimamura (Eds.), Metacognition: Knowing about knowing (pp.
1–25). Cambridge, MA: MIT Press.
Nisbett, R. E., & Bellows, N. (1977). Verbal reports about causal influences on
social judgments: Private access versus public theories. Journal of Personal-
ity and Social Psychology, 35, 613–624.
Nisbett, R. E., & Wilson, T. D. (1977). Telling more than we know: Verbal
reports on mental processes. Psychological Review, 84, 231–279.
Perfect, T. J., & Hollins, T. S. (1996). Predictive feeling of knowing judgements
and postdictive confidence judgements in eyewitness memory and general
knowledge. Applied Cognitive Psychology, 10, 371–382.
Reder, L. M. (1987). Strategy selection in question answering. Cognitive Psy-

chology, 19, 90–138.
Reder, L. M. (1996). Different research programs on metacognition: Are the
boundaries imaginary? Learning and Individual Differences, 8, 383–390.
Reder, L. M., & Ritter, F. E. (1992). What determines initial feeling of know-
ing? Familiarity with question terms, not with the answer. Journal of Experi-
mental Psychology: Learning, Memory, and Cognition, 18, 435–451.
Reder, L. M., & Schunn, C. (1996). Metacognition does not imply awareness:
Strategy choice is governed by implicit learning and memory. In L. M.
Reder (Ed.), Implicit memory and metacognition (pp. 45–77). Mahwah, NJ:
Lawrence Erlbaum.
Schwartz, B. L. (1994). Sources of information in metamemory: Judgments
of learning and feelings of knowing. Psychonomic Bulletin and Review, 1,
357–375.
Schwartz, B. L., & Metcalfe, J. (1992). Cue familiarity but not target retrievabil-
ity enhances feeling-of-knowing judgments. Journal of Experimental Psychol-
Shaughnessy, J. J. (1979). Confidence-judgment accuracy as a predictor of test
performance. Journal of Research in Personality, 13, 505–514.
humans and Rhesus monkeys (Macaca mulatta) in a psychophysical same-
Shimamura, A. P., & Squire, L. R. (1986). Memory and metamemory: A study
of the feeling-of-knowing phenomenon in amnesic patients. Journal of Ex-
perimental Psychology: Learning, Memory, and Cognition, 12, 452–460.
Smith, J. D., Schull, J., Strote, J., McGee, K., Egnor, R., & Erb, L. (1995). The
uncertain response in the bottlenosed dolphin (Tursiops truncates). Journal
Tulving, E. (1994). Foreword. In J. Metcalfe & A. P. Shimamura (Eds.), Meta-
cognition: Knowing about knowing (pp. vii–x). Cambridge, MA: MIT Press.
Tulving, E., & Madigan, S. A. (1970). Memory and verbal learning. In P. H.
Mussen & M. R. Rosenzweig (Eds.), Annual review of psychology (pp. 437–
484). Palo Alto, CA: Annual Reviews.
Underwood, B. J. (1966). Individual and group predictions of item difficulty
for free-recall learning. Journal of Experimental Psychology, 71, 673–679.
13
The Self and Other: A Missing Link

in Comparative Social Cognition
Josep Call
The research area of mental state attribution has experienced great

change in recent years. Topics such as the understanding of intention,
knowledge, and belief are receiving increasing research attention (see
Heyes, 1998; see also Call & Tomasello, 2003, for a recent review).
However, the area that has developed most in the last few years is
the study of the understanding of visual perception in others, particu-
larly in chimpanzees. Studies show that chimpanzees follow the gaze
of conspecifics and humans, follow it past distractors and behind bar-
riers, “check back” with humans when gaze following does not yield
interesting sights, use gestures appropriately depending on the visual
access of their recipient, and select different pieces of food depending
on whether their competitor has visual access to them (see Call &
Tomasello, in press, for a review). These results, which cannot be ex-
plained by invoking a conspecific’s line of sight or orientation as a
discriminative stimulus, show that, at a minimum, chimpanzees know
what others can and cannot see.
Interest in mental state attribution in others contrasts with the mea-
ger research devoted to the study of mental attribution in the self.
Since chimpanzees appear to know about what others can and cannot
see, one can also ask whether they also know about what they them-
selves have seen. Partly, an explanation for the little development of
this research area can be found in the methods that have been used
to answer questions about the self in comparative social cognition.
Two approaches have traditionally dominated the study of the self in
nonhuman animals: imitation and mirror self-recognition. Although
these two paradigms have offered some valuable information about
the self, I argue that these two paradigms have exhausted their ex-
planatory power regarding the self. This is particularly true of the
conceptual self as opposed to the ecological self. Neisser (1988) defines
the ecological self as that who interacts with the environment and
321
receives information about it via the senses, whereas the conceptual

self consists of the mental representation of the individual’s personal
features including his or her knowledge.
Recently, however, new paradigms to investigate metacognition in
nonhuman animals have contributed significantly to the renewed in-
terest in this area (as this volume testifies). These paradigms are
broadly based on presenting animals with uncertain situations and
measuring their responses in an attempt to gauge how much an ani-
mal knows about what it knows and whether it can apply that knowl-
edge to exercise control over the information it seeks to obtain (Call
& Carpenter, 2001; Hampton, 2001; Smith et al., 1995; Smith, Shields,
Schull, & Washburn, 1997). The goal of this chapter is to depict these
methods as valid alternatives to more traditional methods and to indi-
cate the significance of studying the self and others for understanding
comparative social cognition.
This chapter is organized as follows. First, I present the main find-
ings produced by the imitation and the mirror self-recognition para-
digms in relation to the question of the self in social cognition, and I
highlight their limitations. Second, I describe some new methods that
use uncertainty as alternatives and extensions of the imitation and
self-recognition paradigms. I concentrate on the findings of a para-
digm based on presenting incomplete information to see if subjects
seek additional information before making a choice. Third, I discuss
the possibility that the seeking-information paradigm can be used to
gauge metacognitive functioning across various species. I then return
to the questions I raised at the beginning of the chapter about mental
state attribution in others and propose a closer connection, both em-
pirically and theoretically, between the study of mental attribution in
the self and in others.
TWO TRADITIONS IN THE STUDY OF THE SELF IN NONHUMAN ANIMALS

The comparative study of the self has been based on two paradigms:
imitation and mirror self-recognition. Each of these paradigms comes
from two different traditions, developmental and comparative psy-
chology, respectively. Due to their different origins, each tradition has
imprinted its particular features on the comparative study of the self.
Consequently, the questions that each tradition seeks to answer are
not the same, even though they bear some resemblance. Here I review
the main findings of each field in relation to the self and note some
of their limitations.
Imitation
Copying someone else’s actions involves making a correspondence
between the observed actions and producing actions on the self (Mitch-
The Self and Other 323
ell, 1993). In other words, imitation involves a translation of the visual

information (observed actions of others) into kinesthetic information
(produced actions by the self). For that reason, imitation is commonly
used as evidence of self-recognition (Lewis, 1994) or self-concept (Hart
& Fegley, 1994). Moreover, many developmental psychologists see in
the imitation of others one of the most fundamental processes by
which individuals construct their self-concept (Baldwin, 1902; Lewis
& Brooks-Gunn, 1979). This process may be initiated at a very early
age when very young infants imitate the facial expressions of others.
Gopnik and Meltzoff (1994) have argued that by imitating the facial
expressions of others, very young infants get to experience what emo-
tions are associated with certain facial expressions, thus giving them
a glimpse of what others feel when they display those facial expres-
sions. Later, the reproduction of others’ behavior through imitation
will give children additional information about other aspects of oth-
ers’ minds that children will eventually assimilate into their concept of
self, and by extension to others will form their mental state attribution
abilities (theory of mind).
Some authors have extended the connection between imitation and
self-concept to the great apes (Parker & Milbrath, 1994). However, the
evidence that apes copy actions is fragile (see Tomasello, 1996, for a
review). Currently, there is no consensus regarding whether chimpan-
zees or other apes spontaneously imitate the actions of others. Some
studies suggest that the most likely learning mechanism is for one
observer to attempt to reproduce an outcome similar to that produced
by a demonstrator rather than copying the demonstrator’s actions
(Call, Carpenter, & Tomasello, 2002; Call & Tomasello, 1995; Myowa-
Yamakoshi & Matsuzawa, 1999; Nagell, Olguin, & Tomasello, 1993;
Tomasello, Davis-Dasilva, Camak, & Bard, 1987). Others claim that
chimpanzees and other apes are capable of spontaneously copying
actions of others to solve problems, although these studies do not ef-
fectively distinguish actions from results (Stoinski, Wrate, Ure, &
Whiten, 2001; Whiten, 1998; Whiten, Custance, Gómez, Teixidor, &
Bard, 1996). It is true that apes raised by humans are more skilled
at copying the actions of others (Bering, Bjorklund, & Ragan, 2000;
Bjorklund, Yunger, Bering, & Ragan, 2002; Tomasello, Savage-Rum-
baugh, & Kruger, 1993). However, these actions usually involve some
sort of result such as a noise or a change of state in an object, so that
it is hard to know precisely what is the contribution of actions or
results to the reproduction of the target behaviors. Some studies have
eliminated the objects altogether and focused on the reproduction of
gestures (Call, 2001; Custance, Whiten, & Bard, 1995; Miles, 1990).
After extensive training, apes can reproduce novel body movements
on command, but their accuracy is far from perfect, and there is some
indication that apes reproduce those actions with some tangible result
more readily than those without results (Call, 2001).
Mirror Self-Recognition
From its inception, the mirror self-recognition paradigm has been
closely linked to self-awareness and a concept of self (Gallup, 1970,
1982). According to Gallup, self-directed mirror inspection is evidence
for self-awareness (Gallup, 1982), or a representational self (Anderson
& Gallup, 1999). The empirical evidence that has been used to support
this claim is as follows (see Anderson, 1999; Anderson & Gallup, 1999,
for reviews). When confronted with a mirror, chimpanzees and orang-
utans show social responses first, followed by contingent behavior in
front of the mirror, and finally mirror-guided self-directed inspection
of body parts not visible otherwise (Lethmate & Ducker, 1973; Suarez
& Gallup, 1981). When administered the mark test, in which subjects
are anesthetized and an odorless dye is applied to some body part
that is not directly visible, these species touch the mark more often in
the presence of a mirror than in its absence. Developmentally, the
ability to show self-directed behavior appears between 3 and 5 years
of age, with adolescents and young adults showing the highest levels
of self-directed behavior in front of the mirror. In contrast to orang-
utans and chimpanzees (and bonobos, for whom there is also some
positive evidence; see Hyatt & Hopkins, 1994), most gorillas tested
have produced no clear evidence of mirror self-directed inspection.
Two possible exceptions to these negative results are two human-
reared gorillas that also seem to have passed some version of this test
(Patterson & Cohn, 1994; Swartz & Evans, 1994).
Similar tests done with monkeys have produced negative results
(see Anderson & Gallup, 1999, for a review). Monkeys also produce
social responses to the mirror image, but they do not show clear evi-
dence of self-directed mirror-guided behavior despite the numerous
studies that have attempted to find self-recognition in monkeys. Nev-
ertheless, monkeys, like apes, can use mirrors to find hidden food that
cannot be observed directly (Anderson, 1986).
Although there has been continued controversy regarding the pro-
cedural details of the mirror self-recognition paradigm across the
years (e.g., Epstein, Lanza, & Skinner, 1981; Heyes, 1994), the basic
findings are generally accepted. Nevertheless, the question of what
the mirror self-recognition experiments show regarding self-aware-
ness is still a matter of intense debate. While some authors maintain
that mirror self-recognition experiments cannot answer questions
about self-awareness (Heyes, 1994), others see mirror self-recognition
as evidence of self-awareness (Gallup, 1982), a representational self
(Anderson & Gallup, 1999), or a concept of me (Lewis, 1994). I argue
that no major advances in answering this question have been made in
the last 20 years, even though there is more information about more
species. Particularly lacking are answers for the question of the more
psychological self rather than the perceptual self. Mirror self-recogni-
tion (and imitation) experiments are eerily silent regarding what the
self knows, remembers, or sees. But this is not surprising, because
these paradigms were not designed to answer such questions in the
first place. In my view, these traditional paradigms have offered as
much as they can regarding the question of the psychological self, and
I agree with other authors (Anderson, 1999; Gallup, 1994; Itakura,
2001) that novel approaches should complement these more tradi-
tional approaches. These novel approaches, however, should not
merely be covert mirror self-recognition tests. The study of shadow
reflections (e.g., Cameron & Gallup, 1988), video representations of
the self (Anderson, 1999), or the use of computer joysticks (Jorgensen,
Suomi, & Hopkins, 1995), though interesting in themselves, are not
enough. A truly novel approach to the question of the psychological
self is needed. Only recently have new studies attempted to get some
answers on the more psychological level of the self. Next, I review
these studies.
RECENT APPROACHES TO THE STUDY OF THE UNDERSTANDING OF THE SELF

Research on human metacognition is burgeoning. In general, research-
ers have used two methods to explore this area in nonhuman animals.
One is to create uncertainty and measure escape responses. The other
is to provide incomplete information and measure whether subjects
seek additional information. Unfortunately, most studies with hu-
mans are not applicable to nonhuman animals because they rely heav-
ily on language. Recently, however, two lines of research have used
nonverbal measures to assess escape responses in cetaceans and pri-
mates, including humans.
Smith et al. (1995) presented an auditory discrimination task to dol-
phins and humans. They presented subjects with auditory stimuli that
varied between 1200 Hz and 2100 Hz. For a given stimulus, subjects
had to decide whether the stimulus was low or high pitch. If the stim-
ulus was between 1200 and 2099 Hz, they had to press the low key,
whereas if the stimulus was 2100 Hz (the highest pitch) they had to
press the high key. For stimuli at the low end of the distribution this
discrimination was easy, but as stimuli drew closer to the 2100 Hz
mark, the discrimination became increasingly difficult. Subjects were
only rewarded for correct responses. Incorrect responses produced a
time-out period of 9 to 12 seconds. In addition to the low or high key
choice, subjects had a third option: the escape key. Subjects could
press this key to skip the current trial and go to the next one, which
was always reinforced. However, to avoid excessive use of this key,
the computer gradually delayed the presentation of trials on which

reinforcement was guaranteed if the escape key was used too fre-
quently. Results showed that dolphins and humans increased the use
of the escape key when the discrimination became increasingly diffi-
cult. They rarely used the escape key for low-pitch stimuli, and the
percentage of escape responses also decreased for the highest pitch
stimulus. Thus, the distribution of escape responses was an inverted
U-shaped curve that peaked slightly before the 2100-Hz mark.
In another study, Smith et al. (1997) tested rhesus monkeys and
humans with a visual discrimination task and found results similar to
those of the previous study. The stimuli varied in the density of pixels
presented, and subjects had to discriminate low-density boxes (450–
2,949 pixels) from high-density boxes (2,950 pixels). Subjects also had
the option of the escape key available. The contingencies for correct
and incorrect responses and for the use of the escape key were analo-
gous to those of the previous study. The pattern of results was nearly
identical to that found in the previous study. Rhesus macaques and
humans increased the use of the escape key when the discrimination
became increasingly difficult. They rarely used the escape key for low-
density stimuli, and the percentage of escape responses also decreased
for the highest pitch/density stimuli. The distribution of escape re-
sponses was an inverted U-shaped curve that peaked before the 2,950-
pixel mark. Smith and colleagues argued that these studies show that
dolphins, rhesus monkeys, and humans behave in similar ways in
uncertain situations. All species showed some control over their deci-
sion to avoid a given trial under perceptual uncertainty.
Hampton (2001; see chapter 11, this volume) used a different ap-
proach to study metacognition in a nonverbal animal. Two rhesus
monkeys were trained on a delayed matching-to-sample task that was
presented on a touch-sensitive computer screen. Following the presen-
tation of a sample stimulus and a delay of 30 seconds, the subject was
presented with the sample and three distractors. On one third of the
trials, a desirable reward (a peanut) was contingent on selecting the
sample. Incorrect choices resulted in no reward and a time-out period.
On the remaining two thirds of the trials, subjects were offered the
option of declining the test. Choosing that option resulted in the deliv-
ery of monkey chow, a less desirable reward. Whereas declining a test
always produced a reward, opting to take the test produced a peanut
only if the subject was correct. Hampton observed that monkeys per-
formed better on those trials in which they were free to decline a test
compared with the forced test. This suggested that monkeys may have
known when they had forgotten the correct answer.
Two additional tests reinforced this conclusion. In a second experi-
ment, Hampton increased the delay interval between the presentation
of the sample and the alternatives. This was done to foster forgetting
in the monkeys. Results showed that monkeys’ performance declined

proportionally to the amount of delay between the sample and the
alternative stimuli. This was expected. More important, the propor-
tion of trials on which the monkeys chose to decline the trial increased
as the delay between the sample and the alternatives increased. In a
third experiment, Hampton presented some trials without a sample
(a blank picture, to be exact). As expected, the monkeys declined a
high proportion of tests when no sample had been offered. Hampton
concluded that rhesus monkeys know when they have forgotten and
that they can remedy this lack of information by escaping the situa-
tion.
An Initial Study on Seeking Information

The two previous studies showed that cetaceans and primates can
escape situations of uncertainty. They knew when stimuli were too
hard to discriminate or when they did not have information about the
sample stimuli. Another avenue to explore metacognition in nonhu-
man animals consists of presenting incomplete information and seeing
whether individuals seek additional information. Call and Carpenter
(2001) presented chimpanzees, orangutans, and 2.5-year-old children
with a situation in which they had to choose one of two containers to
obtain a reward. Previous to this study, subjects had been presented
with two opaque containers, and they had to displace one of them to
indicate their choice. In this study, however, we replaced the opaque
containers with hollow 30-cm tubes that were placed perpendicular
to the subject. We hid food in the side of the tube closest to the experi-
menter so that the subjects were not able to see the food from their
location. However, subjects could see the food inside the tube if they
bent down and looked through the tube (see figure 13.1). Using this
Figure 13.1 Experimental setup to investigate whether subjects will look inside the tubes
before selecting one of them when subjects are prevented visual access to the baiting
procedure. E, experimenter; S, subject. From Call and Carpenter (2001). Copyright 
Springer-Verlag.
setup, we presented two conditions. In the visible condition, subjects

had visual access to the baiting procedure so that they could see in
which tube the experimenter had placed the reward. In the hidden
condition, the baiting process was conducted behind a screen so that
subjects were prevented from seeing in which tube the experimenter
had placed the reward. We recorded which tube they selected, but
more important, we recorded whether they looked inside the tube
before choosing.
Results indicated that subjects looked significantly more often in-
side the tubes when they had been prevented from witnessing the
baiting process. Typically, in hidden trials, subjects looked inside the
tube, and if they saw the reward they stopped their search and se-
lected that tube. If they did not see the reward inside the tube, they
checked the second tube and then made their choice (figure 13.2). Al-
though in most trials apes chose after seeing the food, in approxi-
mately 20% of the trials in which their first inspection revealed an
empty tube, they selected the other tube without inspecting it. This
means that subjects’ choices (and their looking behavior) were not
rigidly and solely controlled by the sight of food. Instead, they could
make some inferences regarding the location of food without having
seen it. Interestingly, children rarely used this inferential strategy but
preferred to inspect all tubes until they found the reward. We also
observed that if subjects were not allowed to choose one of the tubes
right away, that is, when a 5-second delay was inserted between the
end of the baiting and the opportunity to choose, subjects looked sig-
nificantly more often inside the tube.
Extending the Initial Findings to Other Species

The previous study showed that chimpanzees, orangutans, and chil-
dren looked more often inside the tubes on trials in which they had
not seen the location of the reward than on trials in which they saw
where the experimenter placed the reward. Thus, they had access to
information regarding the location of the reward, and when they did
not have such information, they sought it before they selected a con-
tainer. This initial finding prompted us to probe this skill further. We
saw two ways of broadening the scope of these initial findings. One
way consisted of probing further the mechanism responsible for this
performance. A second way was to study other species, both closely
and distantly related to orangutans, chimpanzees, and children. We
chose to start broadening the scope by comparing species.
From a comparative point of view, it was not surprising that chim-
panzees and orangutans solved this problem, because they also pass
the mirror self-recognition test. A more interesting comparison is
gorillas because, unlike chimpanzees and orangutans, they do not
readily pass mirror tests and do not show mirror-guided self-directed
Figure 13.2 Typical sequence of events in the hidden and the visible conditions. In the
hidden condition, the reward is hidden inside one of the tubes behind a screen (A), the
orangutan looks inside the empty tube (B), then the baited tube (C), and makes a choice
(D). In the visible condition, the reward is placed inside one of the tubes in full view of
the subject (E), and the subject chooses (F). Note the absence of looking behavior in the
visible condition compared to the hidden condition. See color insert.
329
behavior. Povinelli (1993) has argued that gorillas may have lost the
capacity for self-recognition (but see Patterson & Cohn, 1994; Swartz
& Evans, 1994), whereas Gallup (1994) has claimed that gorillas may
differ in cognition from the other great apes. In addition to studying
gorillas, we completed the investigation of the great ape clade by
studying bonobos. In particular, we studied 5 orangutans, 12 chim-
panzees, 6 gorillas, and 4 bonobos housed at the Wolfgang Köhler
Primate Research Center in Leipzig. There were 18 females and 9
males ranging from 4 to 30 years of age. All the bonobos except the
female and all the adult chimpanzees were nursery reared, whereas
all other subjects were mother reared. All subjects lived in social
groups of various sizes, with access to indoor and outdoor areas. Sub-
jects were individually tested in their indoor cages and were not food
or water deprived.
We followed the delayed procedure used in the initial experiment
(Call & Carpenter, 2001) with both visible and hidden trials. Thus, the
experimenter placed two tubes on the platform and either baited them
while the subject watched (visible trials) or conducted the baiting be-
hind a screen (hidden trials). After the baiting was completed, the
experimenter held the tubes in view but outside the reach of the sub-
ject for 5 seconds. The experimenter then pushed the platform toward
the subject so that it could select one of the tubes. All subjects had
learned prior to this experiment to touch one of the two objects avail-
able to request that container. Subjects received a total of 24 visible
and 24 hidden trials presented in four 12-trial sessions. Each session
consisted of 6 visible and 6 hidden trials. All trials were videotaped
and later coded by the experimenter. We measured two variables
across trials: (1) looking inside the tube, and (2) tube selection. A sec-
ond coder who was unaware of the experimental conditions blindly
coded 20% of the trials for reliability purposes on each of the two
measures used in this study. Reliability for both measures was excel-
lent (looking inside the tube: Cohen’s kappa = 0.87; choosing the tube:
Cohen’s kappa = 1.0).
Figure 13.3 presents the percentage of trials in which subjects
looked inside the tubes for each of the two conditions. The four great
apes looked significantly more in the hidden than in the visible condi-
tion, F(1,23) = 38.7, p < .001. There were no differences across species
and no interaction effects. Thus, gorillas were indistinguishable from
the rest of the apes.
Table 13.1 presents the individual performances of each of the apes
in the hidden trials, with particular attention devoted to the pattern
of responses across trials. We distinguished three types of acquisition
of looking inside the tubes: no acquisition, gradual acquisition, and
sudden acquisition. No acquisition consisted of failing to look consis-
tently during testing. Gradual acquisition consisted of progressively
Figure 13.3 Mean percentage of trials in which apes looked inside the tube
as a function of the baiting condition and the species.
increasing the looks inside the tube across trials. This type of acquisi-
tion included a period of alternation between trials, with and without
looks, until the looking response appeared in every trial. Sudden ac-
quisition consisted of looking inside the tube from the first trial (i), or,
if subjects did not look in the first trial, once they started to look they
looked consistently for the remaining trials (ii). Eighteen of the 27
subjects (67%) showed sudden acquisition of the looking behavior. Of
those 18 subjects, 13 (48%) looked from the first trial, whereas 5 others
(19%) first did not look, but once they started they did not stop look-
ing. Five subjects acquired the looking response gradually (19%),
whereas four other subjects showed no acquisition.
The search pattern was analogous to that found in our initial study
(see figure 13.4). Upon finding a baited tube after their first look, sub-
jects stopped their search and selected that tube in the great majority
of trials. In contrast, if they found no food after their first look, they
continued looking. Nevertheless, in a sizeable number of trials (16–
38%, depending on the species) subjects selected the other tube upon
finding an empty one, without looking into it. These “blind” choices
resulted in a performance that was above chance, t(15) = 2.80, p = .014.
In sum, we found the same search pattern as before, and gorillas
were undistinguishable from the other apes. Thus, the ape clade pre-
sents a homogenous performance when seeking information in uncer-
tain situations. One question is how much this skill extends across
other taxa. To answer this question, we contrasted the ape results with
those of another species distantly related to the apes but which shows
some ability to solve problems that require the use of gaze following,
visual communication, or sensitivity to the state of the human’s eyes
Table 13.1 Subject Information and Percentage of Trials in Which Subjects Look Inside
the Tube in Hidden Trials.
Age % Trials Acquisition

Species/Name Sex (yrs) w/Looks Pattern Looks Across Trials
Gorillas
Bebe F 22 46 Gradual NNNNYNNNYYNNNYNYYYNNYYYY
Viringika F 6 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Vizuri F 6 83 Sudden(ii) NNNNYYYYYYYYYYYYYYYYYYYY
Ruby F 4 88 Sudden(i) YYNNNYYYYYYYYYYYYYYYYYYY
Gorgo M 20 0 No NNNNNNNNNNNNNNNNNNNNNNNN
Vimoto M 6 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Orangutans
Dunja F 30 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Pini F 13 96 Sudden(i) YYYYYYYYYYYYYYYYYYYYNYYY
Toba F 7 92 Sudden(ii) NYYYYYYYYYYYYYYYYYYNYYYY
Padana F 4 54 Gradual NNNNNNYNYYYYYYYYYYNNYYNN
Bimbo M 21 83 Sudden(i) YYYYNYYYYYYYYNYYYYNNYYYY
Chimpanzees
Fraukje F 25 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Riet F 25 88 Sudden(i) YYYYYYYYNYYYNNYYYYYYYYYY
Ulla F 24 75 Gradual NNNYNYYNYYYYYYYYNYYYYYYY
Natascha F 21 71 Gradual NNYNYYYYYYYYYYYNYNNYYYNY
Dorien F 21 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Jahaga F 8 88 Sudden(ii) NYYYYYYYYYYYYYYYYYYYYNYN
Gertruida F 8 0 No NNNNNNNNNNNNNNNNNNNNNNNN
Fifi F 8 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Sandra F 8 17 Sudden(ii) NNNNNNNNNNNNNNYNNNNNNYYY
Robert M 26 8 No NNNNNNNYYNNNNNNNNNNNNNNN
Frodo M 8 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Patrick M 4 0 No NNNNNNNNNNNNNNNNNNNNNNNN
Bonobos
Ulindi F 8 79 Gradual NNNYYNNYYYYYYYYYYYYYYYYY
Joey M 19 96 Sudden(ii) NYYYYYYYYYYYYYYYYYYYYYYY
Limbuko M 6 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Kuno M 5 100 Sudden(i) YYYYYYYYYYYYYYYYYYYYYYYY
Notes. Includes the acquisition pattern and whether the subject looked on each of the 24 tri-
als. Y, Yes; N, No. Sudden(i) indicates that subjects looked inside the tube in the first trial;
Sudden(ii) indicates that looks began after the first trial.
Figure 13.4 Mean percentage of trials in which apes made a choice after
their first look as a function of what they saw inside the tube and the species.
(see Call, 2004, for a review). This species is the domestic dog. Bräuer,
Call, and Tomasello (in press) presented dogs with the same problem
that we used with apes but with a modified apparatus. Food was
hidden inside a box that subjects could designate by pressing a lever
with their paw. Opposite the lever was a window with holes through
which dogs could look and smell the contents of the box before select-
ing it. As with the apes, there were visible and hidden trials. Figure
13.5 presents the percentage of trials in which dogs inspected the con-
tents of the box (either by looking into it or sniffing it) and the per-
centage of correct trials. Unlike the apes, dogs did not look (or sniff)
before choosing significantly more often when they had not witnessed
Figure 13.5 Mean percentage of trials in which dogs looked inside the tube
as a function of the baiting condition.
the baiting. The percentage of correct choices in each condition reflects

this fact because they were above chance when they saw the food
location but not when they did not see it. So, unlike apes, dogs did
not seek information before making a choice.
SEEKING INFORMATION AS A METACOGNITIVE INDEX

Recent studies have shown that our initial finding that chimpanzees,
orangutans, and children seek information when they have not seen
the baiting of the reward are replicable with another group of subjects
of the same species and can also be extended to bonobos and gorillas.
Thus, apes and children (unlike dogs) preferentially seek information
when this information is missing. It is important to emphasize that
the behavior they use to seek information is not the same that they
use to select one of the alternatives present on the platform. They
choose by touching the container, not by bending down to look
through it. In fact, bending down to look is not a behavior that they
usually display in our testing situation. When they are presented with
opaque containers on a platform, they do not bend down or engage
in other behaviors that may produce information that they do not
currently have. Yet one can ask whether this seeking behavior is evi-
dence of metacognition or can be explained by means of other mecha-
nisms.
Two mechanisms that do not require a metacognitive explanation
could explain the apes’ performance. One mechanism is that subjects
simply use an automatic hard-wired search routine based on execut-
ing search responses when information about the food location is
missing. Thus subjects would search for food until they found it, then
stop their search and select the appropriate container. There are some
problems with this explanation, however. First, sometimes subjects
stopped searching and made appropriate choices even without seeing
the food. Gorillas and orangutans selected the container they had not
inspected in 38% of the trials in which they did not see the food after
their first look. In addition, recent experiments have shown that when
subjects are forced to choose between two stimuli and only partial
visual or auditory information is offered, they can still infer the loca-
tion of food (Call, in press). This argues against a rigid algorithm for
food finding. Moreover, not all subjects executed this routine, since
some of them never sought information, and some of them that had
seen where the food was located still looked inside the tube, albeit
less often than on trials in which they had not seen the location of the
food.
Another alternative that may explain our results is that subjects
may have learned (rather than having a hard-wired response) some
fixed rule that has produced positive consequences in the past.
Namely, in the absence of food subjects may have produced a variety

of search responses indiscriminately and once some of those responses
were rewarded, they became established. For instance, during the test
subjects may have learned to bend down when a barrier was present
during the baiting because that increased the chances of selecting the
correct tube. Unlike the previous explanation, this could account for
the individual differences observed. However, this explanation cannot
account for the successful performance of those subjects who showed
no evidence of learning during the test.
Of course, one can argue that learning took place before the experi-
ment began. Note that we have no way to verify this possibility, but
invoking the learning of some fixed rule is often postulated but rarely
demonstrated. Although it is true that chimpanzees can associatively
learn conditional discriminations based on the presence of a certain
cue, it is not a foregone conclusion that they always do so. In fact,
apes may not be as skillful as commonly thought about learning such
discriminations in social and nonsocial problems (e.g., Call, Hare, &
Tomasello, 1998; Tomasello, Call, & Gluckman, 1997). Moreover, note
that subjects had never before experienced the tubes and barrier com-
bination as it was presented in the current test. Although one could
argue that subjects may have experienced similar situations in the past
and they transferred those experiences to adapt to the new situation,
this explanation opens the possibility that subjects’ choices are guided
by the properties of the stimuli rather than the particular stimuli them-
selves. In other words, subjects may encode information as “when my
visual access is blocked, then do something appropriate to gain visual
access to it” rather than “when a cardboard screen (or a log, wall, or
rock) is present, then do X.”
This more interpretive and flexible rule is a far cry from the initial
fixed rule of “when the cardboard barrier is present during baiting,
then bend down.” Unlike a fixed rule, it is based on knowledge that
was created by experiencing a variety of situations, none of which
matched exactly the one that we presented in our test. Additional
experiments will be needed to specify more precisely the kind of
knowledge that underlies these rules. For instance, what would hap-
pen if subjects witnessed the baiting and, once it was completed, a
barrier was interposed between the subject and the tubes? Or what
would happen if the baiting occurred behind a barrier, but instead of
hollow tubes the experimenter used opaque cups? If subjects were
using this more interpretive strategy, we would not expect subjects to
bend down and look for the food.
A final argument against merely learning fixed rules: Dogs are
skillful in other domains of visual perception in social situations, even
more than chimpanzees (Call, 2004; Hare, Brown, Williamson, & Tom-
asello, 2002), but they do not pass this test. Clearly, dogs’ failures do
not convert apes’ successes into metacognition, but it shows that an

animal who can learn cues to find food does not seek information
when it is ignorant regarding the exact location of food. In any case,
the dog-ape comparison highlights the contrast between these species
and perhaps indicates a difference between the cognitive mechanisms
that govern these two different animal clades.
Thus, I argue that the recourse to fixed rules such as “bend down in
the presence of a barrier” is not likely to explain some of the evidence
presented here. This does not mean that past experience and learning
are unimportant in developing novel solutions to novel problems. On
the contrary, past experience is probably crucial to develop appro-
priate solutions to novel problems. What it is disputed here is the
validity of using fixed rules to explain phenomena to the detriment
of flexible rules based on knowledge accumulation created through
multiple experiences, none of which exactly matches our test situa-
tion.
In sum, neither of the previous two explanations seems fully satis-
factory. One key feature in which both explanations falter is that sub-
jects show flexibility of action in their seeking behavior. This means
that subjects have some access to what they have and have not seen,
and this access is not likely to be controlled by a rigid program or
learned after trial and error. The nature of this access is still unclear,
and additional research is needed to pinpoint the exact mechanism
responsible for the behavior here described and its scope. Some addi-
tional experiments could probe the effect of increasing delays between
witnessing a baiting and choosing (like Hampton, 2001), increasing
the quality of the reward, the cost of choosing wrong, or the need to
take intermediate steps to find out the location of the reward. For
instance, subjects could lift lids to peek inside tubes and once the food
has been found, they would have to use tools. One of the advantages
of this method is its simplicity. It does not require any training (be-
sides training to touch an object to select) because it capitalizes on
spontaneous behavior.
THE SELF AND THE OTHER IN SOCIAL COGNITION

Positive evidence is accumulating regarding what chimpanzees know
about what others can and cannot see. They follow gaze, look around
barriers, use visual gestures mostly when others are looking, are sensi-
tive to the body orientation and the faces of others when using visual
signals, and know what others have and have not seen in competitive
situations. The findings described in this chapter add another piece to
the social cognition puzzle. Chimpanzees and the other great apes
also know what they themselves have and have not seen. It is there-
fore possible to begin to establish a link between mental attribution in
others and the self. For instance, the cognitive mechanism that gives
rise to following the gaze of another individual behind a barrier to
see what the other individual is looking at may be the same as the
cognitive mechanism that gives rise to looking through the tubes
when one has not seen where the experimenter has deposited the
food. Future research will bring about a closer connection between
these two complementary aspects of research on mental state attribu-
tion.
When discussing the attribution of psychological states to others
(and to the self), it is important to be precise about what has been
shown. I argue that these studies have shown that at least chimpan-
zees (and presumably other apes) know what they have seen, and
they know what others can or cannot see. Some may be tempted to
translate this into meaning that they know what they know and what
others know or do not know. Currently, however, there is no study
that can tease apart seeing from knowing. In any case, both seeing
and knowing are psychological states (see Call & Tomasello, in press,
for a more detailed discussion). Yet they differ in the degree of ab-
straction that they entail. Whereas perceptual information is a key
component of seeing, such perceptual support is less relevant in the
case of knowing. Individuals may represent the idea of seeing by visu-
alizing someone facing in a certain direction and observing a given
event. Knowing is harder to represent at such a perceptual level. This
is why knowing is often described as a more opaque mental state than
seeing. An analogy with the area of categorization may help to clarify
the distinction between seeing and knowing and the different levels
of abstraction that they represent. Perceptual categories would be
analogous to seeing, whereas conceptual (or functional) categories
would be analogous to knowing.
Despite its lower abstract load, it is important to emphasize that
seeing should not be equated to purely observable behavior. Seeing,
like knowledge, is a psychological state, not just observable behavior.
Again, using the analogy with categories can help illustrate the differ-
ence between seeing and observable behavior. Seeing, like perceptual
categories, allows individuals to solve problems with new exemplars
of a given category, whereas purely observable behavior, like rote
memory, only allows individuals to remember previously seen exem-
plars. Although seeing falls within the realm of psychological states,
it is also important to recognize that it is not equivalent to other psy-
chological states such as knowledge or belief. It is also important to
recognize that various psychological states follow different develop-
mental trajectories. For instance, some theorists argue that beliefs owe
their special status to the property that the individual can entertain
the possibility that a given fact may be false, and that individuals may
be induced to hold false beliefs by supplying them with inaccurate
information (Perner, 1991). Note that seeing does not afford this “du-
plicity” property. Individuals either see or do not see a particular
event, but they do not necessarily entertain the possibility that what
they saw may be false. One of the most critical advances in the study
of the attribution of psychological states has been the realization that
different species or members of a same species at different ages vary
in their ability to attribute the various psychological states to them-
selves and others. From a comparative perspective, we still do not
know whether nonhuman animals attribute knowledge to themselves
or others (let alone beliefs) or just visual perception. Moreover, even
within the area of visual perception, many questions remain unre-
solved. As mentioned before, previous studies have shown that chim-
panzees know what others can and cannot see. Yet other aspects such
as perspective taking (how an individual would see an object from a
given perspective) or whether individuals recognize that others can
be attending to two different aspects of an object (e.g., color vs. shape)
remain totally unexplored. Future studies should be devoted to de-
velop this field and strengthen the link between the self and the other
in comparative social cognition.
ACKNOWLEDGMENTS This manuscript benefited tremendously from my dis-
cussions with Malinda Carpenter and Mike Tomasello on this topic. I also
want to thank the participants of the conference on the evolution of self-reflec-
tive consciousness for their valuable comments on this work.
REFERENCES
Anderson, J. R. (1986). Mirror-mediated finding of hidden food by monkeys
(Macaca tonkeana and Macaca fascicularis). Journal of Comparative Psychology,
100, 237–242.
Anderson, J. R. (1999). Primates and representations of self. Cahiers de Psycho-
logie Cognitive, 18, 1005–1029.
Anderson, J. R., & Gallup, G. G., Jr. (1999). Primates and representations of
self. Cahiers de Psychologie Cognitive, 18, 1005–1029.
Baldwin, J. M. (1902). Social and ethical interpretations in mental life. New York:
Macmillan.
Bering, J. M., Bjorklund, D.F., & Ragan, P. (2000). Deferred imitation of object-
related actions in human-reared juvenile chimpanzees and orangutans. De-
velopmental Psychobiology, 36, 218–232.
Bjorklund, D. F., Yunger, J. L., Bering, J. M., & Ragan, P. (2002). The general-
ization of deferred imitation in enculturated chimpanzees (Pan troglodytes).
Animal Cognition, 5, 49–58.
Bräuer, J., Call, J., & Tomasello, M. (in press). Visual perspective taking in
dogs (Canis familiaris) in the presence of barriers. Applied Animal Behaviour
Science.
Call, J. (2001). Body imitation in an enculturated orangutan. Cybernetics and
Systems, 32, 97–119.
Call, J. (2004). The use of social information in chimpanzees and dogs. In L.

Rogers & G. Kaplan (Eds.), Are primates special? (pp. 263–286). New York:
Kluwer Academic.
Call, J. (in press). Inferences about the location of food in the great apes. Jour-
nal of Comparative Psychology.
Call, J., & Carpenter, M. (2001). Do chimpanzees and children know what
they have seen? Animal Cognition, 4, 207–220.
Call, J., Carpenter, M., & Tomasello, M. (2002). Focusing on outcomes and
focusing on actions in the process of social learning: Chimpanzees (Pan
troglodytes) and human children (Homo sapiens). Manuscript submitted for
publication.
Call, J., Hare, B. H., & Tomasello, M. (1998). Chimpanzee gaze following in
an object-choice task. Animal Cognition, 1, 89–99.
Call, J., & Tomasello, M. (1995). The use of social information in the problem-
solving of orangutans (Pongo pygmaeus) and human children (Homo sapi-
ens). Journal of Comparative Psychology, 109, 308–320.
Call, J., & Tomasello, M. (2003). Social cognition. In D. Maestripieri (Ed.), Pri-
mate psychology (pp. 234–253). Cambridge, MA: Harvard University Press.
Call, J., & Tomasello, M. (in press). What chimpanzees know about seeing
revisited: An explanation of the third kind. In N. Eilan, C. Hoerl, T. McCor-
mack, & J. Roessler (Eds.), Joint attention. Oxford, UK: Oxford University
Press.
Cameron, P. A., & Gallup, G. G., Jr. (1988). Shadow recognition in human
infants. Infant Behavior and Development, 11, 465–471.
Custance, D. M., Whiten, A., & Bard, K. A. (1995). Can young chimpanzees
(Pan troglodytes) imitate arbitrary actions? Hayes & Hayes (1952) revisited.
Behaviour, 132, 837–859.
Epstein, R., Lanza, R. P., & Skinner, B. F. (1981). “Self-awareness” in the pi-
geon. Science, 212, 695–696.
Gallup, G. G., Jr. (1970). Chimpanzees: Self-recognition. Science, 167, 86–87.
Gallup, G. G., Jr. (1982). Self-awareness and the emergence of mind in pri-
mates. American Journal of Primatology, 2, 237–248.
Gallup, G. G., Jr. (1994). Self-recognition: Research strategies and experimen-
tal design. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Eds.), Self-aware-
ness in animals and humans: Developmental perspectives (pp. 35–50). Cam-
bridge, UK: Cambridge University Press.
Gopnik, A., & Meltzoff, A. N. (1994). Minds, bodies, and persons: Young chil-
dren’s understanding of the self and others as reflected in imitation and
theory of mind research. In S. T. Parker, R. W. Mitchell, & M. L. Boccia
(Eds.), Self-awareness in animals and humans: Developmental perspectives (pp.
166–186). Cambridge, UK: Cambridge University Press.
Hare, B., Brown, M., Williamson, C., & Tomasello, M. (2002). The domestica-
tion of social cognition in dogs. Science, 298, 1634–1636.
Hart, D., & Fegley, S. (1994). Social imitation and the emergence of a mental
model of self. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Eds.), Self-
awareness in animals and humans: Developmental perspectives (pp. 149–165).
Cambridge, UK: Cambridge University Press.
Heyes, C. M. (1994). Reflections on self-recognition in primates. Animal Behav-

iour, 47, 909–919.
Heyes, C. M. (1998). Theory of mind in nonhuman primates. Behavioral and
Hyatt, C. W., & Hopkins, W. D. (1994). Self-awareness in bonobos and chim-
panzees: A comparative perspective. In S. T. Parker, R. W. Mitchell, &
M. L. Boccia (Eds.), Self-awareness in animals and humans: Developmental per-
spectives (pp. 248–253). Cambridge, UK: Cambridge University Press.
Itakura, S. (2001). The level of self-knowledge in nonhuman primates: From
the perspective of comparative cognitive science. In T. Matsuzawa (Ed.),
Primate origins of human cognition and behavior (pp. 313–329). Berlin:
Springer-Verlag.
Jorgensen, M. J., Suomi, S. J., & Hopkins, W. D. (1995). Using a computerized
testing system to investigate the preconceptual self in nonhuman primates
and humans. In P. Rochat (Ed.), The self in infancy: Theory and research (pp.
243–256). Amsterdam: Elsevier.
Lethmate, J., & Ducker, G. (1973). Untersuchungen zum selbsterkennen im
spiegel bei orangutans und einigen anderen affenarten. Zeitschrift für Tierp-
sychology, 33, 248–269.
Lewis, M. (1994). Myself and me. In S. T. Parker, R. W. Mitchell, & M. L.
Boccia (Eds.), Self-awareness in animals and humans: Developmental perspec-
tives (pp. 20–34). Cambridge, UK: Cambridge University Press.
New York: Plenum.
Miles, H. L. W. (1990). The cognitive foundations for reference in a signing
orangutan. In S. T. Parker & K. R. Gibson (Eds.), “Language” and intelligence
in monkeys and apes (pp. 511–539). Cambridge, UK: Cambridge University
Press.
Mitchell, R. W. (1993). Mental models of mirror-self-recognition: Two theo-
ries. New Ideas in Psychology, 11, 295–325.
Myowa-Yamakoshi, M., & Matsuzawa, T. (1999). Factors influencing imitation
of manipulatory actions in chimpanzees (Pan troglodytes). Journal of Compar-
ative Psychology, 113, 128–136.
Nagell, K., Olguin, R. S., & Tomasello, M. (1993). Processes of social learning
in the tool use of chimpanzees (Pan troglodytes) and human children (Homo
sapiens). Journal of Comparative Psychology, 107, 174–186.
Neisser, U. (1988). Five kinds of self-knowledge. Philosophical Psychology, 1,
35–59.
Parker, S. T., & Milbrath, C. (1994). Contributions of imitation and role-play-
ing games to the construction of self in primates. In S. T. Parker, R. W.
Press.
Patterson, F. G. P., & Cohn, R. H. (1994). Self-recognition and self-awareness
in lowland gorillas. In S. T. Parker, R. W. Mitchell, & M. L. Boccia (Eds.),
Self-awareness in animals and humans: Developmental perspectives (pp. 273–
290). Cambridge, UK: Cambridge University Press.
Perner, J. (1991). Understanding the representational mind. Cambridge, MA: MIT
Press.
Povinelli, D. J. (1993). Reconstructing the evolution of mind. American Psychol-

ogist, 48, 493–509.
Smith, J. D., Schull, J., Strote, J., McGee, K., Egnor, R., & Erb, L.. (1995). The
uncertain response in the bottlenosed dolphin (Tursiops truncatus). Journal
Stoinksi, T. S., Wrate, J. L., Ure, N., & Whiten, A. (2001). Imitative learning by
captive western lowland gorillas (Gorilla gorilla gorilla) in a simulated food-
processing task. Journal of Comparative Psychology, 115, 272–281.
Suarez, S. D., & Gallup, G. G., Jr. (1981). Self-recognition in chimpanzees and
orangutans, but not gorillas. Journal of Human Evolution, 10, 175–188.
Swartz, K. B., & Evans, S. (1994). Social and cognitive factors in chimpanzee
and gorilla mirror behavior and self-recognition. In S. T. Parker, R. W.
Press.
Tomasello, M. (1996). Do apes ape? In C. M. Heyes & B. G. Galef, Jr. (Eds.),
Social learning in animals: The roots of culture (pp. 319–346). New York: Aca-
demic Press.
Tomasello, M., Call, J., & Gluckman, A. (1997). Comprehension of novel com-
municative signs by apes and human children. Child Development, 68, 1067–
1080.
Tomasello, M., Davis-Dasilva, M., Camak, L., & Bard, K. (1987). Observational
learning of tool-use by young chimpanzees. Human Evolution, 2, 175–183.
Tomasello, M., Savage-Rumbaugh, E. S., & Kruger, A. C. (1993). Imitative
learning of actions on objects by children, chimpanzees, and enculturated
chimpanzees. Child Development, 64, 1688–1705.
Whiten, A. (1998). Imitation of the sequential structure of actions by chimpan-
zees (Pan troglodytes). Journal of Comparative Psychology, 112, 270–281.
Whiten, A., Custance, D. M., Gómez, J. C., Teixidor, P., & Bard, K. A. (1996).
Imitative learning of artificial fruit processing in children (Homo sapiens)
and chimpanzees (Pan troglodytes). Journal of Comparative Psychology, 110,
3–14.
Author Index
Adolphs, R., 65–66, 67, 78 Baron-Cohen, S., 67, 72, 78, 83, 111,
Aggleton, J. P., 36, 49 103
Agnes, M. M., 240 Barrett, B., 244, 269
Agnetta, B., 109, 112 Bates, E., 88, 104, 111
Akerman, C., 290, 295 Baudonniere, P., 63, 78
Alexander, M. P., 37, 56, 66, 83 Bauer, P. J., 128–29, 137
Allen, J., 43, 56, 226 Bayley, P. J., 192, 220
Allendoerfer, K. R., 74, 82, 156, 253, Beck, B., 148, 155
271, 273, 295, 302, 320 Becker, J. T., 7, 53
Alvarez, P., 36, 49 Bellows, N., 317, 319
Alvarez-Royo, P., 293 Bem, D. J., 159, 171
Amsterdam, B. K., 63, 78 Benjamin, A. S., 299, 318
Andersen, S. M., 159, 170 Bennett, R. T., 32, 54
Anderson, J. R., 72, 81, 155, 324, 325, Bennetto, L., 76, 78
338 Benson, J. B., 17, 51
Anes, M. D., 227, 240 Beran, M. J., 192, 198, 208, 217, 220,
Angell, F., 246, 251, 269 230, 239
Ansay, C., 73, 78 Beran, M. M., 192, 220
Applebee, A. N., 134, 137 Bergson, H., 16, 49
Arbeit, W. R., 244, 269 Bering, J. M., 323, 338
Arbuckle, T. Y., 298, 318 Bernati, T., 25, 54
Aristotle, 188, 219 Bertenthal, B. I., 118, 138, 162, 171
Asch, S. E., 160, 170 Bever, T. G., 88, 114
Asendorpf, J. B., 63, 78 Bhatt, R. S., 43, 56, 226
Astington, J. W., 120, 137 Bickerton, D., 220
Astur, R. S., 230, 234 Bidell, T., 117, 138
Atance, C. M., 17, 40, 49 Bierschwale, D. T., 86, 113
Au, W. W., 244, 269 Bischof, N., 220
Bjork, E. L., 230, 239
Baker, L. A., 250–251 Bjork, R. A., 78, 230, 239, 299, 318
Baker, M., 214, 220, 271 Bjorklund, D. F., 323, 338
Balda, R. P., 219 Black, S. E., 52, 54
Baldwin, J. M., 323, 338 Blades, M., 74, 79
Balfour, S. P., 242, 271 Blaha, J., 285
Bandura, A., 163, 165, 170, 171 Blake, M., 73, 78
Bard, K. A., 323, 339, 341 Bloom, L. M., 88, 111
343
344 Author Index
Boesch, C., 217, 220 223, 237, 290, 291, 292, 294, 295,
Boesch-Achermann, H., 217, 220 321, 322, 323, 324, 327–328, 330,
Bookheimer, S. Y., 38, 50 333–334, 335, 337, 338, 339, 341
Boring, E. G., 246, 269 Callanan, M. A., 132, 138
Botkin, P. T., 163, 171 Camak, L., 323, 341
Botto, C., 65, 79 Cameron, P. A., 339
Boucher, J., 74, 76, 78, 79 Campione, J. C., 242, 269
Bowlby, J., 158, 171 Campos, J. J., 118, 138
Boysen, S. T., 191, 221 Cant, J. G. H., 295
Braggio, J. T., 231, 239 Capps, L., 120, 138
Braine, M. D. S., 88, 111 Carpenter, A. F., 96, 111, 290, 291,
Braitenberg, V., 143, 155 322, 323, 327–328, 330
Brakke, K. E., 198, 217, 220, 222, 230, Carpenter, M., 292, 294, 339
239 Carruthers, P., 103, 111
Brannon, E., 90, 114, 240 Carver, C. S., 165, 168, 171
Bransford, J. D., 242, 269 Cary, M., 300, 317, 318
Bräuer, J., 333–34, 338 Case, R., 147, 155, 163, 165, 166, 171
Brédart, S., 25, 56 Cass, J., 239
Bretherton, I., 87, 111 Castro, C. A., 253, 269
Briese, T. L., 240 Ceci, S. J., 132, 138
Briggs, G. E., 285, 294 Cermak, L. S., 14, 50, 278, 295
Brooks-Gunn, J., 31, 52, 63, 80, 162, Chalmers, D. J., 15, 50, 293, 294
172, 323, 340 Chan, R. L., 76–77, 80
Brown, A. L., 242, 269 Chen, S., 90, 114
Brown, A. S., 246, 250, 269 Chittka, L., 222
Brown, J. M., 242, 271 Choi, S., 129, 138
Brown, M. W., 36, 49, 335, 339 Chomsky, N., 84, 85, 97, 101, 103,
Brown, R., 87, 111 112, 120, 138, 143
Brown, W., 269 Church, R. M., 88–89, 114
Bruck, M., 132, 138 Cieplucha, C., 239
Bruner, J. S., 86, 103–104, 111, 114, Clark, R. E., 272, 294
134, 137, 171 Clayton, N. S., 7, 10, 17, 37, 38, 41,
Buchanan, J. P., 231, 239 45–46, 47, 50, 51, 149, 155, 191,
Buckner, R. L., 10, 54 192, 220, 225, 226, 228–229, 232,
Budwig, N., 129, 137 239, 240, 274, 294
Bugiani, O., 25, 51 Clement, T. S., 43, 56, 226
Bullock, P. R., 67, 82 Clinton, B., 176, 177
Bundy, R. A., 64, 79 Cochran, B., 214, 220
Burke, D., 239 Cohen, D. H., 103, 111
Bussey, T. J., 36, 41, 50, 53 Cohen, N. J., 12, 36, 50, 272, 294
Butterfield, E. C., 73, 78 Cohn, R. H., 65, 81, 324, 330, 340
Byrne, R. W., 72, 78, 83, 103, 111, Colombo, M., 96, 111
163, 166, 173 Colon, M. R., 43, 54, 192, 223, 226,
240
Cabaret, M., 25, 54 Combs, A. W., 173
Cabeza, R., 52, 53, 226, 239 Connelly, A., 56
Calabrese, P., 25, 49, 52 Conway, M. A., 33–34, 50, 239
Call, J., 63, 64, 65, 69, 75, 78, 83, 103, Cook, R. G., 36, 56, 253, 271
109, 111, 112, 114, 119, 141, 184, Cooley, C. H., 158, 160, 171
Author Index 345
Corballis, M. C., 13, 17, 38, 39, 40, Drummey, A. B., 32, 50
42, 54, 62, 83, 191, 223, 226, 241 Ducker, G., 324, 340
Corkin, S., 272, 294 Dunbar, R., 103, 112
Cornil, V., 56 Dunham, P. J., 103, 113
Coslett, H. B., 37, 53 Dunlosky, J., 81, 242, 269
Cosmides, L., 71, 78 Dunning, D., 236, 239
Costermans, J., 73, 78 Durkheim, E., 160
Courage, M. L., 34, 51, 128, 130, 139 Durwen, H. F., 49, 52
Coussi-Korbel, S., 72, 79 Duval, S., 30, 50
Cowey, A., 247, 269 Düzel, E., 38, 50
Coyette, F., 25, 56
Creelman, C. D., 255, 269 Easton, A., 290, 294
Cross, D., 66, 83 Eddy, T. J., 69, 82, 105–109, 113
Cuddy, L. L., 298, 318 Egnor, R., 269, 320, 341
Cultice, J. C., 73, 79 Ehlers, S., 79
Custance, D. M., 323, 339, 341 Eichenbaum, H., 10, 12, 36, 50, 226,
240, 272, 294
Dalla Barba, G., 17, 50 Eickhoff, A. E., 96, 111
Damasio, A., 116, 123, 128, 138 Einstein, G. O., 214, 220
Damon, W., 171 Eldridge, L. L., 38, 50
Darwin, C., 3, 5, 18–19, 35, 49, 50, Emery, N. J., 41, 50, 191, 220
84–85 Engel, S. A., 38, 50
Davis-Dasilva, M., 323, 341 Epstein, R., 339
Deacedo, B. S., 272, 294 Epstein, S., 171
de la Sayette, V., 51 Epstein, W., 299, 319
DeLeonardis, D. M., 214, 221 Erb, L., 269, 320, 341
DeLong, G. R., 67, 79 Erbetta, A., 25, 51
Dennett, D. C., 58, 79, 144, 149, 150, Esbensen, B. M., 32, 54
155, 156, 185, 186, 273, 293, 294 Estes, W. K., 147, 155
Depew, D. J., 117, 138 Eustache, F., 51
Depoorter, N., 25, 56 Evans, P. D., 144
Descartes, R., 57–58, 79, 142, 293 Evans, S., 43, 54, 75, 82, 155, 192,
De Schonen, S., 63, 82 223, 225, 226, 227, 232, 233, 235,
Desgranges, B., 51 240, 241, 324, 330, 341
de Waal, F. B. M., 72, 79, 180, 186
Dewey, J., 267, 269 Farrant, A., 74, 79
Dewsbury, D. A., 225, 239 Feffer, M., 163, 171
Dickinson, A., 7, 10, 37, 41, 46, 50, Fegley, S., 323, 339
51, 149, 191, 192, 220, 225, 226, Ferguson, S. A., 214, 221
228–229, 239, 240, 274, 294 Fernberger, S. W., 246, 250, 251, 269
Dickinson, P., 239 Ferrara, R. A., 242, 269
Dixon, R. A., 73, 80 Ferrari, M., 64, 79
Dolan, R., 79 Fischer, K. W., 116, 138, 162, 165,
Donald, M., 62, 79, 102, 112, 116, 166, 171
121–123, 130–131, 133, 135, Fischhoff, B., 303, 319
137n3–4, 138, 190, 220 Fitch, W. T., 84, 112
Donne, J., 155 Fivush, R., 30, 33, 50, 53, 124, 125,
Dorus, S., 155 138, 140
Dropik, P. L., 128, 137 Flagg, S. F., 189, 222
346 Author Index
Flavell, J. H., 124, 138, 163, 171, 242, Gillberg, C., 79

269, 290, 294, 296, 318 Giovagnoli, A. R., 25, 51
Fodor, J. A., 120, 138 Glenberg, A. M., 299, 313, 319
Fogleman, T., 86, 115 Gluckman, A., 341
Fontaine, R. G., 134, 138 Godfrey, O., 25, 54
Forestell, P. H., 196, 221 Goldman, S., 140
Fortin, N. J., 226, 240 Goldsmith, M., 73, 80
Frackowiak, R., 79 Gómez, J. C., 323, 341
Fraser, A. C., 186n4 Goodall, J., 45, 51
Frazier, L. D., 235, 240 Gopnik, A., 31, 32, 51, 103, 112, 120,
Freedman, M., 212, 223 129, 138, 339
French, J. A., 240 Gould, S. J., 40, 84–85, 112, 119,
Freud, S., 160, 161 137n1, 138
Friedrichs, A. G., 124, 138 Graf, P., 31, 32, 51
Friston, K. J., 56 Gray, J. A., 267, 269
Frith, C., 79 Green, M. L., 163, 172
Frith, U., 65, 67, 71, 74, 78, 79, 82 Griffiths, D. P., 10, 37, 51, 191, 220,
Fry, C. L., 163, 171 225, 228, 239, 240
Fry, G., 239 Gross, C. G., 96, 111
Fung, H., 133, 139 Gruendel, J., 124, 140
Furmanski, C. S., 38, 50 Gruneberg, M. M., 73, 79
Guillery, B., 37, 51
Gabrieli, J. D. E., 278, 295 Gupta, R., 65, 79
Gadian, D. G., 25, 36, 52, 56, 192, 222
Gaffan, D., 10, 51, 192, 220, 290, 294 Habib, R., 53
Gallistel, C. R., 196, 220 Haden, C. A., 125, 140
Gallup, G. G., 63, 64, 65, 66, 79, 83, Haith, M. M., 17, 51
191, 220, 324, 325, 330, 338, 339, Hamann, S. B., 37, 51, 272, 294
341 Hamilton, J. M. E., 96, 112
Gao, F., 54 Hampton, R. R., 19–20, 21, 41, 75,
Garber, P., 220 79, 142, 149, 156, 192, 220, 253,
Gardiner, J. M., 33, 38, 51, 227–228, 278, 291, 292, 294, 302, 303, 317,
240 319, 322, 326–327, 336, 339
Gardner, B. T., 86–87, 88, 98, 100, Han, J. J., 125, 138
112 Handelmann, G. E., 7, 53
Gardner, R. A., 86–87, 88, 98, 100, Happe, F., 65, 66–67, 79
112 Harbluk, J. L., 32, 54
Garrett, M., 65, 79 Hardin, C., 161, 171
Garry, M., 225, 236, 240 Hare, B. H., 69, 83, 86, 109–110, 112,
Gehlen, W., 49, 52 115, 335, 339
Geiger, K., 222 Harlow, H. F., 112, 269
George, S. S., 246, 269 Harper, D. M., 225, 236, 240
Georgopolous, A. P., 96, 111 Harris, P. L., 120, 137, 139
Gerard, H. B., 161, 172 Hart, D., 323, 171, 339
Gerhand, S., 37, 52 Hart, J. T., 73, 79, 242, 269, 298, 299,
Gerler, D., 298, 319 319
Gilbert, S. L., 155 Harter, S., 162, 171
Gill, T. G., 231, 239 Hashiya, K., 208, 221
Gill, T. V., 87, 114 Hashtroudi, S., 214, 221
Author Index 347
Haupts, M., 49, 52 Jacoby, L. L., 240

Hauser, M. D., 65, 79, 84, 112 James, W., 15, 16, 51, 125–126, 139,
Hayes, C., 86, 112 160, 189, 215, 217, 221, 267, 269,
Hayes, K., 86, 112 296, 319
Hayman, C. A. G., 23, 24, 37, 51, 55 Janowsky, J. S., 74, 80, 296, 319
Hayne, H., 36, 54, 129, 140 Janson, C. H., 221
Heider, F., 66, 80 Jarvis, P. E., 163, 171
Heinze, H. J., 38, 50 Java, R. I., 227–228, 240
Henseler, S., 131, 140 Jenkins, H. M., 274, 294
Herman, L. M., 196, 213, 221, 244, Joaquim, S. G., 74, 81, 299, 319
269, 271 Joerges, J., 222
Hertzog, C., 73, 80 Johansson, M., 79
Heyes, C. M., 63, 80, 103, 112, 321, Johnson, C. N., 33, 51
324, 340 Johnson, M. K., 31, 52, 214, 221
Hicks, R. E., 154, 156 Jolly, A., 72, 80
Higgins, E. T., 58, 158, 161, 162, 164, Jones, E. E., 161, 172
165, 167, 169, 171, 172, 173 Jorgensen, M. J., 325, 340
Hill, S. D., 64, 79, 81 Joyce, J., 185–186n3, 186
Hiroshige, A., 60, 61 Juno, C., 222
Hobson, P. R., 67, 80, 103, 112
Hochstein, S., 96, 113 Kagan, J., 129, 139
Hodges, J. R., 37, 51 Kamil, A. C., 219, 292, 294
Hoffman, H. G., 236, 240 Kandel, E. R., 12, 16, 54
Hoffman, M., 235, 240 Kant, I., 183–184, 185, 186
Hofstadter, D. R., 143, 144, 148, 156 Kaplan, B., 173
Hogrefe, G.-J., 31, 56 Kapur, N., 25, 51
Holinka, B., 49 Karoly, P., 267, 269
Hollins, T. S., 303, 319 Katis, S., 51
Hood, L., 88, 111 Kaufmann, L., 148
Hoogstra, L., 133, 139 Kawai, N., 191, 192, 221
Hopkins, R. O., 37, 52 Kawanishi, D. T., 246, 271
Hopkins, W. D., 196, 222, 324, 325, 340 Kelley, J. W., 198, 220, 230, 239, 240
Houle, S., 53 Kellogg, L. A., 86, 112, 217, 221
Howe, M. L., 34, 51, 128, 130, 139 Kellogg, S. C., 246, 271
Hoyt, J. D., 124, 138 Kellogg, W. N., 86, 112, 217, 221
Hrdy, S. B., 121, 139 Kendrick, D. F., 36, 56, 253, 271
Hudson, J., 30, 50 Kermoian, R., 118, 138
Hughes, C., 67, 80 Kihlstrom, J. F., 26, 52
Hume, D., 183, 186 Kinsbourne, M., 146, 149, 150, 151,
Humphrey, N. K., 58–59, 62, 73, 80, 153, 154, 155, 156
175, 186 Kircher, T. T. J., 15, 51
Hunter, W., 193, 216, 221 Kitchener, E. G., 37, 51
Hyatt, C. W., 324, 340 Klein, S. B., 17, 26, 28, 52, 76–77, 80
Klin, A., 66, 80
Ingvar, D. H., 17, 51 Knight, R. T., 67, 83
Inhelder, B., 166, 172 Knowlton, B. J., 12, 38, 50, 52, 177,
Inman, A., 275, 278, 290, 291, 294 186, 272, 294
Itakura, S., 325, 340 Kober, H., 19, 26, 145, 297
Ivanoiu, A., 56 Kohler, S., 54
348 Author Index
Köhler, W., 39, 52, 150, 156, 190, Lindsay, D. S., 31, 52
192–193, 216, 221 Liu, K., 65, 79
Kojima, S., 208, 221 Locke, J., 175–179, 180, 181–183, 186
Kopelman, M. D., 37, 52 Lockhart, R. S., 18, 30, 52
Koriat, A., 73, 74, 80, 242, 269, 303, Loftus, E. F., 236, 240
317–318, 319 Loftus, J., 17, 26, 28, 52, 76–77, 80
Kornell, N., 75 Lories, G., 73, 78
Koseff, P., 37, 56 Lucas, P. A., 246, 271
Kraemer, P. J., 253
Kralik, J., 65, 79 Macdonald, C. A., 23, 24, 51, 55
Kruger, A. C., 323, 341 MacDonald, S. E., 238
Kuhlmeier, V. A., 191, 221 MacMillan, N. A., 255, 269
Kunze, J., 222 Macphail, E. M., 5, 52
Kwon, P., 31, 52 Madigan, S. A., 296, 301, 320
Malcolm, C. M., 155
Lachman, J. L., 73, 80 Mandler, J. M., 128–129, 137
Lachman, R., 73, 80 Manns, J. R., 37, 52
Ladygina-Kohts, N. N., 86, 112, 217, Manzi, M., 214, 220
221 Marino, L., 292, 295
Lahn, B. T., 155 Markowitsch, H. J., 9, 10, 12, 25, 49,
Lamborn, S. D., 166, 171 52, 55, 190–191, 218, 223, 226, 241
Landau, K. R., 26, 53 Markus, H., 159, 172
Landwehr, R. S., 73, 81 Marsh, E. J., 10, 54
Langley, C. M., 292, 295 Matsuzawa, T., 96, 111n1, 113, 191,
Lanza, R. P., 339 192, 221, 340
Larsen, T., 253, 269 Matthews, W. S., 64, 79
Lashley, K. S., 146, 156, 182, 186 Maunthner, N., 71, 82
Lawson, J., 191, 197, 222 May, D., 158, 172
Leahey, T. H., 6, 52 McAndrews, M. P., 25, 56
Lecky, P., 172 McCann, E., 178, 186n5, 186
Lee, S. C., 10, 54 McCarthy, R., 37, 51
Leekam, S., 67, 82 McClure, M. K., 64, 79
Leibniz, G., 180–181, 186 McDaniel, M. A., 214, 220
Leichtman, M. D., 125, 138 McDonald, K., 196, 197, 222, 223
Leonesio, R. J., 73, 80, 81 McDonald, S. E., 240
Lepage, M., 227, 241 McGee, K., 269, 320, 341
Leslie, A. M., 67, 78, 80, 82, 120, 139 McGuire, W. J., 172
Lethmate, J., 324, 340 McIntosh, A. R., 52, 53
Leube, D. T., 15, 51 McKenna, P., 37, 52
Levine, B., 25, 52 McLachlan, D. R., 23, 32, 54, 56, 212,
Lewin, K., 160, 161 223
Lewis, M., 31, 52, 80, 123, 129, 139, McMillan, C. T., 25, 56
162, 172, 323, 325, 340 McNew, S., 88, 111
Lewis, V., 63, 76, 78 Mead, G. H., 125, 129, 139, 158, 160,
Lichtenstein, S., 303, 319 161, 172
Lieberman, P., 86, 112, 113 Meck, W. H., 274, 294
Liess, J., 52 Meissner, C. A., 235, 240
Linden, E., 87, 98, 113 Meltzoff, A. N., 120, 138, 339
Author Index 349
Menzel, C. R., 45, 52, 192, 197–215, Nagell, K., 323, 340
217, 221, 226, 228, 230–231, 232, Narens, L., 60, 62, 73, 81, 269, 297–
234, 240, 289, 290, 295 298, 303, 319
Menzel, E. W., Jr., 189, 191, 194–196, Neisser, U., 139, 321–322, 340
197, 213, 221, 222 Nelson, K., 13, 26, 30, 31, 33, 37, 50,
Menzel, R., 222 53, 58, 60, 62, 66, 75, 81, 88, 123,
Metcalfe, J., 19, 26, 73, 74, 75, 80, 81, 124, 125, 130, 131, 134, 137n2, 139,
82, 145, 242, 269, 297, 299, 301, 140, 151
313, 319, 320 Nelson, T. O., 73, 78, 80, 81, 242, 243,
Meulemans, T., 56 269, 278, 287, 295, 297–298, 303,
Mikami, A., 111n1, 113 317, 319
Milbrath, C., 323, 340 Neuman, C. J., 64, 81
Miles, H. L. W., 87, 88, 113, 323, 340 Newcombe, N. S., 32, 50
Miller, G. W., 154, 156 Nisbett, R. E., 182, 186, 317, 319
Miller, P. J., 132–133, 139 Nishimura, T., 111n1, 113
Miller, S. M., 163, 172 Nissen, H., 213, 222
Miller, T. C., 65, 79 Noelle, D. C., 15, 53
Minda, J. P., 256, 269 Nyberg, L., 9, 10, 53
Miner, A. C., 73, 81, 299, 313, 319
Mintz, J., 133, 139 Oakes, L. M., 132, 138
Mischel, W., 163, 172 O’Connor, M., 14, 50
Mishkin, M., 25, 36, 38, 50, 52, 56, Olguin, R. S., 323, 340
192, 222, 288, 295 Ollendick, T. H., 72, 81
Mitchell, R. W., 72, 81, 322–323, 340 Olson, D., 120, 137
Monks, J., 73, 79 Olton, D. S., 7, 53, 192, 222
Moore, B. R., 163, 274, 294 O’Neill, D. K., 17, 40, 49
Moore, C., 103, 113 Orlov, T., 96, 113
Moore, P. W., 244, 269 Oser, J., 65, 79
Moretti, M. M., 167, 172 Oswald, D. P., 72, 81
Morgan, C. L., 35, 41, 44, 52, 257, Otto, T., 36, 50
269 Ozonoff, S., 67, 81
Morris, C. C., 67, 81, 299, 319
Morris, R. G. M., 43, 52, 73, 82, 225, Pack, A. A., 213, 221
226, 240 Padawer-Singer, A., 172
Moscovitch, M., 23, 24, 25, 54, 56, Palmer, M., 213, 221
212, 223 Pappas, B. C., 7, 53
Mowrer, O. M., 98, 113 Parfit, D., 177, 186
Mukobi, K. L., 191, 221 Parker, A., 290, 294, 295
Mullen, M. K., 125, 139 Parker, S. T., 323, 340
Muller, U., 222 Parkinson, J. K., 192, 222
Murphy, J., 222 Pate, J. L., 197, 222, 239, 268, 269
Murray, E. A., 36, 52, 53, 192, 222, Patterson, C. J., 163, 172
288, 290, 295 Patterson, F. G. P., 65, 81, 87, 98,
Murray, H. A., 161 113, 324, 330, 340
Myowa-Yamakoshi, M., 323, 340 Pavlov, I., 213, 222
Pearce, J. M., 36, 49
Nadel, L., 9, 53 Peck, V., 73, 78
Nagel, T., 118, 139 Pellizzer, G., 96, 111
350 Author Index
Pennington, B. F., 17, 51, 67, 76, 78, Riley, D. A., 292, 295
81 Ritter, F. E., 299, 313, 320
Perfect, T. J., 303, 319 Roberts, A. C., 190, 192, 222
Perilloux, H. K., 26, 53 Roberts, R. J., Jr., 17, 51
Perlmutter, M., 124, 140 Roberts, S., 54
Perner, J., 13, 31, 32, 53, 56, 66, 67, Roberts, W. A., 38, 39, 53, 54, 222,
81, 82, 83, 338, 340 225, 226, 240, 253
Persson, J., 53 Rocissano, L., 88, 111
Petitto, L. A., 88, 114 Rodriguez, I. A., 43, 54, 192, 223,
Piaget, J., 126, 140, 146, 163, 165, 166, 226, 240
172 Roediger, H. L., III, 10, 33, 38, 53, 54,
Pillemer, D. B., 124, 140 177, 186
Pinker, S., 98, 113 Rogers, S. J., 67, 76, 78, 81
Pittman, T. S., 162, 173 Romanes, G. J., 35, 54
Piven, J., 67, 78 Rosenbaum, R. S., 23, 26, 54
Platt, M. L., 240 Rosenbaum, S., 197, 222
Plesa, D., 131, 140 Rosenblum, L., 237–238
Pleydell-Pearce, C. W., 33–34, 50 Ross, G., 140
Poldrack, R. A., 12, 50 Rousseaux, M., 25, 54
Polkey, C. E., 67, 82 Rovee-Collier, C., 36, 54
Potts, R., 113, 133, 139 Rowe, A. D., 67, 82
Povinelli, D. J., 26, 35, 53, 63, 68–69, Rozin, P., 148, 156
70, 82, 85, 86, 105–109, 113, 162, Rubert, E., 196, 222
165, 173, 295, 330, 341 Ruble, D. N., 165, 173
Premack, A. J., 69, 82 Ruffman, T., 13, 32, 53
Premack, D., 68, 69, 82, 86, 87, 90, Ruggiero, F. T., 189, 222
91, 96, 98, 99, 102–103, 105, 113, Rulf, A. B., 86, 113
114, 119, 140, 184, 186, 213, 222 Rumbaugh, D. M., 87, 90, 91, 96, 98,
Presler, N., 140 114, 197, 198, 216, 217, 220, 222,
Priel, B., 63, 82 223, 230, 239, 249, 268, 269, 271
Prince, C. G., 86, 113 Russell, B., 82
Pruvo, J. P., 25, 54 Russell, F., 239
Pugh, E., 217 Russell, J., 57, 67, 71, 80, 82
Ryle, G., 159, 173
Quine, W. V., 183, 186
Sacks, O. W., 74, 82
Raby, P. R., 250–251, 271 Salmon, E., 56
Ragan, P., 323, 338 Salmond, C. H., 56
Rajaram, S., 33, 37, 38, 53, 177, 186 Sanchez, I. C., 43, 54, 192, 223, 226,
Ramsey, D., 139 240
Ramus, S., 272, 294 Sanders, R. J., 88, 114
Reder, L. M., 73, 81, 299, 300–301, Sands, S. F., 36, 56, 253, 269, 271
313, 317, 318, 319, 320 Sanford, A. J., 96, 112
Reed, S. K., 73, 82 Sanocki, T., 299, 319
Reese, E., 125, 132, 140 Santiago, H. C., 36, 56, 253, 271
Reiss, D., 292, 295 Savage-Rumbaugh, S., 99–100, 114,
Richardson, W. K., 239 191, 196–198, 217, 220, 221, 222,
Richardson-Klavehn, A., 33, 38, 51 223, 230, 239, 323, 341
Ricks, M., 63–64, 83 Scaife, M., 86, 114
Author Index 351
Schacter, D. L., 9, 10, 14, 23, 32, 54, Somerville, S. C., 73, 79
56, 73, 82, 190, 192, 212, 223, 228, Son, L. K., 75, 82, 90
240, 272, 295 Spiker, D., 63–64, 83
Scheier, M. F., 165, 168, 171 Squire, L. R., 12, 16, 28, 36, 37, 49,
Schlosberg, H., 218, 224 51, 52, 54, 74, 80, 177, 186, 192,
Schneider, W., 151, 156, 262, 269 220, 223, 272, 288, 293, 294, 295,
Schroeder, B., 88, 114 296, 319, 320
Schull, J., 249, 260, 269, 271, 302, 320, Statham, A., 173
322, 341 Stern, L. B., 239
Schunn, C., 300–301, 317, 320 Stoerig, P., 247, 269
Schusterman, R. J., 244, 269 Stoinski, T. S., 323, 341
Schuz, A., 143, 155 Stone, V. E., 67, 83
Schwartz, B. L., 15, 43, 45, 74, 81, 82, Straub, R. O., 90, 114
180, 192, 223, 225, 226, 228, 233– Strauman, T. J., 169, 173
236, 237–238, 240, 242, 269, 298, Strote, J., 269, 320, 341
299, 313, 318, 319, 320 Stryker, S., 173
Sears, L., 67, 78 Stuss, D. T., 9, 52, 56, 57, 66, 74, 75,
Selman, R. L., 163, 166, 173 83
Sevcik, R. A., 196, 217, 222 Suarez, S. D., 65, 83, 324, 341
Shanker, S. G., 114 Suddendorf, T., 13, 17, 38, 39, 40, 42,
Shantz, C. U., 173 54, 62, 83, 191, 223, 226, 241
Sharpe, S., 71, 82 Sullivan, H. S., 158, 173
Shaughnessy, J. J., 303, 320 Suomi, S. J., 325, 340
Sherry, D. F., 9, 14, 54, 223, 241 Sutton, J. E., 278
Shettleworth, S. J., 5, 54, 191, 223, Suzuki, J., 111n1, 113
273, 275, 278, 290, 291, 294, 295 Suzuki, W. A., 25, 52, 192, 222
Shields, W. E., 74, 82, 156, 249, 253, Swartz, K. B., 90, 114, 232, 241, 324,
259–260, 265, 269, 270, 273, 295, 330, 341
302, 320, 322, 341
Shiffrin, R. M., 151, 156, 262, 269 Tager-Flusberg, H., 76, 83, 103, 111
Shimamura, A. P., 73, 74, 80, 81, 242, Taylor, M., 32, 54
269, 296, 297, 301, 319, 320 Taylor, T. J., 114
Shulman, C., 71–72, 83 Teixidor, P., 323, 341
Simcock, G., 129, 140 Terrace, H. S., 58, 87, 88, 89, 90, 91,
Simmel, M., 66 98, 99, 100, 114, 142, 145, 238, 277,
Simon, H. A., 73, 82 295
Sinden, M., 52 Thompson, C. R., 88–89, 114
Skinner, B. F., 85, 98, 114, 339 Thomson, G. H., 246, 251, 271
Smith, J. D., 74, 82, 152, 156, 237, Thorndike, E. L., 191, 216, 219, 223
244–246, 247–248, 250–251, 252, Thronesbury, C., 73, 80
253, 254–255, 256, 259, 260, 266, Tidswell, T., 71, 82
269, 270, 273, 275, 290, 291, 295, Tinklepaugh, O. L., 191, 193–194,
302, 320, 322, 325, 326, 341 210, 223
Smith, P. K., 103, 111 Tolman, E. C., 248, 267, 268, 271
Smith, S. M., 242 Tomasello, T., 63, 64, 65, 69, 70, 78,
Smith, S. T., 197, 222 83, 86, 103, 109, 111, 112, 114, 115,
Snyder, L., 88, 111 119, 120, 140, 141, 162, 165, 173,
Snygg, D., 173 223, 292, 295, 321, 323, 333–334,
Solomonica-Levi, D., 71–72, 83 335, 337, 338, 339, 340, 341
352 Author Index
Tooby, J., 71, 78 Wehner, R., 219, 224

Toth, J. P., 52 Weiskrantz, L., 5, 56, 247, 271, 273,
Townsend, J. T., 285, 295 275, 295
Tulving, E., 6, 7, 9, 10, 12, 13, 14, 15, Wellman, H. M., 33, 51, 66, 73, 79,
17, 18, 24, 26, 27, 38, 40, 42, 48, 51, 83, 120, 138
52, 53, 54, 55, 56, 57, 60, 62, 75, 83, Wenner, J. A., 128, 137
116, 121, 130, 136, 137n3, 141, 142, Werner, H., 163, 173
149, 168, 176–177, 179, 187, 188, Westmacott, R., 24, 54, 56
190–191, 192, 196, 211, 212, 215, Wewerka, S. S., 128, 129, 137
218, 223, 226, 227, 228, 238, 241, Wheeler, M. A., 9, 10, 12, 25, 29, 30,
272, 273, 289, 295, 296, 301, 320 56, 57, 62, 75, 83, 216, 224, 226,
Turner, R. H., 164, 173 241
White, S. H., 124, 140
Underwood, B. J., 298, 320 Whiten, A., 62, 72, 78, 83, 323, 339,
Ure, N., 323, 341 341
Wicklund, R. A., 30, 50
Vallender, E. J., 155 Widlitzek, H., 49
Van der Linden, M., 25, 37, 56 Wiebe, D., 73, 81
Vander Wall, S. B., 223 Wilding, E., 290, 295
Van Hook, E., 169, 173 Williams, G. C., 84–85, 115
Van Paeschen, W., 56 Williams, S. L., 222
Van Zandt, T., 285, 295 Williamson, C., 335, 339
Vargha-Khadem, F., 25, 36, 38, 50, Wilson, T. D., 182, 317, 319
52, 56, 192, 222 Wimmer, H., 31, 56, 67, 83
Vauclair, J., 213, 223 Winitz, H., 98, 115
Verfaellie, M., 37, 56, 228, 240, 278, Woodruff, G., 68, 82, 102–103, 114,
295 119, 213, 222
Viader, F., 51 Woodworth, R. S., 218, 224, 246, 251,
Viskontas, I. V., 25, 56 271
von Glasersfield, E. C., 87, 114 Wrate, J. L., 323, 341
Vonk, J., 35, 53, 70, 82 Wright, A. A., 36, 56, 253, 269, 271
Vrba, E. S., 84–85, 111 Wright, J. W., 163, 171
Vygotsky, L., 132, 141, 161
Yakovlev, B., 96, 113
Wagner, A. D., 10, 54, 278, 295 Yerkes, D. N., 193, 224
Walkenfeld, F. F., 140 Yerkes, R. M., 193, 210, 216, 224
Walters, R. H., 163, 171 Yi, S., 139
Wang, Q., 125, 138 Yirmiya, N., 71–72, 83
Warkentin, V., 63, 78 Yu, K. S., 226, 239
Warrington, E. K., 76, 78 Yunger, J. L., 323, 338
Washburn, D. A., 74, 82, 156, 216, Yunker, M. P., 244, 271
217, 230, 234, 241, 249, 253, 259,
269, 271, 273, 295, 302, 320, 322, Zelazo, P. D., 123, 135, 141
341 Zeman, A., 293, 295
Watkins, K. E., 56 Zentall, T. R., 43, 56, 226, 229–230,
Watson, C. S., 246, 271 232, 234, 241
Watson, J., 66, 83 Zohary, E., 96, 113
Weber, B. H., 117, 138, 160 Zola, S. M., 28, 36, 54
Weber, M., 161, 173 Zola-Morgan, S., 49, 288, 293, 295
Subject Index
Page numbers in italics refer to figures and tables.
accessibility mechanisms, of meta- and meta-confidence judgments,

cognition, 318 303–317
adaptation, and memory studies, and reinforcement contingency,
273–274 167–168
See also evolution and traditions in study of self,
American Sign Language (ASL), 86– 322–325
87, 100–101 See also apes; birds; dogs; dol-
amnesia phins; monkeys; primates;
and autobiographical memory, rats
134 apes
and neuropsychological evidence differences in cognition between
for episodic memory and au- gorillas and, 330
tonoesis, 24, 26, 28 and episodic memory, 232–236
animals evolution of compared to humans,
and continuity of self-awareness, 236–237
153–155 imitation and self-concept in,
debate on similarities and differ- 323–324
ences between humans and, and language projects, 97–101
3–5, 14–16, 47–49, 85, 110–111, and recognition of subjective limi-
136–37, 143, 162, 170, 188–192 tations, 184
and dominant configuration and seeking of information as met-
model of consciousness, acognitive index, 334
150–152 self and other in social cognition
episodic memory and autonoesis of, 336–337
in, 5, 6, 35–49, 228–232 social conditions of human in-
functional definition of memory fancy compared to, 120–121
awareness in, 275–276 See also chimpanzees; gorillas; pri-
and Locke’s theory of conscious- mates
ness, 180 Asperger’s disorder, and theory of
mental contents and behavioral ca- mind, 67
pability in study of cognition in, association theory
274 and uncertainty monitoring, 258–
and metacognition, 145–149, 244– 260, 265–266
257, 264–266 and vocabulary development, 98
353
354 Subject Index
auditory discrimination, and meta- brain

cognition in dolphins and hu- and continuity of consciousness,
mans, 244–248, 257, 266, 302, 150–152, 154
325–326 declarative memory and hippo-
autism campal system, 36
and deception, 71–72 episodic and semantic memory in
and episodic memory, 75–77 brain-damaged patients, 22–29
and false belief test, 146–147 frontal cortex and metacognition,
and metacognition, 74 74
and self-recognition, 63–64 prefrontal cortex and theory of
and theory of mind, 67, 68, 70 mind, 66–67
autobiographical memory
and brain injury, 24 card/food associations, and episodic
and infantile amnesia, 134 memory in gorillas, 233–234,
and social-cultural theory of devel- 237, 238
opment of episodic memory, children
33–34 and deception, 71–72
autonoesis development of autonoesis and
and concept of moral being, 19 episodic memory in, 29–35, 46
development of in children, 29–35 and escape from stimulus control,
and differences between humans 146–147
and animals, 5, 6, 47–49 experience and development of
and future, 17–18, 20–22 consciousness in, 118–119
neuropsychology of, 22–29 and gaze-following, 107
and self-consciousness in modern and information seeking, 334
philosophy, 176–177, 179 and metacognition, 74
and semantic memory, 15–16 recent research on memory in,
and time, 16–17 123–125
and reinforcement contingency,
becoming, and motivational perspec- 167–168
tive on self-consciousness, and self digest, 158
167–169 self-regulation, and shared reality,
behavioral capability, and animal 161–167
cognition, 274 and theory of mind, 119–120
betting paradigm, and meta- working memory and self-
confidence judgments in rhesus awareness in, 147
monkeys, 304–314, 316 See also autism; development;
birds, and episodic memory, infant(s) and infancy
228–230 chimpanzees
See also scrub jay and deception, 72
Bischof-Köhler hypothesis, and men- and episodic memory, 230–232
tal time travel, 39, 42 imitation and self-concept in, 323
bonobos and information seeking, 327–330,
and deception, 72 334, 335
and study of information seeking, and intentionality, 107–110
330–332 and learning of language, 86–90,
See also primates 99–100
boundary conditions, and episodic memory in symbol-competent,
memory, 147–149 196–198
Subject Index 355
and mental awareness of future, and projectable self, 62

45 See also autonoesis; self-awareness
and metacognition, 75 continuity, of self-consciousness
progress in research on memory definition of, 142
in, 215–219 and dominant configuration
review of studies on memory in, model of consciousness,
192–196 150–155
self and other in social cognition emergence of, in brain, 150
of, 336–337 and false negative demonstrations
and self-recognition, 64–65 of metacognitive skills, 145–149
and theory of mind, 69–70, 102– and inequalities in cognitive skills,
103, 105–107 143–144
See also apes; primates and modularity assumption, 143
choice phase stimuli, and study of and self-awareness, 149–150
memory in rhesus monkeys, 280 culture
cognition and autonoetic awareness, 21–22,
continuity theory and inequalities 40
of skills, 143–144 and cross-cultural differences in re-
and emergence of cognitive con- search on memory in young chil-
sciousness in young children, dren, 125
126, 128–130, 135, 136 and developmental perspective
other and self in social, 336–338 on evolution of consciousness,
and uncertainty monitoring, 123
262–264 and emergence of consciousness
See also memory; metacognition in young children, 126, 135–136
collective memory, and self- mimesis, and forms of communica-
consciousness in animals, 181 tion, 121–122
communication
and joint attention, 103–110 deception, and projectable self,
mimesis and cultural forms of, 71–73
121–122 decisional processes, and conscious
and vocalizations in chimpanzees, cognition, 266–268
200, 205, 214, 215 declarative memory
See also gestures; language; lexi- discrimination between implicit
grams; symbols memory and, 291
conceptual self, and comparative and features of episodic and se-
social cognition, 321–322 mantic memory, 12
confidence judgments, and metacog- and hippocampal system, 36
nition in rhesus monkeys, delayed nonmatching to sample
303–317 (DNMS) tasks, and research on
consciousness episodic memory, 36–37,
developmental perspective on, 326–327
116–119 development
emergence of in evolution, of consciousness in young chil-
119–123 dren, 116–119, 125–137
emergence of levels of, in young and emergence of consciousness
children, 125–136 in evolution, 119–123
language, self, and, 136–137 and lines of evidence for episodic
Locke’s understanding of, 175–179 memory, 29–35, 48
356 Subject Index
development (continued) relation of, to time, 16–17

and recent research on memory in and self-awareness, 145, 147
young children and infants, semantic memory compared to, 7,
123–125 10–14
self-awareness and continuing, and study of memory in chimpan-
153 zees, 218–219
of semantic memory, 48 uniqueness of, in humans, 14–16,
of theory of mind, 66 47–49
See also children; infant(s) and in- epistemology, and role of self-
fancy consciousness in modern philos-
direct-access mechanisms, and meta- ophy, 175, 183–185
cognition, 298–299, 317, 318 evolution
dogs, and study of information seek- of autonoesis and episodic mem-
ing, 333–336 ory, 8, 19–22, 40, 47
See also animals comparison of in humans and
dolphins, and research on uncer- apes, 236–237
tainty and metacognition, 244– and emergence of consciousness,
248, 257, 266, 302, 325–326 119–123
dominant configuration model, of and expansion of neural network
consciousness, 150–155 in behaviorally sophisticated
Down’s syndrome, and deception, species, 151
71 of language, 84–85, 97, 101–103,
110–111
ecological self, and comparative so- of memory awareness, 291–292
cial cognition, 321–322 of problem-solving skills, 86–97
empiricism, and self-consciousness of projectable self, 58–60
in modern philosophy, 182–185 of self-awareness, 144
enrichment activities, and study of exaptations, and evolution of lan-
memory in apes, 238 guage, 84–85
episodic memory expectant self, and self-regulation in
and boundary conditions, 147–149 children, 159, 166, 167, 169
definition of, 9–10, 226 experience, and development of con-
development of, in children, 29–35 sciousness in infants and young
and differences between humans children, 117–118, 125–128
and animals, 5, 6 expert systems, and cognitive devel-
evolution of, 8, 19–22 opment, 148
and evolution of consciousness, explicit mechanisms, and meta-
121 judgments, 301, 318
and future, 17–18 eyewitness memory, and studies of
and human cognition, 273 episodic memory, 235–236
human literature on, 226–228
neuropsychology of, 22–29 feeling-of-knowing judgments
and personal self, 60 (FKJs), and metacognition, 298,
present status of research on chil- 299
dren and, 125 forgetting, awareness of in rhesus
and projectable self, 75–77 monkeys, 282–287
question of in animals, 35–47, free-choice trials, and memory in rhe-
228–232 sus monkeys, 280–286, 303
Subject Index 357
future language and uniqueness of, 85

and autonoesis, 17–18, 26–28, 40 literature on episodic memory in,
and evolution of episodic mem- 226–228
ory, 20–22 memory and uniqueness of 14–16,
and question of episodic memory 188–192
in animals, 40, 44, 45 and metacognition, 73–74, 242–
See also mental time travel 243, 246–247, 252–257, 264–266,
296–400, 301–303
gaze motivational perspective on self-
autistic children and avoidance of, consciousness and uniqueness
147 of, 170
following of and theory of mind narrative consciousness and
in primates, 69, 105–107 uniqueness of infancy in,
See also mirror image 136–137
genetics, and cognitive differences and prompting in studies of recall
between primates and humans, memory, 213–214
144 self-awareness as unique to, 143
Georgia State University. See Lan- self-recognition and consciousness
guage Research Center in, 63–64
gestures, and study of memory in social referencing as unique to,
chimpanzees, 198, 202, 204, 215 162
gorillas and theory of mind, 65–67
and episodic memory, 232–236 and uncertainty response, 249–
and information seeking, 328, 330, 251
334 See also children; evolution; in-
mirror images and self-recognition fant(s) and infancy
in, 64–65, 324
See also apes; primates identity
grammar and cultural consciousness in
and evolution of language, young children, 135
101–103 Leibniz on moral, 180–181
of uncertainty monitoring, 264 Locke’s understanding of per-
See also vocabulary sonal, 178–179, 181
imitation, and study of self in ani-
humans mals, 322–324
cognitive differences between pri- implicit levels, of metacognition,
mates and, 144 300–301, 317, 318
debate on similarities and differ- implicit memory, and declarative
ences between animals and, memory in animals, 291
3–5, 14–16, 47–49, 85, 110–111, indirect cues, and memory in chim-
136–137, 143, 162, 170, 188–192 panzees, 208
and deception, 71–72 individual differences, in uncertainty
dependency on other humans as response, 251
key characteristic of, 157 infant(s) and infancy
and episodic memory, 5, 6, 47–49, and development of conscious-
273 ness, 118–119
and evolution of language, emergence of levels of conscious-
110–111 ness in, 125–136
358 Subject Index
infant(s) and infancy (continued) evaluation of through memory

recent research on memory in, awareness, 276
123–125 and self in social cognition,
uniqueness of social dimensions 337–338
of human, 136–137 Köhler Primate Research Center
See also children; development (Leipzig), 330
infantile amnesia, and study of mem-
ory, 124, 134 language
inferential mechanisms, and meta- and cognitive consciousness,
cognition, 298, 299–300 129–130
information and information pro- comprehension of spoken by chim-
cessing panzees, 217
encoding of, in semantic versus and episodic memory in animals,
episodic memory, 13–14 41–43
seeking of, in chimpanzees and evolution of, 84–85, 97, 101–103,
orangutans, 327–330 110–111
seeking of, as metacognitive in- individual memory and written,
dex, 334–336 122–123
semantic memory and retrieval of, and joint attention, 103–110
15–16 learning of, by chimpanzees, 86–
and uncertainty monitoring, 263– 90, 99–101
264, 267 mimesis and expansion of con-
inner eye, and metacognition, 58–59, sciousness, 122
62, 74 and reflective consciousness,
instrumental self, and self-regula- 130–133
tion, 159, 161–162, 166, 167 and research projects with apes,
intentionality 97–101
and evolution of language, self, consciousness, and, 136–137
102–103 and semantic memory, 12–13
and natural environment of chim- and theory of mind in children,
panzees, 107–110 120–121
internal sense, and Locke’s theory of and training of monkeys, 90–97
self-consciousness, 183 and uniqueness of humans, 85,
interpersonal significance, and levels 110–111
of self-consciousness in chil- See also communication; grammar;
dren, 163 vocabulary
language acquisition device (LAD),
judgments, and studies of metacogni- 84, 103, 104
tion, 298–318 language acquisition support system
(LASS), 103
Kanzi (chimpanzee), 99–100, Language Research Center (Georgia
197–198 State University), 196, 198, 230,
King (gorilla), 232–238 249
knowledge laser pointer, and study of memory
and concept of self-knowledge as in chimpanzees, 209–210
self digest, 159–160 learning, and development of epi-
cultural consciousness and acquisi- sodic memory in young chil-
tion of cultural, 135–136 dren, 32–33
Subject Index 359
lexigrams neurobiology of, in rhesus mon-

and learning of language by chim- keys, 290–291
panzees, 87, 89 psychological mechanisms of in
and studies of memory in chim- rhesus monkeys, 287–290
panzees, 196, 202–205, 231 simple model of, 276–278
literary consciousness, emergence of memory flag hypothesis, in rhesus
in young children, 136 monkeys, 288–289
mental contents, and study of ani-
mands, and language development, mal cognition, 274
98, 99 mental state attribution, and compar-
massed trials, and study of memory ative social cognition, 321
in chimpanzees, 211–212 mental time travel
memory and Bischof-Köhler hypothesis, 39,
episodic memory as functionally 42
different from other kinds of, and differences between animal
6–7 and human memory, 190
Leibniz on moral identity and, and episodic memory, 6–7, 14–15,
180–181 27, 29, 44, 45, 47, 145, 218, 226
and Locke’s theory of conscious- See also future
ness, 177–179, 181 metacognition
monitoring of, in serial probe rec- definition of, 242, 317
ognition (SPR) task, 252–257 and evolution of language, 110
progress in research on chimpan- and false demonstrations of skills
zees and, 215–219 in animals, 145–149
and recent research in young chil- information seeking as index of,
dren and infants, 123–125 334–336
review of prior studies on chim- and inner eye, 60, 62
panzees, 192–196 and judgments in animals,
in symbol-competent chimpan- 303–317
zees, 196–198 and memory awareness in rhesus
and uniqueness of humans, 14–16, monkeys, 287, 290
188–192 and projectable self, 62, 73–75
written language and individual, studies of, in animals, 244–268
122–123 studies of, in humans, 242–243,
See also autobiographical memory; 246–247, 252–268, 296–300,
declarative memory; episodic 301–303
memory; memory awareness; and study of the self, 325–334
semantic memory; working tacts and development of lan-
memory guage, 98–99
memory awareness See also cognition
critique of, in monkeys, 292– mimesis
293 and cognitive consciousness in
and discrimination between re- children, 130
membering and forgetting in and evolution of consciousness,
rhesus monkeys, 278–287 121–122
evolution of, 291–292 mirror image
and functional definition of aware- and development of autonoesis in
ness in animals, 275–276 infants, 30–31
360 Subject Index
mirror image (continued) Nim Chimpsky (chimpanzee), 87–

and gaze avoidance in autistic chil- 88, 89
dren, 147 nonconservation demonstrations,
and self-recognition in animals, and escape from stimulus con-
64–65, 233, 324–325 trol, 146
See also gaze normativity, and discussions of self-
misinformation study, and episodic consciousness in modern philos-
memory in gorillas, 236 ophy, 175
monitored self, and self-regulation null hypothesis, and self-knowing
in children, 159, 165, 167 consciousness in studies of
monitoring memory in chimpanzees,
of memory in serial probe recogni- 216–219
tion (SPR) task, 252–257
of uncertainty responses, 257–268 object-level metacognitive processes
monkeys and confidence judgments, 304
and consciousness of subjective and studies in humans, 297–298
limitations, 184–185 and uncertainty, 302
and deception, 72 object recognition tasks, and re-
and metacognitive studies, 74– search on episodic memory,
75, 253–255, 257–259, 266, 302, 36–37
303 operant conditioning, and learning
and self-recognition, 65, 324 of language, 85
simultaneous chains and training orangutans, and study of seeking in-
of, 90–97 formation, 327–330, 334
See also primates; rhesus monkeys See also primates
morality other, and self in social cognition,
and Darwin’s concept of moral be- 336–338
ings, 18–19
and role of self-consciousness in palinscopic memory
modern philosophy, 175–182 See episodic memory
motivational perspective, on self- Panzee (chimpanzee), 198–215, 217–
consciousness 219, 230–231, 234
and becoming, 167–169 perceptual categories, and self in so-
description of, 157–158 cial cognition, 337
and self digest, 158–160 philosophy
and shared reality, 160–167 and differences between animal
and uniqueness of humans, 170 and human memory, 188–
motor development, in human in- 189
fants, 118 normative roles of self-conscious-
mythic culture, and narrative con- ness in modern, 174–185
sciousness, 133 photographs, and studies of memory
in apes, 197, 235, 236
narrative physical consciousness, emergence
and emergence of consciousness of in young children, 126, 127
in young children, 126, 133–136 physical time, and episodic memory,
and research on memory in young 16–17, 27–29
children, 124 pigeons (Columbia livia), and studies
neurobiology, of memory awareness, of memory, 230, 236, 274, 278,
290–291 280, 289–291
Subject Index 361
primates reality, motivational perspective on

cognitive differences between hu- self-consciousness and shared,
mans and, 144 160–167, 169–170
and deception, 72–73 recognition, and studies of memory
and metacognition, 74–75 in chimpanzees, 193–196, 212,
and self-recognition, 64–65 214
and social intelligence, 85–86 See also self-recognition
and theory of mind, 68–71 reflective consciousness, emergence
See also apes; bonobos; chimpanzees; of in young children, 126, 130–
gorillas; monkeys; orangutans 133, 135, 136
problem-solving skills rehearsal, in nonhuman primates,
evolution of, 86–97 234
and self-awareness, 145–146 representational language, and sym-
projectable self bolic systems, 130, 133
and deception, 71–73 resource allocation, and motivational
definition of, 57–58 perspective on self-consciousness
and episodic memory, 75–77 in children, 164, 170n1
evolution of, 59–60 rhesus monkeys
and metacognition, 62, 73–75 and consciousness of subjective
and self-recognition, 63–65 limitations, 184
and theory of mind, 65–71 and delayed matching-to-sample
Project Nim, 87–88 task, 326–327
proscopic memory and discrimination between re-
See semantic memory membering and forgetting,
protodeclarative behavior, in hu- 278–287
mans, 104 and metacognition, 75, 249–251
protoimperative behavior, in nonhu- and meta-confidence judgments,
man species, 104 303–318
psychology psychological mechanisms of
of memory in rhesus monkeys, memory in, 287–290
287–290 and visual discrimination task, 326
states of and self in social cogni- See also monkeys
tion, 337–338 road signs, and study of memory in
of uncertainty responses and un- chimpanzees, 197–198
certainty-monitoring perfor- rouge test, and self-recognition, 63
mances, 257–268
psychophysical studies, of metacog- Sally-Anne task, and theory of mind,
nition in humans and animals, 67, 69
246–251 scrub jay (Aphelocoma coerulescens),
punishment, and study of meta- and episodic memory, 37, 45–
confidence judgments in rhesus 46, 47, 149, 180, 228–229, 274
monkeys, 308–309 self
and cognitive consciousness,
rats, and uncertainty monitoring, 128–130
260–262, 266 and episodic memory, 14–15
See also monkeys language, consciousness, and,
reaction times (RT), and study of 136–137
meta-confidence judgments in metacognition and study of,
rhesus monkeys, 311, 313, 314 325–334
362 Subject Index
self (continued) uncertainty monitoring and meta-

and narrative consciousness, 134 cognition in animals and hu-
and reflective consciousness, mans, 242–268
132–133 See also consciousness
and social cognition, 336–338 self-description, and self-knowledge,
and theories of self-knowledge, 159 159–160
and traditions in study of animals, self digest, and motivational perspec-
322–325 tive on self-consciousness,
See also projectable self; standpoint 158–160
self-awareness self-recognition, and projectable self,
and autonoetic awareness in brain 63–65
injury case, 25–26 See also mirror image; recognition
and continuity of consciousness, self-reference
149–155 definition of, 149
definition of, 142, 153 and intentional behavior in ani-
and episodic memory in young mals, 150
children, 31 self-reflection
and escape from stimulus control as biological phenomenon,
in animal studies, 145–147 273–274
and metacognitive states, 243 definition of, 153
as unique to humans, 143 evolution of, 142, 273
See also memory awareness self-regulation
self-concept, and imitation in great and self digest, 158–159
apes, 323–324 and uncertainty monitoring, 267
self-consciousness self-representation, and self-regula-
and comparative social cognition, tion, 160, 166
321–338 semantic memory
development of in young children, and brain-damaged patients,
116–137 22–29
episodic memory, and autonoesis compared to episodic memory, 7,
as uniquely human, 3–49 10–14, 227
evolution and continuum of, and evolution of episodic mem-
142–155 ory, 48
and metacognitive judgments in and retrieval of information,
rhesus monkeys, 296–318 15–16
motivational perspective on, and time, 18
157–170 serial expertise, of monkeys, 96–97
normative roles of, in modern phi- serial probe recognition (SPR) task
losophy, 174–185 and memory monitoring, 252–257,
and question of episodic memory 266
in primates, 225–239 and meta-judgments, 316
and remembering and forgetting sex differences, in uncertainty re-
in rhesus monkeys, 272–293 sponse, 251
and role of metacognition in evolu- signal detection theory (SDT), and
tion of language, 84–111 uncertainty monitoring, 255–257
and projectable self, 57–77 simultaneous chains, and training of
and study of recall and episodic monkeys, 90–97
memory in chimpanzees, social cognition, self and other in,
188–219 336–338
Subject Index 363
social intelligence Sultan (chimpanzee), 193

of nonhuman primates, 85–86 symbols
and theory of mind, 66 apes and learning of, 101
See also social interaction; socializa- and memory in chimpanzees,
tion 196–198
social interaction and representational language,
and dependence of human infant 130
on social world, 118–119, and socialization of children, 163
136–137
and emergence of social conscious- tacts, and development of language,
ness in young children, 126, 98–99
127–129 television, and study of memory in
and human development, 136 chimpanzees, 208–209, 213
and role of social skills in evolu- theory of mind
tion of language, 85 and chimpanzees, 69–70, 102–103,
social comparison and self- 105–107
regulation in children, 165– and narrative consciousness, 134
166 and projectable self, 65–71
social modeling, and self- studies of, in children, 119–120
consciousness of becoming in time
children, 168 memory awareness and forgetting
and social referencing as unique in rhesus monkeys, 285–287
to human children, 162 and question of episodic memory
See also social intelligence; social- in animals, 37–40
ization self-awareness and estimation of,
socialization, self-regulation and 153–154
shared reality in, 161 and semantic memory, 18
See also social intelligence; social unique relation of episodic mem-
interaction ory to, 16–17, 27, 28, 29
special-purpose mechanisms, and See also future; subjective time;
cognitive development, 148– mental time travel
149 tip-of-the-tongue state (TOT), and
spontaneous recall, in language- studies of metacognition, 298
trained chimpanzee, 230–231 transfer tasks, and confidence judg-
spoon test, and autonoetic conscious- ments in rhesus monkeys,
ness of future, 43–47, 48–49 314–17
standpoint, on self and development
of self-regulation in children, uncertainty
164–167 and implicit cues for meta-
stimulus control, and false nega- judgment, 301, 317
tive demonstrations of self- and metacognitive research in ani-
awareness, 145–147 mals, 244–251, 302
subjective limitations, and self- psychology and monitoring of,
consciousness in modern philos- 257–268
ophy, 182–185 and psychophysical studies of
subjective time metacognition in humans,
and autonoetic awareness, 40 246–247
and episodic memory, 16–17, 20, signal detection theory (SDT) and
27, 28 monitoring of, 255–257
364 Subject Index
videotapes, and study of memory in Washoe (chimpanzee), 86–87, 100–101

chimpanzees, 208–209, 212–213, wasp (Sphex), and special-purpose
215 mechanism, 148–149
visual discrimination task, and rhe- working memory
sus monkeys, 326 in brain-damaged patient, 23
vocabulary, and evolution of lan- and self-awareness in animals and
guage, 101–103 young children, 147
See also grammar studies of, in monkeys, 95
Figure 2.2 Taira Sees the Ghosts of His Enemies, by Ando Hiroshige. From the collection of J. Metcalfe.
Figure 3.2 Simultaneous chaining paradigm. A. Example of a seven-
item list. Subjects are required to respond to each color photograph in
a prescribed order (A → B → C → D → E → F → G). B. Variation of the
configuration of list items. List items were presented on a touch-sensi-
tive video monitor in randomly configured displays. The task was to
touch the items in the prescribed order irrespective of their position on
the monitor. Prior to each trial, a new configuration was selected at
random from the more than 5.8 million configurations that could be
generated by presenting seven items in any of 16 positions. C. Types of
forward and backward errors. The shaded boxes show the types of log-
ical errors that could occur before a subject discovers the correct
sequence. See text for additional details. D. Determining the ordinal
position of a list item. The probability of guessing A by making logical
errors to eliminate the A− items. See text for additional details.
Figure 8.1 After touching the lexigram on her keyboard (visible in background) corre-
sponding to the type of object hidden, Panzee points toward the location of the object.
Photo by Charles R. Menzel.
Figure 9.1 King knuckle-walking in his outdoor area. He has .6 hectares of Florida ham-
mock forest.
FACING PAGE
Figure 10.2 A. Frequency discrimination performance by a bottlenosed dolphin (Tursiops
truncatus) in the procedure of Smith et al. (1995). The horizontal axis indicates the fre-
quency (Hz) of the tone. The response high was correct for tones at exactly 2100 Hz, and
these trials are represented by the rightmost data point for each curve. All lower tones
deserved the response low. The percentages of trials ending with the high response
(dashed line) or low response (dotted line) are shown. B. The dolphin’s performance in the
same auditory discrimination when he was also given the response uncertain. The solid
line represents the percentage of trials receiving the uncertainty response at each difficul-
ty level. The error bars show the lower 95% confidence limits. C. The intensity of the dol-
phin’s incidental hesitation and wavering behaviors for tones of different frequencies (Hz).
A. 100
Low
90
Response Percentage 80 High
70
60
50
40
30
20
10
0
1950 1975 2000 2025 2050 2075 2100
Pitch Height (Hz)
B. 100
90 Low
80
Response Percentage
70
High
60
50
40
30
Uncertain
20
10
0
1950 1975 2000 2025 2050 2075 2100
Pitch Height (Hz)
C. 0.8
0.6
Hesitation and Wavering
0.4
0.2
-0.2
-0.4
-0.6
-0.8
1950 1975 2000 2025 2050 2075 2100
Pitch Height (Hz)
Figure 10.3 A. The screen from a trial in the dense-sparse discrimination of Smith et al.
(1997). B. The performance of monkey Abel in the dense-sparse task. The response dense
was correct for boxes with exactly 2,950 pixels—these trials are represented by the right-
most data point for each curve. All other boxes deserved the response sparse. The hori-
zontal axis indicates the pixel density of the box. The solid line represents the percent-
age of trials receiving the uncertainty response at each density level. The error bars show
the lower 95% confidence limits. The percentages of trials ending with the dense response
(dashed line) or sparse response (dotted line) are also shown. C. The performance of mon-
key Baker in the dense-sparse discrimination depicted in the same way. D. The perfor-
mance of seven humans in the dense-sparse discrimination. To equate discrimination
performance across participants, the data have been normalized to place each partici-
pant’s discrimination crossover at a pixel density of about 2,700. The horizontal axis indi-
cates the normalized pixel density of the box.
A. Humans B. Humans
100 100
90 90 Dense
Response Percentage
Response Percentage
80 Dense 80
70 70
60 60 Sparse
50 Sparse 50
40 40
30 30
20 20
Uncertain
10 10
0 0 Uncertain
450 550 650 750 450 550 650 750
C. Baker D. Abel
100 100
90 90 Dense
Response Percentage
Response Percentage
80 80
70 Dense Sparse
70
60 Sparse 60
50 50
40 40
30 30
20 Uncertain
20
10 10
0 0 Uncertain
450 550 650 750 450 550 650 750
Figure 10.4 A. The performance of eight humans in the second kind of dense-sparse dis-
crimination used by Smith et al. (1997). The lower and upper half of a range of densities
deserved the sparse and dense response, respectively. The percentage of trials receiving
the uncertainty response at each density are shown by the solid line. The error bars show
the lower 95% confidence limits. The percentages of trials ending with the dense response
(dashed line) or sparse response (dotted line) are also shown. B. The performance of one
human in the same dense-sparse discrimination. C. The performance of monkey Baker
in this discrimination. D. The performance of monkey Abel in this discrimination.
A. Monkey B. Humans
100 100
90 90
Response Percentage
Response Percentage
80 80
70 Correct 70 Correct
60 60
40 40
30 30
20 20
10 10
0 0
1 2 3 4 NT 1 2 3 4 NT
Serial Position Serial Position
C. SDT Model D. Simulant

0.4 100
NT T 90
0.35
Response Percentage
Response Percentage
80
0.3
70 Correct
0.25 60
0.2 50 Uncertain
0.15 40
30
0.1
20
0.05 10
NT U T 0
0
-4 -3 -2 -1 0 1 2 3 4 5 1 2 3 4 NT
Trace Strength Impression Serial Position
Figure 10.6 A. Serial probe recognition (SPR) performance by monkey Baker in the task
of Smith et al. (1998). NT denotes not there trials. The serial position (1–4) of the probe
in the list of pictures is also given along the x-axis for the probes on there trials. The per-
centage of trials of each type that received the uncertainty response is shown (bold line).
The percentage correct (of the trials on which the memory test was attempted) is also
shown (dotted line). B. Performance by 10 humans in a similar SPR task used by Smith et
al. C. A signal detection theory (SDT) portrayal of monkey Baker’s decisional strategy in
the SPR task of Smith et al. (1998). Unit-normal trace-impression distributions are cen-
tered at the locations along the trace-strength continuum corresponding to the animals
d for probes of the four serial positions in the memory lists (there, T), and at 0.0 for the
NT probes. These normal curves are overlain by the decision criteria that define the ani-
mal’s three response regions (from left to right, NT, uncertain [U], and T). D. Performance
by the SDT simulant that fit best monkey Baker’s performance.
Rats
100
90
Response Percentage
80
Alternating
70 Repeating
60
50
40
30
20
Uncertain
10
0
40 35 30 25 20 15 10 5 0
Pitch Disparity (Hz)
Figure 10.8 Performance by six rats in the frequency discrimi-

nation used by Smith and Schull (1989). The horizontal axis
indicates the frequency difference between the alternating
pitches on a trial. The repeating response was correct for trials
with a frequency difference of 0, and these trials are represent-
ed by the rightmost data point for each curve. All other trials
deserved the alternating response. The solid line represents the
percentage of trials receiving the uncertainty response at each
pitch disparity. The percentages of trials ending with the
repeating response (dashed line) or alternating response (dot-
ted line) are also shown.
Figure 11.1 Measuring memory awareness

behaviorally. Viewing an object (S) causes
the formation of a mental representation,
or memory, of the object (M) that persists
in the brain even when the object is
removed from view. The memory later
guides the response (R) to the remembered
object in a memory test. In the same way
that M is about the object in the world, m
is a secondary mental representation that
is about the memory M. A secondary
response (r) is required to infer the pres-
ence of m. A procedure for doing so is
shown in figure 11.2.
Figure 11.2 Method for assessing whether monkeys know when they remember. Each
colored panel represents what monkeys saw on a touch-sensitive computer monitor at a
given stage in a trial. At the start of each trial, subjects studied a randomly selected
image. A delay period followed, during which monkeys often forgot the studied image.
On two thirds of trials, animals chose between taking a memory test (right panel, left-
hand stimulus) and declining the test (right panel, right-hand stimulus). On one third of
trials, monkeys were forced to take the test (left panel). Better accuracy on chosen mem-
ory tests than on forced tests indicates that monkeys know when they remember and
decline tests when they have forgotten, if given the option.
Figure 11.4 Schematic of a normal trial (upper panel) and a no-sample
probe trial (lower panel). The white lines represent the passage of time.
During the intertrial interval (ITI), the screen was black. On normal tri-
als, a sample appeared and the monkeys had to touch it to advance the
trial. After a delay, the monkeys could choose to either take or decline
the memory test. Probe trials were identical, with the exception that no
sample appeared, and the monkey did not have to touch the screen to
advance the trial.
Figure 11.7 Cartoons of candidate processes underlying memory aware-
ness. The left column represents the contents of cognitive processing dur-
ing trials on which monkeys choose to take the memory test. The right
column depicts the same on trials on which the test is declined. The mem-
ory flag hypothesis posits an indicator for the presence of memory.
Monkeys can respond on the basis of the indicator but are not aware of
the memory itself. In the case of the other two hypothetical mechanisms,
the decision to take the memory test is based on the richness of the mem-
ory retrieved. See text for more details.
Figure 12.2 Contingency table showing the four possible outcomes of correct or
incorrect response to the line task and high- and low-risk bets, starting with two
tokens already present in the reservoir. When high risk was pressed, two tokens
fell into the reservoir after correct responses, but flew out of the reservoir after
incorrect responses. When low risk was pressed, one token fell into the reser-
voir regardless of whether the line response was correct or incorrect. Note that
in Experiment 2, choosing high risk resulted in a grain or loss of three, rather
than two, tokens.
Figure 12.3 A. A sample trial for Experiment 1. In the first phase, nine lines appeared on
the screen. A token reservoir appeared in the bottom right-hand corner of the screen, set
at nine at the beginning of each session. The monkey’s task was to press the longest line.
For each touch, a green border appeared around the pressed item. Then the lines disap-
peared, and two risk icons appeared. The monkey’s task was to report his confidence by
betting either low or high risk. Once he made his bet, the token reservoir changed
accordingly. In this sample trial, the monkey has bet high risk, and was correct on the
line task. Thus, two tokens were added to the reservoir. B. Sample numerical stimuli for
Experiment 2. The task was to press the stimulus with the most (Lashley) or the least
(Ebbinghaus) items in them. Otherwise, the general methods remain the same as those
in Experiment 1.
Figure 13.2 Typical sequence of events in the hidden and the visible conditions. In the
hidden condition, the reward is hidden inside one of the tubes behind a screen (A), the
orangutan looks inside the empty tube (B), then the baited tube (C), and makes a choice
(D). In the visible condition, the reward is placed inside one of the tubes in full view of
the subject (E), and the subject chooses (F). Note the absence of looking behavior in the
visible condition compared to the hidden condition.

(Herbert S. Terrace, Janet Metcalfe) The Missing L

Uploaded by

Copyright:

Available Formats

(Herbert S. Terrace, Janet Metcalfe) The Missing L

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

(Herbert S. Terrace, Janet Metcalfe) The Missing L

Uploaded by

Copyright:

Available Formats

The Missing Link in

OXFORD UNIVERSITY PRESS

Copyright  2005 by Oxford University Press, Inc.

1. Episodic Memory and Autonoesis: Uniquely Human? 3

2. Self-Reflective Consciousness and the Projectable Self 57

3. Metacognition and the Evolution of Language 84

4. Emerging Levels of Consciousness in Early

5. A Continuum of Self-Consciousness That Emerges in Phylogeny

6. Humans as Applied Motivation Scientists: Self-Consciousness

7. Two Normative Roles for Self-Consciousness 174

8. Progress in the Study of Chimpanzee Recall

9. Do Nonhuman Primates Have Episodic Memory? 225

10. Studies of Uncertainty Monitoring and Metacognition in Animals

11. Can Rhesus Monkeys Discriminate Between Remembering

12. Metaconfidence Judgments in Rhesus Macaques: Explicit

13. The Self and Other: A Missing Link in Comparative

Episodic Memory and Autonoesis:

of these answers is terribly informative. A more appropriate question

1. Thesis: Mental time travel exists in humans only.

THEORY: EPISODIC MEMORY AND AUTONOETIC CONSCIOUSNESS

the concept needs to be sharpened. Just saying that it concerns per-

Episodic memory is a recently evolved, late developing, and early dete-

Note that episodic memory is conceptualized as a (hypothetical)

Like all other systems, episodic memory consists of a number of

Properties of Episodic Memory

Semantic/Episodic Memory—Common Features Episodic Memory—Unique Features

• Key function: Knowing—Registering, stor- • Key function: Remembering—

• Transmodal storage • Makes possible mental time

on the right gives “unique” properties of episodic memory, properties

of the world.” Language may have played an important role in the

Relations Between Episodic and Semantic Memory

mation may be stored in both systems. And independence of retrieval

Uniqueness of Episodic Memory

memory occurs whenever an organism thinks about, or mentally rep-

Episodic Memory, Autonoesis, and Time

havior and future-oriented thought (cognition). Frequently, the two

Darwin’s Moral Beings

A moral being is one who is capable of reflecting on his past actions

Darwin’s additional writing provides a rather detailed description

Why Did Episodic Memory Evolve?

Other early expressions of future-oriented thought and planning

AUTONOETIC EPISODIC MEMORY: DATA

The Case of K.C.

tion, even exceeds that of an average University of Toronto student

K.C.’s Episodic Memory

he is doing, that is, he can answer questions appropriately about these

K.C. and the Future

his personal future as he is incapable of seeing himself in his personal

has a little left of one or more of the properties of episodic memory

Why Is the Case of K.C. Relevant?

tion between semantic (noetic) and episodic (autonoetic) mental time

Autonoetic Memory in Children

The observations concerning the differentially delayed develop-

Pearce, 2000). Autobiographical memory is said to entail a “personal

can be and are satisfied without remembering autonoetically what

ANALYSIS: EPISODIC MEMORY IN ANIMALS

ity of animals to perform the task, and the effects of experimentally

The Fourth Problem

In an article titled “Mental Time Travel and the Evolution of the