Toward a Richer View of the Scientific Method

The Role of Conceptual Analysis

Armando Machado Universidade do Minho
Francisco J. Silva University of Redlands
Within the complex set of activities that comprise the sci-
entic method, three clusters of activities can be recog-
nized: experimentation, mathematization, and conceptual
analysis. In psychology, the rst two of these clusters are
well-known and valued, but the third seems less known and
valued. The authors show the value of these three clusters
of scientic method activities in the works of the quintes-
sential scientist Galileo Galilei. They then illustrate how
conceptual analysis can be used in psychology to clarify
the grammar and meaning of concepts, expose conceptual
problems in models, reveal unacknowledged assumptions
and steps in arguments, and evaluate the consistency of
theoretical accounts. The article concludes with a discus-
sion of three criticisms of conceptual analysis.
Keywords: conceptual analysis, criticism, Galileo, scien-
tic method, theory
M
odern science is characterized by a rich variety
of activities. Researchers observe and experi-
ment in the eld and laboratory, devise new
techniques and rene old ones, invent and elaborate con-
cepts, quantify functional relations between variables, build
models and theories, debate controversial issues, write and
review manuscripts and grant proposals, and so on. Within
this set of activities we can identify clusters or groups of
related activities. Observation/experimentation comprises
the actions researchers perform to generate theories and test
hypotheses. This cluster includes operationally dening
variables, manipulating and controlling them, implement-
ing a research design, selecting research participants and
allocating them to groups, and conducting the experiment
itself. Quantication/mathematization consists of analyz-
ing data and formulating laws, models, and theories on the
basis of empirical ndings and previous knowledge. Other
activities in this cluster include discovering the mathemat-
ical function that may link two variables, nding the best-
tting parameters of the function using statistical tech-
niques, integrating two or more quantitative functions
into higher order laws, and testing the statistical signi-
cance of the relation between variables. We recognize in
these two clustersexperimentation and mathematization,
for shortmost of the activities subsumed under the broad
conception of scientic methods in psychology (e.g., Slife
& Williams, 1995; Stanovich, 1998). To a large extent,
these two clusters match the activities described in most
textbooks and courses on research methods (experimenta-
tion) and applied mathematics and statistics for psycholo-
gists (mathematization).
Although experimentation and mathematization are
well-known components of the scientic method and their
importance is unquestionablein fact, psychologists have
acknowledged their importance since the very beginnings
of their science in the last quarter of the 19th century
(Burnham, 1987)there is a third cluster of activities that
also is an essential component of the scientic method.
Theoretical/conceptual analysis comprises the actions re-
searchers engage in when they evaluate the language of
their science. These actions include but are not limited to
assessing the clarity or obscurity of scientic concepts,
evaluating the precision or vagueness of scientic hypoth-
eses, assessing the consistency or inconsistency of a set of
statements and laws, and scrutinizing arguments and chains
of inferences for unstated but crucial assumptions or steps.
In general, then, the goal of conceptual analysis is to
increase the conceptual clarity of a theory through careful
clarications and specications of meaning (Laudan,
1977, p. 50).
Although psychologists undoubtedly engage in con-
ceptual analysis, this aspect of the scientic method does
not seem as well-known or as acknowledged as experimen-
tation and mathematization (see Machado, Lourenco, &
Silva, 2000). First, whereas most introductory psychology
textbooks include a chapter on research methods and basic
statistical ideas, and most instructors of introductory
courses cover this material within the rst week or two of
classes, there is no comparable chapter or even sections of
a chapter devoted to conceptual analysis. Indeed, in con-
trast to methods, research methods, research design, sta-
tistics, statistical analysis, data analysis, and the like, there
is no term or set of terms in introductory psychology or
research methods textbooks to describe conceptual analy-
sis. The only related topics that are occasionally mentioned
Armando Machado was funded by a grant from the Portuguese Founda-
tion for Science and Technology. We thank Susan Goldstein for sharing
her expertise about construct validity.
Correspondence concerning this article should be addressed to Armando
Machado, Instituto de Educacao e Psicologia, Universidade do Minho, Cam-
pus de Gualtar, 4710 Braga, Portugal or Francisco J. Silva, Department of
Psychology, University of Redlands, P. O. Box 3080, 1200 East Colton
Avenue, Redlands, CA 92373-0999. E-mail: armandom@iep.uminho.pt or
francisco_silva@redlands.edu
671 October 2007

American Psychologist
Copyright 2007 by the American Psychological Association 0003-066X/07/$12.00
Vol. 62, No. 7, 671681 DOI: 10.1037/0003-066X.62.7.671
are critical thinking and logic. Second, research methods
and statistics courses are universally required of psychol-
ogy majors and graduate students, but few if any psychol-
ogy courses on conceptual analysis exist or are required.
Third, conceptual analysis is rarely the subject matter of
conference presentations, articles, chapters, or books for
experimental psychologists. Fourth, when psychologists
are asked about the most important concepts or topics that
undergraduates should learn, they respond research meth-
ods, independent and dependent variables, and the like
(see Boneau, 1990); conceptual analysis or anything like it
is unmentioned. Fifth, methodological and statistical
awsnot conceptual problemsare the reasons most of-
ten invoked for rejecting papers submitted for publication
(e.g., Peters & Ceci, 1982; see also Broad & Wade, 1982;
Ghiselin, 1989; Harcum & Rosen, 1993). Although any of
these facts can be interpreted in other ways (e.g., method-
ological and statistical aws are fatal aws and therefore
are more likely than conceptual aws to be invoked as
reasons for rejecting a manuscript; introductory psychology
textbooks are meant to provide an overview of the major
subject areas of psychology), it is difcult to explain this
constellation of facts without concluding that conceptual
analysis is, at least to some degree, devalued in psychology
(Machado et al., 2000; Slife & Williams, 1995).
In the rst part of this article, we show the importance
of conceptual analysis in the work of the quintessential
scientist of the modern era, Galileo Galilei. We show that
the Italian scientist, often portrayed as personifying the
epitome of experimentation and mathematization, used
conceptual analysis to formulate clear concepts and testable
explanations of the physical world and to reject their op-
posites. In the second part of the article, we discuss con-
ceptual analysis with reference to examples in psychology.
Throughout our presentation of the two sets of examples,
we adopt a concrete perspective to illustrate various aspects
of conceptual analysis and its targets, to identify recurrent
themes across distinct domains, and to describe the limits
of the activity. In the third and nal part, we adopt a more
abstract perspective and address a few criticisms of con-
ceptual analysis.
Galileos View of the Scientific
Method: Controlled Experiments,
Mathematical Analysis, and
Conceptual Investigations
Galileo Galilei (15641642) was a leading gure of
the scientic revolution. Some remember his polemics with
the Church concerning the relative merits of the Coperni-
can and Ptolemaic systems and his trial by the Inquisition;
others remember his discoveries of the trajectory of pro-
jectiles and the swinging of the pendulum or his rene-
ments of the telescope. For physicists, perhaps his major
accomplishments were the experiments he performed and
the theories he developed on free-falling bodies near the
surface of the Earth. We begin with a brief review of that
for which Galileo is most well-known (i.e., his experiments
and mathematical discoveries) and then illustrate in some
detail that for which he is less well-known (i.e., his con-
ceptual analyses in physics and astronomy).
Experimentation
Controlled experiments are a legacy of the scientic revo-
lution. In Galileos descriptions of his experiments, it is
easy to recognize some of the main characteristics of the
modern scientic method, namely, manipulating the inde-
pendent variable, controlling extraneous variables, measur-
ing the dependent variable, and addressing measurement
reliability. Consider the following description of Galileos
famous studies with inclined planes.
A piece of wooden moulding or scantling, about 12 cubits long,
half a cubit wide, and three nger-breadths thick, was taken; on its
edge was cut a channel a little more than one nger in breadth;
having made this groove very straight, smooth, and polished, and
having lined it with parchment also as smooth and polished as
possible, we rolled along it a hard, smooth, and very round bronze
ball. [Galileo wanted to study the interrelation between distance
traversed and time taken by a falling body, an interrelation not
contaminated by friction. Hence his attempt to reduce as much as
possible the inuence of this extraneous variable.] Having
placed this board in a sloping position, by lifting one end some
one or two cubits above the other, we rolled the ball, as I was just
saying, along the channel, noting, in a manner presently to be
described, the time required to make the descent. [Because the
motion of free-falling bodies occurs very quicklyand therefore
is hard to characterizeGalileo slowed their motion by means of
the inclined plane.] We repeated this experiment more than once
in order to measure the time with an accuracy such that the
deviation between two observations never exceeded one-tenth of
a pulse-beat. Having performed this operation and having assured
ourselves of its reliability, we now rolled the ball only one-quarter
the length of the channel. . . . Next we tried other distances,
comparing the time for the whole length with that for the half, or
Armando
Machado
672 October 2007

American Psychologist
with that for two-thirds, or three-fourths, or indeed for any frac-
tion. [Galileo varied systematically the independent variable
distanceand measured repeatedly the dependent variabletime
of descent.]. . . . For the measurement of time, we employed a
large vessel of water placed in an elevated position; to the bottom
of this set was soldered a pipe of small diameter giving a thin jet
of water, which we collected in a small glass during the time of
each descent, whether for the whole length of the channel or for
a part of its length; the water thus collected was weighed, after
each descent, on a very accurate balance; the differences and
ratios of the weights gave us the differences and ratios of the
times, and this with such accuracy that although the operation was
repeated many, many times, there was no appreciable discrepancy
in the results. (Galilei, 1638/1914, pp. 178179)
Mathematization
Another legacy of the scientic revolution is the attempt to
formulate mathematically the relation between independent
and dependent variables. To illustrate, let us return to
Galileos use of inclined planes. After many experiments of
the sort described above, some astute thought experiments,
a few false starts, and several decades of thinking about
free-falling motion, Galileo arrived at the correct mathe-
matical formulation of the law: In such experiments, re-
peated a full hundred times, we always found the spaces
traversed were to each other as the squares of the times, and
this was true for all inclinations of the plane (Galilei,
1638/1914, p. 179). Galileo had discovered that d
1
/d
2

t
1
2
/t
2
2
or, in modern terminology, that distance is propor-
tional to the square of time, d k t
2
. For some historians
of science, the coupling of mathematical reasoning with
controlled experiments and observations was Galileos
chief contribution to the development of scientic method
in the 17th century (Gower, 1997, p. 37).
Conceptual Analysis
But Galileos scientic method consisted of more than
controlled experiments and mathematics. There was also
concern for the clarity of concepts, the coherence of hy-
potheses, and the consistency of theoretical arguments. In
several passages of Galileos writings, one can see him
analyzing the language of science, that is, conducting con-
ceptual investigations. These investigations may be what
Stillman Drake (1990), arguably Galileos best scientic
biographer, had in mind when he mentioned reasoning
alongside observation and mathematics as the three funda-
mental aspects of Galileos approach:
[Galileo] agreed that in order to become science, philosophy must
throw out blind respect for authority; but he also saw that neither
observation, nor reasoning, nor the use of mathematics could be
thrown out along with this. True philosophy had to be built upon
the interplay of all three, and no combination could supply the
absence of any one of them. (Drake, 1957, pp. 223224)
Next, we examine examples of conceptual analysis in
Galileos work. In each case, one can see the Italian sci-
entist exposing a problem or difculty associated with a
concept, hypothesis, or account. The sources of the prob-
lems seem to fall into distinguishable categories.
Inappropriate or illogical classification.
A major aim of conceptual analysis is to identify concep-
tual problems, which often arise from inappropriate or
illogical classication of concepts. In his scientic mani-
festo The Assayer, Galileo provided the following
straightforward example when addressing Sarsi, his oppo-
nent, about the nature of comets and the functioning of the
telescope:
If we go back to examine his [Sarsis] argument more closely, we
nd it to be defective because it takes as absolute that which must
be understood relatively, or as bounded that which is unbounded.
In a word, Sarsi has created an incomplete dichotomy (as logi-
cians call this error) when he divided visible objects into far and
near without assigning limits and boundaries between these. He
has made the same mistake as a person who should say, Every-
thing in the world is either large or small. This proposition is
neither true nor false, and neither is the proposition objects are
either near or far. . . . In order to avoid equivocation Sarsi needed
to give his classication at least three parts. . . . Nor should he
even stop there; he should give an exact determination of this
limit, saying for example: I call medium a distance of one
league; far, that which is more than one league; and near, that
which is less. (Galilei, 1623/1957, p. 249)
The point of the example is not whether Galileos
theories were rightin fact, his theory of comets was
wrongbut that a particular argument was conceptually
defective for the reasons he identied. Galileos remark
that the proposition Everything in the world is either large
or small is neither true nor false but equivocal is a hall-
mark of conceptual analysis.
Excessively vague and ad hoc explana-
tions. In this next example, Galileo took issue with
Sarsis excessively vague and ad hoc explanation to ac-
count for the paths of comets. When Sarsi remarked that
Francisco J.
Silva
673 October 2007

American Psychologist
Galileos account on the nature of comets failed to consider
various alternative motions, particularly motions along ir-
regular paths, Galileo pointed out that the concept irreg-
ular path could not be specied, and he concluded as
follows:
Lines are called regular when, having a xed and denite descrip-
tion, they are susceptible of denition and of having their prop-
erties demonstrated. Thus the spiral is regular, and its denition
originates in two uniform motions, one straight and the other
circular. . . . Irregular lines are those which have no determinacy
whatever, and are indenite and casual and hence undenable. . . .
Hence to say, Such events take place thanks to an irregular path
is the same as to say, I do not know why they occur. The
introduction of such [irregular] lines is in no way superior to the
sympathy, antipathy, occult properties, inuences, and
other terms employed by some philosophers as a cloak for the
correct reply, which would be: I do not know. That reply is as
much more tolerable than the others as candid honesty is more
beautiful than deceitful duplicity. (Galilei, 1623/1957, p. 241)
Galileos sensitivity to the denition of the key terms in the
argument (regular and irregular lines) is another expression
of his keen ability for conceptual analysis.
Irrefutable hypotheses. In this third example,
Galileo rejected an irrefutable hypothesis. Sarsi wanted
him to believe that the Babylonians cooked their eggs by
whirling them in slings. Galileo claimed he could not
obtain the effect experimentally, but apparently this was
insufcient to change Sarsis mind. Galileo remarked,
All too prudently you [Sarsi] have secured your position by
saying that there is needed for this effect violent motion, a great
quantity of exhalations, a highly attenuated material, and what-
ever else conduces to it. This whatever else conduces to it is
what beats me, and gives you a blessed harbour, a sanctuary
completely secure. (Galilei, 1623/1957, p. 273)
The nominal fallacy (Naming is not ex-
plaining). When we use a familiar name to label an
unfamiliar phenomenon, our understanding of that phe-
nomenon seems to increase automatically. Galileo was
aware of this cognitive illusion and joked about it in the
following passage:
[Sarsi believed that comets are planets or quasi-planets but Gali-
leo found his arguments fallacious and concluded:] I am not so
sure that in order to make a comet a quasi-planet, and as such to
deck it out in the attributes of other planets, it is sufcient for
Sarsi or his teacher to regard it as one and so name it. If their
opinions and their voices have the power of calling into existence
the things they name, then I beg them to do me the favor of
naming a lot of old hardware I have about my house, gold.
(Galilei, 1623/1957, p. 253)
Similarly, when Aristotles representative in Galileos di-
alogues, Simplicio, stated that everyone knows that gravity
causes bodies to fall downwards, Galileos spokesman,
Salviati, replied,
You are wrong, Simplicio; what you ought to say is that everyone
knows that it is called gravity. What I am asking you for is not
the name of the thing, but its essence, of which essence you know
not a bit more than you know about the essence of whatever
moves the stars around. I accept the name which has been at-
tached to it and which has been made a familiar household word
by the continual experience that we have of it daily. But we do not
really understand what principle or what force it is that moves
stones downward, any more than we understand who moves them
upward after they leave the throwers hand, or what moves the
moon around. (Galilei, 1632/2001, p. 272)
Unjustified extension of a familiar concept
to an unfamiliar domain. Our familiarity with a
set of concepts, usually from everyday speech, breeds
contempt for their scientic analysis. Why worry with
analysis if we all know what these concepts mean? one
might ask. Frequently, though, such uncritical assump-
tions of mutual understanding (Quine, 1936, p. 90) give us
only unjustied feelings of accomplishment. The problem
is particularly acute when familiar terms are extended to
novel domains, as the following interchange reveals. Sim-
plicio knew how to compare two collections of objects and
conclude that Collection A is greater than, equal to, or less
than Collection B. He also knew that a line segment has an
innite number of points. But when he saw two line seg-
ments, one longer than the other, he attempted to determine
which line had more points. That is, he attempted to com-
pare innite sets and, naturally, he extended to this unfa-
miliar and new domain the familiar process of comparing
nite sets:
Since it is clear that we may have one line greater than another,
each containing an innite number of points, we are forced to
admit that, within one and the same class, we may have something
greater than innity, because the innity of points in the long line
is greater than the innity of points in the short line. This assign-
ing to an innite quantity a value greater than innity is quite
beyond my comprehension. (Galilei, 1638/1914, p. 31)
Salviati identied the source of Simplicios puzzle-
ment: The attributes equal, greater, and less, are not
applicable to innite, but only to nite quantities (Galilei,
1638/1914, pp. 3133). The lesson is, when a concept is
extended to an unfamiliar domain, one must check which
properties of the concept hold in the new domain.
Semantic ambiguity (Similar words may
hide different conceptions). If all concepts are
theory-laden, then it is imperative to check whether two
usages of the concept presuppose the same underlying
theory, lest the door be open to misinterpretations and
fruitless disputes. As shown in the next example, Galileo
knew that the same word might hide different conceptions.
According to Sarsi, Galileos idea that friction is the
cause of heat is consistent with Aristotles claim that mo-
tion is the cause of heat. Sarsi reasoned as follows: Al-
though motion, as motion, is not the cause of heat, friction
cannot occur without motion; therefore, motion is at least
derivatively the cause of heat. To which Galileo replied,
But if that is what Aristotle meant, then why didnt he say
friction? When a man can say denitely what he means by
using a simple and appropriate word, why employ an inappropri-
ate one that requires qualication and ultimately becomes trans-
formed into something quite different? But assuming that this was
Aristotles meaning, it still differs from Guiduccis [Galileos
student and spokesman]; for to Aristotle any rubbing of bodies
674 October 2007

American Psychologist
would sufce, even of tenuous ones or of the air itself, whereas
Guiducci requires two solid bodies, for he considers that trying to
pulverize the air is as great a waste of time as grinding water in
the proverbial mortar. (Galilei, 1623/1957, p. 266)
Note that Galileo was not arguing about experimental
or quantitative issues, nor was he saying that motion does
not cause heat. Instead, he was commenting on concepts,
on the clarity or obscurity of their use, and on whether the
denitions of two concepts overlapped.
Galileos works provide several other instances of
conceptual analysis dealing with, for example, the perils of
reasoning by analogy, ad hoc hypotheses, confusing ab-
stract and concrete terms, and the meaninglessness of con-
cepts that express judgments of value (e.g., a circle is the
noblest of shapes). From these examples at the dawn of
modern science we see experiments, mathematics, and
conceptual analysis all playing a signicant role in the
advancement of knowledge.
Examples of Conceptual Analysis in
Psychology
In the examples that follow, we provide a concrete, diver-
sied, example-based view of conceptual analysis in psy-
chology and continue to identify some of its fundamental
characteristics. We begin with an illustration of how con-
ceptual analysis claried the grammar and meaning of a
concept and thereby solved a conundrumhow an in-
verted retinal image yields a righted percept. We then
illustrate how conceptual analysis can be used to expose
problems in models, reveal key assumptions and obscured
steps in arguments, and evaluate the consistency of a the-
oretical account.
Clarifying the Grammar and Meaning of
Concepts
One of the main achievements of the scientic revolution
bears directly on psychological matters and provides a nice
illustration of how conceptual analysis can resolve a long-
standing problem. In 1604, Johannes Kepler correctly de-
scribed how the image of an object is formed in the retina,
a description that is the basis of our modern understanding
of vision. Keplers geometrical analysis assumed that
countless cones of light extend from every point on an
object (the apex) to a common base on the lens of the eye;
as each light ray enters the eye, it is refracted by the cornea
and lens before exciting the retina. As a consequence of
optics, that which is to the right outside is depicted on the
left on the retina, that to the left on the right, that above
below, and that below above (Kepler, cited in Herrnstein
& Boring, 1965, p. 94). But if the retinal image is inverted,
how do we see objects upright in their correct orienta-
tion? The question had disturbed previous researchers and
forced them to abandon any account that resulted in an
inverted image. For example, Leonardo da Vinci compared
the eye to a camera obscura but arranged the optics to
preserve an erect image as a necessary condition for upright
vision (see Crombie, 1964).
The answer to the problem was provided by William
Molyneux when he claried the grammar of the concept of
seeingit is the person who sees, not his or her eyesand the
meaning of concepts such as erect and inverted, up and down,
closest to and farthest from the center of the Earth. His
argument can be recast as follows: Because (a) an eye does
not see the retinal image and (b) the retinal image preserves
the relations among the objects, it follows that (c) there is no
need to force the image to accord with perception. The prob-
lem of how objects are seen upright if their retinal images are
inverted is not to be solved empirically but to be dissolved
conceptually. Two centuries after Molyneux, the physiologist
Volkmann put it thus, The most natural explanation of up-
right vision is that it does not require an explanation (cited in
Wade, 1998, p. 325). Gregory (1996) not only pointed out that
there is no inverted image problem to solve because the image
is not an object that is seen but added the idea that if retinal
images were seen, then there would be an innite regress of
images, images on eyes, images of images on eyes, and so on.
Exposing a Limitation or Problem in a Model
The next example shows how conceptual investigations
may target even elegant and sophisticated mathematical
theories. In this case, the investigation reveals a problem in
the dominant model of animal and human interval timing,
scalar expectancy theory, or SET (Gibbon, 1977, 1981; see
also Gallistel, 1990). To begin, consider the following
discrimination task: A subject (animal or human) is re-
warded for choosing a red key after a short stimulus (e.g.,
a 2-s tone) and a green key after a long stimulus (e.g., an
8-s tone). After some trials, the subject reliably chooses the
red and green keys when the stimuli are short and long,
respectively. To explain how the subject learns the tempo-
ral discrimination, SET postulates an internal clock con-
sisting of a pacemaker that generates pulses, an accumula-
tor that counts the pulses emitted during the stimulus, and
two memory stores that save the counts in the accumulator
at the end of the short and long signals. Although SET has
additional features, for present purposes it is sufcient to
note that the temporal discrimination depends on the for-
mation of two distinct memory stores.
But how does the timing system decide, so to speak,
into which memory to store the count obtained after a
particular stimulus? There is only one plausible answer:
When the subject chooses the red key and is rewarded for
this response, the count in the accumulator is saved to one
memory store (call it A); when the subject chooses the
green key and is rewarded for this choice, the count in the
accumulator is saved into the other memory store (call it
B). Thus, counts are saved to a particular memory store on
the basis of the structural features of the task (e.g., choos-
ing this or that distinctive key and getting a reward). The
theory has no major difculty accounting for the discrim-
ination in this task.
Consider now a second task. A pigeon receives food
for pecking a key after either 10 s or 120 s have elapsed
since the onset of the trial. No cue signals whether the
current trial will be short or long, and the two types of trials
are equally probable. The result of this training is that
during the long trials, the pigeons average rate of pecking
increases from the beginning of the trial until approxi-
675 October 2007

American Psychologist
mately 10 s have elapsed, at which point the rate of pecking
decreases before beginning to increase again and peaking at
120 s (e.g., Catania & Reynolds, 1968; Leak & Gibbon,
1995). SET explains this pattern of responding by assum-
ing that the animal stores the counts obtained at 10 s and
120 s into distinct memory stores (just as in the rst
example above). As Leak and Gibbon (1995, p. 6) put it,
In SET, there is assumed to be a single clock but an
independent memory distribution for each criterion time
interval.
But how does the timing system decide into which
memory to store the count obtained after a particular in-
terval? The answer offered for the rst example will not
work here because there is only one key and no distinct
signal to cue the two trials. That is, the counts cannot be
directed to and saved into different memory stores on the
basis of the structural features of the task. SET explains the
temporal discrimination by assuming that small counts are
saved in Store A and large counts are saved in Store B. But
this explanation is a classic case of begging the question
(petitio principi for logicians): That which was supposed to
be explained by the theory, namely, the discrimination
between two intervals or its equivalent (the discrimination
between two sets of counts) was assumed in the explana-
tionthe system knows the difference between small
and large counts, which is equivalent to the difference
between short and long intervals. Any explanation that
assumes the existence of distinct memory bins begs the
question of how they were initially formed.
Revealing Unacknowledged Assumptions
and Steps in Arguments
Conceptual analysis may help reveal unacknowledged but
fundamental assumptions or steps in an argument or chain
of inferences. Consider the following example. A 5-month-
old baby is shown a doll. Next, an opaque screen rotates up
to hide the doll. Then the baby sees a second doll being put
behind the screen. Finally, when the screen is dropped, the
baby sees either two dolls (the so-called possible result) or
only one doll (the so-called impossible result). The exper-
imenter records the total time the baby looks at the result-
ing doll(s). The typical nding is that babies stare longer at
the impossible outcome than at the possible outcome (e.g.,
Wynn, 1992, 1995). Similar experiments with cotton-top
tamarin monkeys and dogs (substituting food rewards for
dolls) have yielded similar results (e.g., Uller, Hauser, &
Carey, 2001; West & Young, 2002).
These results have been explained as follows: Babies,
dogs, and monkeys (a) compute the outcome of certain
operations (e.g., 1 1 2) and (b) on the basis of that
computation, predict what is behind the screen; (c) when
the screen drops, they compare the number of items they
expected to see with the number that they actually see; (d)
when the comparison yields a mismatch (e.g., 2 vs. 1) they
are surprised and therefore stare longer at the unexpected
outcome. As Wynn (1992) put it,
[When the experimenter placed the second item behind the screen,
the] infants could clearly see the nature of the arithmetical oper-
ation being performed, but could not see the result of the opera-
tion. . . . Infants look longer at unexpected events than expected
ones, thus, if they are able to compute the numerical results of
these arithmetical operations, they should look longer at the
incorrect than at the correct results. (p. 749)
But note that by treating the act of bringing a second
doll to the stage as an arithmetic operation, the physical
action performed by the experimenter and the mental op-
eration executed by the baby are treated as one and the
same. Other authors have treated actions and mental oper-
ations similarly: The results suggest that the dogs were
anticipating the outcome of the calculations they observed,
thus suggesting that dogs may have a rudimentary ability to
count (West & Young, 2002, p. 183, italics added). Al-
though infants and animals might perform computations
and calculations upon seeing the experimenters actions,
computations and calculations are not themselves seen. By
envisaging the experimenters actions as arithmetic opera-
tions or by stating that dogs observe calculations (as op-
posed to perform calculations), two distinct activities (the
experimenters actions and the subjects purported calcu-
lations) were conceptualized as a single one. Perhaps what
the authors assumed was that the experimenters action of
adding a doll or a food reward triggered the subjects
mental operation of addition. Although this assumption
may be correct, it needs to be made explicit and analyzed
empirically before we can discover the properties of the
mapping between the experimenters action and a subjects
mental operation (e.g., What roles do temporal and spatial
attributes of the experimenters actions, the trajectories of
the dolls or food rewards, and the number of times each
doll or food reward is presented play in how the child,
monkey, or dog comes to know that the physical operation
was the occasion for the arithmetic operation?).
Questioning the Consistency of an Account
The next example is potentially more controversial than the
preceding ones. We include it to illustrate additional as-
pects of conceptual analysisin this case, how it questions
the consistency of an account that uses concepts from
different universes of discourse: mental intentions, neural
representations, and motor behaviors. The use of different
languages, as it were, is a perennial source of conceptual
difculties. We also want to show that conceptual analysis
may produce more questions than it answers, stimulate
spirited debate, and inspire further experiments.
An article published in the prestigious journal Science
opened with the following paragraph:
A motor action is voluntary if and only if it is intended. William
James put forward the ideomotor theory of action, which states
that any intention or idea of an action has the tendency to cause
the relevant movements. To prevent ourselves from committing
action errors, it is frequently important that we attend to our
intentions before executing an action. (Lau, Rogers, Haggard, &
Passingham, 2004, p. 1208)
The authors then described an experiment designed to
study the neural correlates of attention to the intention to
perform an act. Using functional magnetic resonance im-
676 October 2007

American Psychologist
aging (fMRI), they compared two conditions in a within-
subjects design. In one condition the participants pushed a
button while paying attention to the act itself; in the other
condition the participants pushed the button while paying
attention to the intention to push the button. Thus the only
difference between the two conditions was the focus of
attention. This was manipulated by requiring them to report
either the time at which they felt the intention to move or
the time at which they actually made the movement (Lau
et al., 2004, p. 1208). The results showed that whereas the
act was reported about 30 ms before it actually took place,
the intention to act was reported about 230 ms before the
act. In addition, the fMRI revealed specic activations
associated with attention to intention (p. 1209), speci-
cally, an enhancement of activity in the pre-supplemental
motor area (pre-SMA) coupled with activations in the right
dorsal prefrontal cortex. The authors concluded that activ-
ity in the pre-SMA reects the representation of intention.
As interesting and provocative as these results are, the
scientic game requires us, as scientists, to analyze the
authors arguments and evaluate their conclusions. Should
we accept Lau et al.s assertions that what people reported
at 230 ms was their intention to act, that these intentions are
represented in the pre-SMA, and that they are causally
related to the behavior of pushing the button? The answers
require conceptual analysis, a reasonably detailed example
of which follows.
Identify the purpose and structure of the
argument. The experiment was designed to show that
attending to the intention of an act has specic brain
correlates and that these correlates differ from the corre-
lates of attending to the act itself. More generally, Lau et al.
(2004) were searching for the neural foundations of volun-
tary or intentional behavior. Their argument has at least
three components in the form of implicit or explicit as-
sumptions: (a) Intention is a mental representation with a
neural instantiation (e.g., increased activity in a particular
brain region), and thus intention is an occurrence, some-
thing that can be the target of effective editing and eval-
uation (p. 1210); (b) intention precedes and efciently
causes voluntary behavior; and (c) given that an intention is
an occurrence, attention can be focused on it.
Critique the structure of the argument.
We mention only critiques to each part separately and
ignore potential critiques of their interconnection. First, if
an intention is a mental or neural occurrence, then we can
ask: What causes an intention? If we reply that an inten-
tion is caused by something intentional, then we have an
innite regress because Intention A would be caused by
Intention B, which would be caused by Intention C, and so
on. But if we reply that intentional behavior is caused by
something unintentional, then in what sense is the inten-
tional behavior voluntary? Second, if intentions efciently
cause voluntary behavior, then the behavior should occur
once it is intended. But this seems unlikely. A tennis
players intention to hit a cross-court forehand on the next
shot does not mean that he will; a return sent to his
backhand side, for example, may cause him to hit a back-
hand down the line. Third, we invite readers to imagine
how they would respond to three different instructions:
Pay attention to the spider, Pay attention to time, and
Pay attention to your intention to push the button. Pre-
sumably, readers know how to follow the rst instruction
(e.g., look at the spider, follow its movements), but how
will they respond to the last two? With regard to the second
instruction, notice that it does not direct the reader to Pay
attention to the clock. With regard to the last instruction,
is the intention to push the button an occurrence that
readers can focus their attention on? If the answer is un-
clear, then consider the following related question: When
one writes a sentence (an intentional act), is there an inner
urge or feeling for each word, a feeling that could be the
focus of the writers attention?
Assess the effects of the critique on the
purposes of the argument. After critiquing the
structure of the authors arguments, one might continue a
conceptual analysis with an assessment of the conse-
quences of the critique. For example, on the one hand, the
critique of Lau et al.s (2004) arguments leads us to ques-
tion the conceptual foundations of the experiment (i.e., the
view that intentions are mental or neural representations
that efciently cause behavior and that could be the targets
of attention). On the other hand, the critique leaves intact
the authors novel and interesting empirical discovery that
the participants behavior and fMRIs changed reliably with
the experimental instructions.
It could be argued that our critique misinterprets the
concept of intention, or at least how the authors intended
the concept of intention to be used. If so, the source of any
misinterpretation is that the authors did not provide a
theoretical denition of intention and/or make explicit how
the concepts of intention, attention, neural representation,
and motor behavior are interrelated. Without some degree
of conceptual explicitness, particularly when the argument
includes concepts belonging to disparate vocabularies and
possessing different grammars (cf. intention and attention,
neural rings, motor actions), the door is left open to
theoretical confusion (see Bolacchi, 2004). The foregoing
aspects of conceptual analysis were embodied in one of
Karl Poppers (1959) epistemological theses: I shall . . .
adopt a rule not to use undened concepts as if they were
implicitly dened (p. 75).
Alternative explanations. If Lau et al.s
(2004) interpretation is problematic, then how did the par-
ticipants reliably comply with the experimental instruc-
tionsto attend to the intention to press a button, an
intention conceived of as an inner urge or feeling to actif
no such feeling or urge exists qua intention to act? One
possible answer is that because the experiment used a
within-subjects design in which the participants received
separate instructions across conditions (pay attention to the
act itself; pay attention to the intention to act) and practiced
during one preliminary session, the participants constructed
a script of how they should behave that was consistent
with the instructions and then followed this script by re-
porting a moment of intention slightly before the action.
Cognitive psychologist Pylyshyn (2003) made a similar
suggestion when discussing whether mental images are
677 October 2007

American Psychologist
pictures. In an article subtitled Are There Really Pictures
in the Brain? he wrote,
Nearly all experimental ndings cited in support of the picture
theory can be more naturally explained by the hypothesis that
when asked to imagine something, people ask themselves what it
would be like to see it, and they then simulate as many aspects of
this staged event as they can and as seem relevant. (p. 113)
Just as mental images may not be pictures, an inten-
tion may not be an inner urge or feeling. Instead, the people
in Lau et al.s (2004) experiment may have attempted to
make sense of the instructions by imagining what it must be
like to have an inner urge before performing an action, and
then followed this script by reporting the moment of
intention. Although this account is speculative, it is plau-
sible and testable. Moreover, it illustrates how a conceptual
analysis may contribute to the growth of knowledge by
checking the conceptual foundations of an argument and
inviting new ways to reinterpret important and provocative
ndings.
Restoring Conceptual Investigations
to Their Rightful Place in
Psychologys Scientific Method
What was done in the preceding examples? We did not
check the methodological soundness of experiments or the
correctness of statistical analyses, for neither of those sets
of activities is the domain of conceptual analysis. What we
did can be classied into three core activities: (a) examin-
ing whether concepts have clear meanings, (b) rendering
explicit the grammar according to which concepts are co-
ordinated, and (c) determining whether a set of theoretical
statements is consistent. If Cronbach and Meehls (1955;
see also Meehl, 1991, chapters 1 and 10) nomological
network comprises nodes (theoretical concepts) that are
connected by strands (lawful relationships holding between
the concepts), then the rst set of activities targets the
networks nodes, the second illuminates the strands, and
the third assesses their consistency. In what follows, we
briey characterize each set of activities.
The rst set consists of identifying a theorys main
concepts, characterizing their domains, and understanding
their meanings by studying how they are used in specic
cases. Thus, when a researcher wonders why we perceive
the world around us as erect if our retinal images are upside
down, the conceptual analysis identies a crucial assump-
tion related to the concept of seeing, namely, that we see
our retinal images. Similarly, when a researcher explains
that generalized or free-oating anxiety is linked to space
and time in the same sense that anxiety can be linked to a
rat (see Drury, 1973), a conceptual analysis detects a shift
in meaning from Little Albert is afraid of rats to Little
Albert is afraid of space and examines what the shift
entails (e.g., Is the theory of free-oating anxiety subject to
empirical refutation if space and time are conceived as
stimuli?). As a rule of thumb, the more open and inferential
the set of theoretical concepts or the more unfamiliar their
domains, the greater the need for the rst set of activities
comprising conceptual analysis (Cronbach & Meehl,
1955).
The second set of activities consists of increasing the
explicitness of a theory by examining the grammar of its
concepts. The importance of this activity derives from the
fact that many theoretical concepts lack explicit denitions
and are instead dened by implicit relations with other
concepts. The role of conceptual analysis is to render these
relations, whatever their source or justication, more ex-
plicit so that they can be quantied and tested (see Cron-
bach & Meehl, 1955; MacCorquodale & Meehl, 1954). As
a rule of thumb, the more commonsensical the vocabulary
of a theory, the more implicit will be its grammar and
therefore the greater the need for the second set of activities
comprising conceptual analysis.
Finally, no scientist can anticipate all consequences of
his or her presuppositions and assertions. The third set of
activities consists of checking the consistency of the theory
as it applies to a particular case. The sorts of questions one
might ask are as follows: Are there unacknowledged as-
sumptions and steps in an argument? Are the theorys
arguments sound? Are the theoretical statements under
scrutiny empirically sensitive or necessarily true (see
Smedslund, 1995)? As a rule of thumb, the greater the
degree of explicitness, articulation, and quantitative for-
malization of a theory, the more likely the theory is to
engage the third set of conceptual analytic activities.
This view of conceptual analysis is related to construct
validity (see Cook & Campbell, 1979; Cronbach & Meehl,
1955). To validate a construct, researchers must dene,
clarify, and explicate the network of meanings attached to
the construct and make explicit its assumptions. These
steps are prior to experimentation and mathematization
because unclear concepts, equivocal propositions, invalid
arguments, and incoherent accounts cannot, by denition,
achieve any form of validity. Cronbach and Meehl (1955)
expressed similar ideas when they reminded us that con-
cepts may or may not be weighted with surplus meaning
(p. 287), but when they are it is incumbent upon the
investigator to make the surplus meaning sufciently clear
that others can accept or reject it (p. 291); that is, the
meaning must be consensible. Moreover, when the [no-
mological] network is very incomplete, having many
strands missing entirely and some constructs tied in only by
tenuous threads, then the implicit denition of these con-
structs is disturbingly loose and the meaning is underde-
termined (Cronbach & Meehl, 1955, p. 294). Conceptual
analysis comprises the set of activities that have the lin-
guistic corpus of theory as their object, specic uses and
applications of theory as their context, and the objectica-
tion and logical evaluation of the semantic and syntactical
patterns of theory as their purpose.
Criticisms of Conceptual Analysis
In this article we have focused on the importance of con-
ceptual/theoretical investigations. And although conceptual
analysis is a distinct component of the scientic method,
the necessity of which is recognizable in either 17th-cen-
tury physics or 21st-century psychology, our preceding
678 October 2007

American Psychologist
analyses and arguments may themselves elicit criticism.
We anticipate three major ones.
Conceptual analysis is obvious and obvi-
ously a part of psychologys scientific method.
Some may argue that the conceptual problems in the pre-
ceding examples were obvious and that, as we have char-
acterized it, conceptual analysis is only sound reasoning.
For these reasons, few psychologists deem it necessary to
discuss conceptual analysis explicitly, dedicate entire
courses to it, or write books and articles about it. Such a
view, though, contradicts the long-standing practice of
writing about, promoting, and teaching what is critical to
experimental psychologists daily professional practices.
With regard to the obvious, we caution against such
claims. In Piagets well-known number conservation task,
a 7-year-old concludes that despite the rearrangement of a
row of candies, the number of items obviously remains the
same because none were added or subtracted. But this same
7-year-old does not remember that two years earlier she
concluded that the number of candies obviously increased
when they were spread out because the row of candies was
longer. Conceptual errors, statistical fallacies, and design
aws are more obvious to a good reviewer than a good
author (see e.g., Abelson, 1995; Campbell, 2002).
We should also be cautious of claims that conceptual
analysis is just sound reasoning. In the broadest sense, the
foundations of experimentation and mathematization may
also just be sound reasoning, a fact that does not diminish
their value. Conceptual analysis is sound reasoning, but it
is sound reasoning applied to the verbal output of scientists,
to their theories, accounts, models, and concepts. Probing
these verbal outputs for their intelligibility is comparable to
probing experiments for their methodological soundness or
quantitative models for their mathematical correctness.
Conceptual analysis is a form of verbal
tyranny. Some may say that conceptual analysis is a
form of policing peoples words and accounts, of repress-
ing the freedom to promote theoretical novelty. Conceptual
analysis, in other words, is in conict with the tolerance
necessary for scientic growth. But to say that conceptual
analysis sties theoretical creativity is akin to saying that
the rules of logic stie thinking, the rules of research design
stie experimental creativity, or that the rules of mathe-
matics stie quantitative creativity.
In a similar vein, we may recognize different tastes
when it comes to the use of constructs, metaphors, and
theories and different tolerances for conceptual and theo-
retical ambiguity. To paraphrase the proverb, trying to
close the door to all sources of ambiguity and error will
also leave truth outside. But tolerance for ambiguity and
even error does not mean that one should not strive to
reduce them. Conceptual analysis helps to detect and re-
duce ambiguity and error and thereby increase theoretical
precision.
Conceptual analysis is negative activity.
This argument contends that conceptual analysis amounts
to destructive criticism, whereas what psychology needs
most is constructive theorizing. A related concern is that
conceptual analysis tells scientists only what not to do,
whereas what scientists need most are guidelines for what
to do. Indeed, conceptual analyses can have a negative
avornot unlike the peer-review process. But it can also
be conceived positively in at least two interrelated ways: as
part of the selection process of science (Hull, 1988; Popper,
1979) and as part of scientic criticism (e.g., Abelson,
1995). Through mental mutations and recombinations, as it
were, scientists engender new hypotheses and theories and
then subject them to two broad types of selection. One is
based on the empirical adequacy of the scientists conjec-
tures, and the other on their conceptual clarity, explicitness,
and consistency. Observations and experiments, on the one
hand, and conceptual analyses, on the other hand, are lters
through which all scientic hypotheses, models, and theo-
ries must pass. In one of his epistemological theses, Karl
Popper (1960/1985) characterized the latter lter as fol-
lows:
There is no criterion of truth at our disposal, and this fact supports
pessimism. But we do possess criteria which, if we are lucky, may
allow us to recognize error and falsity. Clarity and distinctness are
not criteria of truth, but such things as obscurity or confusion may
indicate error. Similarly coherence cannot establish truth, but
incoherence and inconsistency do establish falsehood. And, when
they are recognized, our own errors provide the dim red lights
which help us in groping our way out of the darkness of our cave.
(p. 55, italics added)
And as with all scientic criticism, we need to offset its
perceived, immediate costs with its constructed, long-term
gains (see Lyons, 1965; Pearson, 1892/1957, pp. 5455).
Conclusions
Ziman (1967) stressed the social dimension of science
when he dened it as a search for consensus within a
community of rational and competent researchers. But for
scientic statements to be consensible, they must reconcile
not only facts with facts but also minds with minds. When
they fail to do the former, they are empirically false, and
science becomes social convention; when they fail to do the
latter, they are unintelligible, and science becomes an id-
iosyncratic representation of reality. The way to reconcile
facts with facts and minds with minds is to emphasize all
three clusters of activities involved in the scientic method.
Despite its importance for scientic progress, con-
ceptual analysis is acknowledged less than experimen-
tation and mathematization (see the following philoso-
phers and historians of science: Bachelard, 1934/1984;
Hacking, 1983; Laudan, 1977; Whewell, 1989; psychol-
ogists: Green, 1992; Katzko, 2002; Machado et al.,
2000; Slife & Williams, 1995; and biologists: Ghiselin,
1989; Vogel, 1998). The reasons why are undoubtedly
many and varied (Machado et al., 2000). We mention
one of a historical nature. As heirs of the scientic
revolutionseen mostly through the lenses of the En-
lightenmentpsychologists know the importance of its
two major novelties: experimentation and quantication.
Naturally, there has been less explicit concern for that
aspect of science that existed before the scientic revo-
lution, that aspect present in medieval and classic Greek
679 October 2007

American Psychologist
science, namely, conceptual analysis. Science has al-
ways included the screening of concepts and arguments
for clarity and coherence. Archimedes, one of the great
classical scientists, discoursed about the lever using the
axiomtheorem style of reasoning, in which issues of
conceptual clarity and logic take center stage. However,
even during the course of the scientic revolution, con-
ceptual analysis remained a part of Galileos scientic
method and even of Francis Bacons description of
method (see Hacking, 1983, chapter 15). It was not
abandoned in favor of experimentation and mathemati-
zation.
We return to where we started. The scientic method
comprises a large set of activities. This set may be parti-
tioned in several different ways, but in the end one should
always recognize three subsets or clusters of activities: (a)
performing controlled experiments, systematic observa-
tions, and correlational studies, (b) framing mathematical
or statistical laws, and (c) analyzing concepts and theories.
Any partition that yields only two subsets is an impover-
ished view of method. If our arguments are correct, then
psychologists should replace the currently dominant view
of the scientic method with one that assigns conceptual
analysis its proper weight. This richer view of method
would express historian of science William Whewells
(1989) famous dictum that science consists of the colliga-
tion of facts and the clarication of concepts.
REFERENCES
Abelson, R. P. (1995). Statistics as principled argument. Hillsdale, NJ:
Erlbaum.
Bachelard, G. (1984). The new scientic spirit (A. Goldhammer, Trans.).
Boston: Beacon Press. (Original work published 1934)
Bolacchi, G. (2004). On social sciences and science. Behavior and
Philosophy, 32, 465478.
Boneau, C. A. (1990). Psychological literacy: A rst approximation.
American Psychologist, 45, 891900.
Broad, W., & Wade, N. (1982). Betrayers of truth: Fraud and deceit in the
halls of science. New York: Simon & Schuster.
Burnham, J. C. (1987). How superstition won and science lost: Popular-
izing science and health in the United States. New Brunswick, NJ:
Rutgers University Press.
Campbell, S. K. (2002). Flaws and fallacies in statistical thinking. New
York: Dover.
Catania, A. C., & Reynolds, G. S. (1968). A quantitative analysis of the
responding maintained by interval schedules of reinforcement. Journal
of the Experimental Analysis of Behavior, 11, 327383.
Cook, T. D., & Campbell, D. T. (1979). Quasi-experimentation: Design &
analysis issues for eld settings. Boston: Houghton Mifin.
Crombie, A. C. (1964, May). Early concepts of the senses and the mind.
Scientic American, 210(5), 108116.
Cronbach, L. J., & Meehl, P. E. (1955). Construct validity in psycholog-
ical tests. Psychological Bulletin, 52, 281302.
Drake, S. (1957). Discoveries and opinions of Galileo. New York: Anchor
Books.
Drake, S. (1990). Galileo: Pioneer scientist. Toronto, Ontario, Canada:
University of Toronto Press.
Drury, M. OC. (1973). The danger of words. London: Routledge &
Kegan Paul.
Galilei, G. (1914). Two new sciences (H. Crew & A. Salvio, Trans.). New
York: Dover. (Original work published 1638)
Galilei, G. (1957). The assayer. In S. Drake (Ed.), Discoveries and
opinions of Galileo (pp. 229280). New York: Anchor Books. (Orig-
inal work published 1623)
Galilei, G. (2001). Dialogue concerning the two chief world systems (S.
Drake, Trans.) New York: Modern Library. (Original work published
1632)
Gallistel, C. R. (1990). The organization of learning. Cambridge, MA:
MIT Press.
Ghiselin, M. T. (1989). Intellectual compromise. New York: Paragon
House.
Gibbon, J. (1977). Scalar expectancy theory and Webers law in animal
timing. Psychological Review, 84, 279325.
Gibbon, J. (1981). On the form and location of the psychometric bisection
function for time. Journal of Mathematical Psychology, 24, 5887.
Gower, B. (1997). Scientic method: An historical and philosophical
introduction. New York: Routledge.
Green, C. D. (1992). Is unied positivism the answer to psychologys
disunity? American Psychologist, 47, 10571058.
Gregory, R. L. (1996). Mirrors in the mind. Oxford, England: Oxford
University Press.
Hacking, I. (1983). Representing and intervening. New York: Cambridge
University Press.
Harcum, E. R., & Rosen, E. F. (1993). The gatekeepers of psychology:
Evaluation of peer review by case history. Westport, CT: Praeger.
Herrnstein, R., & Boring, E. G. (Eds.). (1965). A source book in the
history of psychology. Cambridge, MA: Harvard University Press.
Hull, D. (1988). Science as a process: An evolutionary account of the
social and conceptual development of science. Chicago: University of
Chicago Press.
Katzko, M. W. (2002). The rhetoric of psychological research and the
problem of unication in psychology. American Psychologist, 57, 262
270.
Lau, H. C., Rogers, R. D., Haggard, P., & Passingham, R. E. (2004,
February 20). Attention to intention. Science, 303, 12081210.
Laudan, L. (1977). Progress and its problems: Toward a theory of
scientic growth. Berkeley: University of California Press.
Leak, T. M., & Gibbon, J. (1995). Simultaneous timing of multiple
intervals: Implications for the scalar property. Journal of Experimental
Psychology: Animal Behavior Processes, 21, 319.
Lyons, J. (1965). A primer of experimental psychology. New York: Harper
& Row.
Machado, A., Lourenco, O., & Silva, F. J. (2000). Facts, concepts, and
theories: The shape of psychologys epistemic triangle. Behavior and
Philosophy, 28, 140.
MacCorquodale, K., & Meehl, P. E. (1954). E. C. Tolman. In W. K. Estes,
S. Koch, K. MacCorquodale, P. E. Meehl, C. G. Mueller, Jr., W. N.
Schoenfeld, & W. S. Verplank (Eds.), Modern learning theory: A
critical analysis of ve examples (pp. 177266). New York: Appleton-
Century-Crofts.
Meehl, P. E. (1991). Selected philosophical and methodological papers.
Minneapolis: University of Minnesota Press.
Pearson, K. (1957). The grammar of science. New York: Meridian Li-
brary. (Original work published 1892)
Peters, D. P., & Ceci, S. J. (1982). Peer-review practices of psychological
journals: The fate of published articles, submitted again. Behavioral
and Brain Sciences, 5, 187255.
Popper, K. (1959). The logic of scientic theory. New York: Basic Books.
Popper, K. (1979). Objective knowledge: An evolutionary approach (rev.
ed.). Oxford, England: Clarendon Press.
Popper, K. (1985). Knowledge without authority. In D. Miller (Ed.).
Popper selections. Princeton, NJ: Princeton University Press. (Original
work published 1960)
Pylyshyn, Z. (2003). Return of the mental image: Are there really pictures
in the brain? Trends in Cognitive Sciences, 7, 113118.
Quine, W. V. (1936). Truth by convention. In O. H. Lee (Ed.), Philo-
sophical essays for Alfred North Whitehead (pp. 90124). New York:
Longmans, Green.
Slife, B. D., & Williams, R. N. (1995). Whats behind the research?
Discovering hidden assumptions in the behavioral sciences. London:
Sage.
Smedslund, J. (1995). Psychologic: Common sense and the pseudoem-
pirical. In J. A. Smith, R. Harre, & L. Van Langenhove (Eds.), Rethink-
ing psychology (pp. 196206). London: Sage.
Stanovich, K. E. (1998). How to think straight about psychology (5th ed).
New York: HarperCollins.
680 October 2007

American Psychologist
Uller, C., Hauser, M., & Carey, S. (2001). Spontaneous representation of
number in Cotton-Top Tamarins. Journal of Comparative Psychology,
115, 248257.
Vogel, S. (1998). Academically correct biology. American Scientist, 86,
504506.
Wade, N. J. (1998). A natural history of vision. Cambridge, MA: MIT
Press.
West, R. E., & Young, R. J. (2002). Do domestic dogs show any evidence
of being able to count? Animal Cognition, 5, 183186.
Whewell, W. (1989). Theory of the scientic method (R. E. Butts, Ed.).
Indianapolis, IN: Hackett.
Wynn, K. (1992). Addition and subtraction by human infants. Nature,
358, 749750.
Wynn, K. (1995). Infants possess a system of numerical knowledge. New
Directions in Psychological Science, 6, 172177.
Ziman, J. (1967). Public knowledge. Cambridge, England: Cambridge
University Press.
681 October 2007

American Psychologist