Rescher 2001 Paradoxes

Paradoxes
Their Roots, Range, and Resolution
Nicholas Rescher
Paradoxes
Paradoxes
Their Roots, Range,
and Resolution
NICHOlAS RESCHER
OPEN COURT
Chicago and La Salle, Illinois
To order books from Open Court, call toll free 1-800-815-2280

Open Court Publishing Company is a division ofCarus Publishing Company.
Copyright 2001 by Carus Publishing Company
First printing 2001
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 written permission of the
publisher, Open Court Publishing Company, a division ofCarus Publishing Company,
315 Fifth Street, P.O. Box 300, Peru, Illinois 61354-0300.
Printed and bound in the United States of America.
Library of Congress Cataloging-in-Publication Data

Rescher, Nicholas.
Paradoxes: their roots, range, and resolution / Nicholas Rescher.
p. cm.
Includes bibliographical references and index.
ISBN 0-8126-9436-8 (aile paper) - ISBN 0-8126-9437-6 (pbk. : alk. paper)
1. Paradox. 1. Tide.
BC199.P2 R47 2001
165--dc21
2001021224
For Rudolf Haller
Brief Contents
Detailed Contents
Preface
Index of Paradoxes Discussed
Summary of Symbols, Technical Terms,
and Principles
1
2
3
4
5
6
7
8
9
10
11
12
13
Aporetics
Paradox Solution via Retention Prioritization
Plausibility as a Guide to Prioritization
Levels of Paradoxicality
Paradoxes of Vagueness
Paradoxes of Ambiguity and Equivocation
(Insufficient Distinctions)
Philosophical Paradoxes
Paradoxes of Inappropriate Presupposition
(Improper Totalities in Particular)
Mathematical Paradoxes
Semantical Paradoxes that Involve Conflicting
Claims regarding Truth
Inductive Paradoxes (Conflicts of Probability
and Evidence)
Paradoxes of Hypothetical Reasoning
(The Problem of Assumptions that Conflict
with Accepted Beliefs)
Paradoxes of Choice and Decision (Conflicting
Reasons for Action)
Bibliography
Index
VII
XIIl
xv
XiX
3
25
43
57
77
91
117
137
159
193
219
233
255
281
289
Detailed Contents
Preface
Summary of Symbols, Technical Terms, and
Principles
Paradoxes Discussed in Chapter 1
Aporetics
1.1 Apories and Paradoxes
1.2 Paradoxes Root in Overcommitment
1.3 Plausibility
1.4 Plausibility and Presumption
1.5 The Acceptability of Contentions
Xlll
xv
xix
1
3
3
9
15
17
20
23
25

2.1
Addressing Paradox Resolution Via Priority

Rating
2.2 Procedural Considerations
2.3 Disjunctive Resolutions: Conflicting Report
Paradoxes
2.4 Compound Paradoxes
25
32
34
37
41
43

3.1
Theses of Lowest Priority: The Meaningless

and the False
3.2 Comparative Plausibility and Prioritization
3.3 Issues of Plausibility: Routes to Plausibility
Assessment and Priority Determination
3.4 Against Semantic Absolutism
ix
43
46
47
51
Detailed Contents
55
57
57
59
61
65
70
4.1 Four Modes of Paradox Solution
4.2 Dissolvable Paradoxes
4.3 Decisively Resolvable Paradoxes
4.4 Indecisively Resolvable Paradoxes
4.5 The Classification of Paradoxes
75
77
77
83
84
87
5.1
5.2
5.3
5.4
The Paradox of the Heap

The Color-Continuum Paradox
Sir John Cutler's Stockings
The Hermeneutic Circle

6.1 Equivocation as a Road to Paradox
6.2 Zeno's Paradox of Differentiation
6.3 Zeno's Space Paradoxes
6.4 Zeno's Achilles and the Tortoise Paradox
6.5 Zeno's Paradox of the Arrow
6.6 Zeno's Paradox of the Moving Bodies
6.7 The Relation Riddles of Eubulides
6.8 Further Riddles
6.9 The Master Argument of Diodorus
(Kyrieuon)
6.10 Sameness Amidst Change: Mind-Body
Equivocation
6.11 Frege's Morning Star Paradox
6.12 The Categorization Paradox
6.13 The Surprise Examination (or Execution)
89
91
91
94
96
98
100
101
102
104
107
108
110
111
112
Detailed Contents
Xl
115
117
117
123
125
126
7.1 Philosophical Paradoxes
7.2 Philosophical Apories Tie Issues Together
7.3 Philosophical Plausibility
7.4 The Role of Distinctions

(Improper Totalities in Particular)
8.1
8.2
8.3
Unwarranted Presuppositions
Paradoxical Questions
Negatively Totalitarian Self-Involvement:
The Barber Paradox
8.4 Illicit Totalization Paradoxes
8.5 Immanuel Kant's Antinomies
8.6 Predictive Paradoxes
8.7 Summary
135
137
137
141
143
145
147
151
154
157
159
9.1
9.2
9.3
9.4
9.5
9.6
9.7
9.8
9.9
9.10
Cantor's Paradox and the Viable Introduction

Requirement (VIR)
The Successful Identification Principle (SIP)
Pseudo-Objects
Russell's Paradox
Paradoxes Related to Russell's: Curry's and
Grelling's
Berry's Paradox and its Ramifications
Richard's Paradox
Burali-Forti's Paradox
Addressing the Mathematical Paradoxes
The Efficacy of the Successful Identification
Requirement (SIR)
159
164
167
170
173
176
181
184
184
188
Detailed Contents
xu
10 Semantical Paradoxes that Involve

Conflicting Claims regarding Truth
10.1
10.2
10.3
10.4
10.5
Self-Falsification
The Liar and His Cousins
Problems of Self-Reference
Paradoxes of Truth-Claim Dissonance
The Preface Paradox
11 Inductive Paradoxes (Conflicts of

Probability and Evidence)
11.1
11.2
11.3
11.4
11.5
Evidential Indecisiveness
Self- Deception
The Lottery Paradox
Hempel's Paradox of the Ravens
Goodman's Grue/Bleen Paradox
12 Paradoxes of Hypothetical Reasoning

(The Problem of Assumptions that Conflict
12.1
12.2
12.3
12.4
Hypotheses and Burley's Principle

Counterfactual Conditionals
Further Examples
The Crucial Difference Between Factual and
Hypothetical Contexts
12.5 Reductio-ad-Absurdum Reasoning
12.6 Thompson's Lamp as an Illustration of
Reductio Reasoning
12.7 Per-Impossibile Reasoning
191
193
193
199
203
207
213
217
219
219
221
222
224
227
231
233
233
238
242
244
245
247
249
Detailed Contents
13 Paradoxes of Choice and Decision

(Conflicting Reasons for Action)
13.1
13.2
13.3
13.4
13.5
13.6
A Sample Paradox of Decision

The Prisoner's Dilemma
Newcomb's Perverse Predictor Paradox
Allais's Paradox and the St. Petersburg Paradox
The Dr. Psycho Paradox
Retrospect: The Pragmatic Dimension
Bibliography
Index
Xlll
253
255
255
260
264
267
269
275
281
289
Preface
The principal aim of the book is to show that a wide variety of
cognitive issues-not only the standardly recognized paradoxes
but also counterfactual conditionals and various sorts of self-reference problems--can all be accommodated by common analytical
means and have their perplexities resolved by use of the same
machinery of inconsistency management. On this basis aporetics
emerges as a versatile and fruitful instrumentality of issue clarification and problem solving.
More than a hundred and thirty major and minor paradoxes
are analyzed in the course of the discussion-including virtually all
of the best known instances of this phenomenon. Entire books
can and have been written about many of these individual paradoxes, so it should be clear from the outset that none of them will
be dealt with in their full and often complex detail. Our interest
here is in the general theory of paradoxes-in those shared and
common features in virtue of which a generic approach to paradox analysis can be articulated. On this basis, consideration of
individual paradoxes will be merely illustrative. They will be
treated not in their full and idiosyncratic detail but as case studies
that exemplify significant aspects of a general theory of paradox
analysis. So I must plead for understanding and forgiveness if the
reader is disappointed to find his or her own favorite solution of a
particular paradox overlooked. For the aim here is entirely
methodological. It is the generic procedure of dealing with paradoxes that is at the forefront here, and particular paradoxes are
treated merely as illustrations and not with a more specifically
doctrinal end in view. The aim of the present discussion is to present a way of handling paradoxes in general.
The book aims to provide some significant innovations regarding the subject at large. These innovations specifically include:
The standardization of paradox analysis through the
machinery of general aporetics.
The exploration of the distinction between truth and plausibility as an instrument of paradox analysis.
xv
XVl
Preface
The utilization of epistemic prioritization in paradox resolution and the exploration of the role of plausibility rankings in this regard.
The specification of acceptance profiles for the comparative
assessment of alternative resolutions of a paradox.
Introduction of the idea of cycles of inconsistency into
paradox analysis and implementation of the corresponding
distinction between simple and complex paradoxes.
Adoption of the Successful Identification Requirement
(SIR) as an instrument of paradox resolution adequate to
deal with the entire range of mathematical paradoxes.
The use of the Viable Introduction Principle (SIP) as a simpler and more natural replacement for Russell's Vicious
Circle Principle and its congeners in resolving semantical
paradoxes.
All in all, the aim of the book is to show that paradox analysis
is a more straightforward matter than is generally thought-and
yet not just to maintain this in the abstract but to illustrate it in
comprehensive and diversified detail.
I am grateful to several colleagues who have read the book in
manuscript and have suggested corrections and improvements, in
particular Carl Davulis, Peter Davison Galle, Alexander Pruss, and
Neil Tennant.

Absurd Instruction Paradoxes
138-39
98-99
Achilles Paradox (Zeno)
142
Adult-Child Paradox
267
Allais's Paradox
147-151
Antinomies (Kant)
Arrow Paradox (Zeno)
100-01
Bald Man Paradox or Phalakros (Eubulides)
82-83
143-44
Barber Paradox
Berry's Paradox
176-181
144-45
Bibliography Paradox
Bizet-Verdi Paradox
68-70
179-180,209-212
Box Paradoxes
62n2
Bridgekeeper Paradox
184
Buridan's Ass Paradox
273-75
Cantor's Paradox
161-64
Categorization Paradox
III
258
"Chicken" Paradox
Color Continuum Paradox
83-84
Compound Paradox
38-39
Conflicting Report Paradoxes
35-36
63-65
Contract of Protagoras Paradox
Cosmic Explanation Paradox
61
Counterfactual Hypothesis Paradoxes (various)
233-245
Curry's Paradox
173
De Morgan's Paradox
164
14--15
Deduction Paradox
256-57
Demands of Reason Paradox
Differentiation, Paradox of (Zeno )
94--96
34--35
Dilemmatic Paradoxes
Discrepant Information Paradoxes
9-15
122-23
Divine Foreknowledge Paradox
130--32
Divine Omnipotence Paradox
12n
Dog Paradox
269-272
Dr. Psycho Paradox
xvii
xviii
Dualized Liar Paradox

Electra and Orestes Paradox (Eubulides)
Equivocal Evidence Paradoxes
Evil, Paradox of
Exception Paradox
Existence, Paradox of
Existential Paradox
Explanation, Paradox of
Form-Matter Paradox of the Presocratics
Freedom-Causality Paradox
Frege's Morning Star Paradox
Garbage Truck Paradox
God Paradox, Kierkegaard's
Grelling's "Heterological" Paradox
Grue/Bleen Paradox (Goodman)
Hanging Paradox
Happiness Paradox, Mill's
Heap Paradox or Sorites (Eubulides)
Hempel's Ravens Paradox
Hermeneutic Circle Paradox
Heterological Paradox (Grelling)
Hilbert's Hotel Paradox
Horn Paradox (Eubulides)
Hotel Paradox (Hilbert)
Illicit Self-Reference Paradoxes
Illicit Totalization Paradoxes
Incommensurability of the Diagonal
Instantiation Paradox
Justification Paradox, Aristotle's
Kantian Antinomies
Kierkegaard's God Paradox
Liar Chain Paradox
Liar Paradox of Epimenides the Cretan
(Eubulides)
Lottery Paradox
Masked Man Paradox (Eubulides)
Master Argument Paradox
Mikado Paradox
Mill's Happiness Paradox
Millet Seed Paradox (Zeno)
208-09
103
220-21
30-31
17
65-67
120-21
123-24
127-29
124-25
110--11
105
119-120
174-75
227-230
62-63
28-29
77-82
224-26
87
174-75
160-61
12,140
160-61
165-67,203-04
145-151, 172n
246
213
60-61
147-151
119-120
209
81-82,199-202
52-53,222-24
102
107-08
138-139
28-29
83
Mind -Body Paradox

Moore's Paradox
Morning Star Paradox (Frege)
Motion Paradoxes, Zeno's
Moving Body Paradox (Zeno)
Nasty Crocodile Paradox
"Never Say Never"
Newcomb's Pervasive Predictor Paradox
Of-Two-Minds Paradox
Optical Illusion Paradoxes
Overlooked Man Paradox (Eubulides)
Paradox of Change, Aristotle's
Paradoxes of Equivocal Evidence
Parent-Beater Paradox
Per-Impossible Style Paradoxes
Perverse Predictor Paradox (Newcomb)
Plausibility Paradox
Predictive Competition Paradox
Predictive Paradoxes
Preface Paradox
Prisoner's Dilemma Paradox
Promise-Breaking Paradox
Propositionalized Instantiation Paradox
Protagoras, Contract of
Racetrack Paradox (Zeno)
Ravens Paradox (Hempel)
Reductio-style Paradoxes
Reiteration Paradox
Relation Riddles (Eubulides)
Richard's Paradox
Rule- Exception Paradoxes
Russell's Impredicability Paradox
Russell's Paradox of Sets
Russell's Relational Paradox
Russell-Gettier Paradox
Sancho Panza's Hanging Paradox
Sea Battle Paradox, Aristotle's
Self-Advantage Paradox
Self-Contradiction, Paradoxes of
Self-Counterexampling Paradoxes
XIX
108-110
44-45
110-11
98-102
101-02
62n
8
264-66
221-22
46-47
103
93-93
219-221
139-140
249-252
264-66
93-94
266
151-54
213-15
260-63
126
213
63-65
59-60
224-26
245-49
104-05
102-04
181-83
141-42
173-74,213
170--73
176
67-68
62-63
118-19
132-33
20-22
193-94
xx
Self-Deception Paradoxes
Self-Falsification Paradoxes
Self-Negation Paradox
Sense-Deception Paradoxes
Ship of Theseus Paradox
Sir John Cutler's Stockings, Paradox of
Smashed Vase Paradox
Socratic Ignorance Paradox
Socratic Paradox of Ethics
SomethingjNothing Paradox
Sorites or Heap Paradox
Space Paradoxes, Zeno's
St. Petersburg Paradox
Supertruth Paradox
Surprise Examination (or Execution) Paradox
Teacher as Assassin Paradox
Third Man Paradox
Thompson's Lamp Paradox
Time Travel Paradox
Truth Paradox
Truth-Status Assignment Paradoxes
Unachievable Wisdom Paradox
Valid Argumentation Paradox
Voting Paradox
"Who is the Engineer?" Paradox
Zeno's Achilles Paradox
Zeno's Arrow Paradox
Zeno's Motion Paradoxes
Zeno's Moving Bodies Paradox
Zeno's Paradox of Differentiation
Zeno's Racetrack Paradox
Zeno's Space Paradoxes
221-22
194-202
203
8,46-47
86
84-86
27-28
32-33
129-130
106-07
77-82
96-98
268-69
167-69
112-14
93
117-18
247-49
121-22
12-13
209-213
258-260
169-170
255-56
13-14
98-99
100-01
98-102
101-02
94-96
59-60
96-98
Summary of Symbols,
Technical Terms, and Principles
SYMBOLS
Standard logic symbols
propositional logic: -, &, v,
~,~
quantificationallogic: 3, 'if
identity (or definition): =
deductive entailment: r
inductive implication: :}
Priority ordering: >
Set-theoretic notation: I J,
E,
Ix: Fx}
Item identification: (t x) ex
NOTE: Ix: Fx} = (ty)'ifx(xEy~Fx)
TECHNICAL TERMS
aporetic cluster: a set of propositions that are individually plausible
but collectively inconsistent. [NOTE: An aporetic cluster constitutes a paradox.]
MCS (maximal consistent subset): any consistent subset of an
aporetic cluster that becomes inconsistent through the addition of anyone of its other members.
xxi
XXII
Summary of Symbols, Technical Terms, and Principles
inconsistent n-et: a group of n collectively inconsistent propositions

that becomes consistent if anyone of its members is omitted.
RIA-alternatives (retention/abandonment alternatives): anyalternative for rendering an inconsistent set maximally consistent
that is the result of retaining some of its members and omitting others. [NOTE: such an R/A-alternative can be presented
in the format ... / ... , by indicating to the left of the slash
the propositions that are retained and to its right those that are
abandoned.
cycle of inconsistency. a minimally inconsistent subset of an inconsistent set of propositions. [NOTE: Any sU<i:h cycle of inconsistency constitutes an inconsistent n-et.]
priority ordering: a specification of the retention-priority relations
among the propositions of a set. For example the inconsistent
quartet {p, q, r, s} might have the priority order [p, r] > s > q
indicating that p and r have a like priority status, superior to
that of s which in turn is superior to that of q.
retention profile (of an R/A alternative): a specification of the
proportion of propositions in the successive priority categories
whose retention that R/A-alternative permits. [Thus in the
preceding example, of the R/A-alternative p, q, r/s would have
the retention profile {I, 0, I} indicating that it provides for
the retention of all of the priority-l statements, none of the
priority-2 statements, and all of the priority-3 statements.]
PRINCIPLES OF IDENTIFICATORY
MEANINGFULNESS
VIR (Valid Identification Requirement): For something to be
appropriately characterized by an identifactory specification of
the format "the x that satisfies the condition C-( symbolically
(tx)Cx-the existence of a unique item of this description
Summary of Symbols, Technical Terms, and Principles
xxiii
must be independently pre-established. Otherwise this supposed item remains undefined.

SIP (Successful Introduction Principle): For something to be successfully introduced into the discussion as: "the item x that satisfies the condition C," the identifying condition C at issue
must not itself presuppose the existence of this item. A meaningful introduction cannot presume that the item being introduced has already been identified. (This goes back to the
"independently pre-established" of VIR. )
VCP (Bertrand Russell's Vicious Circle Principle): No collection
can contain members defined in terms of itself.
Paradoxes Considered in Chapter 1

"Never Say Never"
Sense-Deception Paradoxes
The Paradox of the Horns (Eubulides)
The Truth Paradox
The "Who is the Engineer?" Paradox
The Deduction Paradox
The Exception Paradox
Paradoxes of Self-Contradiction
Aporetics
1.1
Apories and Paradoxes
Every era in the history of philosophy has seen a concern with

paradoxes. To be sure, the pioneering Zeno of Elea (born ca. 500
B.C.) never called his paradoxes by that name, and even Aristotle
often referred to them simply as "arguments" (logoi), though he
denounced them as fallacies (paralogisms, paralogismoi). Another
term used for such phenomena by Plato and Aristotle was
"sophisms" (sophismata)-though it is doubtful that the Sophists
themselves had employed it to characterize their arguments.
These apories (aporia )-paradoxes by another name--concerned
Aristotle in his Sophistical Refutations and the Stoics also devoted
much interest to this subject-not to speak of the Sceptics to
whose mill it brought much grist. The medieval schoolmen were
also enthralled by it, and some of their principal thinkers wrote
extensively on insolubilia, as they somewhat pessimistically characterized such puzzling modes of argumentation. Kant treated the
cognate issues under the title paralogisms and antinomies. In the
nineteenth century, however, the term "paradox" ultimately
established itself as standard for the phenomena at issue. But even
as a rose by any other name would smell as sweet, so it is that
throughout this long history one and the same thing-paradoxical argumentation-has been at issue throughout. In this regard
the same fundamental problems-and often the same standard
examples-have preoccupied logically acute thinkers since the
dawn of philosophy.
The word "paradox" derives from the Greek para (beyond)
and doxa (belief): a paradox is literally a contention or group of
contentions that is incredible-beyond belief. In the root sense of
the term, paradoxes are thus a matter of far-fetched opinions,
Chapter 1
curious ideas, outlandish occurrences and such-like anomalies in

general that run counter to ordinary expectations. 1 The paradoxes
discussed by the German polymath Max Nordau in his book of
that title 2 are hardly more than eccentric opinions.
One must distinguish between logical and rhetorical paradoxes. The former type is a communicative predicament-a conflict of what is asserted, accepted, or believed. The latter is a
rhetorical trope-an anomalous juxtaposition of incongruous
ideas for the sake of striking exposition or unexpected insight.
That waking life is but a dream or that the painter immobilizes
and renders lifeless the living subject he is trying to portray are
paradoxes alright, but in the merely rhetorical mode. And much
the same is true of Schopenhauer's thesis that suicide is the
supreme expression of the will-to-live. People often say things like
"Freud was no Freudian" or "That poem of his was no poem."
Paradoxes of this rhetorical sort-"Nature imitates art"-are
pieces of descriptive perversity, instances of what G.K. Chesterton
called "Truth standing on her head to elicit attention."3 This
clearly holds for Shakespeare's "There's method in his madness"
as well as such obiter dicta as G.B. Shaw's "The Golden Rule is
that there is no Golden Rule" or Oscar Wilde's "I can resist
everything except temptation," or Chesterton's own "Spies do
not look like spies." As with Sam Goldwin's classic "Include me
out," what we have here are duplicitous formulations that skate
on the thin borderline between sense and nonsense.
Rhetorical paradoxes often pave the way to logical paradoxes.
Consider a dictum like "A driver is just an automobile's means for
transporting itself to another place." This picturesque version of
the usual auto-driver relationship of means and ends runs afoul of
our realization that automobiles are inert instrumentalities and
drivers are purposive agents, so that it is the drivers who are
bound to occupy "the driver's seat."
1. Augustus de Morgan's A Budget of Paradoxes (London: Longmans Green, 1872;
2nd ed. 1915) is replete with material of this sort, largely drawn from mathematics and
natural science. Travel to the moon and shopkeepers who write books on Kant's philosophy are among the innumerable paradoxes proposed by the authors of intellectual curiosities cited by de Morgan.
2. Max Nordau, Paradoxes (Chicago: Laird and Less, 1886).
3. G.K Chesterton, The Paradoxes of Mr. Pond (New York: Dodd Mead, 1937), p.
63. Chesterton went on to observe that "Most fellows who talk paradoxes are only trying
to show off" (p. 95).
Aporetics
Not only are there extensive scholarly discussions about paradoxes, but there is also a literature that revels in paradoxes and
unfolds episodes and narratives based on them. Belletristic writers
such as Rabelais, Cervantes, and Sterne as well as G.K. Chesterton
and J.L. Borges corne to mind here. An excellent study by R.L.
Colie examines the early phases of this literature which does, however, fall outside the focus of present concern. 4
The metaphysical poet George Herbert in his devotional poem
"Providence" wrote: "Thou art in all things one, in each thing
many/For thou art to infinite in one and all" (Lines 43-44).
Christian discourse is full of such rhetorical paradoxes. Thus
Sebastian Franck (1499-1542) published in 1534 his book
Paradoxa, a collection of 280 "paradoxes" culled from biblical
texts. 5 These include such dicta as "Victory is with the vanquished" (Triumphos penes victos) and "God is never nearer than
when far away" (Non est proprior quam procul absens deus) and
"Faith believes in unbelief' (Fides in incredulitate credit). But
Christian discourse has its logical paradoxes as well. St. Paul
explained to the Athenians that he carne to make known to them
an unknown and unknowable God. And at the religion's center
lies the great Christian paradox that death is the gateway to eternallife. However it is not with the preachers but with the theologians that we generally encounter actually communicative
paradoxes, the doctrine of the Trinity being a classic case in point.
It was with a view to issues of this sort that the theologian
Tertullian was led to his notorious dictum: credo quia absurdum
("I believe because it is absurd").
The technical literature of logic, mathematics, and philosophy
is also rife with discussions of various types of paradoxes. But
these are generally dealt with as so many individual and isolated
episodes, each requiring its own characteristic mode of approach.
There has been no attempt at an uniform, across-the-board
approach to the subject of paradoxes and their resolution. By contrast, the present discussion sees paradox as a generically uniform
4. Rabelais, Gargantua and Pentagrual; Cervantes, Don QJtixotc; Laurence Sterne,
Tristram Shandy, G.K Chesterton, The Paradoxes of Mr. Pond; and Jorge Luis Borges,
Labyrinths, are prime examples of paradox-oriented literature. Sec also Rosalie L. Colic,
Paradoxia Epidemica (Princeton: Princeton University Press, 1966).
5. Sebastian Franck, 280 Paradoxes or Wondrous Sayings. Translated by E.J. Furcha
(Lewistown, NY: Edward Mellen Press, 1986).
Chapter 1
phenomenon that calls for a uniform treatment. It sets out to

sketch a discipline (aporetics one may call it) concerned specifically with the general issues that arise with paradox resolution
across-the- board.
In the common usage of everyday discourse a paradox is a
judgment or opinion that is contrary to the general opinion or
"common sense.,,6 A paradox on this basis would be an obviously
anomalous contention that someone seriously propounds despite
its conflict with what is generally regarded as true. Cicero thus
sensibly observed that "what they, the Greeks, call paradoxes, are
what we Romans call marvels."7
Among philosophers and logicians, however, the term has
come to acquire a more specific sense, with a paradox arising
when plausible premisses entail a conclusion whose negation is
also plausible. We thus have a paradox when a set of individually
plausible theses {P , ... , P } validly entails a conclusion C whose
negation non-C is itselfpla~sible. And this means that the set {p),
P , ... P , non - C} is such that all of its members are individually
pfausibl; while nevertheless logically inconsistent overall.
Accordingly, a variant but equivalent way of defining the term is
by specifying that a paradox arises when a set of individually plausible propositions is collectively inconsistent. 8 And the inconsistency
at issue here must be real rather than merely seeming. Paradox is
the product not of a mistake in reasoning but of a defect of sub6. It is ironic that one of the first uses of the word in English recorded in the Oxford
English Dictionary is for the 1616 definition of Bullokar's Chapbook, which reads as follows: "Paradox, an opinion maintained contrary to the commonly allowed opinion, as if
one affinne that the earth doth move round and the heavens stand still."
7. Hac paradoxa illi, admirabilia dicamus, and again, admirabilia contraquc opinioncm ommium; Paradoxa, Prooem, 4. (Scripta quae manserunt omnia, IVjIll, ed.
C.F.W. Mueller (Leipzig: Teubner, 1878), p_ 198. See also Academicorum priorum, II
44, 136 [ibid IV.I, p. 81].
8. An edition of Webster's dictionary defines paradox (in the logical sense) as "an
argument that apparently derives a self-contradictory statement by valid deduction from
acceptable prernisses"(a definition that would be closer to the mark if that "apparently"
were shifted from qualifying deriJles to qualifying acceptable premisses. Webster's Ninth New
Collegiate Dictionary (Springfield, MA: Merriam-Webster, 1983), p. 853. The present
definition differs from this specification in two main respects: (1) It speaks of plausible
propositions or premisses instead of "acceptable" ones, the reason for this being that
"acceptability" is a matter of yes-or-no-a thesis is either apparently acceptable or notwhile plausibility is clearly a matter of degree since a thesis can be more or less plausible;
(2) The present definition would substitute "derives" for "apparently derives." With fallacies we have apparent rather than real derivation, but the reasoning of a genuine paradox must be cogent.
Aporetics
stance: a dissonance of endorsements. A variant way of looking at

a paradox is therefore as a conflict-engendering at;gument, a piece
of deductive reasoning to a contradictory result. The several theses that compose the paradox are its premisses, and its conclusion is
uniformly a statement of the format: "Premiss X is incompatible
with Premiss y''' For already Aristotle characterized as "dialectical" arguments those which create a paradox by deriving a contradiction from generally accepted opinions. 9 The expression
antinomy is also used to characterize the assertions giving rise to
such anomalous situations.
When claims involve a conflict or tension that stops short of
actual logical conflict-of strict inconsistency-this is generally
indicative of irony rather than paradox. It is ironic (rather than
paradoxical) that the competent general must both protect his
soldiery and endanger them by use, and that he cannot do the
one without foregoing the other. It is ironic (rather than paradoxical) that the individual soldier cannot pursue glory without
putting his life at risk.
Moreover, what Communist theoreticians called a "contradiction"-that is, a conflict or tension between discordant factors or
opposing forces, a phenomenon they thought to find in capitalism-is not the sort of "inconsistency" at issue with actual paradoxes. Mere tensions as such do not engender paradox. If the
lesson of Bernard de Mandeville's "Fable of the Bees" is correct,
it will be ironic (rather than actually paradoxical) that the conditions needed for a society to flourish economically demand a
comparatively low level of such traditional economic virtues as
frugality, prudence, and frivolity-avoidance.
Paradoxes thus arise when we have a plurality of theses, each
individually plausible in the circumstances, but nevertheless in the
aggregate constituting an inconsistent group. In this way, logical
paradoxes always constituted aporetic situations, an apory being a
group of acceptable-seeming propositions that are collectively
inconsistent. Viewed separately, every member of such a group
stakes a claim that we would be minded to accept if such acceptance were unproblematic. But when all these claims are conjoined, a logical contradiction ensues.
9. Soph. Bien., 16Sbl-S.
Chapter 1
Sensory illusion paradoxes, for example, occur in situations in

which we cannot even "trust our own eyes," so to speak, as in the
well-known case of the straight stick that looks bent when held at
an angle under water. Sight tells us it is bent; touch insists that it
is straight. Or again consider the situation of looking at a landscape over a fire or a hot radiator. The trees and buildings vibrate
and shake, while nevertheless "we know perfectly well" that they
are immobile. Here we confront a Sense-Deception Paradox where
one sense affirms what the others deny. In all such cases we have
to make up our minds about how to resolve the inconsistency.
Since logical paradox arises through the collective inconsistency of individually plausible propositions, it is clear that the
individual premisses of a paradox must be self-consistent.
(Otherwise they could hardly qualify as plausible.) And this
means that those self-inconsistent propositions that are commonly characterized as paradoxical, have to be viewed in a larger
context before actual logical paradox ensues.
Thus consider the proposition
(N) It is never correct to claim that something is never the
case. ("Never say never!")
To make manifest that an actual (logical) paradox is at issue here
we can elaborate the situation as follows:
(1) N makes tenable claims.
(2) N is a statement of the "it is never the case" format.
(3) If N is correct, then every statement of the "it is never the
case" format is false.
(4) Nentails its own falsehood. By (2), (3).
(5) N does not make a tenable claim. By (4), since no claim
that entails its own falsehood is tenable.
(6) (5) contradicts (1).
It is this expanded account which elaborates the contradiction to
which N gives rise, rather than simply N itself, that represents the
paradox at issue.
Aporetics
1.2 Paradoxes Root in Overcommitment

Since paradoxes arise through a clash or conflict among our commitments, they are the product of cognitive overcommitment. We
regard more as plausible than the realm of fact and reality is able
to accommodate, as is attested by our falling into contradiction.
Paradox thus roots in an information overload, a literal embarrassment of riches.
In theory, three sorts of cases can arise, according as the data
at our disposal provide enough, too little, or too much information to underwrite a conclusion. The general situation at issue
here can be seen in the following example of three pairs of equations in two unknowns, which poses the problem of determining
the values of the parameters x and y.
Case I
(Too Little)
Case II
(Enough)
Case III
(Too Much)
x+y=2
3x+ 3y= 6
x+ y= 3
x- y= 3
x+y=4
2x + 2y= 5
In the first case there is too little information to make a determination possible, in the second case just enough, and in the third
too much. For the information overload of case III creates an
inconsistent situation. This inherent conflict of the claims at issue
renders them collectively inconsistent and thereby paradoxical.
The prime directive of rationality is to restore consistency
in such situations. To be sure, it is a possible reaction to paradox
simply to take contradictions in our stride. With Pascal we might
accept contradictions for the sake of greater interests, and say
that "a la fin de chacque verite, il faut ajouter qu'on se souvient
de la verite opposee" ("after every truth one must be mindful
of the opposite truth. ")10 The Greek philosopher Protagoras
(born ca. 480 B.C.), the founding father of the Sophistic school,ll
10. Blaise Pascal, Pensees, No. 791; Modern Library edition No. 566, p. 185. Recall
the dictum that "consistency is the hobgoblin of small minds."
11. For Protagoras see Zeller, Philosophie der Griechen, Vol. 1/2, pp. 1296-1304. See
also Plato's dialogues Protagoras and The Sophist.
10
Chapter 1
notoriously held that the human situation was in this way paradoxical throughout, and that anything and everything that we
believed could be argued for pro and con with equal cogency. And
he was perfectly prepared to accept the deeper paradoxical consequence that this holds also for the very thesis at issue. 12 But this
sort of resignation in the face of inconsistency is hardly a comfortable-let alone a rational-posture. Even if one's sympathies were
so inordinately wide, inconsistency tolerance should be viewed as
a position of last resort, to be adopted only after all else has failed
US. 13
Yet once consistency is lost, how is it to be regained? Any and

every paradox can be resolved by simply abandoning some or all
of the commitments whose conjoining creates a contradiction. In
principle, paradox management is thus a straightforward process:
to appraise the comparative plausibility of what we accept and
restore consistency by making what is less plausible give way to
what is more so. It is this generic and uniform structure of paradox management that makes a generic and uniform approach to
their rational management possible, paving the way to that single
overarching discipline of aporetics (to avoid using the somewhat
barbarous "paradoxology").14 The exploration of this domain is
the principal task of the present book whose central thesis is that
there indeed is such a general and uniform approach to the rational management of paradoxes.
Consider an example. Let it be supposed that three (reasonably reliable) sources report on a sighting of birds flying by. The
first says there were around three of them; the second says there
were around five; the third says that there was "a small flock" and
that it was odd in number. Now suppose further that it is important for us to have a definite estimate. What is it to be?
12. "Protagoras ait, de omnis re in utrumque partem disputari posse ex aequo, et de hac
ipsa, an ominis res in utrumque partem disputabilis sit." Seneca, Epistola moralia, XIV, 88,
43.
13. Compare N. Rescher and Robert Brandom, The Logic of Inconsistency (Oxford:
Blackwell, 1979).
14. The terminology is adapted from the Greek aporetike techne, the "aporetic art" of
articulating, analyzing, and resolving aporia or paradoxes. The conception of such an
enterprise as central to philosophy figured prominently in the work of the German philosopher Nicolai Hartmann (1882-1950). See his Grundzuge einer Metaphysik dey Erkenntnis
(Berlin: de Gruyter, 1921; 5th ed. 1965).
Aporetics
11
The situation is aporetic. We have three contentions:

(1) There were around three birds (that is, 2,3, or 4).
(2) There were around five birds (that is, 4,5, or 6).
(3) The number of birds was smallish but odd (that is, 3, 5,
or 7).
We clearly cannot accept all of these claims since there is no number common to them all. Though each has a basis of evidential
support, they are collectively incompatible. Any two of them yield
a conclusion that is destabilized by the other. With such
Discrepant Information Paradoxes, as with all paradoxes, we have
a situation of cognitive dissonance.
In confronting an aporetic inconsistency we have no sensible
alternative but to jettison something, to abandon or at least modifY one or another of the theses involved in the collision. At this
point there is a decisive difference between theoretical and practical contexts. In practical contexts there is a possibility of compromise-of effecting a division that enables us in some way and to
some extent "to have it both ways", say to proceed A-wise on
even days and B-wise on odd ones. But we cannot rationally do
this with beliefs. In theoretical contexts we must choose-must
resolve the issue one way or another. 15
However, logic alone will not help us to choose how a conflict
of inconsistency should be resolved. It does no more than tell us
that we must forego one of the claims, but affords no hint as to
which one. The best we can manage here is to accept two of our
three conflicting theses, viz., either (I) and (3) with the resulting
estimate of 3, or (2) and (3) with the resulting estimate of 5, or
(1) and (2) with the resulting estimate of 4. Which route shall we
choose? Here we are simply at sea-unless and until we have some
further guidance. For note that the moment we learn that one of
our sources is less accurate or reliable than the others a definite
resolution is at our disposal.
15. For a fuller treatment of this issue see the author's "Ueber einen zentralen
Unterschied zwischen Theorie und Praxis." Deutsche Zeitschrift fuer Philosophie, vol. 47
(1999),pp.171-182.
Chapter 1
12
Greek philosophers of a whimsical tendency were drawn to

sophisms (sophismata) along the lines of The Paradox of the Horns
(keratines) discussed by the Megarian philosopher Eubulides of
MiletuS. 16 It was based on the reasoning:
(1) You have no horns.
(2) If you have not lost something, you still have it.
(3) You have not lost any horns.
(4) Therefore: you (still) have horns. (From (2) and (3).)
Here theses (1), (2), and (3) are all supposed to be factual truths.
However, (2) of course obtains only under the proviso that the
thing in question is something you have had in the first place.
Seeing that this paradox rests on a presupposition which, as it
stands, is simply false, it is readily resolved. I7
Again the following Truth Paradox affords an instructive
example:
(1) Truth is a property of sentences and only of sentences.
(2) Sentences are elements of human language.
(3) Human language cannot exist without humans.
(4) Truth cannot exist without humans.
Some philosophers are beguiled by this line of thought into
accepting an anthropocentrism of truth: accepting (1)-(3) they
16. For Eubulides see pp. 102-04 below. Diogenes Laertius, Lipes of Eminent
Philosophers, II, III and IV, 39, reports that some think that this sophism is a joke aimed
at the cuckold. It was also attributed to Chrysippus.
17. On the various ancient discussions of The Paradox of the Horns see Prant!,
Geschichte, Vol. I, p. 53. The paradox is almost as bizarre as the sophistical Dog Paradox
that turns on the reasoning:
"That dog is your dog. That dog is a father. Therefore: That dog is your father."
This paradox preoccupied the Megarian paradox-theorists and also the Sophists (see Plato,
Euthydemus, 299E). (Some elaborated the sophism by adding: "You beat that dog.
Therefore: You beat your father.") Another variant of such a paradox was discussed by the
Stoics: "He is a bad person; he is a butcher; therefore, he is a bad butcher." (See Prantl,
Geschichte, Vol. I, p. 492.)
Aporetics
13
insist that (4) must be abandoned. But there are big problems
here. For the reasoning at issue fails to heed the distinction
between sentences and statements or abstract propositions (the
information that sentences purport). Actually, it is false to say, as
per (1), that truth appertains to sentences only: it is also a feature
of statements. And if we substitute "statements" for "sentences"
in the argument then (2) fails. For while human-language sentences can express statements, the facts at issue with potentially
true statements can-at least in theory-outrun the limits of
human language.
As such philosophical paradoxes indicate, the inconsistency at
issue need not be obvious, but may well lie hidden from casual
perusal. The puzzle of "Who is the engineer?" yields a vivid example of a nonphilosophical paradox of this sort. It envisages a series
of reports providing the following background information:
On a train operated by three men, Jones, Smith, and Robinson,
as engineer, brakeman, and fireman (but not necessarily
respectively), are three passengers: Mr. Jones, Mr. Smith, and
Mr. Robinson. The following information is given:
(a) Mr. Robinson lives in Detroit.
(b) Mr. Jones receives a salary of $40,000 a year.
(c) Smith beat the fireman at billiards.
(d) The passenger whose name is the same as the brakeman
lives in Chicago.
(e) The brakeman lives halfway between Chicago and Detroit.
(f) The brakeman's nearest neighbor, one of the passengers,
earns a salary exacdy three time as large as the brakeman's.
(g) Smith beat the engineer at tennis.
Here these (a)-(f) make it possible to identify the engineer. This
can be shown as follows when we use R, J, S to abbreviate the
names of the individuals at issue and J, b, e to indicate the fireman, brakeman, and engineer, respectively:
Chapter 1
14
FACT
REASON
(1) S"# f
(c)
(2) R is not b's next-door neighbor
(a), (e)
(3) J is not b's next-door neighbor
(b), (f)
(4) Sis b's next-door neighbor
(2), (3), (f)
(5) S doesn't live in Chicago
(e), (4)
(6) R doesn't live in Chicago
(a)
(7) J does live in Chicago
(5), (6), (d)
(8) J= b
(7), (d)
(9) J"# f
(8)
(10)R=f
(1),(9)
(11) S= e
(8), (10)
Smith is the engineer! But of course when (g) is added with its
message that S"# e, a contradiction at once results. The information overload that compromises the (g)-augmented puzzle of
"Who is the engineer?" creates an aporetic situation that results in
paradox. This fact, however, is well concealed in the fog of complexity that pervades this puzzle's information base.
Consider the statement "All of my statements made on [suchand -such a date] are false." This is in principle altogether unproblematic. Indeed if the date is one on which the only thing I said
was "2 + 2 = 5" then the statements at issue is both perfectly
meaningful and flat-out true. However, if the date at issue is the
one on which I made that very statement, then paradox is the
clear result. Thus paradox may well be a matter not only of what a
statement says but also of the contingent circumstances in which it
is made.
Take another example, that of the Deduction Paradox which
takes the following dilemmatic form:
(1) If a proposed deductive argument is valid, then the conclusion can contain no information that is not inferentially
contained in the premisses.
Aporetics
15
(2) This circumstance means that a deductively valid argument is essentially uninformative-it adds nothing new
over and above what is already in hand.
(3) If a proposed deductive argument is invalid, then it is useless. A viable conclusion cannot be based upon it.
(4) A deductive argument is thus either uninformative (as per
(2)) or pointless (as per (3)). In either case, it can accomplish no informatively useful work.
That premiss (2) is untenable here is vividly illustrated by the
"\Vho is the Engineer?" Problem. Clearly a distinction must be
drawn. Implicit information containment as per premiss (1) is one
sort of thing, and cognitively discernible informativeness is something very different, seeing that finite intelligences are not deductively omnipotent.
Paradoxes all arise in one and the same generic way, namely,
through aporetic overcommitment. In each case we have a situation in which we are minded to accept as plausible more than the
limits of consistency can manage to contain. We can, accordingly,
approach paradoxes from the angle of the preceding consideration
regarding aporetic analysis subject to the idea that addressing a
paradox is a matter of breaking the chain of inconsistency that it
involves by seeking out the weakest links of this chain.
1.3
Plausibility
It should be stressed from the very start that the acceptance or
assertion of claims and contentions is not something that is uniform and monolithic but rather something that can take markedly
different types or forms. For we can "accept" a thesis as necessary
and certain on logico-conceptual grounds ("Forks have tynes"),
or as unquestionably true, although only contingently so ("The
Louvre Museum is in Paris"), or in a much more tentative and
hesitant frame of mind as merely plausible or presumably true
("Dinosaurs became extinct owing to the earth's cooling when
the sun was blocked by cometary debris").
\Vhen a proposition is so guardedly endorsed we do not have
it that, a la Tarski, asserting p is tantamount to claiming that "p is
16
Chapter 1
true." What is at issue with the guided endorsement of a claim p

as plausible is simply the contention that "p is presumably true."
And what this means is that we will endorse and employ p insofar
as we can do so without encountering problems but are prepared
to abandon it should problems arise. In effect we are not dedicated partisans of such a claim but merely its fair-weather friends.
We see such a claim as meriting provisional endorsement "under
the circumstances." With assertion or acceptance in this third,
tentative or provisional mode, what is maintained is to be seen as
a matter not of actual truth but of mere plausibility-a contention that we would accept as true "if we would get away with
it," that is, if taking this step did not issue in anomalous (that is,
unacceptable) consequences. In sum, the claim is made provisionally, subject to the caveat that its retraction may become
necessary.
A proposition can be plausible and yet actually false.
Consider an example. During the American Civil War there were
a few instances in which women managed to disguise themselves
as men and serve as soldiers in the Union Army. This makes
the thesis "Those who served in the Union army were male"
strictly speaking false. Yet while a generalization that fails (only)
in extraordinary or anomalous cases may be literally false, this
does not annihilate its status as plausible. However, it is a plausible thesis and not a plausible truth. Strictly speaking it should be
characterized as a plausible falsehood-but just that is exactly the
point at issue: a proposition can be plausible and yet actually
false.
Plausible propositions playa very special role in the cognitive
scheme of things. We feel free to make use of them as premisses
when working out answers to our questions. But their use in the
setting of particular question-resolving contexts is not predicated
on an outright and unqualified commitment to these propositions
as true. For we know full well that we cannot accept all those
plausibilities as truths since this can and generally would lead us
into contradiction.
Plausibilities are accordingly something of a practical epistemic device. We use them where this can render effective service.
But we are careful to refrain from committing ourselves to them
unqualifiedly and come-what-may. And we would, in particular,
refrain from using them where this leads to contradiction. In sum,
17
Aporetics
our commitment to them is not absolute but situational: whether

or not we endorse them will depend on the context. To reemphasize: the "acceptance" that is at issue here represents no more
than a merely tentative or provisional endorsement.
The shift from assertion-as-true to assertion-as-plausible offers
us a degree of flexibility with the claims that we make. Thus consider the Exception Paradox which centers on the contention that
"All generalizations have exceptions." This leads to the following
paradox:
(1) All generalizations have exceptions.
(By hypothesis.)
(2) (1) is true.
By (1).)
(3) (1) is a generalization.
(From (1) by
inspection. )
(4) (1) has exceptions.
(From (1), (3).)
(5) Any generalization that admits

of exceptions is false.
(As a principle of
logic.)
(6) (1) is false
(From (3), (4),

(5). )

Note, however, that this paradox is immediately dissolved when
the thesis at issue in (1) is asserted not as true but merely as plausible. For then the inferential step from (1) to (2), which is essential to deriving the contradiction, is automatically invalidated.
1.4.
Plausibility and Preswnption
Unfortunately, life being what it is, we cannot always get away with
accepting the plausible outright because actual truth is something
more selective and demanding than mere plausibility, seeing that
plausibilities-unlike truths--<:an conflict both with truths and with
one another. Distributively true statements are of course collectively
true: we have [T(p) & 11q)] ~ T(p & q). But this is emphatically
not the case with plausible let alone with merely probable state-
18
Chapter 1
ments. For plausible (and probable) statements can actually come

into conflict with one another, and thereby impel us into paradox.
Plausibilities are thus one thing and truths another. We
"accept" plausible statements only tentatively and provisionally,
subject to their proving unproblematic in our deliberations. Of
course, problems do often arise. X says 25 people were present, Y
says 15. Sight tells us that the stick held at an angle under water is
bent, touch tells us that it is straight. The hand held in cold water
indicates that the tepid liquid is warm but held in hot water indicates that it is cold. In such cases we cannot have it both ways.
Where our sources of information conflict-where they point to
aporetic and paradoxical conclusions-we can no longer accept
their deliverances at face value, but must somehow intervene to
straighten things out. And here plausibility has to be our guide,
subject to the idea that the most plausible prospect has a favorable
presumption on its side.
Plausibility is in principle a comparative matter of more or less.
Here it is not a question of yes or no, of definitive acceptability,
but one of a cooperative assessment of differentially eligible alternatives of comparative advantages and disadvantages. We are
attached to the claims that we regard as plausible, but this attachment will vary in strength in line with the epistemic circumstances. And this fact has significant ramifications. For the idea of
plausibility functions in such a way that presumption always favors
the most plausible of rival alternatives.
The concept of presumption at issue here is effectively the
same as the traditional legal idea, accurately explained already in
the eighteenth-century textbook of Archbishop Whately as
follows:
According to the most current use of the term, a "presumption"
in favour of any supposition means, not (as has been sometimes
erroneously imagined) a preponderance of probability in its
favour, but such a preoccupation of the ground as implies that it
must stand good till some sufficient reason is adduced against it;
in short, that the burden of proof lies on the side of him who
would dispute it. 18
18. Robens Whately, Elements of Rhetoric (Urban, 1963; first edition 1846), Pt. I, ch.
III, sect. 2.
Aporetics
19
It is in just this sense that in Anglo-American law a "presumption of innocence" operates in favor of the accused. The "burden
of proof' is carried on the other side, and the presumption stands
until it is overthrown by stronger countervailing considerations. In
virtually all probative contexts there is, for example, a standing
presumption in favor of the normal, usual, customary course of
things.
Against the claims of the senses or of memory automatically to
afford us the truth pure and simple one can deploy all of the traditional arguments of the sceptics, and take as guide the dictum of
Descartes: "All that up to the present time I have accepted as most
true and certain I have learned either from the senses or through
the senses but it is sometimes proved to me that these senses are
deceptive, and it is wiser not to trust entirely to any thing by
which we have once been deceived."19 But of course, such arguments in support of the vulnerability of sensory data serve only to
re-emphasize their role as promising truth-candidates in the sense
of presumptions rather than outright truths as such. A presumption, after all, is to be given some credit, even if not a totally
unqualified endorsement. Its epistemic standing is not rock-solid,
to be sure, but it is enough to call for a sufficient effort for its dislodging. Our commitment to a presumption may not be absolute,
but it is nevertheless a commitment all the same. 20
The plausibility tropism inherent in the principle of a presumption in favor of the most plausible alternative is an instrument of
epistemic prudence. It amounts to the sort of safety-first approach
that is known in decision theory and the theory of practical reasoning as a minimax potential-loss strategy. The guiding idea of
such a strategy is to opt in choice-situations for that alternative
which minimizes the greatest loss that could possibly ensue. 21
19. Rene Descartes, Meditations on First Philosophy, No. I, translated by R.M. Eaton.
20. Israel Scheffer makes a similar point in the temporal context of a change of mind
in the light of new information: "That a sentence may be given up at a later time does not
mean that its present claim upon us may be blithely disregarded. The idea that once a
statement is acknowledged as theoretically revisable, it can carry no cognitive weight at all,
is no more plausible that the suggestion that a man loses his vote as soon as it is seen that
the rules make it possible for him to be outvoted." Science and Subjectivity (New York,
1967), p. 118.
21. For the relevant issues at the level of decision theory in general see R.D. Luce and
H. Raiffa, Games IJnd Decisions (New York: Wtley, 1957).
20
Chapter 1
(The strategy is obviously different from an expected-value

approach.) The presumption of veracity is fundamental to the
ay in which we draw cognitive benefit from our sources of information-human and instrumental alike, our own sensory
instrumentalities included. Once we recognize and acknowledge
an informative source as such we stand prepared to take its declarations at face value until such time as problems come to view.
1.5.
The Acceptability of Contentions
In practical matters we can afford to be flexible and tolerant

because the possibility of compromise is in prospect.22 But matters
of truth are something else again; they are hard-edged and inflexible (at any rate in all ordinary cases and situations). As far as
knowledge is concerned, falsehoods must be consigned to the
trash-heap: it no longer makes a useful contribution to our
knowledge as such. But as far as practice is concerned, the matter
is otherwise.
Nevertheless, presumption provides us with the prospect of a
kind of epistemic compromise. If it transpires that only almost all
A's are B's, then it is clearly incorrect to claim "AU A's are B's."
However, in operating with this statement in practice we will (ex
hypothesi) almost never encounter problems. Or again, if the A's
and the B's are alike in those respects in which we generally deal
with them, it makes little difference how different they may be in
other respects. If we are cooks or dieticians rather than botanists
then we can implement the (false) idea that "Tomatoes are vegetables, not fruits" and almost never run into trouble. No doubt
the real and full truth is preferable, but life being what it is, there
will be times when it makes sense to let the merely plausible do
duty in its place.
Recognizing that this sort of thing can happen with statements
that are merely false, it is worthy of note that this situation can
even obtain with statements that are literally self-contradictory.
22. The case for this contention is set out in detail in the author's paper "Ueber einen
zentralen Unterschied zwischen Theorie Wld Praxis," Deutsche Zeitschrift fuer Philosophie,
vol. 47 (1999), pp. 171-182.
Aporetics
21
Even such paradoxical theses can have their cognitive use.

Consider, for example, the following array of self-contradictory
statements:
Universal claims are never true.
The unqualified contentions are exaggerations.
Never-maintaining statements are never correct.
Predictions are always off the mark.
General rules always have exceptions.
Construed literally, purely as cognitive theses, such statements
are self-defeating and thus seemingly absurd. But the situation is
different if we look on the matter from the angle of practice. For
from this angle we can view the issue as one of practical ("proceed as though") epistemic policy.
Never to make universal claims.
Never to make unqualified contentions.
Never to make never-claims.
Never to make unqualified predictions.
Never to lay down altogether unqualified rules.
Never to make dogmatic claims to knowledge.
Such practical policies of cognitive procedure are certainly not
absurd. They may be of questionable logic but they are not without a certain practical wisdom. For the self-defeating nature of
such recommendations does not preclude the prospect of their
being constructively procedure-underwriting at the level of operative policies. And it is through this fact that they are able to make
good a claim to be acknowledged as plausible.
But of course we are skating on thin ice here and must be
careful. Thus the medieval philosopher John Buridan discussed
22
Chapter 1
the anomaly posed by the statement: "Every proposition is affirmative. "23 Though false, this self-instantiating claim seems otherwise harmless. But it nevertheless has the logical consequence
"No proposition is negative." And this statement is not only false
but absurd, seeing that it itself provides a counter-instance to its
own assertion.
23. See G.E. Hughes, John Buridan on Self-Reference (Cambridge: Cambridge

University Press, 1982), p. 34.

The Smashed Vase Paradox
The Paradox of Evil
The Socratic Ignorance Paradox
Dilemmatic Paradoxes
Conflicting Report Paradoxes
Compound Paradoxes
Paradox Solution via

Retention Prioritization
2.1 Addressing Paradox Resolution via
Priority Ratings
Since a paradox is generated by an inconsistent set of propositions, it is useful to begin with a brief review of some relevant logical ideas regarding such inconsistent sets.
Every set of individually consistent but collectively incompatible propositions engenders a family of maximal consistent subsets
(MCS's). These are those consistent subsets that are rendered
inconsistent by the addition of any of the omitted propositions.
Further every set of individually consistent but collectively
inconsistent propositions also engenders a family of minimal
inconsistent subsets (MIS's). There are those inconsistent subsets
that are rendered consistent by the deletion of anyone of their
members. These minimal inconsistent subsets constitute the
"cycles of inconsistency" that afflict an inconsistent set of propositions. l A set will be more (or less) extensively inconsistent when it
has more (or fewer) cycles of inconsistency. Thus an inconsistent
set is maximally inconsistent if every single pair of its members is
inconsistent. It will then have as many mcs's as it has members,
because all but only its individual members constitute MCS's.2
1. Most of the paradoxes discussed in the extensive literature of the subject involve
only a single cycle of inconsistency and accordingly have as Iowan index of inconsistency as
is possible for an inconsistent set of the size at issue.
2. On this basis, the comparative extent to which a set is inconsistent can be measured
by the ratio (k - 1) -;- (n - 1), where k is the number of MCS's and n the number of setmembers. This inconsistency index of a set can range from 0 in the case of a consistent set
to 1 in the case of a maximally inconsistent set of individually consistent propositions.
25
26
Chapter 2
An inconsistent n-et (quartet, quintet, etc.) is a set of n collectively inconsistent propositions that is minimally inconsistent in
that the deletion of anyone of which will restore overall consistency, so that every n-l sized subset of such a set is a maximal
consistent subset (MCS). Thus consider the situation where both
{( 1), (2), (3)} and ((2), (3), (4) I are inconsistent triads, and we
now take into view the set {(I), (2), (3), (4)}. Owing to the overlaps involved this is not an inconsistent quartet, since instead of
having the requisite four MCS's (one for each available proposition) it has only three: {(I), (2), (4)}, {(l), (3), (4)}, and {(2),
(3) }.
It is a readily demonstrable fact of logic that any inconsistent
set of propositions can be decomposed into (potentially overlapping) subset-components each of which is an inconsistent n-et.
Accordingly, an inconsistent set has as many "cycles of inconsistency" (as we have called them) as there are inconsistent n-ets to
this decomposition.
Paradox solution is in general a matter of bringing consistency
to an inconsistent set of propositions. On this basis, there is a
generic, across-the-board methodology for analyzing paradoxes of
all kinds. For apory resolution becomes an exercise in epistemic
damage control: confronting an inconsistent cluster we have to
restore a cognitively viable situation. And the object is to achieve
this at minimal cost-with the least possible sacrifice among the
theses towards which we were, in the first instance, favorably
inclined. Yet given the conflict among the propositions involved in
an aporetic situation, it is clear that they cannot all be true. (The
truth, after all, must constitute a consistent whole.) And so when
confronted with an aporetic situation there is really only one way
out: some of the theses that engender the conflict must be abandoned-if only by way of restriction or qualification. We have here
overextended ourselves by accepting too much-more than a due
heed for consistency can possibly support. And this means that one
and the same recourse-the abandonment of some of the propositions involved on the basis of their absolute or comparative untenability-can resolve the problem. It also means that paradox
resolution never comes cost-free: there is always something that
we must give up for the sake of recovering consistency.
Logic can tell us that our modes of reasoning are valid-that
when applied to truths they must lead to truths. But it does not-
27
cannot-tell us that when applied to plausibilities, valid arguments cannot yield implausible (or even self-contradictory) conclusions. (Mter all, a conjunction is often less plausible than
its conjuncts.) Logic thus provides no insurance against paradox. And the reason is simple. Inconsistency requires a choice
among alternatives for premiss abandonment. But logic is valuefree. It will dictate that we must make choices in the interests of
consistency resolution, but not how. It can criticize our conclusions but not our premisses. And so, as long as we are conjuring
with plausibilities, the threat of paradox dogs our steps, irrespectively of how carefully and cogently we may proceed in point of
logic.
The long and short of it is that paradox management requires
an extra- or supra-logical resource. For the way to restore consistency to an aporetic situation is to implement some sort of prioritization principle that specifies how, in a case of conflict, we
should proceed in making some of the relevant claims give way to
others. What is needed is a rule of precedence or right of way.
Considerations of priority are needed for breaking the chain of
inconsistency at its weakest link. The guiding ideas of this
approach are accordingly two:
that paradoxes of the most diverse sorts can be viewed in a
uniform way as resulting from an aporetic overcommitment to
theses which, albeit individually plausible, are nevertheless collectively incompatible. And on this basisthat paradoxes of the most diverse sorts can be resolved
through a uniform process of weakest-link abandonment in
view of the fact that some of the conflicting theses take precedence or priority over others.
And with this second point, considerations of episternic evaluation--of priority determination-become an inevitable part of
paradox management.
The general process at issue is best conveyed via some particu1ar examples, beginning with the Smashed Vase Paradox based on
the contention "There's no real harm done by breaking the
vase-after all, it's all stiU there." Now consider the theses:
28
Chapter 2
( 1) If we smash the vase into bits, the vase no longer exists as
such.
(2) There is nothing to the vase over and above the mass of
ceramic material that constitutes it.
(3) When the vase is smashed, all the ceramic material that
constitutes it still remains in existence.
(4) By (2) and (3), the vase still remains in existence after it is
smashed, contrary to (1).
Thus {( 1), (2), (3)} constitutes an inconsistent triad. And (1) and
(3) are both incontestable facts while (2) is no more than a plausible-sounding principle. We thus have no alternative but to reject
(2), as the weakest link in the chain of inconsistency, plausibility
notwithstanding. Here presumably we would say something like:
"There is more to the vase than merely the ceramic material that
constitutes it, namely the organization of that material into a certain sort of vase-shaped configuration." On this basis we reject (2)
as untenable, despite its surface plausibility.
In paradox analysis it becomes crucial to list in detail all of the
theses and principles that are essential to establishing the aporetic
collection at issue. For until the links of that chain of inconsistency are all spelled out explicidy, it is not possible to determine
with confidence where exactly is the proper place to break the
chain. If we do not have all the links set out we cannot determine
which of them is the weakest.
With a view of this methodology, consider the Happiness
Paradox of John Stuart Mill which goes as follows: 3
(1) His own happiness is the natural end of any rational agent.
(2) A rational agent will adopt whatever end is natural for a
being of his kind.
(3) Therefore (by (2) and (3)) a rational agent will adopt his
own happiness as an end).
(4) A rational agent will adopt only those ends that he can
realistically expect to achieve by striving for their
realization.
3. See Chapter 5 of his Autobiography.
Paradox Solution pia Retention Prioritization
29
(5) Therefore (by (3) and (4 a rational agent can realistically

expect to achieve his own happiness by striving for its
realization.
(6) But it is a fact of life that rational people realize that they
cannot expect to achieve their own happiness by striving
for its realization.
Here {(I), (2), (4), and (6)1 constitute an inconsistent cluster, (3)
and (5) being merely derivative-as indicated. The maximal consistent subsets (MCS's) of this inconsistent quartet are: {(I), (2),
(4)}, {(I), (2), (6)}, {(I), (4), (6)1, {(2), (4), (6)1. These engender four corresponding retention/abandonment alternatives: (1),
(2), (4)/(6) and (1), (2), (6)/(4), and (1), (4), (6)/(2) and (2),
(4), (6)/(1). The problem is to choose one of these R/A-alternatives in preference to the rest.
As regards the status of the propositions involved, Mill himself
saw (1) and (2) as basic principles of rationality. And he accepted
(6) as a crucial insight into human nature. But (4) is not more
than a highly plausible supposition. The ranking of those four
theses in point of precedence and priority would accordingly be as
follows: [(1), (2)] > (6) > (4). (Note that theses of equal priority
are bracketed together.) Accordingly Mill deemed it necessary to
abandon any unqualified endorsement of thesis (4) however plausible and sensible it may otherwise seem to be. For the pursuit of
even unrealizable ends can-in certain circumstance-yield other
incidental benefits.
This example is typical and the process of paradox resolution
accordingly has the following generic structure:
1. Setting out the aporetic group at issue in the paradox

at hand, via an inventory of the aporetic (collectively
inconsistent) propositions involved, and exhibiting how
the logical relations among them engender a
contradiction.
2. Reducing this inconsistent set to its inferentially nonredundant foundation.
30
Chapter 2
3. Making an inventory of the maximal consistent subsets
(MCS's) of the resulting aporetic cluster. And on this
basis4. Enumerating the various alternatives for retention/abandonment combinations (R/A-alternatives) through which
aporetic inconsistency can be averted.
5. Assessing the comparative precedence or priority of the

propositions at issue.
6. Determining which among the resultant retention/abandonment alternatives for consistency restoration is optimal
in the light of these priority considerations.
It is, of course, possible that some propositions of an aporetic
cluster will not figure in the inconsistency that arises. Such "innocent bystanders" will stand apart from the clash of contradiction
and thereby belong to every maximal consistent subset (and to no
minimal inconsistent subset). For them priority does not matter:
they will emerge as immune for rejection, irrespective of any rankings of procedure and priority.
"Paradoxes," one student of the subject writes, "are selfenclosed statements with no external reference point from which
to take a bearing upon the paradox itself." 4 And this is true
enough. For those paradoxical inconsistencies themselves afford
no resources for their resolution-mere logical analysis of their
assertoric content is no way out. To resolve paradoxes we need an
external vantage point-a means for assessing the cognitive viability of the mutually incompatible theses that are involved, something about which those propositions themselves do not inform
us. It is, in general, the relevant principles of precedence and priority that fill the bill here.
Thus consider the following aporetic cluster-the Paradox of
Evil-from the domain of philosophical theology:5
4. Rosalie L. Colie, Paradoxia Epidemica: The Renaissance Tradition of Paradox

(Princeton: Princeton University Press, 1966).
5. On this paradox and related issues see Richard M. Gale, The Nature and Existence
of God (Cambridge: Cambridge University Press, 1991).
31
(1) God is not responsible for the moral evils of the world as
exemplified by the wicked deeds of us humans, his creatures.
(2) The world is created by God.
(3) The world contains evil: evil is not a mere illusion.
(4) A creator is responsible for whatever defects his creation
may contain, the moral failings of his creatures included.
(5) By (1 )-( 3) God is responsible for the world as is-its evils
included.
Let us step back and consider the systemic standing or status of
these contentions. Thesis (5) can be put aside as a consequence of
others that carry its burden here. Thesis (3) comes close to qualifying as a plain fact of everyday life-experience. Theses (1) and (2)
are (or could reasonably be represented as being) fundamental
dogmas of Judeo-Christian theology. But (4) is at best a controversial thesis of moral theology, involving a potentially problematic denial that creatures act on their own free will and
responsibility in a way that disconnects God from moral responsibility for their actions. We thus have the priority ranking [(1), (2)]
> (3) > (4). 6 On this basis it is reasonable to take the line that one
of these aporetic theses must be abandoned in this context. And it
is easily (4) that should give way seeing that only the R/A alternative (1), (2), (3)/(4) abandons a lowest-priority thesis.
By deleting the less plausible items from an aporetic cluster so
as to restore it to consistency we purify it, as it were, in the dictionary sense of this term of "diminishing the concentration of less
desirable materials." But note that those less desirable materials
are not necessarily worthless. For even they themselves are, by
hypothesis, supposed to be more or less plausible. Though oflowest priority in one context, a plausible thesis that is caught up in
an aporetic conflict may nevertheless be able to render useful epistemic service?
6. To be sure, a committed atheist would have a very different priority ranking-one
in which (1) would presumably be the low man on the totem pole.
7. For a concrete illustration see pp. 37-38 below.
32
Chapter 2
2.2
Procedural Considerations
The basic rule for choosing among the appropriate retention/

abandonment alternatives is:
AI: Minimize the extent to which you abandon propositions of

the highest relevant priority level. (Or equivalently: maximize the extent to which you retain propositions of the
highest priority level.)
Of course, this measure alone may not suffice, since several competing RjA alternatives may tie in the regard. The next rule to
apply is:
A2: When (owing to a tie) rule Ai does not settle matters reapply it at the next lowest priority level.
And when this does not do the job, we simply proceed to the next
level,
A3: Whenever application of the preceding rules of this sequence
does not settle matters, proceed to apply the same process at
the next priority level.
Every retention/abandonment alternative thus has a retention

profile in point of prioritization determined by the proportion of
statements whose retention it allows at the successive priority levels.
And the optimal R/A-alternative is that whose profile stands last
in the (numerical) lexicographic ordering when one compares
all of the available possibilities. For this means performing optimally at making low-priority theses give way to theses of higher
priority. The resultant procedure minimizes implausibility or-less
elliptically-maximizes the minimal plausibility of the propositions we propose to endorse. (What we have here is accordingly a
maximin plausibility principle of assessment.)
An example of this process is provided by the Socratic
Ignorance Paradox. Let us suppose that Socrates maintains that he
knows nothing. To be sure, Socrates presumably said "I know
only that 1 know nothing [more]" (Nihil scio praeter hoc, quod
nihil scio) which is nowise paradoxical. But let us amend reality
33
somewhat for the sake of an example. Then we confront an

aporetic situation which stands somewhat as follows:
(1) Socrates claims to know nothing.
(2) By (1), Socrates makes the claim: "I know nothing."
(3) To claim something is to hold it to be true.
(4) To hold something to be true is effectively to claim to
know that it is so, and so to claim to know something.
(5) Sensible people do not make inconsistent claims.
(6) Socrates is a sensible person.
(7) By (5) and (6), Socrates does not make inconsistent
claims.
(8) By (1) and (4), Socrates makes inconsistent claims.
Here {( 1), (3), (4), (5), (6)} constitutes an inconsistent quintet.
Now (1) and (6) are putative facts that constitute part of the
stagesetting of the problem: their position is secure. And (3) and
(5) are plausible principles of rational process, as is (4) albeit at a
lower level of plausibility. We thus arrive at the plausibility ranking
[(1), (6)] > [(3), (5)] > (4), with the result that the circle of
inconsistency should be broken at (4). The rationale for so proceeding is that even conscientious people can stake claims at the
level of probable or plausible conjecture or supposition rather
than at the level of outright knowledge.
Here we have available the five RjA-alternatives, of abandoning exactly one of the theses (1), (3), (4), (5), (6). These RjAalternatives have the retention profiles {t, 1, I}, { h hI}, (1, 1,
t}, {l,h I} and
1, I}, respectively. (As indicated above, these
numbers indicate the proportion of retentions in the three categories at issue.) The optimal profile is that which stands last in a
lexicographic ordering, and it is this which indicates the optimal
RjA-alternative that is available in the circumstances. It is clear
that the third of these offers the best option, and the sacrifice of
(4) results.
{h
34
Chapter 2
2.3
Disjunctive Resolutions: Conflicting

Report Paradoxes
To be sure, the preceding rules will fail to decide matters when

several retention/abandonment alternatives have the same retention profile. We then come to the end of our preferential tether
without reaching a unique result. When this occurs and further
preferential elimination is impracticable, we must fall back on a
disjunctive resolution: For then all that we can then say is that one
or another of those equally eligible alternatives must obtain. This
sort of thing is rare among the familiar paradoxes but can of
course happen.
The journey into paradox sometimes proceeds by way of
dilemma. A dilemma-engendered paradox takes the following
generic form:
(1) If C, then P. A given.
(2) Ifnot-C, then Q A given.
(3) Either Por Q From (1), (2).
(4) Not-P. (Pis by hypothesis unacceptable).
(5) Not-Q (Qis by hypothesis unacceptable).
(6) (3) contradicts (4)-(5).
Here an unavoidable disjunction (Either Cor not-C) leads to an
unacceptable result (Either P or Q).
Thus consider the following as an illustration of a Dilemmatic
Paradox: Suppose you borrowed $10 from Tom and $10 from
Bob. On your way to repaying them you are robbed of everything
but the $10 you had hidden in your shirt pocket. By no fault of
your own, you now face the following paradoxical dilemma:
(I) You are obligated to repay Tom and Bob.
(2) If you pay Tom you cannot repay Bob.
(3) If you repay Bob you cannot repay Tom.
(4) You cannot honor all your obligations: in the circumstances this is impossible for you. (By (I )-( 3).)
35
(5) You are (morally) required to honor all your obligations.

(6) You are not (morally) required to do something you cannot possibly do (ultra posse nemo obligatur).
Here 1(1), (2), (3), (5), (6) form an inconsistent cluster. (1), (2),
and (3) are supposedly fixed facts of the case. Both (5) and (6)
are-or seem to be-basic general principles of morality. The
resulting plausibility ranking is [(1), (2), (3)] > [(5), (6)]. The
conclusion is the one of (5) and (6)-or both-must be abandoned. We here have an instance of the objective situation contemplated above.
To be sure, instead of outright rejection we could endeavor to
save them in some duly qualified form, perhaps somewhat as follows:
(5') You are morally required to honor all your obligationsalbeit only insofar as the circumstances of the case permit
(provided those circumstances are not of your own
making).
(6') You are not (morally) required to do something you cannot possibly do (provided the impossibility at issue is not
of your own making).
Subject to such qualifications the paradox can be overcome
through premiss abandonment along the by now familiar lines.
Other examples of disjunctive solutions are forthcoming in the
case of Conflicting Report Paradoxes. Thus suppose that three
equally reliable contemporaneous polls report the following paradoxical result regarding how people stand on a certain issue of
public policy:
(2)
(3)
(1)
For
Against
Unwilling to respond
30%
30%
40%
55%
30%
30%
55%
15%
15%
Here polls (2) and (3) are incompatible: there is no way to redistribute the nonrespondents of these polls so as to bring them into
36
Chapter 2
accord. We must thus choose between (1), (2)/(3) and (1),

(3)/(2) whose condition in point of retention/priority is identical. A disjunctive result, along the lines of (1) & [(2) v (3)] now
ensues, suggesting the inconclusive conclusion: For 30-70%,
Against 30-70%. & far as our polling information goes, the situation is simply a tie-even though this was perhaps not clearly
apparent on first view. And this resolution is the best that we can
achieve here, since we are caught up in a tie in point of retention
priorities. (If we had reason to discount poll (2) or poll (3), the
result would be quite different.)
A further example is also instructive. Suppose a 3 x 3 tic-tactoe configuration and suppose that three sources (A, B, and C)
give us reports about this to the following effect:
(1) There are two X's in row 1.
(2) There are three X's in row 2.
(3) There are two X's in row 3.
(4) No column has three X's.
These four claims are clearly incompatible: given (1)-(3), there is
no way in which (4) can be realized. We have a paradox on our
hands.
In view of the collective incompatibility of (1 )-( 4) we must
break the chain of inconsistency. Now let it be that (1) and (3)
stem from source A, (4) from source B, and (2) from source C.
And let it be that sources A and C are equally reliable while
source B is somewhat less so. Given the envisioned difference in
source reliability, (4) abandonment offers the best prospect
here. Theses (1), (2), and (3) will thus be retained and (4) jettisoned, so that we arrive at (1), (2), (3)/(4) as the optimal RjAalternative.
However, if we are sensible we will not simply reject B's evidence altogether. Conceding it some modicum of plausibility, we
would at least try to minimize the number of columns that have
three X's. And this leads to the conclusion that just one single
column has only three X's, thereby going beyond the information
conveyed by (1), (2), (3) alone.
37
Plausible claims, frail thought they are, still count for something. In endeavoring to accommodate insofar as possible even
those theses that we abandon-and not simply throwing them on
the trash-heap of rejected falsehoods-we position ourselves to
profit from the information they provide. To abandon a plausible
proposition in the context of resolving a paradox is not to dismiss it
as false and thereby pervasively and context-indifferently useless.
Abandonment is accordingly not something absolute but rather
contextual. And it does not abolish completely the cognitive utility that a proposition can have by virtue of its inherent plausibility.
2.4
Compound Paradoxes
A fundamental fact of paradox theory---of aporetic~is that every

paradox is resolvable in principle by abandoning commitments.
For if we dismiss (and thus withhold endorsement from) sufficiently many of the theses that constitute the aporetic cluster at
issue there will no longer be an inconsistency among our actual
commitments. However, even the best available resolution may
well be an indecisive one that leaves us confronting a disjunctive
plurality of alternatives no particular one of which can be preferentially justified in the circumstances.
If we abandon a thesis in the context of one paradox, must we
not also reject it in the context of another? Not necessarily.
Consider an inconsistent cluster compound of two inconsistent
triads {(I), (2), (3)} and {(3), (4), (5)} where (1) > (2) > (3) >
(4) > (5), Looking to the second triad we will retain (3) and
abandon (5). But when we take the first triad also into view, we
realize that it is (3) that must be abandoned. When concerned for
coherence overall, we must (obviously enough) take a broad,
inclusive view of the relevant paradoxes. (Nevertheless, here-as
elsewhere-taking everything into account may prove to be somewhere between impracticable and impossible. 8 )
8. On this theme the words of Nicholas ofCusa still ring true: "A finite intellect, therefore, cannot by means of comparison reach the absolute truth of things. Being by nature
indivisible, truth excludes 'more' or 'less,' so that nothing but [the whole 1 truth itself can
be the exact measure of truth .... The relationship of our intellect to the truth is like that
of a polygon to a circle; the resemblance to the circle grows with the multiplication of the
38
Chapter 2
The illustration of the preceding paragraph yields an example

of a compound paradox constituted by an aporetic cluster that has
more than one cycle of inconsistency. For when we group all of
the relevant propositions together and contemplate the inconsistent cluster {(I), (2), (3), (4), (5)} we find that this gives rise to
the two aforementioned inconsistent triads.
A concrete example of such a compound paradox is as follows.
Suppose the following four collectively inconsistent reports:
(1) There are no males in the room.
(2) There is only 1 person in the room.
(3) Tom Smith is in the room.
(4) Jack Jones is in the room.
This aporetic cluster consists of three cycles of inconsistency: {( 1 ),
(3)}, {(I), (4)}, {(2), (3), (4)}. Thus consistency-restoration overall could be effected by the following R/A-alternatives: (1),
(2)/(3), (4) and (2), (3)/(1), (4) and (3), (4)/(1), (2) and (2),
(4)/(1), (3). Thus if we had reason to see report (1) as more reliable than the rest we would be led to the resolution (1), (2)/(3),
(4) which alone enables us to restore consistency everywhere
while still retaining the only top-priority thesis at hand. We would
then arrive at the definitive result that there is just one female in
the room.
Suppose however that we had a compound paradox consisting
of three cycles of inconsistency (A, B, C) constituted as follows by
three interlocking inconsistent quartets as shown on p. 39.
Note that we could break all three chains of inconsistency at
issue here by dismissing premiss (5). But whether or not this
would be the appropriate solution would depend entirely on the
priority-status of the propositions involved. Suppose, for example,
that the priority situation were:
(5) > [(2), (4), (6)] > [(1), (3), (7)]

angles of the polygon; but apan from its being reduced to identity with the circle, no multiplication, even if it were infinite, of its angles will make the polygon equal [0 the circle
.... The more profoundly we learn this lesson of ignorance, the closer we draw to truth
itself." Nicholas of eusa, On Learned Ignorance I, iii; translated by Germain Heron (New
Haven: Yale University Press, 1954.)
39
c
Then we would (and should) unhesitatingly abandon the trio (1),
(3), (7), even notwithstanding the fact that dismissing (5) alone
would restore consistency across the board.
Though interesting from a theoretical point of view, compound paradox is an uncommon phenomenon. None of the standard paradoxes encountered in the extensive literature of the
subject involves more than a single cycle of inconsistency-that is,
they all relate to inconsistent n-ads.

Moore's Paradox
The Lottery Paradox
Plausibility as a Guide
to Prioritization
3.1
Theses of Lowest Priority: The Meaningless and

the False
The simplest way to eliminate an aporetic paradox is to determine

that one or another of its essential premisses is flat-out untenable
by way of being either meaningless or false. Here falsity is simple
enough but meaninglessness is rather more complicated because
there are significantly different routes to this destination. The
hermeneutically meaningless is literal nonsense, unintelligible gibberish; the informatively meaningless is absurd and conveys no
usable information, and the semantically meaningless is devoid of
any stable truth status, so that here there can be no fact of the
matter one way or the other. "Yellow weighs wooden tentacles" is
hermeneutically meaningless: one can make no sense of it. "He
drew a square circle" is informatively meaningless: it anomalously
takes away with one word ("circle") what it gives us with another
("square"); we understand-sort of-what is being said but deem
it absurd. Finally, "This sentence is false" is semantically meaningless: neither truth nor falsity can be ascribed to it.l
In principle, various different avenues can lead to meaningless
statements: grammatical anomaly, definition conflict, category
confusion, presupposition violation, and referential vacuity. The
1. The distinctions at issue here go back to Husserl's distinction between the
hermeneutically and the informatively meaningless, between non-sense (Unsinn) and contrasense (Widersinn). The former is radically meaningless; the latter simply confused and
self-conflicted. See Edmund Husser!, Logische Untersuchungen (Halle: Niemeyer, 1891),
Vol. I, pp. 110-16, translated by J.N. Findlay (London: Routledge; New York:
Humanities Press, 1970).
43
44
Chapter 3
first two generally produce hermeneutic meaninglessness, the next

two generally engender informative meaninglessness, and referential vacuity generally leads to semantic meaninglessness.
While meaninglessness is perhaps the gravest flaw a paradoxical
premiss can have, mere falsity is yet another, somewhat less grave
failing. In this light, consider the example afforded by Moore's
Paradox which centers on the paradoxical nature of locutions of
the form "P but I do not believe it." 2 This contention conveys
the useful lesson of a complexity of "acceptance." The problem
here turns on the fact that we may be "of two minds" about some
things: that there are some facts that we acknowledge as such with
one side of our minds but nevertheless find them somewhere
between hard and impossible to accept. ("I find it incredible that
some Confederate generals should eventually become generals in
the U.S. army," or " I find it incredible that a President of the
U.S. should survive in office despite a widely publicized affair
with a White House intern".) Moore thus emphasized the paradoxical nature of sentences of the form "P but I do not believe
that P' ("The entire physical universe was once smaller than the
head of a pin but I do not really believe it"). To be sure, there is,
in general, no anomaly to the assertion "P but X does not believe
it". The difficulty arises only in one's own personal case. And, as
Moore sees it, the difficulty roots in the conflict between what the
speaker implies by his assertions, and what he states therein,
between what is implicit and what is explicit.
To formulate the paradox that arises here, let S be a statement
of the form "P but I do not believe that P." And now consider:
2. The paradox was discussed by G.E. Moore in his book Ethi.s (Oxford: Clarendon,
1912) and also in P. Schilpp (ed.), The Philosophy of G.E. Moore (La Salle: Open Court,
1942), pp. 430-33, and in P. Schilpp (ed.), The Philosophy of Bertrgnd Russell (La Salle:
Open Court, 1944), p. 204. (The former of these books contains an article on Moore's
Paradox by Morris Lazerowitz [pp. 371-393].) The paradox was already known in
medieval times. Albert of Saxony discussed the paradoxical nature of propositions of the
form "Socrates believes himself deceived in believing a certain proposition A," seeing that
in believing A Socrates maintains it as true and in believing himself deceived he rejects it as
false. (See Kretzmann and Stump 1988, p. 361.) On this basis, Albert developed a com
plex argument to the effect that it is impossible for Socrates (that is, anyone) to know that
he is mistaken about a concrete matter of fact (ibid, pp. 363-64). And analogously,
William of Heytesbury (ca. 131O-ca. 1370) argued in his ReguliJe so/vendi sophismlJtg (ca.
1335) that a person cannot doubt something he knows. (See Boh 1993, pp. 48 and
67-68.) Nihil s.ire potest nisi verum became an accepted dictum.
45
(1) S makes a meaningful statement that conveys coherent

information. (A plausible supposition.)
(2) In making an assertion of the form "Pbut also Q' a
speaker indicates by implication his acceptance of P. (A
logico-linguistic fact.)
(3) In making an assertion of the form "Pbut I don't believe
that P" a speaker overtly states his rejection of P (i.e., his
nonacceptance thereof). (A logico-linguistic fact.)
(4) It follows from (2) and (3) that in making a statement of
the form "Pbut I don't accept that P" a speaker contradicts himself by indicating both that he accepts P and that
he rejects it.
(5) (4) is incompatible with (1).
In the circumstances, we have little alternative but to abandon
(1). Assertions of the form" P but I don't believe that P" are
accordingly untenable in contexts of rational communication on
grounds of self-contradiction. 3
The paradox contemplated by Moore is thus dissolved,
because one of its crucial premisses-namely (1 )-is simply
untenable. For while paradoxical statements of the form S are perfectly intelligible (hermeneutically meaningful) and people doubtless do say that sort of thing, they are nevertheless not
informatively meaningful-there is no usable information that we
can manage to draw from them. (Perhaps a more generous reaction would be to say that people do say "P but I do not believe
it" are simply adopting a somewhat incautious shortcut to saying
something like "Though I know full well that p, I nevertheless
still find it difficult to accept its being so," which is a perfectly
meaningful and unproblematic assertion.)
However, not every paradox can be written off so conveniently
by rejecting some key premiss as false or, worse yet, meaningless.
For various paradoxes resist dissolution by this strategy. Sometimes
we need to invoke considerations of comparative plausibility.
3. Think of Spinoza's scornful dismissal of "one who says that he has a true idea and
yet doubts whether it may not be false" (Ethics, bk. I, prop. 8, sect. 2).
46
Chapter 3
3.2
Comparative Plausibility and Prioritization
Comparatively low plausibility is yet another serious deficiency of

paradoxical premisses. The priority status of propositions is often
crucial to the question of conflict resolution. Thus consider:
(1) All A's are B's.
(2) This A is not a B.
If (1) is seen as a law of nature and (2) as a mere guess, then we
must of course jettison (2) as circumstantially untenable. On the
other hand if (1) is seen as a mere theory-a candidate law of
nature, as it were-while (2) reports the discovery of a patent fact
of observation (for instance, the Australian black swan that counterinstances the thesis that "All swans are white"), then it is (1)
that must be abandoned. Its priority rating is thus a matter not of
what a proposition says, but of the epistemic status that it has in
the larger scheme of things.
Assessments of precedence and priority accordingly do not
relate to the declarative content of a proposition; a more probable
or plausible proposition does not somehow say "I am more probable/plausible." Its precedence and priority ranking bears upon
the systemic standing or status that we take a claim to occupy in
the cognitive situation at hand. (Think here of the analogy of an
axiom in mathematics or formal logic. Axiomaticity is not a matter of what the proposition says but rather of its cognitive status or
standing in the system of propositions at issue.)
A thesis is minimally plausible when it is compatible with what
we know. Beyond that, plausibility depends on the extent to
which we are committed to it through evidence and the centrality
of its position within our overall body of cognitive commitments.
We would and should be reluctant to see as plausible a claim that
did not qualify as probable in the epistemic circumstances at issue.
Consider an example of how plausible contentions can engender paradoxes. Optical illusions always engender mini-paradoxes.
Take for example, the illusion created by the stick which looks
bent when held at an angle under water. Here we have two
reports:
47
(1) The stick is bent. (As attested by sight.)

(2) The stick is straight. (As attested by touch.)
The two reports are clearly incompatible. To resolve the incompatibility at issue with such Optical Illusion Paradoxes we must, as
usual, assess the comparative plausibility of the propositions
involved. Thus since touch is more reliable than sight in matters
regarding the shape of smallish things we resolve the apory by
making (1) give way to (2).
Again, consider the optical illusion created when oppositely
oriented arrowheads are put at the end of equal line segments.
(A)
EO
(B) >
<
Here we have
(1) Line segment (A) is shorter than (B). (As reported by
sight.)
(2) Line segment (A) and (B) are equally long. (As reported
by measurement.)
Since measurement is generally more reliable in such matters
than sight, we can straightforwardly resolve this apory in favor of
(2).
In this way, all optical illusions give rise to aporetic situations
that can be resolved through preferential prioritization on the
basis of larger background considerations regarding the plausibility of claims.
3.3
Issues of Plausibility: Routes to Plausibility

Assessment and Priority Determination
The propositions we endorse have two dimensions. There is the

hermeneutic or internally content-oriented dimension of what the
proposition affirms-its meaning. And there is the epistemic or
48
Chapter 3
externally status-oriented dimension of the epistemic standing

or status of those theses within the wider framework of our
commitments.
The former (hermeneutic) dimension is a matter of what the
proposition affirms-of the range of questions it answers. The
second (epistemic) dimension relates to questions that can or
should be answered about it: Is it axiomatic or derivative? Is it
seen as certain or probable or merely plausible? Is its standing that
of a mere hypothesis or supposition or is it an established fact?
Now as the examples we have been considering show, the prime
requisite for the resolution of paradoxes and antinomies is often
not some logical mechanism for more elaborately elucidating the
inner assertoric content of the propositions involved, but a device
for assessing the external status of a proposition by way of a priority rating that reflects its contextual tenability. What matters here
is the availability of principles of precedence or priority for deciding which propositions should be made to give way to rivals in the
conflicting circumstances at hand.
It is a crucial fact of paradox management that paradoxes can
be resolved by abandoning their low-priority premisses. And there
are two main routes to determining the propositional priorities
required for paradox resolution, namely fiat and inquiry-or,
equivalently, stipulation and investigation. As regards the former,
one distinction is very important for present purposes: the difference between a stipulation (or stipulative hypothesis) and a supposition (or fictive hypothesis). As this terminology will be used
here, a "stipulation" is a postulate which, by conventionalized
fiat, lays down the conditions of a problem. Stipulations accordingly can and should be viewed as fixed facts even though custom
allows it to be said that such stipulations are true "by hypothesis"
[ex hypothesi]. Full-fledged hypotheses by contrast are assumptions
made "for the sake of discussion." They do not qualifY as truths;
we are simply invited to go along with them, indulging them for
the time being with a "suspension of disbelief." Thus consider the
piece of reasoning:
(1) Suppose that three men are all sitting in a room which is
otherwise unoccupied. (2) Let one of them be called
"Henry." Now if(3) the other two are also called
"Henry," then (4) everyone in the room will have the
same first name.
49
Here (1) and (2) represent stipulations-truths by fiat so to

speak. By contrast, (3) is a mere hypothesis. In this context of discussion, we are committed to (1) and (2), while (3) is no more
than a provisional supposition. And (4), as a consequence that is
contingent upon (3), is also of a provisional and "merely hypothetical" status. The second pair of statements has a standing very
different from that of the first pair.
In case of an aporetic conflict, problem-definitive stipulations
will of course enjoy the right of way short of their being logically
inconsistent. 4 By issue-definitive fiat they-and their ramifications
by way of implication and presupposition-are something we have
to be prepared to accept. In adopting the issue-definitive stipulation H we ipso facto thereby render its denial not-H untenable in
the context of discussion at issue. And whatever conflicts with the
natural consequences of H in the presence of other high-priority
claims must now also be abandoned. For example, the lawful ramifications of hypotheses dominate over the lawful ramifications of
mere facts. Thus if this lead bar (which does not conduct electricity since lead does not conduct electricity) were instead made of
copper, then it would conduct electricity (since copper conducts
electricity) .
Stipulations apart, however, priority ranking is a matter of plausibility-of how firmly the contentions at issue are rooted in the
cognitive scheme of things. And this is not just a matter of evidence but of the overall systematization of one's beliefs. Thus if I
know but little about elephants it seems plausible to think they
would have a longer lifespan than gorillas once I know that as a
rule large mammals live longer than small ones. But the rule that
larger animals eat more than smaller ones is firmer, and so that elephants eat more than gorillas is substantially more plausible.
In contexts of factual inquiry a contention is plausible to the
extent that there are good grounds for thinking it to be true
(good grounds for maintaining it, cogent evidential or probative
reasons for it). Indeed this is a feature that is characteristic of
"contexts of inquiry" as such. There are, to be sure, other sorts of
contexts where other sorts of considerations can come into play in
matters of plausibility and cognitive priority. And in due course
4. For complications introduced by logical inconsistency see Chapter 12 below.
50
Chapter 3
we shall see how priorities that are not plausible come into play
and what sorts of ground-rules are operative there. The basis for
these matters is in the end pragmatic and rests on the particular
purposes that are at issue in the various contexts of deliberation.
Overall we arrive at the following priority ordering in contexts
of inquiry. Issue-definitive stipulations come first. These have right
of way come what may. But from there on in it is a matter of plausibility. And here the order of decreasing priority stands as follows:
Definitions and acknowledged conceptually necessary
truths. (Linguistic conventions, mathematical relationships
and principles of logic included.)5
General principles of rational procedure in matters of
inquiry (inductive science) and world outlook.
Patent observational or experiential "facts of life" regarding
the world's concrete contingent arrangements.
General laws and well-confirmed generalizations.
Highly probable contentions regarding matters of contingent fact.
Reasonably warranted suppositions.
Merely provisional assumptions and tentative conjectures.
Speculative suppositions.
The middle range of this register is occupied by the sorts of
propositions that Aristotle called endoxa in the opening chapter of
his Topics-that is to say, generally acknowledged beliefs and
widely accepted opinions. (This linkage reinforces the idea that it
need not be specifically truths that are at issue.)
Certain general principles of precedence implement the preceding ordering through prioritizing
-
the comparatively more basic and fundamental.
the comparatively more factual or less conjectural.
5. That stipulations can predominate even over these will become dear in the discussion of per impossibile reasoning in section 12.6.

-
the comparatively more probable or more reliable.
the comparatively better-evidentiated
the comparatively more commonplace or less far-fetched.
51
Precedence and priority in point of plausibility are thus largely

determined by the criteria and principles inherent in the teleological goal structure of factual inquiry, namely to aid us in doing the
best we can to secure viable answers to our questions about how
matters actually stand.
Plausibility is one of our principal guides to paradox resolution. But in deliberation, as in life, damage control is generally
not to be realized free of charge. \Vhen the propositions that we
abandon are meaningless or false, their abandonment is effectively
cost-free. But not so with plausible propositions. Clearly the more
plausible a thesis is, the greater the price of its abandonment. In
foregoing a plausible thesis we always lose something we would
(ideally) like to have. And when this must be done for want of a
cost-free alternative, we have to view the loss we sustain as the
price of the best-available bargain.
3.4 Against Semantic Absolutism

\Vhen the premisses of a paradox represent known facts (and their
denials accordingly known falsehoods) there is then no arguing
with them: we have to take them in our stride. But when they are
mere plausibilities rather than known truths they will not be gift
horses into whose mouths we dare not look.
The presently contemplated treatment of aporetic situations is
thus radically different from the traditional approach, whose only
recourse is to dismiss some of the conflict-generating theses as
false when without recourse to plausibility. \Vhen acceptance-astrue is the only available mode of endorsement, conflict resolution demands the identification of falsehoods. However if
acceptance is envisioned more broadly to include also acceptanceas-merely-plausible, then thesis-rejection need not be categorical
(as flat-out false). Dismissal is accordingly not something final and
absolute, but merely a matter of abandonment in a context on
groWlds of here being the weakest link. Being the weakest link in
52
Chapter 3
one context does not mean that a thesis will also be the weakest
link in others. A plausible thesis that we must reject with regret in
one setting may yet live to fight another day elsewhere. And as we
saw in Section 2.3, in the context of paradox management the
abandonment of a thesis does not necessarily preclude us from
deriving some informational benefit from it.
To be sure, a logical rigorist might object as follows:
It makes no sense to move beyond "acceptance as true" to
envision also a more tolerant and less restrictive conception of

"acceptance as plausible." For the contradictions that arise
here show that such laxity opens the doorway to paradox and
perplexity.
However, the rigorism of rejecting all considerations of mere
plausibility outright carries a price. For when we carry our
demands for categorical certainty too far we do so at the price of
ignorance-of losing promising answers to our questions. The
main aim and purport of the present deliberations is to show in
detail that inconsistency is no fatally final flaw and that paradox
need not issue in perplexity. As long as we have a sensible mechanism of precedence and priority determination at our disposal-as
we indeed do-we are in possession of an Ariadne's thread to
guide us out of the labyrinth of paradox.
We here come to what is perhaps the most serious challenge to
the present plausibilistic approach. This is the semantic absolutism
of those who would argue as follows:
We agree with the idea of consistency restriction through
omission. But rather than seeing that weakened link in the
chain as merely contextually untenable on grounds of comparatively minimal plausibility we would instead insist upon dismissing it as outright false.
However this variant rigoristic approach has serious drawbacks.
Many paradoxes resist dissolution by way of premiss falsification.
This is illustrated by such situations as, for example, that involved
in the Lottery Paradox. For consider a lottery based in drawing an
integer in the range 1 to 1,000,000. And now consider the following apory-engendering cluster.
53
(I) Some integer in the range 1-1,000,000 will be drawn.

(A defining condition of the problem-situation.)
(2) The particular integer iwill not be drawn. (A very plausible, because so highly probable, proposition.)
( 3) Theses (2) holds for any-and therefore every-particular
value of i in the range 1-1,000,000. (A consequence of
(2) and the defining conditions of the problem.)
(4) By (2), (3), no integer in the range 1-1,000,000 will be
drawn.
(5) (4) contradicts (I).
Here {( I), (2), (3)} form an inconsistent triad. In view of the
indicated considerations, the priority situation is (I) > [( 2), (3)].
In view of this we have to abandon the idea that a particular integer will not be drawn-though of course we are left totally in the
dark about how this will happen. All that we can say is that some
(otherwise unspecifiable) integer in the range 1-1,000,000 will be
drawn. (Pp. 222-24 below will return to this paradox.)
Accordingly, a semantic absolutism that says "abandon only
those theses whose falsity you can specifically establish" would
plunge us into difficulty. Sometimes we can only establish that
one of a group of theses must be false but cannot carry this
through to the specificity of saying which one is false. 6 And here
lies the advantage of dealing in variable plausibility rather
than truth/falsity alone. For this enables us to resolve paradoxes
through dismissing comparatively implausible prepositions
rather than having to find propositions that are identifiably
false (as it may well prove impracticable to do). In paradox resolution we face a dilemma of choice: in the interests of consistencymaintenance some of the pertinent claims must be abandonedsome item made to give way to others. We cannot have it all and
must settle for minimizing the loss. And in this matter of damageassessment plausibility suffices to do the job.
6. The Sorites Pllrlldoxdiscussed in section 5.1 affords another good example of this.

Zeno's Racetrack Paradox
Aristotle's Justification Paradox
Cosmic Explanation Paradox
Sancho Panza's Hanging Paradox
The Nasty Crocodile Paradox
The Contract of Protagoras Paradox
The Paradox of Existence
The Russell-Gettier Paradox
The Bizet-Verdi Paradox
4.1
Four Modes of Paradox Solution
Solving a paradox by eliminating the inconsistency of an aporetic

cluster calls for abandoning some of the contradiction-engendering premisses involved. But there are importantly different bases
for doing this, as follows1. An independent determination that some of the premisses
essential to the aporetic contradiction at issue are either
meaningless or flat-out false and can therefore be dismissed
altogether-with the effect of restoring consistency. (In the
former [meaninglessness] case the paradox is interpretatively dissolved; in the latter (falsity) case it is probatively dissolved on grounds of the untenability at issue.)
2. A determination that some of the premisses essential to the
contradiction at issue are so interrelated that in every case
the interpretative or applicative circumstances under which
some of them are tenable are such that some others are
untenable (that is, false or meaningless), so that there is no
possible way in which all of these essential premisses can be
uniformly true together. (In such cases of interpretive dissonance the paradox is lexically dissolved on ground of
ambiguity or equivocation.)
3. A determination that some of the premisses essential to
the contradiction are comparatively so implausible in relation to the others that they can be rejected in the context
at issue.
57
58
Chapter 4
And moreover3a. The consistency-restoring dismissal of minimally plausible theses can be achieved in one single way. (In this
case the paradox is decisively resolved by plausibility
considerations.) Or3b. The consistency-restoring dismissal of minimally plausible theses can be achieved in several different but
equally eligible ways. (In this case the paradox is indecisively resolved through a mere disjunction of alternative possibilities.)
In line with item 1 of the initial list, it must be recalled that there
are two distinct routes to the meaninglessness of aporetic premisses. For it is important to distinguish between hermeneutical
and semantical meaninglessness. Hermeneutical meaninglessness
obtains when there is no way of making informatively interpretive sense of the proposition at issue: it is more or less uninterpretable nonsense. Semantical meaninglessness, by contrast,
arises when there is no possible way of assigning a stable truth
status to the proposition at issue as either definitively true or
definitively false.
We thus arrive at three levels of paradoxicality according as the
paradox in question admits of dissolution, of decisive resolution,
or merely of an indecisively disjunctive resolution. Moreover, in
the first case, we will no longer see that supposed paradox as a
really serious problem because we now view it as actually the
product of an error of some sort-to wit one that rests on the
mistake of accepting theses that are meaningless or false or not
co-tenable.
It must be noted that the difficulty of a paradox is something
quite different from its level of paradoxicality. For this is a matter
of how complicated it is to make the determinations of viability
and comparative priority needed for arriving at a solution. This is
a matter of the conceptual and epistemic hurdles one has to overcome to provide a validating rationale for the particular acceptability determinations and priority assessments that provide the
basis for paradox resolution.
4.2
59
Dissolvable Paradoxes
Zeno of Elea, who was born around 495 B.C., was one of the
most celebrated and influential pre-Socratic philosophers of Greek
antiquity.l Aristotle regarded him as the inventor of philosophical
dialectic. His famous paradoxes of motion were the subject of
innumerable treatises and discussions. Among these arguments
designed to show the impossibility of motion was his classic
Racetrack Paradox, which is represented in the following account:
A runner is to race from the starting point to the goal. But to
reach the goal he must first reach the halfway point. And to
reach this point he must first reach the halfway point to it. And
so on ad infinitum. So the runner can never make a start at
aU-he is immobilized.
The paradox at issue here roots in the following aporetic cluster:
(1) To reach any destination whatever, a runner must first
reach the halfway point en route.
(2) By repetition of this principle at issue in (1) reaching the
ultimate goal requires an infinite (unending) series of
prior goal-reachings that must be achieved in advance.
(3) Since the runner moves at a finite speed, every goal-reaching he manages to achieve requires a finite amount of
time.
(4) So in virtue of (2) and (3) reaching the goal requires an
infinite amount of time, since an infinite number of finite
quantities cannot yield a finite total.
(5) Thus the runner can never reach the goal at aU. (By (4 ).)
(6) But since this argument is general and holds for any goal
whatsoever-and thereby for any point different from the
starting point-the runner is immobilized.
1. On Zeno see Footnote 5 on p. 94 below.
60
Chapter 4
(7) Yet as we know full well, runners are not immobilized: he
can and will reach the goal.
In searching out the fly in this ointment-the weak link in the

chain of inconsistency-suspicion comes to focus on the contention of thesis (4) that an infinity of finite quantities cannot
yield a finite total. For in simply deploying the finite-infinite contrast this contention neglects the matter of the size of the finite
amounts at issue. And the fact of the matter is that an infinite
number of increasingly diminishing quantities can indeed make
up a finite total-as modern mathematics shows in such cases as:
t+t+t+=l
So here the dismissal of premiss (4) as false eliminates the inconsistency and thereby dissolves the paradox. But this is something
that is not obvious on the surface because here the rationale for
falsity-rejection calls for recourse to the mathematics of infinite
series.
Zeno's Racetrack paradox thus illustrates the aforementioned
difference between levels of paradoxicality and levels of difficulty.
Its paradoxicality level is low, since it is unproblematically dissolvable through rejection of a pivotal premiss. On the other hand,
establishing the untenability at issue through a broader rationale
for what it is that makes these flawed premisses untenable is an
issue of substantial subtlety and difficulty. Making a decisive resolution of a paradox stick can be a complex business.
Again, consider the following Justification Paradox that figured prominently in Aristotle's thought:
(1) A rational belief must be based on a rationale of justifying
considerations.
(2) The rationale of a belief cannot include the belief itself
since this will produce a vicious circle. No belief can figure
in its own appropriate justification. The rationale of a
belief must therefore always involve further, different
beliefs.
61
(3) Since this leads to an infinite regress, there can be no such

things as a rational belief.
(4) We know full well that many beliefs are rationally justified.
Aristotle's resolution of this apory was two-sided. In the case of
necessary beliefs he rejected (2), holding that these beliefs are
effectively self-certifying, so that the justifactory regress here ends
with self-evidence. (He insisted that such self-substantiation is not
vicious or vitiating in the case of necessary truths but simply
makes manifest the nature of their necessity.) And in the case of
contingent factual beliefs he rejected (1), maintaining that their
validation need not be mediated discursively by further beliefs but
can be based upon the course of immediate experience.
4.3
Decisively Resolvable Paradoxes
Consider the following Cosmic Explanation Paradox:

(1) Every feature of nature has a natural explanation.
(2) Therefore: nature-as-a-whole, the universe, has a satisfactory natural explanation.
(3) Natural explanations explain some feature of nature in

terms of others.
(4) But in explaining nature-as-a-whole causally we cannot
involve its parts or features, since this would make for a
vitiating circularity.
(5) Therefore: we cannot give a satisfactory natural explanation
of nature-as-a-whole.

However, (1) is ambiguous. It can mean "every partialfeature of
nature" in which case it is true but does not yield (2). Or it can
means "every feature of nature-holistic ones included" in which
case it yields (2) but is problematic and-in the present contextquestion-begging. And so the problematic nature of (1) renders
the paradox decisively resolvable.
Chapter 4
62
A historic example of a sophisticated paradox is afforded by

the Sancho Panza Hanging Paradox of Cervantes's novel Don
Quixote (Book II, Chapter 51). Here Sancho Panza is made governor of an island where he must uphold a curious law that
stipulates that on arrival visitors are to be questioned about their
plans and to be hanged if they answer falsely. One day Sancho had
to resolve the problem posed by a visitor who, when asked what
he would be doing on the island, responded "I am here to be
hanged." Now if he is hanged, the answer will be true and the
penalty inappropriate, while if he is not hanged, the answer will be
false and the law would require his hanging. 2
The situation that results answers to the following summary:
The traveler is
The traveler speaks
The law therefore requires
Hanged
Truly
Not hanged
Not hanged
Falsely
Hanged
In no case can the traveler's fate be brought into agreement with

the law's requirements.
This situation gives rise to the following aporetic cluster:
(1) The law should always be obeyed. Lawful requirements
should always be met. People ought (always) to do as the
laws require.
(2) "Cannot" implies "need not": what people cannot possibly do is never actually obligatory for them. ( Ultra posse
nemo obligatur as the Roman legal maxim maintained.)
(3) Ought implies can: what people legally should do must be
possible for them. From (2) by contraposition.
2. This classic paradox also has the ancient variant of the Nasty Crocodile Paradox
mentioned twice by Diogenes Laertius (Lives of the Philosophers, 44 and 82). Having
snatched a baby the nasty crocodile turned to the father: "Answer carefully," he said, "for
your baby's life depends on your truthfulness: will I eat your baby?" After thinking for a
moment the cagey father replied: "Yes, I do believe you will." For this paradox sec Prantl,
Geschichte, Vol. I, p. 493. (Ashworth 1974, p. 103, discusses the variant Bridgekeeper
Paradox featuring the functionary who threw liars into the water.)
63
(4) In some circumstances-such as those indicated by the

example of "The Gallows" story-people cannot possibly
do what the laws require.
( 5) In some circumstances people need not do as the laws
require. (From [3] and [4].)
Here {(I), (2), (4)) constitute an inconsistent triad. Now (1)
specifies a plausible legal ideal, while (2) is a basic axiom of
juridico-ethical rationality. And (4) is an aspect of the defining
hypothesis of the situation. The overall rating of priorities is thus:
(4) > (2) > (1)
And in consequence (2), (4)/(1) is the appropriate resolution for

the paradox. No other R/A-alternative can match its acceptability
profile of {I, 1, O} utilizing the prospects of retained theses in the
three minority categories. One must turn a blind eye to that law,
presumably arguing (1) is false because occasionally "the law is an
ass" and sometimes requires absurd things.
Sancho's sagacious response took exactly the line "\Vhichever
way I decide, the law is going to be disobeyed, so I might as well
be merciful and let this wretch go free." And this makes perfectly
good sense. In the circumstances the tacit premiss of the situation
"The law is to be obeyed: what is done should be in accordance
with the law" needs to be abandoned in the case at hand precisely
because this case is so constructed that there just is no possible
way for the law to be obeyed. The paradox is thus dissolved by
noting that it rests on the mistaken presupposition that the legalcode's stipulation is a meaningful law that can and should always
be obeyed.
Another juridical example of a decisively resolvable paradox is
afforded by the ancient story of the paradox of The Contract of
Protagoras. This paradox, widely disseminated in classical antiquity, arises from the following story:
The Greek sophist and teacher Protagoras contracted with
some students for a tuition fee to be paid if, but only if, they
won their first case. One of these, the clever Euathlos, sued in
64
Chapter 4
court for free tuition with the argument: "If I win this case
then by the court's judgment I will owe nothing. And if I lose,
then I will owe nothing thanks to the contract, since this is my
first case." Protagoras, being no less clever, of course argued
exactly the reverse. 3
The paradox that emerges here inheres in the following aporetic
paIr:
(1) The court's finding should be in accord with what the
contract stipulates. However(2) In the specified circumstances, that contract and the
court's ruling are bound to disagree on the following
basis:
The Court)s
Finding
A fee is due
No fee is due
Winner of
the Case
P
The contract
Specifies
No fee is due
A fee is due
Thus in no case can the court's finding be brought into line with
the contract's specification.
Now since thesis (2) represents the defining hypothesis of the
problem, there is no arguing with it. The only available option is
to make (1) give way, invoking the legal axiom that an absurd
contract is no contract because it invalidates itself (conventum
absurdum non est conventum). To be sure, the general considerations of aporetics indicating that (1) should be abandoned do not
tell us what-if anything-should be put on its place. This
requires a closer look at the substantive detail of the particular
paradoxical situation that is before us. And on this basis, the
court's only sensible option is to make a ruling and to prioritize it
over what the contract specifies in the circumstances (thereby
abandoning (1)). Now here the salient feature of this case is that
3. On ancient discussions of this paradox see Prantl, Geschichte, Vol. I, pp. 493-94.
65
the situation between the contending parties is entirely symmetric, exactly as the response of Protagoras indicates. Thus by
deploying the basic principle of justice "Treat like cases alike" the
Court's sensible ruling would be to divide the amount in contention (the student's fee) equally between the parties, awarding
just half of the agreed amount to Protagoras. To be sure, this
requires the explicit-and deliberate-sacrifice of thesis (I), but
this is entirely appropriate in the circumstances.
In any event, however, this paradox is decisively resolvable
since it is clear which aporetic premiss must be abandoned. The
only real problem here is that of why--that is, of establishing just
exactly how the justifYing grounds for this abandonment are to be
articulated. And it deserves stress that this sort of thing is often
the most challenging and difficult aspect of paradox management.
In comparing the Sancho Panza Paradox and the Contract of
Protagoras Paradox one observes that they have exactly the same
generic structure, as manifested in those alternative-detailing tabulations. (And this is also the case with the Liar Paradox to be discussed in 10.2 below.) On this basis the Stoic paradoxologists of
classical antiquity characterized such paradoxes as based on
"table-turning" (antistrophe),4 seeing a fundamental symmetry
obtains so that whatever can be said on one side can be answered
by an equivalent counterpart on the other. That is, we have the
situation that not just equally cogent but in fact structurally identical arguments can be given both for a thesis P and for its negation not-Po
4.4
Indecisively Resolvable Paradoxes
Some paradoxes do not admit of a decisive resolution. For example consider the following Paradox of Existence:
(I) The real is rational: for every actual state of affairs there is
a reason why (that is, why it is so rather than otherwise).
[This is the "Principle of Sufficient Reason".]
4. Sc:c: Prand, Geschichte, Vol. I, pp. 493-94.
66
Chapter 4
(2) Therefore: there is a reason for the world's existence-and
indeed for its existing as it is.
(3) The satisfactory explanation ofa mundane (world-appertaining) fact requires a mundane rationale. (For an extramundane explanation would run afoul of reason's
proscription of explaining obscurum per obscurior.)
(4) Therefore: the real has an explanation in terms of mundane
facts.
(5) Any mundane (world-appertaining) fact is either partial or
global: it must either relate to some part or aspect of
nature or to some feature of it.
(6) But the world's satisfactory explanation cannot be partitive: the whole cannot be satisfactorily accounted for
through some part or aspect of itself.
(7) Nor can the world's satisfactory explanation be of the
intra-mundane sort, for then it would be circular, seeing
that it would account for reality-as-a-whole in terms of
itself.
(8) Therefore: there is no rationally satisfactory explanation of
the existence of the-world-as-a-whole.
It is clear that {(I), (3), (5), (6), (7)J constitutes an aporetic cluster. Here (3), stipulates a fundamental principle regarding the
sovereignty and modus operandi of reason, and (5) is an inevitable
principle oflogic. (1), (6), and (7), by contrast, are no more than
plausible principles. We thus arrive at the following priority ranking: (5) > (3) > [(1), (6), (7)]. It emerges on this basis that the
paradox has three viable resolutions based on the preferentially
optimal R/A-alternatives (3), (5), (6), (7)/( 1); (1), (3), (5),
(7)/(6); and (1), (3), (5), (6)/(7). Three distinct pathways to
resolution are thus available:
(1) - rejection. Abandoning the idea of the explicability of the
real.
( 6 ) -rejection. Reliance on cosmic evolution. This envisions a
naturalism that sees the world's entire nature and the
67
world's laws as evolving emergently over the course of

time for some proto-status of physical existence.
(7) - rejection. Rejecting the idea that self-explanation is always
vicious. Accepting a circularity of sorts and distinguishing
between vicious and virtuous circles.
A trio of options confronts us, none of which prevails over its
rivals by considerations of general principle. A disjunctive resolution is the best we can achieve here.
Another paradox that has been extensively discussed in the
recent epistemological literature was initially posed by Bertrand
Russell in Human Knowledge,S and amplified in a widely discussed paper by Edmund Gettier. 6 This Russell-Gettier Paradox
stands as follows:
(1) X believes (accepts) p v q.
(2) X believes this only because he justifiedly believes p; he
disbelieves q.
(3) In actual fact, however, q is true, and p is false--even
though all the evidence at X's disposal points in its direction.
(4) X's belief in p v q is (subjectively) justified (because this
follows deductively from something that he reasonably
believes).
(5) X's belief in p v q is actually true. (From (3) because q is
true).
(6) When a belief is both subjectively justified and objectively
true then it constitutes knowledge. Thus X knows P v q.
(7) Nevertheless we cannot and would not credit X with
knowledge of p v q because his basis for its acceptance is
altogether erroneous and inappropriate.
5. Bertrand Russell, Human Knuwledge: Its Scope and Limits (New York: Simon and
Schuster, 1948), pp. 154-55.
6. Edmund Gettier, "Is Justified True Belief Knowledge?" Analysis, vol. 23 (1963),
pp.122-23.
68
Chapter 4
Given this problem situation we must examine the epistemic status of the theses at issue. Now (1)-(3) are facts fixed by the defining stipulations of the problem. (5) follows from (3). (4) and (6)
are plausible epistemic principles, albeit not as clearly plausible
and appropriate as (7). We thus obtain the following priorities
with regard to the premisses of the aporetic cluster I(1), (2), (3),
(4), (6), (7)} that stands before us:
[(1), (2), (3)] > (7) > [(4), (6)]
In the circumstances we have two optimal RjA-alternatives,

namely those that reject only (4) and only (6) respectively. Both
have the same optimal available retention profile {I, 1, t}. We are
thus driven to abandoning either (4) or (6). Here too straightforward plausibility considerations again do not constrain one particular outcome. We again achieve only a disjunctive resolution.
It is not that in such indecisive cases we cannot see the way
clear about how the paradox can be resolved on the basis of plausibility considerations. Rather, the situation here affords us with a
plurality of options-an embarrassment of riches, so to speak.
Now sometimes these will be reason to prefer one alternative over
the others. But sometimes there is no prospect of this because
consistency can be restored in several equally eligible ways. The
Bizet-Verdi Paradox affords a good example?
Consider the question: "What if Bizet and Verdi were fellow
countrymen?" This confronts us with the counterfactual situation
of the conditional:
If Bizet and Verdi were fellow countrymen, then Bizet would
be (French? Italian?) and Verdi would be (French? Italian?).
The context of the issue is set by the following items of accepted
knowledge:
(1) Bizet was of French nationality.
(2) Verdi was ofItalian nationality.
7. For the paradox see Rescher 1961.
69
(3) Fellow countrymen are persons who have the same

nationality.
(4) Bizet and Verdi were not fellow countrymen.
The hypothesis of the conditional instructs us to jettison (4) and
replace it with
(4') Bizet and Verdi were fellow countrymen.
However, the resulting group {(I), (2), (3), (4')} is still an inconsistent quartet. If we are to maintain the hypothesis (4') and the
meaning-explanation built into (3), then we only have two ways
to resolve consistency: (A) to drop (1) and make Bizet an Italian,
or (B) to drop (2) and make Verdi a Frenchman. We thus confront two maximal consistent subjects of our aporetic cluster,
namely {(I), (3), (4')} and {(2), (3). (4')}.
Since (4') is a defining hypothesis of the pattern and (3) a conceptual truth, we arrive at the overall priority ordering:
(4') > (3) > [(1), (2)]
Here the alternatives of (I)-abandonment and (2)-abandonment
are tied in point of acceptability considerations. In view of this
alternatives (A) and (B) are on exactly the same plane: there is no
reason to give one precedence over the other, and the only safe
course is to disjoin them, so as to arrive at the conditional:
If Bizet and Verdi were fellow countrymen then either both
would be French or both would be Italian.
This noncommittal "either-or" conditional is the best we can
achieve in this case. Here we have to settle for disjunctive solution: in the circumstances any more definite conclusion would be
entirely unjustified. s
8. For more detail on paradoxes engendered by these "counterfactual conditionals"

see Chapter 12 below.
70
Chapter 4
It is clear, however, that "undecidable" paradoxes of such a

sort-where no particular resolution can be privileged on the
basis of general principles-are distinctly more deeply paradoxical
than their simpler compeers where a decisive resolution can be
achieved.
4.5
The Classification of Paradoxes
Like pretty much anything else, paradoxes can be classified

according to various different principles. The prime prospects here
are: subject matter content, etiology in term of what it is that goes
wrong so as to give rise to the paradox, mode of resolution, and
complexity with reference to the extent to which elaborate analytical machinery is required to address the paradox.
Classification by content is the most familiar of these processes.
Its results are indicated in Display 4.1 on pp. 72-73. Content is a
function of the subject matter at issue in the paradox. It turns on
the sort of vocabulary being used. Thus logico-semantical paradoxes would deal with issues of affirmation and denial, predication and relation, designation and reference, and so forth.
Mathematical paradoxes would deal with sets and their membership, or again with quantitative or geometric relationships.
Epistemic paradoxes deal with belief, knowledge, inquiry, or ignorance; ethical paradoxes with conflicts of duty and obligation. And
so on.
Apart from content, the three salient issues regarding a paradox are: how it arises, how it should be resolved, and how much
effort its resolution requires. The key parameters are thus: content, etiology, treatment, and difficulty. For present purposes, etiology will serve as the primary classificatory principle, and the
subsequent discussion will focus primarily upon how it is that the
paradox arises-the root cause of the perplexity at issue. And this
approach proceeds in the footsteps of Aristotle's own practice in
his Sophistical Refutations, where he treated sophism according to
the sort of error or fallacy that he took to be at issue, and thereby
according to the sort of resolution he saw as appropriate.
Now in general, paradox roots in cognitive overextension-it
arises through our accepting too much. And so, what is required
throughout paradox management is premiss abandonment. But
71
this is something that can be validated by different rationales.

There are thus different modes of paradoxicality. All paradoxes go
wrong, but different paradoxes can go to this common destination along different routes:
meaninglessness;
falsity;
vagueness: mistaken assumption of a clarity of meaning or

definition;
ambiguity and equivocation: mistaken assumption of conformity of meaning;
implausibility (insufficient evidentiation or improbability,

flawed induction);
unwarranted presupposition;
truth-status misattribution: inappropriate truth-status

imputations;
untenable hypotheses; assumptions at odds with the cognitive

environment;
value conflicts: perplexing choices.
These seven nagativities represent the major deficiencies through

which paradoxes corne into being. When one asks "What has
gone wrong?" in the case of paradoxes it is in these directions that
one must principally look.
Still, the pathways to paradox, though plural, represent a limited plurality. When we clarify paradoxes in line with "how things
go wrong" we arrive at smallish list-pretty much as indicated
above.
Note, however, that the resultant classification need not yield
mutually exclusive taxa, since several things might go wrong at
once in sufficiently awkward cases. Nor will a taxonomy in terms
of modes of resolution do so. For if the paradox is so paradoxical
that several different things go wrong-and that in consequence
each of several different processes would suffice to effect a resolution' then this one selfsame paradox will fit into several different
categories even within the same classification.
72
Chapter 4
DISPLAY 4.1
CLASSIFICATION OF PARADOXES
BY SUBJECT MATTER
SEMANTICAL PARADOXES
(essentially involving the ideas of truth,

falsity, and reference)
-
truth-oriented paradoxes
reference-oriented paradoxes
MATHEMATICAL PARADOXES
set theoretic paradoxes (essentially involving the ideas of sets

and set membership)
Russell's Paradox
-
numerical paradoxes (additionally involving the idea of

counting and ordering)
Cantor's Paradox
Berry's Paradox
Richard's Paradox
- geometrical paradoxes
-
probabilistic paradoxes
PHYSICAL PARADOXES
Zeno's Paradoxes of Motion

EPISTEMIC PARADOXES
(essentially involving the ideas of

knowledge and belief)
Moore's Paradox
The Socratic Ignorance Paradox
Aristotle's Justification Paradox
The Lottery Paradox
PHILOSOPHICAL PARADOXES
moral paradoxes

-
metaphysical paradoxes
The Paradox of Existence

-
paradoxes ofphilosophical theology
The Paradox of Evil
73
74
Chapter 4
Unquestionably, disagreement will often arise among paradox

analysts. One of them may think that the problem lies with one of
the aporetic theses (say due to equivocation) while another thinks
that something else (say a problematic presupposition) is at fault.
And it is, of course, perfectly possible that both may be right
because there are multiple flaws within the aporetic group at
lssue.
Fortunately, however, this sort of thing does not usually happen and paradox classification in terms of "what goes wrong" is
often a straightforward matter. Thus of paradoxes those considered in Chapter 1, discrepant report paradoxes and sense-illusion
paradoxes both belong to the category of problematic inductions,
seeing that evidential dissonance is involved. The who-is-theengineer paradox involves the unwarranted presupposition that
the person-description at issue-which is actually self-contradictory-succeeds in identitying someone. Philosophical paradoxes
frequently trade on an ambiguity or equivocation that needs to be
removed by using distinctions.
The ensuing chapters will examine a wide range of paradoxes
in line with the those above-indicated etiological groupings. And
the discussion will endeavor to show that despite substantial variation as regards the source and nature of the paradoxes at issue,
the single uniform approach to resolution described in these initial chapters can be used throughout the entire course of paradox
management. Paradox resolution, in sum, is a matter of pursuing
a unified strategy by means of diversified tactics.

The Heap Paradox or Sorites (Eubulides)
The Bald Man Paradox or Phalakros
(Eubulides)
The Millet Seed Paradox (Zeno)
The Color Continuum Paradox
The Paradox of Sir John Cutler's Stockings
The Ship of Theseus Paradox
The Hermeneutic Circle Paradox
5.1
The Paradox of the Heap
Vagueness in the application of terms is a major source of paradox. Vague terms will generally have a more-or-Iess well-defined
central core of application surrounded by a large penumbra of
indefiniteness and uncertainty. And so when a term T is vague,
non- T will automatically also be so. There will accordingly be a
fuzzy region of ambivalent overlap between T-situations and nonT situations where our plausibilistic inclination is to see the issue
both ways. The aporetic inconsistency of paradox then becomes a
tempting possibility.
The eristic (from he eristike as wrangling or dialectic) that the
Megarian school of Greek antiquity took over from the Sophists
in the fourth century B.C. was an early version of dialectic--or of
spurious dialectic according to Aristotle, l its principal opponent,
who dealt with the matter extensively in his Sophistical
Refutations. 2 It was thematically oriented towards puzzles, paradoxes, and sophisms. Aporetics is the descendant of this enterprise
as it continued through the mediation of the medieval study of
sophisms and insolubilia.
Eubulides of Miletus (born ca. 400 B.C.) was the most prominent and influential member of the Megarian school of dialecticians as whose head he succeeded its founder, Euclid of Megara, a
1. Aristotle, Suph. Elen., II, 6.

2. On the Megarians in general see Zeller, Philosophie der Griechen, Vol. II/I, pp.
244-275. For Eubulidcs see Diogenes Laenius, Lives of the Philosophers, II x lOB-lID;
Prantl, Geschichte, Vol. I, pp. 21, 50--5B; Aristotle, Soph. Elen., 179a32ff.
77
78
ChapterS
pupil of Socrates. 3 The Megarians derived this concern from the

ancient Sophists who reveled in reason-baffling paradoxes because
they brought grist to the mill of their teaching that reason is inadequate for enabling us to grasp the reality of things. (How-they
asked-can there be a point in rational inquiry if one must
already know what is true in order to recognize it when one finds
it?4) Like the Sceptics later on, the Sophists were not concerned
with paradoxes for their own sake but because they saw them as
making manifest a fundamental aspect of the human condition.
The Megarians took the view that what was pivotal for dialectic was not so much logic, the theory of correct inference, as eristics the theory of error avoidance. s As they saw it, conflict is the
avenue to health. Knowing how to deal with sophisms-how to
manage thought in situations of conflict and competition-is as
important for tuning the mind as sporting competition is for tuning the body.
Eubulides did more to promote concern for paradoxes than
any other single thinker in the history of the subject, with the
exception of Zeno of Elea. He is credited with seven important
paradoxes: The Liar (pseudomenos), The Overlooked Man (dialanthanan) , Electra and her Brother, The Masked Man (egkekalummenos) , The Heap (sorites), The Horns (keratines), and The Bald
Man (phalakros). At present only The Heap and The Bald Man
will concern us; the other paradoxes of Eubulides will be considered in due course.
The "Paradox of the Heap"-the Sorites Paradox (from the
Greek saros = heap )-is posed in the following account:
A single grain of sand is certainly not a heap. Nor is the addition of a single grain of sand enough to transform a non-heap
into a heap: when we have a collection of grains of sand that is
not a heap, then adding but one single grain will not create a
heap. And so by adding successive grains, moving from I to 2
to 3 and so on, we will never arrive at a heap. And yet we
3. Pretty well all that is known about Eubulides derives from Diogenes Laertius, Lives
of the Philosophers, bk. II, sect. 106-120. See Zeller, Philosophie der Griechen, Vol. II11,
pp.246.
4. Zeller, Philosophie der Griechen, Vol. 1/2, p. 1380. On the Sophists in general see
the larger context of this passage on pp. 1371-84.
5. On this aspect of the Megarian teaching, see Prantl, Geschichte, Vol. I, pp. 41-58.
79
know full well that a collection of 1,000,000 grains of sand is a

heap, even ifnot an enormous one. 6
To achieve a more perspicuous formalization of the argument, it
is helpful to use some abbreviate symbolism. Let g,I represent a
collection of i grains of sand and let us adopt H(g) to abbreviate
the statement: "The group g of sand-grains is a heap." We then
have:
(1) -H(gI)
an observable fact
(2) H(gI,OOO,ooo)
an observable fact
(3) (Vi)[ -H(g) ~ -H(gi+l)]
a seemingly evident general

principle
(4) -H(gI,OOO,ooo)
from (1), (3) by iteration
(5) (4) contradicts (2)

Here (1) and (3) together logically entail (4). And (4) yields (5)
which contradicts (2). The triad {(1), (2), (3)} accordingly constitutes an aporetic cluster and one of its members must be abandoned if the paradox is to be resolved.
But what of the acceptability-status of these propositions?
What sorts of considerations of priority and precedence are operative here?
As the above indications make clear, (1) and (2) are observable
facts relative to our understanding of what a "heap" is. On the
other hand, thesis (3) is cast at a high level of abstract generality
and is really no more (though also no less) than an eminently
plausible generalization-a theory. The theses of our inconsistent
triad accordingly have the following priority ranking:
[(1), (2)] > (3).
6. On this paradox and its ramifications see Chapter 2 of R.M. Sainsbury, Paradoxes
(2nd cd., Cambridge: Cambridge Universiry Press, 1995), pp. 23--51. Originally the paradox also had a somewhat different form, as follows: Clearly 1 is a small number. And if n is
a small number so is n + 1. But this leads straightway to having to say that an obviously
large number (say a zillion billion) is a small number. (See Prantl, Geschichte, Vol. I, p. 54.)
80
ChapterS
We have three retention/abandonment alternatives here, whose

retention profiles stand as indicated:
(1), (2)/(3)
with retention profile <1, 0>
(1), (3)/(2)
with retention profile < hI>
(2), (3)/(1)
with retention profile < h 1>
The first alternative alone enables us to retain all of the top priority theses and the optimal option here is accordingly to abandon
(3). Plausible though it may seem, this thesis is less so than its
rivals: it is the weakest link in this chain of inconsistency.
And this is perfectly reasonable. For (3) is actually rather problematic. When dealing with such vague concepts that have a slippery slope, as it were, it is clear that we are in a more
confidence-inspiring situation with particular claims like (1) and
(2) and (5) than we are with an unrestricted generalization like
(3). (After all "we know a heap when we see one.") With factual
issues of this sort we stand on firmer ground with particular and
concrete matters than we do with matters of abstract generality.
And yet despite the comparative frailty of (3) we would not
(and should not) simply reject it as flat-out false. For its negation
entails (3i)[ -H(g)
& H(g,1 + 1)], and we would find it very difficult
1
indeed to conceive of sudl an i. But we do not (and need not) set
up (3)'s negation as true; instead, we have the option of seeing
(3) as plausible even though we propose to abandon it pro tem as
"contextually untenable" in the present context.
The course of traditional paradox resolution via the dismissal
of premisses as false is here made impracticable by disassembling
premiss (3) into the following multitude:
(3.1) -H(gI) ~ -H(g2)
(3.2) -H(g2) ~ -H(g3)
( 3.999999) - H(g999.999) ~ - H(gl,OOO.OOO)
81
The paradox at hand means that we must abandon some of these

as unacceptable, but we would be hard-pressed to identify any
one specific culprit. Yet in seeing (3) as merely plausible rather
than flat-out true-we can avert the difficulty which otherwise
arises here. The Sorites Paradox is thus particularly important
because it shows that the doctrinal resource of paradox resolution represented by the strategy of dismissing premisses as false
has its difficulties. But plausibility is something else again.
Clearly, propositions of the format "If C(g), then 9 is not a heap"
grow increasingly implausible with an increase in the size of g,
irrespective of the substance of condition C.
The Greek philosopher Chrysippus (ca. 280-208 B.C.), was
one of the most prolific authors of the Stoic school and its ablest
logician. His extensive writings did much to bring Stoicism into
prominence. Chrysippus saw the analysis of sophism as a cardinal
training-ground for dialectics? The Stoics offered an extensive
treatment of the entire range of the paradoxes known in antiquity
(in particular those of Zeno, the Sophists, and the Megarians). 8
Already Zeno of Citium in Cyprus (born ca. 320 B.C.), the
founder of the Stoic school, made the study of paradoxical
sophisms a key element of his teaching on dialectics. 9 He discussed all of the Megarean paradoxes. 10 And Chrysippus, his principal follower, devoted entire tracts to several of the most
important paradoxes-five of them dealt with the Liar Paradox
alone.l 1 Somewhat in the spirit of the present book, the Stoics
taught that the paradoxes did not involve formal fallacies of reasoning, but rooted in the substantive untenability of essential premisses. With the Liar in particular, Chrysippus offered the
resolution (characterized as cassatio, that is, "null and void," by
the medievals) that the proposition at issue is simply meaning-
7. On Chrysippus see Zeller, Philosophie der Griechen, III/I, pp. 40--44 and 111-18.
His discussions of sophism are detailed in Prant!, Geschichte, Vol. I, pp. 487-496.
8. See Prant!, Geschichte, Vol. I, 485-496 (esp. p. 490).
9. On Zeno see Diogenes Laertius, Lives of the Philosophers, VII, 1-160. See also
Zeller, Philosophie der Griechen, Vol. III/I, pp. 28-34.
10. Ibid. 43-44 and 82. See also Zeller, Philosophie der Griechen, Vol. III/I, p. 116.
11. Zeller, Philosophie der Griechen. Vol. III/I, p. 116. For this paradox see Chapter
10 below.
82
ChapterS
less. 12 However, with only the triochotomy of true/false/meaningless in hand-in contrast to the presently contemplated mechanism of comparative plausibility-this led the Stoics into difficulty
with the analysis of The Heap Paradox where this machinery is
clearly insufficient. Thus it is greatly to his credit that Chrysippus
made the important point that the Heap Paradox (Sorites) cannot
be resolved on the basis of classing the premisses as true or false
or meaningless but requires a total suspension of judgment as to
truth status. 13
What makes this Heap perplexity a "paradox of vagueness" is
that the imprecision of what a "heap" is undermines the accept~ -H(g., + I)]
ability of a generalized principle like V'i[ -H(g.)
,
which rides roughshod over the fact that a slippery slope is at
issue here. This sort of generalization rests on the mistaken
impression that the concept at issue is more sharply defined than
is the case. This can be seen graphically by looking at the situation
from a variant angle. Thus consider the (overly brief) series:
HHH??? HHH
Note here that the insertion Qf a group of indeterminate (neither
H nor H) cases between the H's and the H's enables us to retain
the rule:
Whenever gj is a non-heap, thengj + 1 is not a heap that is, a
non-heap is never succeeded by a heap.
This thesis is, of course, subject to the interpretation that there
ar~ now two ways of failing to be a heap, namely to be non-heap
(H) or to be an indecisive or borderline heap (as indkated by?).
It is now clear that (1) a single step forward from H will never
carry us over to H, while nevertheless (2) -.a sequence of steps
along the series will eventually carry us from H's to H's.
The equally notorious Phalakros (or Bald Man) Paradox (from
the Greek phalakro~a bald man) is closely related since it simply
12. See Riistow 1908, p. 115: Cassantes autem dicunt, quod dicens se dicere falsum
nihil dicit, or again non err TJerum nec falsum, quia nullam tale est propositio. See also
Prantl, Geschichte, Vol. IV, p. 41.
13. See Prantl I, 489. Sextus Empiricus, AdTJ. Math, VII, 416. Cicero, Academica, II,
29 and 93. See Zeller, Philosophic der Griechen, Vol. III/I, p. 116.
83
moves in the direction opposite from the Heap Paradox while

otherwise being strictly analogous.
A man with a full head of hair is obviously not bald. Now the
removal of a single hair will not turn a non-bald man into a
bald one. And yet it is obvious that a sufficient continuation of
that process must eventually result in baldness.
The treatment of this paradox parallels that of the Heap Paradox.
And here too those slippery slope iterations just do not work
smoothly with vague concepts such as "being a heap"--or for that
matter being tall or rich or bald. After all, the key to paradox resolution is the search for the weak spots-those chinks in the
armor of paradox where its defenses are the weakest. And with the
Bald Man Paradox this will once more be the iteration thesis
which lays down that general rule: "loss of a single hair does not a
bald man make. "14
The Millet Seed Paradox of Zeno of Elea is an ancestor to
these paradoxes. It too roots in the effect of unnoticeably small
("subliminal") differences, arising from the question of how it is
that while dropping a single millet seed makes no sound, dropping a bushel of them makes a loud thud. IS Leibniz used this
example as a case in point for his theory of "minute," unnoticeably small, perceptions beneath the threshold of consciousness. In
any case, in this instance also that iteration rule akin to mathematical induction becomes the weakest link in the plausibilistic chain.
5.2
The Color-Continuum Paradox
Consider the following situation which characterizes what might

be called the Color-Continuum Paradox. We layout a long row of
color patches: say 100 of them. Any two adjacent ones are color14. Again, breaking up the general rule into a large series of particulars will make this
paradox more baffling. For now while we can identifY the region in which the remedy of
premiss rejection must fall, we cannot pinpoint the exact location at which this must be
effected.
15. The paradox also took a variant fonn in antiquity: a single drop of water effects no
change on the fonn of a stone, while a long-continued succession of drops wears a hole in
it. (See Aristode, Physics, 253bI4.)
84
ChapterS
wise indistinguishable to the unaided eye. But gradually and

glacially we shift over to quite a different color by the time we get
to the end of the series. We thus arrive at the aporetic cluster represented by the following four theses:
( 1) Patches that are color wise visually indistinguishable (to a
normal observer in normal circumstances) have the same
color.
(2) Patches [1] and [2] are colorwise visually indistinguishable, as are [2] and [3]. And so on up to patches [99] and
[100].
(3) Hence-all these patches have the same color (by (1)).
(4) Nevertheless, patches [1] and [100] are visually distinguishable as having quite different colors.
Here {(l), (2), (4)} constitute an inconsistent triad. And since (2)
and (4) are straightforward facts, it is the more suppositional (1)
that must be abandoned. Color identity is something more complex than what can be settled by visual means alone.
However, we would again presumably not wish to abandon
(1) outright-and there is no need to do so. But we would have
to demote it from the realm of the flat-out true to that of the
merely plausible. This would provide for its continued availability
in other deliberations despite its contextual untenability in the
present case.
5.3
Sir John Cutler's Stockings
The Paradox of Sir John Cutler}s Stockings is based on a story that

goes as follows:
Sir John Cutler had a favorite pair of stockings to which he
was greatly attached. But over the years they wore out. Still he
would not be parted from them. Bit by bit and piece by piece
the stockings were repaired and rewoven over the years, with
one successive repair superimposed on its predecessor until at
last nothing of the original material remained. The puzzle
85
arises whether what remains at the end is still the same pair of
stockings.
This story leads to the following paradox:
( 1) When a smallish repair is effected on a stocking what
results is still the same stocking. (Yet carried on sufficiently this can result in a total replacement of that stocking's fabric.)
(2) The final stocking results from the initial one through a
succession of smallish repairs.
( 3) Ergo (by (1) and (2)) the final stocking is the same as the
original.
(4) When an ongoingly "repaired" material object like a
stocking (or a car or a piece of furniture) no longer has a
single bit of material in common with what was there initially it cannot count as the same object.
( 5) Ergo (by (2) and (4)) the final pair of stockings is no
longer identical with the original pair-contrary to (3).
The group {(I), (2), (4)) obviously constitutes an inconsistent
triad, (3) and (5) being derivative claims. Now (2) is an observable fact. And (1) and (4) are general principles which, though
both eminently plausible, are nevertheless such that (1) enjoys
some advantage in point of its plausibility. (This advantage roots
in the fact that (4) is the more ambitious in its range and generality.) We thus arrive at the priority ranking:
(2(1(4)
In seeking to break the chain of inconsistency at its weakest link

we note that it is (4) that affords this point of greatest vulnerability. We thus arrive at the result that the R/A-alternative (2),
( 1)/( 4) that is optimal, seeing that its retention profile {I, 1, O}
wins out over the available alternatives. To feel comfortable about
this resolution we would certainly like to have a well-articulated
rationale for (4 )-dismissal-a fuller account of just why it should
ChapterS
86
be that it deserves to be relegated to the bottom of the plausibility ranking. And this demand can presumably be met by noting
that familiarity with automobile repair and with the gradual
replacement of the cellular material of a human body inure us to
the prospect of identity despite replacement.
This paradox also takes the version of the Ship of Theseus
Paradox, which is repaired plank by plank and spar by spar until
not a bit of the original ship remains. Plutarch tells us that on
Theseus's return journey to Athens from slaying the Minotaur, his
men preserved the ship by replacing its rotten timbers with new
ones. 16
The following paradox is at issue here:
(1) Whenever only a single part is interchanged, the resulting
ship remains the same, as was.
(2) When the totality of parts is interchanged the shops are
interchanged.
(3) By (1) the initial and final ships in the possession of each
of the parties are unchanged at the end of the operation.
(4) By (2) the initial and final ships in the possession of each
of the parties have been interchanged.
Here (1) and (2) are inconsistent. And seeing that (2) seems
more plausible than (1), the best plan would be to abandon (1) in
its full generality, viewing it as a principle of limited flexibility that
"can be stretched only so far." To be sure, it is somewhere
between difficult and impossible to be precise about the limits of
this flexibility since where a limit is not precise one cannot indicate it precisely. And it is this circumstance that marks the paradox
as one of vagueness.
16. Plutarch's Lives, "Life of Theseus," 22-23. Clearly when only one or two planks
are replaced it is still the same ship. But as the process continues, how long does this situation persist. Is it still the same ship when all of its fabric has been replaced?
87
5.4 The Hermeneutic Circle

Matters of degree often engender paradoxes of vagueness. Thus
consider the Hermeneutic Circle Paradox encapsulated in the following theses: l7
(1) We can get a firm grip on the meaning of verbal texts.
(2) One cannot firmly determine the meaning of a word without a firm grip on the meaning of its sentential (or even
larger) context.
(3) One cannot firmly determine the meaning of a seman tical
(or even larger) context without a firm grip on the meaning of its component words.
(4) In view of(2) and (3), it is not possible to get a firm grip
on the meaning of a verbal text( contrary to (1).
Here (1)-(3) present an inconsistent triad. And since (2)-(3)
stand on essentially the same plausibility footing we effectively
have to choose between abandoning (1) and abandoning the pair
(2 )-( 3).
The pathway to resolution here lies in the consideration that
the idea of a "firm grip on meaning" is not an on-off dichotomy
but is actually a matter of degree, of more or less. And on this
basis we can enter into a cyclic feedback alternation between word
and context interpretation that enables us to achieve a stepwise
escalation that provides an increasingly firm grip on the meaning
of word and context alike.
An instructive lesson results from the sorts of examples that
have been surveyed here. For they show that the conception of
17. The idea (and terminology) of a hermeneutic circle is due to Wilhelm Ditthey
(1833-1911), who saw this conception ofa reciprocal interpretative dependency of part
and whole as recurring at many levels, not just sentence/paragraph, but paragraph/book,
book/genre, genre/cultural tradition. Dilthey regarded this phenomenon as characteristic
of interpretation in the human sciences in general, not just at the textual level, but in plastic arts, cultural mores, ways of thinking, and other reactions of the human sciences. On
the hermeneutic circle see D.C. Hoy, The CriticlJl Circle: LiterlJture IJnd Hinory in
ContemporlJry Hermeneutics (Berkeley: University of California Press, 1978), and Paul
Ricoeur, Hermeneutics IJnd the HumlJn Sciences (Cambridge: Cambridge University Press,
1981).
88
Chapter 5
comparative plausibility is a serviceable instrumentality which

enables us to have our cake and eat it too, as it were. In contexts
where plausible propositions are unproblematic we can make use
of them in pursuing our question-resolving inquiries. And where
they prove to be contextually untenable we can abandon them in
this context without necessarily incurring a total loss that must
carryover to other contexts as well. The crucial point here lies in
the previously stressed difference between acceptance as plausible
and acceptance as true.

Aristotle's Paradox of Change
The Teacher as Assassin Paradox
The Plausibility Paradox
Zeno's Achilles Paradox
Zeno's Arrow Paradox
Zeno's Moving Bodies Paradox
Relation Riddles ofEubulides
The Masked Man Paradox (Eubulides)
Electra and Orestes (Eubulides)
The Overlooked Man Paradox (Eubulides)
The Reiteration Paradox
The Garbage Truck Paradox
The Master Argument Paradox
The Mind-Body Paradox
The Categorization Paradox
The Surprise Examination (or Execution)
Paradox
Paradoxes of Ambiguity
and Equivocation
6.1
Equivocation as a Road to Paradox
Nicholas of eusa (1401-64), who propounded a doctrine of the

coincidence of opposites (coincidentia oppositorum), maintained
the paradox that all being is, as it were, nothing. But distinctions
must be drawn in this regard since closer scrutiny indicates that
what he had in view is that the maximum of being (God) is the
minimum of comprehensibility because "nothing has no properties" (nihil sunt nullae proprietates) and we cannot ascribe properties to God since none of the properties of which we mere
mortals can form an adequate conception can be attributed to
God. l His words took on rather unaccustomed senses, however,
and his minimum-maximum paradox is resolved by the fact that
different things are being measured. And this illustrates a more
general situation.
One of the most common routes to paradox is through the
mistakenly uniform use of a term or concept that actually has a
plurality of senses and applications. Already Aristotle regarded
such equivocation and ambiguity as a pathway to fallacy and paradox,2 and his own example, "I saw him being beaten with my
1. Nicholas of Cusa, Of Learned Ignorance (translated by Fr. Germaine Heron (New

Haven: Yale University Press, 1954); see pp. 17 and 39. True learning, Nicholas held, lies
in grasping the inevitability of our ignorance: "The relationship of the intellect to the truth
is like that of a [regular1 polygon to a circle; the resemblance of the polygon to the circle
grows with the multiplication on its angles, ... but no multiplication of its angles--1:ven if
it were infinite-will make the polygon equal to the circle."
2. Aristode, Soph. Elen., 165b25ff and 177a40ff.
91
92
Chapter 6
own eyes," affords a vivid illustration of the phenomenon. 3 This

sort of statement is all too often a mere sleight of hand (or,
rather, pen!) of this general sort.
In considering property-attribution claims of the usual format,
Fx, the ancients noted that ordinary properties admit of qualification. Smith is young for a senator, but old for a carpenter's
apprentice. Jones is heavy for a jockey but light for a Sumo
wrestler. And to avert the logical conflict of the paradoxical claim
that x is both F and not-F, Aristotle stressed that we have to construe that property attribution as something that is relativized
both as to time and as to respect. This led him to a version of the
Principle of Contradiction that nothing can be both F and not- F
at the same time and in the same respect. Distinctions of time and
of respect here come into playas crucial paradox-averters.
Aristotle analogously contemplated what might be called the
Paradox of Change which runs somewhat as follows: 4 "The young
Socrates is clearly not old. But Socrates is always one and the
same person. Hence the old Socrates is not old." The paradox
here runs as follows:
(1) Socrates is always the same person.
(2) By (1), Young Socrates is Old Socrates.
(3) Old Socrates is old.
(4) Young Socrates is not old.
(5) By (2), (4) Old Socrates is not old.
Aristotle's analysis of the inconsistent triad represented by {( 1),
(3), (4)J is that paradox results from ignoring the factor of time.
The verb "is" is equivocal. It can represent a timeless or time-
3. Aristotle, Soph. Bien., 177blO-12. Another example he offered was: "This dog is a
father. This dog is yours. Therefore, this dog is your father" (ibid. 179b14-1S; but the
example occurs already in Plato's Buthydemus 298D-E). Or again: "One can only give
what one has. But one can give somebody a kick. And yet a kick is not the sort of thing
one can have" (ibid, 171aS.)
4. Aristotle, Metaphysics, Book IV, Chapters 3-4, and compare Categories, Chapter 5.
Paradoxes ofAmbiguity and Equivocation
93
indifferent relationship like that of the "is" of identity in (2), or a

temporalized is-at-the-time relationship as in (3) or (4), now with
"Old Socrates is old" to be construed as "Old Socrates is old at
the time when he is Old Socrates." In the former (timeless) sense
of "is" (1) is true but (3)-(4) false, while in the temporalized
sense of "is" (1) is false (3)-(4) are conjointly true. And so the
premisses of the perplex are never all-true-at-once. A failure to
implement distinctions in point of time is, as Aristotle maintained,
a fertile source of paradox as a failure to heed differences in
respect. His Principle of Contradiction comes to the fore here. It
is indeed true that "Something cannot both be and not be thusand-so"; but one has to add "at the same time and in the same
respect."
Consider in this light the Sophistical Teacher as Assassin
Paradox of Plato's Euthydemus (283D) which runs: "Education is
little better than murder, for to bring into being Alpha as he
would like him to be (educated) is to do away with Alpha as he is
(ignorant)." Here we must, of course, distinguish between the
person as such-the individual Alpha as he is from birth to
death-and his various temporal and descriptive stages in the
course of change as per young/old or ignorant/learned. To
change Alpha is certainly not to do away with him.
The scholastics of the middle ages, John Buridan preeminent
among them, reveled in sophisms (sophismata) such as "Water has
five letters; one can drink water; therefore, one can drink letters,"
a paradox that cries out for dissolution via a distinction that
removes the equivocation as between the word "water" and the
substance it stands for. Or again, the ancient paradox "Man
belongs to the animal kingdom, so the animals possess man,"
obviously two senses of "belongs to," namely "is part of" and "is
owned by."
Consider what might be called The Plausibility Paradox engendered by the following inconsistent triad:
(1) Plausible propositions can be accepted: they can appropriately be endorsed.
(2) Plausible propositions can engender apories: they can
form inconsistent clusters of collectively incompatible
propositions.
94
Chapter 6
(3) It is never appropriate to accept (all of) a group of inconsistent propositions.
The exit from paradox in the face of this inconsistent triad lies in
noting the equivocation of the term "accept" as between (i) the
stronger sense of "accept as true" and (ii) the weaker sense of
"accept as credible/plausible" or "accept as a promising truthcandidate" or the like. In sense (i), thesis (3) is correct but (1) is
false, while in sense (ii) thesis (1) is correct but (3) is false. (And
different things are at issue here: provisional acceptance as true is
no more actual acceptance as true than a bronze owl is an actual
owl.) Thus in neither case will all of those incompatible propositions obtain together, and the paradox is thus resolved. Any piece
of well-articulated reasoning is subject to the (usually tacit) presupposition that identical terms are being used in a uniform sense:
that where different senses or uses of a term are at issue this will
be made explicit by some appropriate terminological complication. Equivocal premisses become disconnected from one another.
In this sort of way, the identification of equivocal, ambiguous, or
vague terms can often unravel an apparent paradox.
Equivocation paradoxes can accordingly be seen as involving a
failure to satisfy a requisite communicative presupposition.
Working our way out of a distinction-neglecting paradox
requires close attention to the use of words in order to overcome
the verbal miscalculations that can arise. It is, after all, clear that
one principal avenue to paradox proceeds by running together
things that should be separated.
6.2
Zeno of Elea, born around 490 B.C., devised some of the most
celebrated paradoxes of all time-widely discussed and debated
already in classical antiquity. 5 Eight of these have come down to
5. On Zeno see Zeller, Philosophie der Griechen, Vol. I/I, pp. 746-765, as well as
Gregory Vlastos, "Zeno ofElea" in P. Edwards (cd.), The Encyclopedia of Philosophy, Vol. 8
(New York: Macmillan, 1967), pp. 368-379, and also G.E.L. Owen, "Zeno and the
Mathematicians," Proceedings of the AristoteliIJn Society, Vol. 58 (1957-58), pp. 199-222.
An informative collection of discussions of Zeno's paradoxes is Wesley C. Salmon (cd.),
Zeno's PIJrIJdoxes (Indianapolis: Bobbs Merrill, 1970), which also offers a bibliography of
some 140 entries.
95
us. One is the Millet Seed Paradox already considered on p. 83,

and another the Racetrack Paradox discussed on pp. 59-60.
Additionally, there are three directed against space and three
against motion-all to be considered later in the present chapter-as well as the Paradox of Differentiation to which we now
turn.
Zeno sought to support the doctrine of his teacher Parmenides
that plurality must be denied and that at bottom all is one.
According to Plato, "the first hypothesis of the first argument of
Zeno's book" runs as follows:
If existences are many, they must be both alike and unlikealike in differing from all others, and unlike since they are not
one and the same. But this is impossible, since unlike things
cannot be like, nor like things unlike. Therefore, existences are
not many.6
The paradox at issue here runs as follows:
( 1) As we look about us, we see by the evidence of our own
senses that existence is plural-that it is heterogeneous in
that there are many different kinds of existents, and thus
many existences.
(2) Let it then be supposed that existences-the kinds of existents that there are-are many.
( 3) Then all existents have the common factor of differing
from all the other kinds of things, of all the other existences. They are thus all alike in fundamental regard of
differing from others and "being themselves," so to speak.
(4) But if existents thus have a fundamental feature in common they are all of the same kind, homogenous. So there
is only one king of existent, only one existence.
(5) Since this consequence (4) contradicts (2) we have established not-(2) by a reductio ad absurdum. Thus existences
are not many.
(6) But (5) contradicts (1).
6. See Plato, plJrmenides, 127D.
96
Chapter 6
Zeno saw this attribution to the same things of likeness and

unlikeness as a violation of the idea later encapsulated in
Aristotle's principle of contradiction. For him it threw doubt
upon the extent to which reality honors logic. But as Aristotle
himself (rightly) saw it, the problem here is one of equivocation:
the matter is not one of being the same or different pure and simple but of being the same or different in one or another particular
respect. On this basis, Zeno's Paradox of Differentiation paved
the way towards the Principle of Contradiction.
6.3
Zeno also proposed a series of paradoxes designed to invalidate

the entire idea of plurality of punctiform places in an extended
space. He offered three arguments against an extended space
composed of a plurality of places (or points) which run as follows:
I.
"If there are many (spatial units) they must be both so

small as to have no magnitude at all, and so large as to be
infinite." If, as is said, spatial units have no magnitude,
no extension, they will be nothing at all. So they must
have magnitude if they are to exist and compose space.
But now since they have extension, then they will have
parts and their parts (e.g., their top half) would also have
extension. And this in turn would also require extension,
and so an ad infinitum. So spatial units would be infinite:
they would not be parts of the spatial whole, but the
whole of it.
II. Spatial units are said to be something. But if one is added

to another then this other is (ex hypothesi) enlarged
(that is, increased by what was added. But this, it is said,
is not the case with spatial units: their addition does not
enlarge, nor their subtraction diminish. Hence they are
nothing.
III. Spatial units are said to be the constituents of larger
(finite) wholes. But these larger wholes must then have
97
some fixed number of units (neither more nor less. But

between any two distinct units there is yet another. So
there cannot be any fixed number of them: they must be
infinite in number?
These perplexes are all paradoxes of equivocation:
No. I involves two problems. First is the equivocation of
"spatial unit" as between a single point and a small (atomic)
region which leads to in the questionable supposition that no
extension implies "no existence" rather than simply "no measurable size." Second is the mistaken idea that having "infinitely many extended constituents" implies "infinite in
extent." As with Zeno's Achilles paradox this overlooks the
fact that an infinity of (ever-diminishing) extensions can nevertheless run up to a finite total as per t + + ~ + ... = 1.
No. II pivots on the idea that the addition of spatial units

implies an increase in measurable size. If these "spatial units"
are genuinely punctifrom then this need not be so. (The interval from 0 to 1 is just as long in the way of measurement
whether or not one includes the endpoint.)
No. III comes to grief on the fact that finite wholes can have
infinitely many constituents (e.g., through successively taking
away half of what is left). Moreover, with infinite numbersunlike finite ones-the addition of more units need not lead to
a different (larger) while. When Nis infinite we can (supposedly) achieve N + 1 = N. (Thus the series 1,2,3,4,5, ... and
2,3,4,5,6, ... have equally many members-as can be seen
by pairing the first of each with that of the other, and then the
second, and then the third, etc.)
To be sure, any adequate resolution of Zeno's spatial paradoxes
calls for raising the level of mathematical sophistication beyond
7. This statement of the paradoxes is paraphrased from the texts given in John Burnet,
Early Greek Philosophy, 4th ed. (London: Macmillan, 1930), pp. 315-16.
98
Chapter 6
the level of ideas current in his time. For these paradoxes all pivot
on the erroneous supposition that what is infinitely divisible must
be infinitely great. This would indeed be a telling argument
against the atomist if they were to hold that spatial units admit of
infinite divisibility-which it was the very reason for being of their
theory to deny. But it certainly does not hold against the geometers-once they developed their theory of space to a level of sufficient sophistication.
6.4 Zeno's Achilles and the Tortoise Paradox

Zeno of Elea also devised a series of ingenious arguments
designed to show the impossibility of motion. One of these was
the Paradox ofAchilles and the Tortoise, based on the following
narrative:
Notoriously fleet-footed Achilles has a race with a proverbially
slow tortoise. Naturally enough, the tortoise demands a head
start. But now by the time that Achilles reaches the tortoise's
starting point the tortoise will have moved on and will be
somewhat ahead. And when Achilles reaches that position the
tortoise will have moved on and will still be ahead a bit. And
so on. Thus Achilles will never catch up with the tortoise. s
The paradox at issue here goes as follows:
(1) At no stage of the endless sequence of positional catchups, will Achilles have succeeded in reaching the tortoise.
Therefore(2) Achilles will never pass the tortoise.
(3) But-as we know full well-Achilles will soon pass the
tortoise.
8. Kirk, Raven, Schofield, The Presocratic Philosophers, 2nd ed. (Cambridge:

Cambridge University Press, 1983), p. 272.
99
Here theses (1) and (3) are perfectly in order. But the chain of
inconsistency is broken at (2) which actually does not foUow from
(1) at all because the move from "at no stage of the sequence" to
"never" is simply inappropriate.
The fact of it is that a deep underlying equivocation is at work
in the paradox as between "the sequence of catch-ups is unending
(limitless) in step?' and "the sequence of catch-ups is unending
(limitless) in time." The two expressions have a different sense.
For the catch-up sequence just does not cover the whole of the
future, seeing that it converges to a final limit even as t + + t +
... converges to 1. Thus the steps of that sequence, even in endless totality, wiU only cover a finite timespan.
Zeno's Achilles paradox is thus decisively resolvable through
the recognition that an equivocation unravels the aporetic inconsistency through which it arises. (Whoever thinks that this unfavorable series does not sum up to one but always falls short of it
just does not understand the function that those three little dots
are serving here: they stand for et cetera, meaning and all the
rest. )
Along these same lines, there is a cognate paradox already
discussed by Aristotle (Physics, 239b 9-14). It runs essentially as
follows:
Suppose you wish to move from point A to point B. Before

you can accomplish this motion, you must first reach point C,
the halfway point between A and B. But before you can reach
point C, you must first reach point D, the halfway point
between A and C. And so on. Thus before you can accomplish
any motion you must first accomplish an infinity of prior
motions, which means that you can never accomplish any
motion at all.
The situation here is, of course, identical with that of Zeno's
Racetrack Paradox. And the road to resolution is thus the same
in both cases. One is also reminded of Achilles and the Tortoise.
There we had it that a limitless series of steps do not require a
limitless amount of time; here we have it that a limitless series of
steps do not require a limitless amount of space.
100
6.5
Chapter 6
Zeno's Paradox of the Arrow
Another of Zeno's paradoxes of motion is that of The Arrow. 9

What is at issue here is an aporetic collision among the following
theses:
(1) At any given instant the arrow does not move. (A clear
fact: instants are too short to allow movement.)
(2) A span of time consists of (nothing but) instants.
(A salient aspect of our understanding of time as a onedimensionally linear manifold.)
(3) The arrow is immobile: it never moves in any given span
of time. (From (1) and (2).)
(4) Arrows can and do move. (A fact of observation.)
Since we know (4) as a fact, it is clear that (3) must give way to it.
Accordingly, there is something seriously amiss within the pair
constituted by (1) and (2) from which (3) supposedly derives.
And yet both of these seem undeniably true, so that the paradox
is a really troublesome one.
The problem here actually roots in the first premiss because a
vitiating equivocation is at work. The fact that an arrow does not
move DURING an instant (an instant, being of its very nature,
too brief to allow the accomplishment of a movement) does not
mean that an arrow is not in motion AT this instant. And once this
distinction is drawn we come to realize that premiss (1) is equivocal as between:
(1.1) At any given instant an arrow does not accomplish a
movement.
(1.2) At any given instant an arrow is not in motion.
Here premiss (1.1) is true but does not join with (2) to yield the
paradoxical (3). (The fact that arrow-motion does not accomplish
a movement at an instant does not mean that it is not accomplished over a great array of them-any more than the fact that a
9. Kirk, Raven, Schofield, op. cit., pp. 272-73.
101
single letter conveys no information means that a text inducing a

great many of them cannot do so.) On the other hand, premiss
(1.2), which does actually yield (3), is simply false. So neither way
will an actual paradox result. The distinction between being in
motion and accomplishing a movement is crucial for resolving this
paradox because the former is actually compatible with ins tanteneity whereas the latter is not.
Zeno's Arrow paradox is geared to time. Its challenge is "But
just when is the arrow moving? There is really no instant of time at
which it moves." The Megarian Diodorus Cronus reformulated
this paradox with respect to space, configuring it as a dilemma:
That which moves must either move where it is or where it is
not. But it cannot move where it is, because that is a fixed
location. And it cannot do anything-let alone move-where
it is not. Hence it cannot move at al1. 10
The problem with this dilemmatic Motion Paradox of Diodorus
is again one of equivocation. A distinction once more saves the
day. The fact that an arrow does not move at its fixed, now-determinate position (but is only in motion there) does not mean that
it does not move from this fixed, now-determinate position to
another (exactly by virtue of being in motion).
6.6
Zeno's Paradox of Moving Bodies
Zeno's Paradox of Moving Bodies runs as follows: l l

Suppose we have three bodies, each of the same unit size.
Initially they are in alignment. Then two are set into motion.
While one of them (A) remains stationary, the second (B)
passes it from left to right, and the third (C) passes it from
right to left. The two moving bodies move at the same speed,
say one unit per minute. Then after one minute B will have
traveled one unit of distance since it will have passed by all of
10. See ZeUer, Philosophie der Griechen, Vol. II/I, p. 266.

11. Kirk, Raven, Schofield, op. cit., pp. 274-76.
102
Chapter 6
A. On the other hand, B will have traveled two units of distance since it will not only have passed all of C, but have trav-
eled one unit beyond it.

Initial
~~
~ti=j
After 1 Minute
A
B
Thus we have the contradiction both that B travels at 1 unit

per minute and that B travels at 2 units per minute.
The resolution here lies in noting the equivocation of the idea of
speed between an unrelativized and a relativized mode. Thus B's
velocity relative to A is 1 unit per minute, and B's velocity relative
to Cis 2 units per minute. There is no contradiction in this when
that essential relativization is heeded. We must simply avert the
equivocation as between speed relative to a fixed frame of reference
and speed relative to the position of other particular objects.
6.7 The Relation Riddles of Eubulides

Let us now turn from Zeno to the later Eubulides who flourished
around 400 B.C. and taught at Megara, near Athens, conducting a
prolonged controversy with Aristotle about the value of sophisms
and paradoxes. Three of his paradoxes turn on the same issue, and
can all be represented in the form of riddles: 12
The Masked Man (egkekalummenos): "Do you know this
masked man?" "No." "But he is your father. So--do you not
know your own father?"
12. For Eubulides see Section 1 of Chapter 5. Diogenes Laertius Lives of the
Philosophers, Bk. II, Sect. 10, 108-110 and Bk. VII, Sect. 7, 187. See also Prantl
GeschiclTte, Vol. I, pp. 50-58, and Aristotle, Soph. Bien., 179a32ff.
103
Electra and Orestes. Electra saw that a man was approaching

her. The man was her brother, Orestes. So--Did Electra see
that Orestes was approaching her?
The Overlooked Man (dialanthanon): Alpha ignored the man
approaching him and treated him as a stranger. The man was
his father. So--did Alpha ignore his own father and treated
him as a stranger?
The same equivocation is at work throughout these Relation

Riddles of Eubulides. In each case, there is, firstly, the individual as
the protagonist of the narrative takes him to be, and, secondly, the
individual as the narrator takes him to be (or as he actually is). In
this regard, the paradoxes at issue are typified by that of the
Masked Man, which in its elaborated forum runs somewhat as follows:
(1) You know who your father is.
(2) You do not know who the masked man is.
( 3) The masked man is your father.
(4) By (2), (3), you do not know who you father is.
The problem here is that the "knowing who" of premiss (2) is
equivocal and admits of a variety of modes. In mode No.1 (subject's perspective) you do "know" who the veiled man is: he is
that person over there. That is, under that description "the man
in the mask standing over there" you do indeed know the masked
man. And so premiss (2) is false. However, in mode No.2 (narrator's perspective) you do not know who the masked man is,
namely, you do not know that he is your father-and, consequently, you do not know your father's identity with the masked
man. In mode No.2, premiss (2) is thus true, but premiss (1) is
now false. In this manner, all three of these riddles present typical
paradoxes of equivocation. The point is that what a person knows
can be stated both from the subject's and from the reporter's perspective. When I say that Caesar knew intimately the men who
were to slay him, it is-clearly-the reporter's, and not the sub-
Chapter 6
104
ject's perspective that is at issue. (Caesar would never have

thought of Brutus as "my good friend and assassin.")
Eubulides was an opponent and critic of Aristotle. For
Aristotle dismissed such sophisms of equivocation as mere errors,
and careless mistakes of a sort. Eubulides, on the other hand,
seems to have thought that a deep point is at issue, namely that
we do not know things pure and simple but only in a relativized
way, only under a certain description or from the vantage point of
a particular cognitive perspective. And in this regard it would
seem that Eubulides was entirely on the right track. Their insistence that things can only be grasped from the vantage point of a
conceptual standpoint led to the Megarians being characterized as
"friends of ideas (ton eidon philous)."13
6.8
Further Riddles
Consider the riddle: "Tom and John both had (different) cars.
John said that he had the fastest car in town and Tom said the
same thing. Did Tom agree with John?" A paradox arises here
which might be called The Reiteration Paradox:
(1) John said that he had the fastest car in town.
(2) Tom said so too: that is, Tom said that John has the
fastest car in town.
(3) Tom said so too: that is, Tom said that he (himself) had
the fastest car in town.
(4) By (1) and (2) Tom and John are in agreement.
(5) By (1) and (3) Tom and John are in disagreement.
As (2) and (3) make transparently clear, this paradox roots in
equivocation. For the expression "the same thing" as used in stating the riddle can mean either the same contention or statement
(viz., "John has the fastest car in town") or the same sentence or
13. Prantl, Geschichte, Vol. I, pp. 37-38.
Paradoxes ofAmbiguity and Equi7Jocation
105
form of words (viz. "I myself have the fastest car in town"). Here
the paradox is resolved by noting that one cannot have it both
ways, and that with either alternative one member of the pair
(5 )-( 4) becomes untenable so that the paradox is dissolved.
And this sort of mismatch is generally the case with such riddles. Thus consider the classic riddle:
Q: What has four sets of wheels and flies?
A: A garbage truck.
This riddle is supposed to baffle us because things that fly just do
not have four sets of wheels. And so the following Garbage Truck
Paradox looms up before us:
(1) The object at issue has four sets of wheels and flies.
(2) Natural objects that fly (such as birds) do not have wheels.
( 3) Artifacts that fly (such as airplanes) do not have four sets
of wheels-they have two.
(4) The object at issue is either natural or artificial.
(5) Therefore: the object at issue just cannot both have "sets
of wheels" and also fly, contrary to (1).
Escape from this riddle paradox lies in remarking the ambiguity of
"flies." The riddle creates the expectation that "flies" is a verb
here. But of course this could also be a noun, with "has flies" =
possess or is accompanied by flies. Thus (1) and (5) are disconnected by the fact that (1) sees "fly" as an insect-indicative word
while (5) sees it as a transport-indicative verb. Again, that crucial
(if tacit) supposition of a uniformity of meaning is violated. And
once the appropriate distinctions are introduced, the premisses of
this Garbage Truck Paradox are disconnected with the result that
the paradox dissolves. As Aristotle stressed, the drawing of distinctions affords an effective device for the management of equivocation paradoxes. 14
14. See Soph. Elen., 17Sb27-38.
106
Chapter 6
The sophists of pre-Socratic Greek antiquity relished something/nothing puzzles. IS They deliberated over such questions as
"If nothing ever comes from nothing, how can anything come to
be?" And they were drawn to such riddles as:
Q: What is that which isn't?

A: Nothing!
or in a somewhat more elaborate formulation:
Q: How can anything be something that is not something?

A: By being nothing. I6
Plato's dialogue The Sophist is full of reflections on this sort of
Something/Nothing Paradox, whose formal structure is roughly
the following:
(1) Everything is self-identical: anything whatsoever is whatever it is.
(2) Therefore: Nothing is nothing. (From (1).)
(3) To say that X is Yis to say that X is something-or-other.
(4) Nothing is something or other. From (2), (3).
(5) Whatever is something-or-other is not nothing.
(6) Nothing is not nothing. From (4), (5).
Here {( 1), (3), (5)} constitutes an inconsistent triad. And (3) is
the Achilles Heel of this aporetic situation. For it confuses the is
of predication (as per X is Y, "Fiction is unreal") with the is of
being (with is as exists, that is, as being something-or-other in the
realm of reality) .
15. Gorgias even entitled his treatise "On Nature or That Which Is Not."
16. Lewis Carrol's Alice in Wonderland offcrs thc following variant: "I see nobody on
the road," said Alice. "I only wish I had such eyes," The king remarked in a fretful tone.
"To be able to see Nobody! And at that distance too! Why, it's as much as I can do to see
real people, by this light!"
107
The fact of it is that nothingness as an abstract condition exists

in the same way that any other abstraction does-self-identity or
complexity or the like. But nothing is just that, utter vacuity, the
total absence of things, and not the presence of some peculiar sort
of thing, a "nothing." To say "Nothing has the feature F" is to
say that "no existing thing has this feature," symbolically -3xFx
or equivalently "dx-Fx. It is not to say that there is some strange
sort of existing or quasi-existing thing, a non-thing or Unding,
called Nothing, that possesses this feature (symbolically 3x(x = n
& Fn).)
Plato's puzzle question how can we can regard "nothing" as
total vacuity when it is clearly something, to wit nothingness, the
object of theoretical concern, consideration and discussion-can
be resolved on this basis. What we are talking about is indeed
something-viz. nothingness, the condition of utter vacuity. But
this does not mean that by saying (for example) that nothing lacks
all properties (nihil sunt nullae proprietates) we are denying properties to a certain sort of thing-a particularly negative object
called "nothing." All of those something/nothing paradoxes are
based on an equivocation in "nothing" that leads to a confusion
between the absence of any objects designated by the term to the
inevitably present sense or meaning of the term itself.
As such examples indicate, riddles will often engender a paradoxical situation that we can avert by unmasking an equivocation.
Such riddles pivot on conceptual (rather than optical) illusions.
All of the preceding examples are instances of what might be
called the dilemmatic equivocation collapse of a paradox. This
occurs when the paradox in question has premisses so interconnected by a common equivocal term that in every case the sense
which makes a given premiss true renders some of the others false,
so that in trying to make the premisses true one is caught in a
defeating dilemma.
6.9
The Master Argument of Diodorus (Kyrieuon)
Perhaps the most important ancient continuator of Zeno's assault

on motion was Diodorus Cronus (born ca. 350 B.C. )17 He devel17. See Zeller, Philosoplne der Griechen, Vol. Il/l, pp. 247-48 and 266-271.
108
Chapter 6
oped four arguments against motion that were, in effect, variations on the theme of Zeno. However, the most famous paradox
proposed by Diodorus Cronus is the following Master Argument
Paradox: 18
(1) There is contingency in nature, that is, there are actual
states of affairs that are not necessary but merely contingent
(in that their non-realization is, theoretically, possible).
(2) The possible cannot engender the impossible: no possible
state of affairs can produce a consequent one that is
impossible.
(3) Once a state of affairs is realized it is impossible to change
it: whatever is actual thereby becomes necessary. And so .
after the fact the nonoccurrence of anything actual is
impossible.
(4) By (2) and (3) no possible state of affairs can result in the
nonoccurrence of anything actual. For the nonoccurrence
of anything actual is (ultimately) impossible by (3) and the
possible cannot engender the impossible (by 2). In consequence, anything actual is necessary.
What we have here is a paradox of equivocation. We must distinguish between what is necessary or impossible after the fact and
what is so even in prospect.
Thus (2) is problematic and can obtain only if suitably construed. For what is possible before the fact can indeed engender
something that is impossible after the fact on grounds of being
incompatible with it.
6.10 Sameness amidst Change: Mind-Body

Equivocation
Paradoxes of equivocation often arise in philosophical contexts.
Consider, for example, a hypothetical situation of the sort illustrated by the following Mind-Body Paradox:
18. For the Master Argument and its literature see the author's "A Version of the
'Master Argument' ofDiodorus," The Journal o/Philosophy, vol. 63 (1966), pp. 438-445.
109
X and Ywere hooked up to a thought-transfer machine. And

as a result all of their thoughts-knowledge and memories, likings and tastes-were interchanged. As a result the bodies of
these people continued unchanged. It was just their minds that
were interchanged-the entire content thereof. Which one is
X and which one is Y?
The issue turns on whether it is bodily continuity that is pivotal
for personal identity or continuity of thought and personality.
And the following paradox arises:
(1) After the transfer process, the possessor of X's body is also
the possessor of rs mind, and conversely.
(2) Bodily continuity is the definitive factor that is crucial for
personal identity.
(3) Ergo-the post-operation possessor of X's body is X, and
the same for Y.
(4) Mental continuity is the definitive factor that is crucial for
personal identity.
(5) Ergo-the post-operation possessor of X's mind is X, and
the same for Y.
(6) Two different individuals are at issue. No one can be both
Xand Y.
Here {(I), (2), (4), (6)} constitute an inconsistent quartet. But
( 1) is fixed by the defining hypotheses of the problem, and (6) is
an incontrovertible fact. So we have the priority ranking:
(1) > (6) > [(2), (4)]
On this basis, we will have to abandon at least one of(2) and (4).
However, since in the existing circumstances these two contentions are to all visible intents and purposes "in exactly the same
boat," we have little option but to abandon both of them, plausible though they seem.
And the justification for doing so could plausibly run as follows: Both (2) and (4) rest on the "definitive factor" supposition
that "personal identity" is a clearly defined, precisely articulated
llO
Chapter 6
concept whose application rests on a single determinative factor

(bodily continuity in the one case and mental continuity in the
other). But if we regard this supposition as false and view the concept of "the same person" at issue in personal identity as a complex concept that involves an admixture of bodily and mental
considerations-a mix whose nature is not altogether clear and
definitive-then the props are pulled out from under theses (2)
and (4). The paradox can thus serve as a roadway towards an
instructive lesson: personal continuity is something complex and
many-sided-no one single factor should be seen as decisive here.
Perhaps the lesson is even that personal sameness is an equivocal
idea and that-at least in theory--one should distinguish between
physical and mental sameness.
6.11
Along these lines consider the Morning Star Paradox due (in
essence) to the German mathematician GottIob Frege:
(1) The Morning Star is not the same as the Evening Star.
(2) The Morning Star is the same as the planet Venus.
(3) The Evening Star is the same as the planet Venus.
(4) The Morning Star is the same as the Evening Star. (From
(2) and (3)-contrary to (1).)
Here again there is a paradox of equivocation. The expression
"the Morning Star" and "the Evening Star" refer to the same
object by different characterizations associated with different
identificatory-descriptive properties. Hence (1) is appropriate
because distinctly different modes of identification are at issue. But
the objects of reference-the things being referred to---are one and
the same, and both (2) and (3) are appropriate on this basis. The
items at issue with "the Morning Star" and "the Evening Star"
are different identificationally but identical referentially, and we
must correspondingly distinguish between the sense of the identificatory expression at issue and its objective reference. With "is the
III
same as" construed identificationally (2) and (3) are false but (1)
is true; with "is the same as" construed referentially (1) is false
but (2) and (3) are true. Accordingly, this distinction dissolves the
paradox since the sense of the relevant terms needed to render
certain of its premisses true will falsify others.
The situation is substantially the same as that of the Relation
Riddles of Eubulides. We have to do with one and the same thing
under different identifying characterizations.
6.12 The Categorization Paradox

Again, consider what might be called the Categorization Paradox.
Some items can be categorized as defining a certain collection or
group: rivers crossed by George Washington, for example, or
things that have been transported by train. Others we cannot categorize-or indeed even form a conception of them, much as
Caesar could not form a conception of electrical appliances. Yet is
not the idea of things we cannot collect together in a categorization group a contradiction in terms? For in speaking and thinking
of them in this sort of way, have we not already categorized them,
namely as Uncategorizable? To all appearances there is a paradox
here, due to the fact that the uncategorizable ipso facto becomes
categorizable in just exactly that way.
On this basis we appear to be trapped within the aporetic duo:
(1) There are uncategorizable things.
(2) Anything uncategorizable can be categorized.
The resolution of this paradox lies in the fact that there is an
equivocation at work here. We have to distinguish between the
initial category-impoverished situation and the subsequent category-enriched situation that ensues upon introducing the pseudocategory of what is U ncharacterizable in terms of those initially
envisioned categories. With respect to pre-categorizability thesis
(1) is true but (2) is false, while with respect to post-categorizability (1) is false but (2) is true. Neither way will both premisses
crucial to the paradox be co-tenable. A distinction has again saved
the day.
112
6.13
Chapter 6
The Surprise Examination (or Execution)
What has become known as The Surprise Examination ParadoxI 9

further illuminates the role of equivocation as a source of paradox. It arises in the prosaic-sounding situation of the following
narrative:
There is a schoolteacher who announces: "Class, you had best
do a lot of reviewing this week-end. For there will be an examination on the topic next week. However, I am not going to
tell you on what day the exam will come; its timing will be a
surprise-the exam will not be foreseeable in advance." Now
in the class there is one over-clever student who reasons as follows: "The teacher has promised us a surprise examination.
But the exam cannot be on Friday, the last school day of the
week. For it is supposed to be a surprise, and if it has not taken
place by Thursday, then its coming on Friday, the last possible
day, would no longer be surprising. It will become foreseeable
after school on Thursday. So Friday is out. But if Friday is out,
then so is Thursday. For if the exam has not taken place by
Wednesday, then (since Friday is out) it would have to be on
Thursday. And so its falling on Thursday would also conflict
with its stipulated surprisingness. And the further steps along
this road are now easy. For if Thursday and Friday are both
out, then-by parity of reasoning-the exam cannot come on
Wednesday either. And then similarly with Tuesday and with
Monday. In consequence, the whole idea of a surprise examination is an impossibility and we have nothing to worry
about." So reasons the over-clever student, who accordingly
predicts that there will be no examination at all, taking the
teacher to have promised surprise and concluding that in the
circumstance this rules the examination out.
19. The Surprise Examination Paradox was initially presented as a surprise hanging
paradox and eventually also recast into other guises (such as a surprise military inspection
or a surprise wanime blackout). Its original version was presented in a paper by W.V.
Quine circulated in the early 1940s but not published until a decade later as "On a So
Called Paradox" (Mind, vol. 62 [1953], pp. 65--67; reprinted in Quine 1966). In the
meantime, it was discussed by D.]. O'Connor (1948), L.J. Cohen 1950, M. Scriven 1951,
P. Weiss 1952, and others. For a survey of proposed solutions see Avishai Margalit and
Maya Bar-Hillel, "Expecting the Unexpected," Philosophia, vol. 13 (1983), pp. 263-288.
113
What is one to make of this predictive puzzle, which some writers

present in terms of a surprise execution with a judge taking the
place of the teacher?
The aporetic situation at issue can be represented by means of
the following theses:
(1) The examination will occur next week.
(2) The examination may occur on any day of the next week,
bar none.
(3) When it occurs, the examination will be a surprise to the
students up to the very day that it is given.
Here (2) implies that the examination may occur on Friday. But
(1) and (3) imply that the examination cannot occur on Friday
(since it will then obviously be predictable when it has not yet
occurred by Thursday).
How is this chain of inconsistency to be broken? Let us examine the possibilities. As regards plausibility, we have it that (1) and
(2) state fixed facts that are parts of the defining condition of the
problem. What is at issue here lies (by hypothesis) entirely within
the teacher's power. Thus (3) is the weakest link in the chain and
we have it that the plausibility situation is [(1), (2)] > (3), with
(1), (2 )/( 3) as the appropriate resolution thanks to its optimal
retention profile of {I, OJ.
There is in fact good reason to see (3) as problematic. Its vulnerability roots in the fact that it suffers from a vitiating equivocation by failing to fix the temporal as-oJ-when specification of the
impredictability or surprise at issue. (After all, no event is surprising after the fact.) Now on Friday, on the morning of the schoolweek's last day, one can no longer see as surprising the threatened
examination that has not occurred up to that time. That is, the
timing of the examination can no longer be surprising as of then.
But at the time of the announcement prior to the week at issue--or
indeed on the Monday of that week itself-the fact of the exam's
falling on Friday would assuredly be a surprise since it certainly
could not then be predicted on the basis of evidence in hand. To
be sure, as of the end of school on Thursday the surprise will be
gone. But the choice of Friday would certainly be a surprise during the week before. And the same with any other day of the
114
Chapter 6
week. And so, if the day of the examination is kept concealed at

the time of announcement, then we have a surprise examination
alright (rather than a pre-scheduled one)-though of course the
surprise inevitably vanishes with the passage of time. The long and
short of it is that inability to pinpoint occurrence times is nothing
absolute but is itself a matter of timing. One must acknowledge
both (i) that there cannot (in the logical nature of things) be an
examination that is surprising at or after the fact, and (ii) that
there cannot be a day-deadlined examination that is surprising up
to the day of administration, even after the next-to-the last day
has passed. The fact, however, remains that an examination whose
timing is a surprise sufficiently early on thanks to its impredictability at that earlier time is an unproblematic possibility.
And so the crux of this supposed paradox is that one must give
due heed to the temporal reference-point relative to which the
timing of the "surprise examination" is indeed surprising. All that
the teacher has promised is that the day of the examination will be
an unpredictable surprise at the time of the announcement, and not
necessarily at various later junctures (which, as the paradox shows,
is in some cases impossible). Owing to the equivocation of that
pivotal aporetic thesis, there is no actual paradox of prediction
here, only the vitiating flaw of equivocation inherent in neglecting
the time as of which surprise obtains-that is, a failure to grasp
the norms of surprise. And so, that over-clever student who pits
himself against the instructor via the prediction "There will be no
examination at all" is thereby in reality setting himself up for an
all the more assured surprise when the examination actually
occurs-whatever the day may be.
The value of paradoxes as an instrument of inquiry is strikingly
exhibited by these paradoxes of ambiguity and equivocation. For
the discovery and validation of the distinctions needed to eliminate inconsistency is and ever has been a powerful instrument for
fostering clarity of thought and perceptiveness of understanding.

"Third Man" Paradox
Aristotle's Sea Battle Paradox
Kierkegaard's God Paradox
The Existential Paradox
The Time Travel Paradox
The Divine Foreknowledge Paradox
The Paradox of Explanation
The Freedom/Causality Paradox
The Promise-Breaking Paradox
The Form/Matter Paradox of the Presocratics
The Socratic Paradox of Ethics
The Divine Omnipotence Paradox
The Self-Advantage Paradox
7.1 Philosophical Paradoxes
Paradoxes arise in philosophy principally because of discord within
the wider community of philosophers. Conflicting doctrines find
their advocates. One philosopher or school builds up a case for
one contention and another philosopher or school builds up a
case for an rival contention incompatible with the first. If (or,
rather since) the plausibility of a philosophical contentions must
be understood on the liberal democratic basis of having some
philosophers on its side (rather than approval of one authoritative
philosopher or school), it's little wonder that plausible contentions can come into aporetic conflict in this domain.
Opponents of Plato's Theory of Ideas, deployed the notorious
"Third Man)) Paradox which stands effectively as follows: 1
(1) What makes distinct objects into items of the same type is
assimilation to an Idea. Individual men are alike as men
because they all participate in similarity-kinship to the
ideal and eternal archetype of Man. (This, in essence, it
the core of Plato's theory.)
(2) But if this is so in general, then what assimilates this or
that individual man to the Man-idea must be their common participation in yet another super-idea, Man No.2,
that assimilates a man and the Man-idea to one another.
1. The Third Man argument was launched against the Theory of Ideas by Parmenides
in Plato's dialogue of that name. It was emphatically endorsed by Aristotle in Metaphysics,
990blS-23 and 1031bI9-1032a7. See Prantl, Geschichte, Vol. I, pp. 18-19.
117
us
Chapter 7
(3) And this process will continue ad infinitum: We now need

Man No.3 to liken Man No.2 to the first-order Manidea( and so on.
(4) A viable doctrine cannot involve itself in an infinite
regress. Thus~ontrary to (1 )-Plato's Theory ofIdeas is
not viable.
As Plato's critics saw it, the sensible way out of this paradox is to
jettison (1) thereby abandoning Plato's Theory of Ideas. (To be

sure, a defender of Plato's theory would doubtless abandon (2)
and see the participation relationship between individuals and
ideas in terms other than type-similarity.)
Consider next the following Sea Battle Paradox put on the
agenda of philosophical deliberation by Aristotle:
(1) Truth is timeless. If a (fully definite) statement is ever
true, then it is true always and everywhere, and is as true
the day after tomorrow as it is today.
(2) If a statement about what happens tomorrow is already
true today, then what it claims will be inevitable, an
accomplished fact, as it were, that is effectively fated and
predestined.
(3) One of the two statements "A sea battle will take place in
the bay on [tomorrow's date]" and "A sea battle will not
take place in the bay on [tomorrow's date]" will be true
on the day after tomorrow.
(4) It is already fixed, certain, and inevitable that the sea battle is or is not going to occur. (From (2), (3).)
(5) But since free will exists and the battle arrangements and
place are not as yet fixed and decided, that putative sea
battle tomorrow is not something that is already fixed,
certain, and inevitable~ontrary to (4).
These statements clearly constitute an aporetic circle of inconsistency that will have to be broken. And-naturally enough-the
philosophers who have wrestled with the pattern have disagreed
about the plausibilities of the case.
119
Aristotle himself saw (1) as the culprit-the weak link in the

chain. And the reason-as some interpreters see it-is that propositions like "A sea battle occurs on [tomorrow's date]" are not
true timelessly but become true as of the date of the events in
question and then remain so. (Earlier on they are in a truth
limbo.)
The Megarian theorists of classical antiquity opted to reject
thesis (5). They were determinists who sought to exclude chance
and contingency from the world.
By contrast, modern theorists often opt for rejecting thesis (2).
The truth of temporalized statements is, they would argue, solely
and exclusively dependent on what happens at the time talked
about in the statement. And on this basis truth is wholly and
entirely detached from matters of determinism, inevitability, and
predetermination.
However, what is perhaps the most sensible approach is as follows. Clearly {(I), (2), (3), (5) constitutes an aporetic quartet.
Now here (1) and (3) are fundamental principles of the theory of
truth. However (2) and (5) are more problematic and speculative
theses about fate, inevitability, and predestination. So we arrive at
the priority ordering [( 1), (3)] > [(2), (5)], with the result that
the RIA-resolutions (I), (2), (3)/(5) and (1), (3), (5)/(2) are
tied for optimality with a retention profile of {I, t} We have, in
effect, a standoff between two possible exits from inconsistency:
abandoning (2) with its idea that preset truth entails future
inevitability, and abandoning (5) thereby accepting the idea of
predeterminative inevitability and predestination. The resolution
of this choice lies beyond the reach of general principles and calls
for the deployment of a substantive philosophical doctrine.
Philosophical theology provides a further illustration of the
sort of complexities involved. According to S0ren Kierkegaard it
is a paradox of Judeo-Christian philosophical theology that
humans cannot conceive of God as he really is. Kierkegaard)s God
Paradox envisions the following inconsistent triad:
( 1) It is rationally appropriate to worship God.
(2) A rational being will not~annot-worship something he
does not properly understand.
( 3) Man cannot understand God properly.
120
Chapter 7
Seeing (1) as rendered problematic through its linkage of religion

to rationality, while accepting (2) and (3) as evident facts of life,
Kierkegaard envisioned the priority ranking: [(2), (3)] > (1).
Accordingly he maintained (1) must be rejected. As he saw it, the
demand for a rational comprehension in matters of religion is
untenable and inappropriate. Thus Kierkegaard did not take the
atheist's easy route to (1 )-rejection. Instead, with St. Paul and
Pascal, he viewed religious faith as something above and beyond
reason-something that, with Tertullian, might have to be
acknowledged in the face of its inherent contradictions.
To be sure, a prioritization of commitments can provide a way
of resolving a dilemma. (In thinking about religion-as contrasted
with thinking within religion-Kierkegaard would like us to
maintain consistency.) However, as the example indicates, the priorities cannot always be seen as dear-cut, and theorists operating
with different priorities might be perfectly content to abandon (2)
or (3 ).2
Philosophers of the "existentialist" school are drawn to the
The Existential Paradox, which turns on the idea that birth (the
entry into life) is automatically the start of an unavoidable transit
towards death (the exit from life).3 Formally elaborated, the paradox runs somewhat as follows:
(1) The processes of organic nature are all geared to the realization and amplification of life.
(2) Life is indissolubly linked to death: every living organism
is on an inescapable journey towards death. Nature is
characterized by the death tropism of its organic beings.
(3) (2) is incompatible with (1).
Here (1) is dearly the more speculative and problematic contention, seeing that (2) is effectively a "fact of life." But (1 )'s
abandonment would presumably be coordinated with a distinc-
2. On theological paradoxes see H. Schr6cr, Die Denkform der Paradoxaljtiit als theologisches Problem (Gottingen: Vandenhoeck and Ruprecht, 1960). See also Gale 1991.
3. On this paradox and its theological and psychological ramifications see Howard A.
Slaatte, The Persistence ofParadox (New York: Humanities Press, 1968).
121
tion between life at the level of individuals and life at the level of
species. Thus thesis (1) would be accepted with regard to largescale aggregates of individuals but not with regard to particular
individuals. Indeed, the mortality of individuals could/would be
represented as a means to foster the well-being of the mass. Here,
as so often, distinctions provide a means towards salvaging something in the wake of thesis abandonment.
The Time Travel Paradox is a science fiction fixture: 4
(1) Time travel is theoretically possible.
(2) By going into the past, the time traveler can encounter
his own ancestors.
(3) The time traveler can interact with the people he meets,
and-in particular-can kill them.
(4) In view of (3), the time traveler can prevent his own birth
by killing one of his antecedents.
(5) Contrary to (1), time travel is impossible since it involves
the absurd prospect of preventing one's own birth.
The possible modes of resolution are readily surveyed:
(1) -abandonment. Rejecting the possibility of time travel
altogether.
(2) - abandonment. Rejecting the possibility of time-travel
access to certain spatial sectors of the temporal past.
(3) -abandonment. Enjoining the time traveler to causal
impotence in at least some domains.
(4) - abandonment. Precluding the time traveller from certain
kinds of self-interference.
4. On time travel and its philosophical ramifications see John Earman, "Recent Work
on Time Travel," in S. Savitt, ed., Time's Arrow Today (Cambridge: Cambridge University
Press, 1995).
122
Chapter 7
In view of the highly conjectural nature of the issue, the solutions

of the paradox must be highly conjectural as well. But as the preceding indications show, the speculative philosophy of science
offers various doctrinal possibilities for resolving the time-travel
paradox. Again, the assessment of comparative plausibilities
requires taking a stance on the substantive issues at stake.
The conceptual complexity of philosophical issues is such that
it is easy to be caught up in problems of terminology. For this reason the philosophical paradoxes are a fertile source of examples of
paradoxes of equivocation. Thus consider the classic Divine
Foreknowledge Paradox from philosophical theology based on the
following inconsistent quartet: 5
(1) Man is a free agent.
(2) Being omniscient, God must know--even "in advance,"
so to speak-how human decisions will eventuate.
(3) But if human decisions are foreknown they are thereby
temporally predetermined. Given that what one does is
foreknown to God, one cannot do otherwise.
(4) Accordingly, man is not a free agent, since the capacity to
decide and to act differently is the crux of freedom.
The resolution of this paradox depends on one's basic views of
God, man, and their relation. Theological determinists see (1) as
the weakest link and opt for a position of deterministic predestination. Other theorists abandon (3) and propose a more sophisticated theory according to which divine foreknowledge does not
entail predetermination and is compatible with the freedom of
man's will. Still others reject (4) and see freedom as residing not
as a capacity to act differently but rather in the accord between
one's actions and one's deliberative decisions, foreknowledge of
which would therefore not impede freedom. Still others dismiss
(2), arguing that foreknowledge must be abandoned because
omniscience is a matter of knowing not literally everything but
5. For ramifications of this paradox and many related paradoxical issues see Gale
1991.
123
only everything that can be known-which is not the case with

yet unresolved free human decisions.
As this plethora of possibilities indicates the topic is fodder for
dispute and John Milton's frustrated devils were constrained to
debate in vain this issue of "Fixt Fate, Free Will, foreknowledge
absolute/And found no end, in wand'ring mazes lost" (Paradise
Lost, II, 559-560). All their endeavors were fruitless, since alone
among God's rational creatures those devils were themselves subject to a "fixt Fate"; losing their free will after its inappropriate
exercise led to their expulsion from heaven.
7.2
Philosophical Apories Tie Issues Together
In philosophical epistemology the need arises to come to grips

with the following Paradox of Explanation:
( 1) Principle of Sufficient Reason: Every fact needs and (in
principle) has a satisfactory explanation.
(2) Principle of Non circularity: No fact is self-explanatory.
And none can appropriately figure in its own explanatory
regress.
3)
Principle of Comprehensiveness: No explanation of a fact is

satisfactory so long as (any of) its explanatory materials
themselves go unexplained.
(4) But then since the facts at issue at any stage of the explanation must be explained, and this explanation itself
requires something new and yet unexplained (by (4)), it
follows that no fact can ever be explained satisfactorily.

Here (1 )-( 3) constitute an inconsistent triad. At least one of the
plausible principles of this aporetic cluster has to be abandoned in
its unqualified generality. There are three exits from this situation:
(1) - rejection. The acceptance of surds, of brute facts that
must be accepted without themselves having or needing
explanation.
124
Chapter 7
(2) -rejection. Allowing some facts to playa role in their own

explanation-perhaps by adopting a nonlinear ("coherentist") model of explanation that distinguishes between
vicious and virtuous explanatory circles.
(3) -rejection. Accepting an explanation as satisfactory once it
reaches a point where the materials at issue are substantially clearer and more perspicuous then the fact being
explained-a point where sufficient explaining has been
done that we are entitled to call it a day.
Three very different philosophical positions are involved here:
surdism, coherentism, and explanatory pragmatism. And it is clear
that the apory at issue interlocks them into a coordinated interrelationship.
Consider the Freedom/Causality Paradox of the following
aporetic cluster, which sets the stage for controversy about freedom of the will:
(1) All human acts are causally determined.
(2) Humans can and do act freely on occasion.
(3) A genuinely free act cannot be causally determined-for if
it were so determined then the act is not free by virtue of
this very fact.
These theses represent an inconsistent triad in which consistency
can be restored by any of three distinct approaches:
(1) - rejection: "Voluntarism"-the exemption of free acts of
the will from causal determination (Descartes).
(2) - rejection: "Determinism" of the will by causal constraints
(Spinoza).
(3) -rejection: "Compatibilism" of free action and causal determination-for example, via a theory that distinguishes
between inner and outer causal determination and sees the
former sort of determination as compatible with freedom
(Leibniz).
125
Again we have a variety of philosophical positions interlocked

through their common role in an aporetic situation. And in this
case the issues can be addressed by disentangling our knotted terminology through suitable distinctions.
7.3 Philosophical Plausibility

In point of strategy, the way to resolve a paradox in philosophy is
the same that we must adopt anywhere else, to wit: thesis abandonment, hopefully accompanied by some partially salvaging distinction. And, of course, the more plausible we deem the
thesis-the better we deem the case for its retention to be-the
harder we have to "bite the bullet" when faced with its loss.
The problem with philosophical paradoxes is that the crucial
business of determining the comparative priority and precedence
of theses is seldom altogether straightforward in this domain. It is
something that depends on assessments and evaluations that are
likely to be sensitive to the specific ideological sensibilities of the
problem solver.
But how to proceed? What is our standard of priority to be?
With philosophy, our guidance for making these curtailments lies
in the factor of overall systematicity. The operative principle at
work here is that of achieving the optimum alignment with experience-the best overall balance of informativeness (answering
questions and resolving problems) with plausibility (keeping to
claims which on the basis of our relevant experience there is good
reason to regard as true). We want answers to our questions but
we want these answers to make up a coherent systematic whole. It
is neither just answers we want (regardless of their substantiation)
nor just safe claims (regardless of their lack of informativeness)
but a reasonable mix of the two-a judicious balance that systematizes our commitments in a functionally effective way. 6 And this
goes for paradoxes as well. Here too it is not just locally but globally adequate solutions that one would ideally want.
6. The aporetic nature of philosophy and its implications are explored in detail in
Nicholas kscher, The Strife of Systems (Pittsburgh: University of Pittsburgh Press, 1985).
The book is also available in Spanish, Italian, and German translations.
126
7.4
Chapter 7
The Role of Distinctions
Whenever the resolutions of philosophical paradoxes requires the

abandonment of some of the theses involved in the conflict there
just is no easy way out that is altogether cost-free. The issue is
always one of a choice among alternatives where, no matter how
we turn, we find ourselves having to abandon something which
on the surface seems to be plausible. And at this point distinctions
serve another important function. For whenever one of the premisses of a paradox has to be abandoned in the interest of a consistency-restoring solution, damage-content efforts can be made
to salvage some element of that non-abandoned premiss by introducing a distinction to divide it into an untenable and a tenable
part. 7
Consider an example. The Promise-Breaking Paradox has
played a significant role in ethical theory. It runs as follows:
(1) Promise breaking is morally wrong.
(2) It is never morally wrong to do what we cannot possibly
help doing.
( 3) In some circumstances one cannot help breaking a
promise: promise breaking is circumstantially unavailable.
One way of proceeding here is to distinguish between voluntary
and involuntary promise breaking. In the voluntary sense of the
term "promise breaking" (1) is true but (3) false, while in the
involuntary sense of the term (3) is true but (1) false. The paradox suffers a dilemmatic collapse on grounds of equivocation. But
we could now salvage something from the wreckage. Retaining
(2) and (3) one could then still salvage something of (I)'s contention that promise breaking is always morally wrong by restricting such condemnation to the special case of voluntary promise
breaking.
And this strategy of damage containment by means of distinctions has a wide applicability. Since each thesis of an aporetic cluster is individually attractive, unqualified rejection lets the case for
7. See Rescher, The Strife of Systems (op. cit.), especially Chapter 3, pp. 64-77.
127
a rejected thesis go unacknowledged. By merely modifYing rather

than outright rejecting a problematic thesis we may be enabled to
give proper recognition to the full range of considerations that
initially led us into contradiction.
Distinctions accordingly enable the philosopher to remove
inconsistency not just by the brute negativism of thesis abandonment but by the more subtle and constructive device of thesis
qualification. The crux of a distinction is not mere negation or
denial, but the amendment of an untenable thesis into something
positive that does the job better. Thus consider an inconsistent
triad having the following structure:
All A's are B's.
All B's are Cs.
Some A's are not Cs.
Note that these are incompatible only if the connecting terms
that link these theses together, A, B, C, are used in precisely the
same sense in each occurrence. If careful scrutiny manifests the
least deviation-if, say, the A of one of those theses is Al and that
of the other is A 2-then the inconsistency is removed and the
problem abolished. Distinctions are thus a crucial tool for philosophical problem solving; they are the natural means for eliminating inconsistency.
Let us examine the workings of this process more closely.
Consider the Form/Matter Paradox based on the following
four contentions, all of which were viewed with favor by various
Pre socratic philosophers:
(1) Reality is one: real existence is homogeneous.
(2) Matter is real (self-subsistent).
(3) Form--considered geometrically-is real (self-subsistent).
(4) Matter and form are distinct (heterogeneous).
Here (2)-(4) entail that reality is heterogeneous, thereby contradicting (1). The whole of the group (1 )-( 4) accordingly represents an aporetic cluster that reflects a cognitive overcommitment.
128
Chapter 7
And this situation is typical: the problem context of philosophical

issues standardly arises from a clash among individually tempting
but collectively incompatible overcommitments. The issues standardly center about an aporetic cluster of this sort-a family of
plausible theses that is assertorically overdeterminative in claiming
so much as to lead into inconsistency.
In such situations, whatever favorable disposition there may be
toward these plausible theses, they cannot be maintained in the
aggregate. Something has to give. In particular, we can proceed:
To reason from (2)-(4) to the denial of (1),
To reason from (1), (3), (4) to the denial of(2),
To reason from (1), (2), (4) to the denial of(3),
To reason from (1 )-( 3) to the denial of (4).
Accordingly, the ancient Greek philosophers confronted the following range of possibilities:
(1) - denial: Pluralism (Anaxagoras) or form/matter dualism
(Aristotle)
(2) - denial: Idealism (the Eleatics, Plato)
(3)-denial: Materialism (atomism)
(4) -denial: Dual-aspect theory (Pythagoreanism)
As this aporetic analysis indicates, the doctrines of these various
major schools of Greek metaphysics are locked into a coordinated
unity by the paradox at issue.
Of course, each alternative has its own problems as well as its
own opportunities. Consider, for example, the situation that arises
with rejecting thesis (2). It is instructive to note that this could be
accomplished by not simply abandoning it altogether but rather
by replacing it-on the idealistic precedent of Zeno and Platowith something along the following lines:
(2 ') Matter is not real as an independent mode of existence;

rather it is merely quasi-real, a mere phenomenon, an
appearance somehow grounded in immaterial reality.
129
The new quartet (1), (2'), (3), (4) is cotenable.

In adopting this resolution, one again resorts to a distinction,
namely that between:
(i) strict reality as self-sufficiently independent existence
and
ii)
derivative or attenuated reality as a (merely phenomenal)

product of the operation of the unqualifiedly real.
Use of such a distinction enables us to resolve an aporetic cluster-yet not by simply abandoning one of those paradox-engendering theses but rather by qualifying it. (Note, however, that
once we follow Zeno and Plato in replacing (2) by (2' )-and
accordingly reinterpret matter as representing a "mere phenomenon"-the substance of thesis (4) is profoundly altered; the old
contention can still be maintained, but it now gains a new significance in the light of new distinctions.)
What is sometimes called the Socratic Paradox in ethical
theory turns on Socrates's thesis that "nobody does wrong willingly." 8 This leads to the following line of argumentation:
(1) Rational and well-informed people often act wronglyand willingly so.
(2) A rational person will only do willingly that which he
believe will-in the circumstances-yield a greater balance
of positive as against negative results.
(3) To choose the wrong deliberately damages the psyche of
the agent to an extent that is automatically greater than
whatever benefits that wrong action may otherwise yield
him.
(4) By (2) it is only through mistakenly believing that-contrary to (3)-a wrong action yields a positive balance of
benefit that a rational person will do wrong willingly.
8. See Plato, Protogoras, 352.
130
Chapter 7
5) A rational and properly informed person will never act

wrongly. By (3)-(4).
Here {( 1), (2), (3)} constitute an inconsistent triad. Prioritizing
(2) and (3), Socrates himself therefore rejected (1), and was thus
led to his "paradoxical" teaching that "Nobody [who is rational
and properly informed] does wrong willingly." The result, for
Socrates, is that wrongdoing (moral transgression) is ultimately a
failure of the intellect (of rationality and its proper application)
rather than a failure of the will.
On this basis, it emerges that, on Socrates's own analysis, the
paradox is essentially one of equivocation. We must distinguish
between actual choice and action and rational choice and action.
With rational action (1) is false and (2) true; whereas with actual
action (1) is true and (2) false.
Again consider the classic conundrum of medieval philosophical theology: can God, an omnipotent being, reduplicate himself?
This question engenders a dilemmatic paradox. For if God cannot
do so, then how can one say that he is omnipotent, seeing that
there is something he cannot do? But if he can do so, then how
can he qualify as omnipotent, seeing that there is, or could be,
another no less powerful being who can check his actions?
This puzzle gives rise to the Divine Omnipotence Paradox
which roots in the following aporetically inconsistent collection of
theses:
(1) There is an all-powerful, omnipotent being, God. (By
assumption. )
(2) An omnipotent being is one who can do literally anything. (This is, supposedly, the definition of "omnipotence.")
(3) God can do literally anything. (From (1) and (2).)
(4) God can reduplicate himself. (From (3).)
(5) A reduplicated God would be equipowerful with the original. (By the definition of "reduplication.")
131
(6) The action of any being can be counteracted-and thus

negated-by the action of an equipowerful one. (By the
definition of "equipowerful.")
(7) A being whose actions can be counteracted by another is
not all-powerful (omnipotent). (From (2).)
(8) God is not omnipotent-contrary to (1). (From (5)-(7).)
The clash between (8) and the starting point (l) indicates that the
overall set is aporetic-that is, collectively inconsistent. Again we
have a paradox. 9
Since various theses of this apory are derivative in that they can
be obtained by logical inference from others, we can trace the
conflict at issue back to the inconsistent quartet I(1), (2), (5), (6).
Because these four theses are the only nonderivative ones (whose
justification is not in terms of "from,") they alone carry the burden of contradiction here.
And so in these circumstances we are driven to a forced sacrifice of one of the four theses at issue. Once again, we have to
probe for the weakest link in the chain of inconsistency.
Let us look at our problematic foursome more closely. Here
(1) is a fundamental dogma of monotheistic theology and (5) and
(6) issue from comparatively unproblematic definitions. However,
the purported definition at issue in (2) is in fact highly problematic because we have to ask ourselves whether omnipotence is the
ability to do literally anything rather than anything that is logically
possible. On this basis, the priority ranking at issue is
(1) > [(5), (6)] > (2).
Here we have just three priority levels. The appropriate resolution
is accordingly (1), (5), (6)/(2) with its here optimal retention
profile. 10
And there is in fact good reason for concern about (2). For the
paradox at issue could be seen to be based on a questionable con-
9. For this paradox and related issues in philosophical theology see Gale 1991.
10. There are, of course, more radical possibilities such as the atheist's all-out rejection
of the very possibility of God.
132
Chapter 7
struction of omnipotence. And just this was the alternative generally favored among the medieval scholars. As they saw it, it is
omnipotence in the "can do literally anything" sense of the term
that is the weak spot here. For "omnipotence" cries out to be
construed as "can do anything that is possible" or "can do anything that a God-like being could reasonably want to do." And if
(2) is abandoned--or, rather, reconstrued in the proposed waythen the inconsistency is resolved.
This example conveys an important lesson. The use of distinctions interacts with plausibility considerations. For there is good
reason for the policy of fixing upon the least plausible aporetic
premiss and dividing it by a saving distinction. It is, after all, more
than likely that the burden of carrying the "bad" side of the distinction was the source of the implausibility involved.
To be sure, distinctions are not needed if all that concerns us is
averting inconsistency; simple thesis abandonment, mere refusal
to assert, will suffice to that end. Distinctions are necessary, however, if we are to maintain informative positions and provide
answers to our questions. We can guard against inconsistency by
refraining from commitment. But that leaves us empty-handed.
Distinctions are the instruments we use in the (never-ending)
work of rescuing our assertoric commitments from inconsistency
while yet salvaging what we can.
The history of philosophy is accordingly shot through with the
use of distinctions to avert aporetic difficulties. Already in the dialogues of Plato, the first systematic writings in philosophy, we
encounter distinctions at every turn. In Book I of the Republic,
for example, Socrates's interlocutor quickly falls into the following
Self-Advantage Paradox:
(1) Rational people always pursue their own best interests.
(2) Nothing that is in a person's best interest can be disadvantageous to their happiness.
(3) Even rational people will-and must-sometimes do
things that prove disadvantageous to their happiness.
Here, inconsistency is averted by distinguishing between two
senses of the "happiness" of a person-namely the rational contentment of what agrees with one's true nature and what merely
133
redounds to one's immediate satisfaction by way of pleasure, in

sum, between real and merely affective happiness. With real happiness, (2) is true but (3) false, while with merely affective happiness, (2) is false, but (3) is true.
For the most part, the Platonic dialogues present a dramatic
unfolding of one distinction after another as Socrates and his
interlocutors seek to work their way out of a maze of inconsistency. And this use of distinctions as a paradox-buster is pervasively common-place in philosophy because distinctions always
make it possible to salvage something from the wreckage.

Absurd Instruction Paradoxes (such as the
Mikado Paradox)
The Parent-Beater Paradox
The Horn Paradox
Rule/Exception Paradoxes
The Adult/Child Paradox
The Barber Paradox
The Bibliography Paradox
Kantian Antinomies
Predictive Paradoxes (various)
Paradoxes of
Inappropriate Presupposition
(Improper Totalities
in Particular)
8.1
Unwarranted Presupposition
Like other instances of exact reasoning, paradoxes routinely, if

only tacitly, involve certain basic formal communicative presuppositions, preeminently including the following two:
The terms involved in formulating these propositions are
meaningful, well-defined, and unequivocal (in other words,
uniform in meaning across different premisses).
The propositions at issue are meaningful, and at least plausible if not actually true.
Of course, such presuppositions may possibly fail to be met-in
ways that can prove fatal to the viability of the argumentation of
the paradoxes at issue.
However, it is not these generic and routinely standard communicative presuppositions that primarily concern us in the present chapter. Rather it is substantive and thereby fact-prejudging
presuppositions that will be at issue here. The difference is significant. Communicative presuppositions such as the above are
generic prerequisites for the cogency of the argumentation of a
paradox. Substantive presuppositions are particular material
assumptions whose untenability (let alone erroneousness) would
137
138
Chapter 8
erase some of the premisses of the argument. For paradoxes often

rest upon inappropriate and unwarranted factual suppositions.
For example, they can result from a suppositional process of construction whose directions are actually impossible to carry outbe it in principle or in practice. Thus consider the instructions:
1. Take brick No.1.
2. Take brick No.2.
3. Place brick No.1 alongside brick No.2.
This set of instructions is in general quite unproblematic. But what
if brick No.1 is to be the same brick as No.2? Clearly the instruction "Take a brick and then put it next to itself" is absurd, and
straightaway leads to paradox. The pictures of M.C. Escher vividly
illustrate this phenomenon, for example, his print of a hand drawing itself, 1 which suggests the even stranger possible variant of a
whitewash wielding figure in process of painting itself out. As with
optical illusions, it is not that such vistas themselves are paradoxical
but rather that the statements we take to provide a "natural"
description of what those pictures depict form an inconsistent set.
Paradoxes of this kind rest on the unstable foundation of the mistaken supposition or presupposition that a certain hypothetical
construction can be carried out-something which is in fact not
the case because an inherent absurdity is involved through a conflict with fundamental facts of physics or of geometry.
Again consider the Mikado Paradox based on Gilbert and
Sullivan's operetta of the same name, a paradox which ensues
upon four rules:
There is to be one execution per month
Execution is to be by beheading with a sword
There is to be no departure from a specified ordering of
prisoners to be executed
1. M.e. Escher, The Graphic Work of M.e. Escher (New York: Duell, Sloan, and
Pearce, 1961), p. 47. There's also Steinberg'S drawing of the same subject in Saul
Steinberg, The Passport (New York: Harper, 1954).
139
Only the official Lord High Executor can carry out an

execution.
And now a paradox obviously results when the next prisoner to be
executed is himself appointed to be Lord High Executioner.
Paradox can thus affect not only contentions and questions
but also instructions. The injunction "Never say never" is every
bit as paradox-engendering as "All generalizations are false." And
the sign posted on the wall that reads "No signs are to be posted
here" is in much the same boat as the graffito that says "All statements on this wall are false." Another example of directions that
cannot be carried out are "Do not read this sentence"-the
request obviously comes too late. A more interesting example is
the following resolution reportedly adopted by the Municipal
Council of Canton, Mississippi: 2
1. Resolved, by this Council, that we build a new jail.
2. Resolved, that the new jail be built out of the materials of
the old jail.
3. Resolved, that the old jail continue to be used until the
new jail is finished.
A nice trick if they can do it!
When we subscribe to an unwarranted presupposition regarding something where "we ought to know better," this all too easily runs into conflict with other things that we know-or think we
do. A good instance of this is provided by the old chestnut "Have
you stopped beating your father?"3 This engenders the classic
Parent-Beater Paradox based on the following contentions:
(1) The respondent must answer the question Yes or No.
2. Patrick Hughes and George Brecht, Vicious Circles and Infinity (Garden City:
Doubleday, 1975).
3. For this paradox see Diogenes Laertius, LiJ'es of the Philosophers, II, 35. The moderns changed the referent here from father to wife, but political correctness calls for reversion to the original form. The American Philosophical Association's "Guidelines for the
Non-Sexist Use of Language" (http://apa.udel.edu/apa/publications/texts/nonsexist/
html) specifically mentions the wife-beater riddle as an instance of the introduction of sexual stereotypes into philosophical discourse. (lowe this reference to Karen Arnold.)
140
Chapter 8
(2) Ifhe replies Yes, that comes to acknowledging that he

used to beat his father.
(3) Ifhe replies No, that comes to acknowledging that he still
beats his father.
( 4) Either way, it follows that the respondent concedes that
he has beaten his father.
(5) But as we know perfectly well, the respondent may never

have beaten his father at all.
(6) Consequently, the respondent can fail to speak truly no
matter how he replies.
(7) However, some direct reply to a Yes/No question must be
true in any circumstance.
(8) (6) and (7) contradict one another.
How can this chain of inconsistency be broken? One promising
prospect here relates to thesis (3). For this thesis ignores the fact
that a negative reply to the question admits of two distinct constructions and covers two different possibilities. By accepted usage
the negative response here covers at least two possibilities: "No, I
haven't stopped beating him because I'm still doing it," or "No, I
haven't stopped beating him because I never even started." The
construction of (3) that supports (4) simply ignores those other
possibilities. It rests on the presupposition-totally unwarranted
in the circumstances-that the respondent has been beating his
father.
The Horn Paradox of Eubulides (probably) is another example
of an inappropriate-presumption paradox much favored in classical antiquity: "Have you lost two horns? No! Well, then you must
still have them!"4 Such paradoxes which rest upon an unwarranted and often even erroneous presupposition have always been
a commonplace in this field.
4. See p. 12 above. For Eubulides see pp. 102-04 above.
8.2
141
Paradoxical Questions
The Parent-Beater Paradox and the Horn Paradox illustrate the

larger phenomenon that not only statements but even questions
can be paradoxical. The question "Is the answer to this question
No?" is an example. For consider the resultant situation:
The answer given is
The answer claimed by ~he

answer that has been given
Yes
No
No
Yes
There is no way of achieving an alignment here. For every direct

answer to the question results in a paradoxical statement: "Yes the
correct answer to this question is No" and "No the correct
answer to this question is Yes." And this state of affairs generalizes. A paradoxical question is one every direct answer to which is
paradoxical through being self-consistent. No (direct) answer to
such a question is possibly correct. And the crux of all such paradoxes lies in the fact that the question at issue involves a presupposition that fails to obtain. 5
A standard presumption of meaningful questions is that they
can be answered directly and that some direct answers are at least
possibly correct. Paradoxical questions violate this presupposition.
On the other hand, a question is crypto-paradoxical if there is
no possible way of giving a direct answer to it that is incorrect.
For example: "Is your answer to this question affirmative?" Here
there is simply no way for the answer given and the answer that is
validated by the given answer to get out of synch.
We may be on paradoxical ground if we assert that "All rules
have exceptions" or-as Oliver Wendell Holmes Jr. put it more
picturesquely-"No general proposition is worth a damn." But
the fact remains that virtually all rules admit exceptions and that
virtually all generalities are subject to qualification, the point at
5. All questions involve some prcsuppositions---cven if only the minimalistic one that
they are answerable. On these issues of question epistemology see the author's Empirical
Inquiry (Totowa: Rowman and Littlefield, 1982).
142
Chapter 8
issue being made unparadoxical by that quantification "virtually."

Now as Aristotle noted in his Sophistical Refutations,6 the primrose path to paradox is paved with questions that involve subscription to a general rule that actually has exceptions: "Ought one to
do what the law requires?" "Ought one to honor the wishes of
one's parents?" As with the Parent-Beater Paradox, you cannot
win either way with such a question. For if you answer No, the
rule will be against you, and if you answer Yes, then the plausible
exceptions will be cited against you? The mistaken supposition
that a sensible rule is absolutely unexceptionable and always
applies is the crux of the problem here.
Consider an example. Frederic, the protagonist of Gilbert and
Sullivan's Pirates of Penzance, is in a predicament. For the question "How old is Frederic?" poses a puzzle owing to the following Adult/Child Paradox:
(1) Frederic is an adult of 25 years, having lived for that many
years.
(2) People are as many years old as they have had birthday
anniversaries.
(3) Frederic was born in a leap year on February 29th, a leap
day. He has thus had only five birthdays.
(4) By the birthday reckoning at issue in (2)-(3), Frederic is a
mere child of only five years.
The culprit here is clearly premiss (2). It states a perfectly plausible rule to which there are only once-in-a-blue-moon exceptionspecifically 1 in 1,460. However, in Frederic's case we have to
come to terms with one of these exceptions. Dissonance arises
here because the specifics of the case point one way, the generalities of the rule another.
6. See Aristotle, Soph. Elen., 176a19-38.

7. Sec: Aristotle, Soph. Elen., 173a19-25 and 174b12-19.
8.3
143
Negatively Totalitarian Self-Involvement:

The Barber Paradox
As noted above, all contexts of rational assertion and deliberation

are governed by an at least tacit supposition that not only are the
terms being used in the assertions at issue univocally meaningful,
but that these assertions themselves are at least plausible if not
actually true. Should this presupposition prove to be unwarranted, the chasm of paradox yawns wide-open.
This is illustrated by The Barber Paradox, 8 a conundrum that
has received much attention. It is based on the following riddle:
A certain village has a barber whose practice it is to shave all

the adult male villagers who do not shave themselves. He himself, of course, is an adult male who lives in the village. Does
he or does he not shave himself?
The evident paradox here is that if he is a self-shaver, then (by
hypothesis) our barber does not shave himself, while if he is not a
self-shaver, then (by hypothesis) he does shave himself. Either way
we are in difficulty.
Putting it more formally, our barber B is (by hypothesis) someone such that:
'<:Ix [S(B, x) iff -S(x, x)]

But when the x at issue is instantiated as B himself, this yields:
S(B, B) iff -S(B, B)

This result plunges us into paradox, seeing that we now can maintain neither S(B, B) nor -S(B, B). It seems that consistency is
beyond our reach here.
The following theses constitute the aporetic cluster that
defines the paradox:
8. For this paradox see Russell 1918 and T.S. Champlin, Reflexive Paradoxes
(London: Routledge, 1988), esp. pp. 172-74. Also called the Barber of Ser>ille Paradox, it
is an informal variant of Russell's Paradox in set theory (for which see pp. 170-73 below).
144
Chapter 8
(1) There is---or can be-a barber who answers to the specifications of the narrative.
(2) This barber of the narrative (like any other) either does or
does not shave himself, but not both.
(3) If the barber shaves himself then, by the narrative's stipulation he is not someone who shaves himself.
(4) If the barber does not shave himself then, by the narrative's stipulation he does shave himself.
(5) Thus either way a contradiction ensues.
The optimal point at which to break the chain of consistency here
is clearly at its very start with that hypothetical barber himself. For
there is not and cannot be a barber who answers to the specified
conditions. The barber paradox thus is vitiated from the very outset, by being predicated on the supposition of a barber who cannot possibly be. That purported introduction of a barber upon
the stage of discussion misfires-it fails to introduce.
Again consider the closely related Bibliography Paradox arising
from the following account:
There certainly are bibliographies of bi bliographies (and lists
of lists and registers of registers). And such a bibliography (or
list of register) is complete ifit lists all bibliographies (etc.) But
of course most bibliographies will be incomplete when they
leave something out. And one bibliography that many of them
will leave out is themselves. Such a bibliography may be characterized as self-omitting. But now consider this idea of selfomitting bibliographies. And let us contemplate the idea of a
complete bibliography of these-a bibliography that lists all
(and only) the self-omitting bibliographies. Will such a bibliography include itself? Clearly this question plunges us into
paradox.
What we have here is once more a paradox based on an erroneous
and untenable presupposition, namely that there can be such a
bibliography (or list or register). There cannot. The very idea of
such an item is absurd, as the articulation of the paradox itself
makes manifest.
145
Again, let us suppose the following definition:

T
the conjunction-of-truths that conjoins all (but only)

those conjunctions-of-truths that do not include themselves as conjuncts.
Then paradox ensues. For if T includes T as a conjunct, then it

does not. And if T does not include T as a conjunct, then it does.
The way out here is, of course, to say that, notwithstanding its
overt specification as a conjunction, a conjunction of the specified
sort simply does not exist. Neither T nor anything else fills the
bill. Which is to say that T is no more than an illicit or improper
pseudo-specification of an item purportedly at issue. Put differently, that purported "conjunction of truths" does not in fact
qualifY as something that can be characterized in this way at all.
There is not and cannot be such a conjunction and so the issue of
what it does or does not include simply does not arise.
8.4 Illicit Totalization Paradoxes

Various paradoxes attest to the impropriety of negation-totalities
arising from expressions which might be called the "X of X's"
format. Items of some sort are seen as self-inclusive by encompassing themselves. For example:
a set of sets
a list of lists
a picture of pictures
a conjunction of conjunctions
a text of texts
a praiser of praisers
a shaver of shavers
146
Chapter 8
All such self-encompassing item-types are homogeneous in the

sense that here an X of X's is itself an X (a set of sets is a set, a list
of lists is a list, and so forth).
With such self-encompassment in view we can now project an
X-totalization of self-nonencompassing X's:
the X of all those X's that do not X-wise encompass themselves (the set of all those sets that do not set-include themselves)
It is clear that such self-exclusive totalization is always paradoxical.
Thus consider the container (set) of all containers (sets) that do
not contain (set-include) themselves, the list of all lists that do not
list themselves, the picture of all pictures that do not depict themselves, etc. The idea here is that if "the x such that, for every y, x
encompasses y iff y does not encompass itself." Putting the usual
symbolism to work, what we have in view is the following totalization item:
(tx) 'lfy (x enc y iff -y enc y).

If (but only ifl) such an item exists-let us continue to call it Xthen we have:
'lfy (X enc y iff -y enc y).
But the modus operandi of X-specification is such that X is supposed itself to be an item of the range of our variable x) yetc. This
being so, we can particularize the preceding generalization to X
so as to obtain:
X enc X iff -X enc X.
Paradox is now upon us. But so is its resolution. For this contradiction itself straightaway negates our preceding assumption that
that initial totalization item does (or can) exist.
The issue is thus straightforward. We confront the question:
Does this totalitarian X-item encompass itself X-wise or not?
Does that set of sets include itself, that list of lists list itself, that
picture of pictures picture itself, and so on? And paradox occurs
147
immediately because in virtue of its very definition that totality of

self-exclusive X's is itself an X that neither can nor cannot encompass itself. The problem, of course, is that the presupposition that
such a negatively totalitarian X does or can exist is simply false.
The item purported to be an instance of the kind at issue (a set, a
list, etc.) just does not exist at all: to call it such does not make it
such. A substantive matter is at issue that cannot be settled by terminological slight of hand. (In the next chapter we shall see that
the violation of a very general requirement is at issue, namely the
Successful Identification Principle, SIP.)
The preceding difficulties relate to the misidentification
of items purporting to be of a type to which they could not possibly belong in view of the way in which that identification is presented. Throughout this range of cases, we have paradox of
inappropriate presupposition that incorrectly holds that something is appropriately identified by a specification which actually
does not and cannot succeed. Any body of reasoning or argumentation presumes, explicitly or tacitly, that its propositions are
meaningful and thus that its relevant referring terms succeed in
doing so. And when terms become unraveled on the basis of an
inappropriate totalitarian self-involvement, this crucial presupposition is falsified and the paradox dissolved. When the charge of
illicit totalization can be made to stick, it is a highly effective presupposition -buster.
8.5
Immanuel Kant's Antinomies
The general idea we have been considering, inappropriate totalization, played an important role in the philosophy of Immanuel
Kant-although he approached the issue in a positive rather than
negative light. As he saw it, the four classic "Antinomies" of
Critique of Pure Reason exemplify the preceding situation. These
Kantian Antinomies run as follows:
I.
The totality ofphysical existence (= the world or the universe as a whole)

is limited in the physical manifold of space and time.
is unlimited in the physical manifold of space and time.
148
Chapter 8
II. The ultimate simples of nature (= the absolutely atomic,

totally indecomposable constituents of physical substance
are pervasive throughout all of physical reality.
are excluded from physical reality (in other words, do
not exist as such).
III. The totality of natural occurrence (= the aggregate of
events in physical realm)
is of necessity determined by physical law.
is free from determination by physical law.
IV. An absolutely necessary) self engendering being (= a being
that itself encompasses the totality of its own cause)
is physically or causally active in the natural world.
is physically and causally excluded from the natural
world.
In each case what we have at issue is an extra-ordinary, somehow
totalitarian (that is ultimate or absolute) object of consideration
regarding which some far-reaching physically geared feature is
both flatly affirmed and categorically denied. What is being
claimed throughout is (structurally speaking) that a certain putative object (A) both does and does not have a certain physically
oriented feature (P). Conflicting predications are thus at issue: "A
is P' and "A is not-F." And, so Kant insists, equally good arguments can be constructed either way, so that both the thesis and
the antithesis can be rendered plausible. 9
On this basis, Kant's antinomies were seen by him as engendering a paradox along the following general structural lines:
(1) Item A (for instance, physical-existence-as-a -whole) is a
legitimate object of predication. [A supposition.]
(2) There is cogent reason for saying that item A (for
instance, physical existence) is F (for example, limited).
9. On Kantian antinomies see N. Hinske, "Kants Begriff der Antinonie und die
Etappen seiner Ausarbeitungen," Kant-Studien, vol. 56 (1965), pp., 485-496, as well as
the several commentaries on the Critique of Pure Reason.
149
[A substantive fact grounded in argumentation that Kant

presents in some detail.]
(3) There is cogent reason for saying that item A (for exampie, physical existence) is not F (for instance, unlimited).
[Also a substantiable fact rooted in presented argumentation.]
(4) (2) and (3) are logically incompatible if (and whenever) A
exists as a legitimate object of predication. [A fact of
logic. ]
(5) Therefore, by (2)-(3) item A does not exist because it is a
fact of logic that no legitimate object of predication can
have incompatible predicates.
As Kant saw it, the natural place to break this chain of contradiction is at thesis (1) which-so he maintains-is the weak spot of
the antinomy. For him the proper lesson to be drawn is that only
identifiable particulars can exist as proper objects of physically
oriented predication whereas abstract totalities are vitiatingly
problematic in this regard.
The four Kantian antinomies arise from a series of corresponding questions of a common format
Is the ultimate unit of A limited or unlimited in point of B?
where the four cases stand as follows:
A
(1) physical existence
spatio-temporal distribution
(2) physical divisibility
physical presence
(3) natural occurrence
lawful determination
(4) existential self-sufficiency
causal participation in nature
And all of these questions make the common supposition that

that ultimate item at issue with A actually exists as a legitimate
150
Chapter 8
object of physically oriented predication-something that Kant

emphatically denies.
Kant thus argues that in each and every case, the resolution of
the antinomy consists in rectitying a mistaken presupposition that
the ultimate A-unit at issue is a bona fide (physical) existent that is
capable of being a proper bearer of (physically oriented) predicates
such as spatiotemporal extent, physical existence, lawful determination, or causal efficacy. Such ultimates are communicative devices
to guide the processes of thought and discussion but not items of
natural existence of which physical features can be predicated.
(This is why the second, negative side of the antinomies always
comes closer to the mark.) Kant's position is not so much that of
Nicholas of eusa that certain totalities are beyond human comprehension as that they are not defined as items that can bear the sorts
of predicates at issue in their putative specification.
Kant's position here is certainly discussible. One could, for
example, contend that those supposed "cogent reasons" at issue
in (2)-(3) constitute the Achilles heel of the argument rather than
thesis (1). Or again one could try to dissolve the contradictions
on grounds of equivocation, arguing (for example) that the sense
in which the physical world is limited in space-time is different
from the sense in which it is not so limited.
All the same, Kant's idea of rejecting certain totalities as incapable of characterization in certain terms of reference is not all
that implausible. No painting of a painting can represent all of its
detail. And even so no thought or theory can encompass the
entire reality to which it itself belongs. The hypothetical projection of such an item envisions a totality that is incapable of cognitive realization-one whose acceptance conflicts with the things
we know about the nature of human knowledge. As Kant saw it,
the synoptic (totalitarian) purview of the claims of classical metaphysics is simply unavailable to us humans. We have to operate
within the physical world and therefore from a particular (specifically human) perspective. We have no cognition-external
Archimedean fulcrum outside the realm of our thought from
which to move the entire domain of thought as a whole.
And this Kantian perspective relates to the relationship
between inner and outer. We humans are emplaced within nature
and by this very fact are limited in the information we can obtain
about nature. All of our observation-based judgments are partial
151
and thereby incomplete. A<; far as actual knowledge goes, absolutistic totalities are off limits-be it the totality of actual experience (the self) or of its objects (the world) or of its rationale
(ultimate purpose or God).10 Such ideas represent useful mindcontrivances for the assessment of actual knowledge but do notcannot-themselves constitute actual objects of knowledge. For
Kant, they are not entities at all, but only practical resources,
thought instrumentalities useful only for contrasting what we
actually have with what we would ideally like. For him, those
improperly totalized items simply cannot constitute objects of
knowledge. And paradox accordingly results because here the presupposition of existence is not satisfied.
8.6
Predictive Paradoxes
I can confidently make the indefinite prediction that some of my

best-considered predictions will corne false, but of course cannot-without paradox-predict specifically which ones. However,
I may very well (and quite unparadoxically) predict that your predictions will corne false-and even specifically identify them.
Inappropriate presumptions about the future are a fertile
source of paradox. One of them relates to the very possibility of
prediction. Could a predictive machine--call it Pythia-predict
everything? Could it be so perfected as to provide a tenable
answer for every predictive question one enough cares to ask?
Clearly not. For one thing, there are essentially paradoxical questions on the lines of: "What is a predictive question that you will
never-ever have to deal with in answering questions put to you?"
Pythia itself must keep silent on this issue-or else confess its
imperfection with an honest CAN'T SAY. But the question itself is
certainly not meaningless, since we ourselves or some other predictor could in principle answer it correctly.
Again, we could wire our supposedly perfect predictor up to a
bomb in such a way that it would be blown to smithereens when
next it answers YES. And we now ask it: "Will you continue active
and functioning after giving your next answer?" Should it answer
10. See the discussion of these ideas in Kant's Critique of Pure Reason, A680-B708 =
A689-B717.
152
Chapter 8
it would be blasted to bits and its answer thereby be falsified.

Should it answer NO it would (ex hypothesi) continue in operation
and its answer thereby be falsified. l l
To say that a predictive question is unresolvable is not, of
course, to say that a predictor cannot answer it. We could certainly program a predictive machine to answer every yes-no question by "YES," every name question by "George Washington,"
every when? question by "in ten minutes," and so on. If it is simply an answer that we want, we can contrive predictors that will
oblige us. But the problem, of course, is to have a predictor
whose predictions are probable or, in the case of a putatively perfect predictor, correct, on a basis that is credible in advance of the
fact.
Specifically, one of the issues regarding which no predictor can
ever function perfectly is its own predictive performance. In the
area of prediction, as elsewhere, paradoxes can arise through
problems of self-reference. Consider the question: "Is the prediction that you are going to respond negatively to this question
going to come true?" The ensuing situation is as follows:
YES,
The answer
given is
This answer means

that the prediction is
This means that

the answer is
YES
Correct
False
NO
Incorrect
True
In no case can the given answer manage to correspond to the

facts.
Analogously, consider the question: "Will you answer
this question-this very question-indefinitely, that is, by CAN'T
SAY (rather than YES or NO)?" We arrive here at the following situation:
11. The example is adapted from John L. Casti, Searching for Certainty (New York:
Morris, 1990), p. 42.

The answer
given is
Its truth status

is automatically
153
So the predictor
YES
false
fails (by falsity)
NO
false
fails (by falsity)
CAN'T SAY
immaterial
fails (by indecision)
Inevitably, our predictor cannot function adequately here either.

Or again consider a somewhat different example:
The predictor starts in state non -C. And it is programmed to
shift to state C if ever it gives a negative answer. At a moment
before time t (with only just enough time for processing one
single last question before t) we ask: "Will you be in state C
at t?"
The answer
gzven zs
Its truth status

is automatically
So the predictor
YES
false
fails (by falsity)
NO
false
fails (by falsity)
CAN'T SAY
immaterial
fails (by indecision)
In no case can the predictor at issue possibly provide a correct

answer to such questions-though we ourselves could in principle
do so.
The long and short of it is that no predictor can function perfectly with respect to its own operations and, in particular, none
can predict all of its own future predictions. 12 The very most one
12. This point has been argued forcefully in other contexts by K.R. Popper. See his
The Poverty of Historicism (London: Routledge, 1957), p. vi. But also compare Peter
Urbach, "Is Any of Popper's Arguments Against Historicism Valid?" British Journal for
the Philosophy of Science, 29 (1978) 117-130 (see pp. 127-28). Popper's argumentation
has its problems, however. For it pivots on Kurt GodeI's famous Incompleteness Theorem
to the effect that no reasoner proceeding on mathematical principles can decide whether a
certain sentence G is or is not a theorem of arithmetic. Popper now proposes asking the
putative predictor the question: "If you are asked at t the question 'Is G a theorem?' (and
we attend until f for an answer) will you or will you not before f respond with 'yes' or
Chapter 8
154
could plausibly ask for might seem to be that predictors predict

correctly everything that is in principle possible for them to predict.
But even here there are problems. Perhaps the most obvious of
them is that of deciding just what is "in principle possible" for a
predictor by way of genuine prediction. For to determine what is
in theory or in principle predictable-and what is not-we have
to look to the deliverances of natural science. For science itself
tells us that some sorts of predictions are in principle infeasiblefor example, the outcome of such noncausal quantum processes as
are at issue in the question: "Exactly when will this atom of a
transuranic element with halflife 283 years disintegrate?" "Exactly
what will be the position and momentum of this particle two minutes from now?" What is needed at this point is a determination
of what is in fact predictable-and not just an indication of what
the science of the day considers (perhaps mistakenly) to be predictable. And this is an issue that could only be settled satisfactorily by perfected or completed science, something we do not have
now and could not be sure of having even if (per impossibile) we
did sometime acquire it.
From the vantage point of aporetics, it emerges that those predictive paradoxes issue from problematic questions that inappropriately presuppose themselves to have a viable answer. 13
8.7
Summary
The overall lesson that emerges from the analysis of these paradoxes of inappropriate presupposition is that these can in general
be regarded from a reverse point of view. That is, they can also be
construed as reductio ad absurdum arguments that establish the
untenability of that inappropriate presupposition itself. 14
'no'?" (See also Popper's "Indeterminism in Classical Physics and in Quantum Physics,"
British Journal for the Philosophy of Science, Vol. 1 (1950) 117-133 and 173-195 (see p.
183).) The problem with this panicular example is that the temporality at issue is spurious.
For the predictor's incapacity to answer the given question--other than by CAN'T SAYroots in its incapacity to answer its detemporalized counterpart: "If asked 'Is G a theorem?'
will you answer 'yes' or 'no'?" Given the speciousness of its temporalization, the question
at issue is not genuinely predictive. Its incapacity to answer may render that predictor cognitiTJely imperfect, but it is not predictiTJely imperfect.
13. On these issues see the author's Predicting the Future (Albany: State University of
New York Press, 1998).
14. For more on such reduction-to-absurdity argumentation see Chapter 12 below.
155
Of course, spotting that inappropriate presupposition is the

easy part. The difficulty generally lies in the subsequent task of
diagnosis-detecting just what there is about that problematic
thesis that makes for its untenability. Determining a rationale for
relegation to inappropriateness is often the most challenging part
of paradox analysis.

Hilbert's Hotel Paradox
Cantor's Paradox
De Morgan's Paradox
Illicit Self-Reference Paradoxes
The Supertruth Paradox
The Valid Argumentation Paradox
Russell's Paradox of Sets
Curry's Paradox
Russell's Imprcdicability Paradox
Grelling's "Heterological" Paradox
Russell's Relational Paradox
Berry's Paradox
Richard's Paradox
9.1
Cantor's Paradox and the Valid Introduction

Requirement (VIR)
Mathematics is a purely theoretical enterprise. And so, when

things do go wrong here, they will go wrong in a fundamental
way. Inconsistency must be taken very seriously in this domain.
Paradoxes have played a major role in the philosophy of mathematics in the twentieth century. Since all of these mathematical
paradoxes are, in effect, paradoxes of unwarranted presupposition,
this chapter could be seen as a mathematical appendix to its predecessor. For the long and short of it is that the mathematical
paradoxes share the common format of pivoting on a presupposition that something-or-other is meaningful and/or true which
actually is not so.
The mathematics of infinity is a fertile arena of paradox
because it leads to results at odds with the finite point of view.
Laurence Sterne's Tristram Shandy, who spent two years in
chronicling the first two days of his life, lamented that at that rate
he would never be able to finish the job. But a self-perpetuating
committee working at the same rate would be able to write up a
complete account of its own history. No day's activities would
have to be omitted~ven though the work would fall ever more
deeply into arrears. The resources of the infinite make strange
states of affairs realizable.
The so-called "paradoxes of the infinite" arose through the
very mistaken supposition that the groundrules for reasoning
about finite qualities can be carried over intact to the infinite case.
It seems paradoxical, although it actually is not, that a mere part
of an infinite set can be every bit as large as the set itself-that
159
160
Chapter 9
(for example) there are just as many even integers as there are
integers all told because every integer can be paired with it double
as per:
1,2,3,4,5, ...
I I I I I
2,4,6,8, 10, ...
This yields the rather more picturesque Hilbert)s Hotel Paradox.
With an ordinary (finite) hotel, no more guests can be accommodated once the rooms are filled. Not so with Hilbert's hotel, with
its infinity of rooms bearing the numbers of the integers:
1, 2, 3, 4, 5, ...
Even though every room is filled-Room No. i with guest No. i,
for i = 1,2,3, ... -further guests can always be accommodated.
When Messrs. A and B check in, the manager simply directs the
desk clerk to make some room changes: "Let's put A in No.1,
and B in No.2, and then shift Guest No. 1 to Room No.3, and
Guest No.2 to Room No.4, and so on." Although full,
Hilbert's Hotel can always take in more guests. Even an infinity
of new quests can be accommodated via the instruction "Let's
just move Guest No. i to Room 2i," for all i. That certainly
opens up a lot of rooms. And there is no real paradox here-no
actual inconsistency-as long as we do no make the mistaken
presumption that the same arithmetical groundrules (such as N +
1 > N) that hold on ordinary finite quantities will hold for infinite ones as well. l
"How can there possibly be an infinite set-for example, the
set of all integers-when such a set must be complete (contain all
integers) and yet represents a collection that can never be completed, since no matter how many integers it already includes
there are yet more of them to be added?" But this objection is
based on the mistaken idea that a set needs to be specified by
1. On the same basis, somebody With an infinite income (a dollar for every integer)
could payout a like amount in income taxes and still have the initial amount left, dollar for
dollar.
161
some sort of sequential listing of its content and yet cannot admit
a complete inventory which is indeed something that can only be
done with finite sets. The reality of it is that sets can be specified
simply in terms of a condition of membership.
In analyzing mathematical paradoxes we can again map out
the possible approaches to resolution in the usual way by identifying the RjA-alternatives. An illuminating starting point here is
afforded by Cantor's Paradox of set theory. Georg Cantor
(1845-1918), the German mathematician and founder of the theory of sets, proved the important theorem that for any and every
set, the set of all of its subsets-its so-called power set-has a cardinality (mathematical size) that is greater than that of the original set. This means that there can be no greatest set, and
therefore also no set of all sets, since such a supposed total would
at once yield a larger one.
This power set business engenders an aporetic situation for the
idea of a set of all sets, symbolically Ix: x is a set}. For this set
leads at once to Cantor)s Paradox which runs essentially as follows:
(1) For any and every predicate F, there exists the set of all
items that have F-symbolically Ix: Fx} or more fully
(ty) Vz (Fz == z E y).
(2) The characterization " - - - is a set" specifies a predicate.
(3) Hence (by (1) and (2)) the set of all sets exists.
(4) Provably (by Cantor's Theorem): Any set gives rise to a set
oflarger cardinality than itself (namely its power set-the
set of all its su bsets ) .
(5) Hence (by (3) and (4)) the power set of the set of all sets
must have a larger cardinality then that set itself, since it
includes all its subsets.
(6) But a set of cardinality greater than that of the set of all
sets, is patendy impossible.
(7) Hence (by (6)) a set of all sets cannot exist.
162
Chapter 9
Here we have an aporetic cluster formed by the theses {( 1), (2),

(4), (6)). Now (2) is securely fixed in place: to abandon it would
be to abandon set theory. And with (4) and (6) firmly in place as
demonstrable mathematical facts, the only remaining alternative is
to abandon (1) and thereby to accept placing some suitable
restrictive limitation upon the predicates (F) that are capable of
engendering sets answering to the definition "the set of all items
that have F."
From this point of view we realize that Cantor's Paradox is a
paradox of unwarranted presupposition. For thesis (1) is predicated on the (deeply problematic) idea that meaningful predication is of itself sufficient to determine a well-formed set
automatically and without further ado-that no further conditions are required for a predicate to define a set.
An auxiliary observation about definite descriptions is crucial
here. With a putative item characterized as "the x such that the
condition Cx obtains"-symbolically (lX)C-two sorts of things
can go wrong. There can be failures of definition and failures of
identification. A failure of definition occurs whenever we do not
have uniqueness, that is whenever it is not the case that:
3y[Cy&V'z(Cz
~z
= y)]
This means that (lX)CX will be defined only when C demonstrably

applies to a single existing object.
On this basis, (lX) Cx is only meaningful in a setting where it is
established that one and exactly one x is such that Cx obtains, so
that this remains undefined whenever:
(i) The condition Cis uninstantiated and applies to nothing
at all (Cx is always false).
(ii) The condition C is multiply instantiated and applies to
several distinct objects (Cx is true for several values of x).
When one of these default conditions obtains, then (lX)CX is

undefined, indeed nonexistent: there just is no such thing.
And this means further that an attribution-statement of the
format F[(lX)CX] will be:
163
true if there is exactly one item that satisfies C and this item
has the feature F.
false if there is exactly one item that satisfies the condition
C and this item does not have the feature F.
meaningless otherwise-( that is, if there is not just exactly

one item that satisfies the condition C, so that (tx)Cx goes
undefined. (It is, of course, the semantical sense of "meaningless" that is at issue here, with its indication of lack of
determinate truth status.)
We thus arrive at what might be called the Valid Identification

Requirement (VIR):
For something to be appropriately brought upon the stage of
discussion by an identificatory specification of the format "the
x that satisfies the condition C" the availability of a unique
item of this description must be independently pre-established.
For example, in set theory the power theorem shows that this is
infeasible in the particular case of "the set that includes all sets."
For including all sets is something that a set just cannot do.
Accordingly, "the set of all sets" is not a meaningful item. It is a
set in name only, something that is called a set, but does not actually qualify as such because it does not--cannot-admit of meaningful identification. And this very fact that "the set of all sets" is
not intelligibly identified-is a mere pseudo-item-abolishes the
perplexity generated by the question of whether or not it is a
member of itself.
Thus consider the definition of "the set of all items that possess the property F':
Ix: Fx} = (ty)VX(XE
y~
Fx)
For an aspiring set-specification of this format to yield a welldefined set we must pre-establish the unique determination of the
item supposedly at issue.
What we have in the case of failure in this regard is a referential singularity of much the same kind as the arithmetical singu1arity that arises when we attempt to divide by zero. In the former
164
Chapter 9
case the propositional formula F(tx)Cx is not defined, while in

the latter the arithmetical function f(y) = 1. is not defined. This
function has an arithmetical singularity at / = 0; Fy has a referential singularity at y = (tx)Cx (when C is unsuitable). On this basis,
that "set of all sets" is an undefined item even as
is an undefined quantity.
When we divide by zero and proceed as though ~ were a welldefined quantity, we fall into paradox. Thus since 2 x 0 = 0 and 3
x 0 = 0 we have 2 = ~ and 3 = ~ so that 2 = 3! And so we fall
into paradoxical arguments like Augustus de Morgan's comic
demonstration that 1 = 2, which runs as follows: let x = 1. Then
x2 = x. Therefore x2 - 1 = x-I. Dividing both sides of this equation by x -1 we have: x + 1 = 1. But since x = 1, this means 2 = 1.
The fallacy at the root of De MowanJs Paradox of course lies in
dividing by x- 1, that is, by zero.
And the same thing happens when we proceed in inappropriate cases as though (ty)Cx were a well-defined object at a point of
the discussion when this simply cannot be claimed.
9.2
The Successful Introduction Principle (SIP)
Besides the identification failure at issue in the Valid Identification

Requirement (VIR) there is also the prospect of an introduction
failure. That is, if (tx)Cx is to be used to introduce an item-to initiate or introduce it as an item upon the stage of discussion-then
it must not happen that the condition C is such that it refers to
(tx)Cx itself-that is, that (tx)Cx should occur within the formulation of C. For example, let C be the condition such that Cx iff x =
(ty)Cy. Here of course there is no way of assuring that this condition applies to a unique individual-or for that matter to any individual-until we have settled the identity of (ty)Cy-that is, unless
we have already accomplished the very thing that we are in process
of trying to do. Not only circular explanations, and circular definitions, but also circular introductions are by nature futile.
The preceding Valid Identification Requirement accordingly
has the inferential consequence that: The identifying condition for
an item must not involve a reference-explicit or tacit-to that item
itself. This might be called to the Successful Introduction Principle
(SIP). The basic idea is that the classic fallacy of "begging the
165
question" must be avoided not only in argumentation,2 but also

in identification. 3
Now observe that Cantor's putative "set of all sets" not only
violates the Valid Identification Requirement (VIR) because there
is no such thing as a totalized set of all sets, but also violates the
Successful Introduction Principle (SIP) because the C-condition
of the here-operative (tx)Cx specification-namely being an allencompassing set that is also supposed to include (tx)Cx itself-is
such that the SIP-prohibited reference to (tx)Cx occurs in it. That
set-totalization mistakenly presupposes that the set at issue is
already available.
The rationale for SIP stands essentially as follows:
If an item is to be introduced into the agenda of discussion
meaningfully, then its introducing specification must not presume that this item is already available for consideration That
is, if something is to be introduced into the discussion by
means of a defining specification that takes the identificatory
format of its being "the (one and only) item x such that Cx
obtains," then this identifying specification C must not refer to
or quantify over that item itself-however indirectly. In other
words, identificatory self-reference is prohibited: if X is by definition to be (tx)Cx then mention of X cannot occur within C.
(Note that what we have here is a propriety of expository procedure that sets a condition for the meaningfulness of the
resulting discourse.)
Identification must answer our questions about what it is that is
to be at issue; an identification cannot proceed in a circular way
that supposes that we already know the answer. It does not help
to identify someone by the vacuous specification that he is "the
person who is husband to the lady who is his wife." 4
2. See C.L. Hamblin, Fallacies, revised ed. (Newpon News: Vale Press, 1993), pp.
32-35 and passim.
3. The ancients already had a good sense of this, in rejecting such circular arguments
(diallelos logos) as: Where does Theon live? ["Identify for me the place where Theon
lives".]; He lives where Dion lives; And where does Dion live? He lives where Theon lives.
(See Prant!, Geschichte, Vol. I, p. 492.)
4. Observe that this principle has as a corollary the Illicit Totalization Principle (ITP):
To be meaningfully and viably introduced into the discussion (that is, adequately specified,
identified, defined) a totality (collection or whole) must not be purponed in its introducing definition to include itself. That is, if X is by definition to be (\.X)G:, the totality of
166
Chapter 9
While violation of the Valid Identification Requirement (VIR)

leads to a failure of reference, violation of the Successful
Introduction Principle (SIP) leads to a failure of specification. In
the former use there is no such thing; in the latter, no proper identificatory introduction has been effected-irrespective of any questions of existence. In neither event do we "know what we are
talking about." When these principles are violated, what we have is
not an inferential defect, a logical fallacy, but rather a referential
defect, a seman tical error. Referential vacuity is just exactly thatthe literal absence of a determinable object of discussion. Byanalogy, consider the proposition: "The statement in the (adjacent)
box is complex." Now there just is no statement in the box!-and
there is nothing for the statement at issue to be about. We have

as much reason to say that "it" is false as to say that "it" is
true, there simply being no it at issue. That supposedly identifying expression "The statement in the box" is vacuous and
claims about "it" (that it is true, or false, or complex) are literally
meaningless.
This Successful Introduction Principle (SIP) swings a powerful ax that fells a wide range of paradoxes. The basic idea that
underlies this principle is already encountered in a tract attributed
to William of Occam (1285-1347). This stipulates that in institutio, the inauguration of a term into the setting of a discussion
(disputatio), one must follow the rule numquam pars potest significare totum ("part [of an expression] can never signify the
whole"). This principle was illustrated via the statement "Every
statement is true" (Omnis propositio est vera), part of whose formulation ("every statement") brings the whole of itself-the
entire statement at issue-within its own range of designation.
objects answering to the condition C, then X must not itself be one of the items whose
status in relation to C must be settled enroute to setting the boundaries of (lX)Cx itself.
This principle is, in its effect, closely akin to Russell's Vicious Circle Principle (VCP)
already discussed in the preceding chapter. It too can serve to sideline various mathematical paradoxes.
167
The Occamite author then applies this principle to the solution of

the paradox engendered by:
(A) A is false.
Here the specification of A must be rejected (institutio non est
admittenda) because A's self-invoking specification violates the
principle at issue: A significat hoc totum A significat falsum" is an
illicit specification (institutio). For the idea that "A stands for 'A
is false'" leads straightaway into the paradox (sophisma) that A is
true if false and false if true. 5 This Occamite analysis is yet another
example of the subtle and original contributions of thirteenthcentury logic. Its focus on the introduction of terms rather than
the existence of things puts our medieval author's approach ahead
of the treatment of cognate ideas in the early twentieth century
(as, for example, Bertrand Russell's Vicious Circle Principle which
will be discussed shortly). 6
It should be observed that the rationale of VIR and SIP is not
that paradoxes can result from their violation. (That is simply an
ex post facto aspect of their nature.) Their justification lies in the
fundamental prior considerations of the same qua non condition
under which processes like identification and introduction can
conceivably accomplish their intended work. What is at issue are
fundamental principles of rational procedure and not just ad hoc
devices for paradox-elimination.
9.3
Pseudo-Objects
To see how pseudo-objects which violate the principles under

consideration can engender paradox let us consider in detail the
idea of a supertruth that encompasses all truths'? Such a
5. See Prantl, Geschichte, Vol. IV, p. 42. The text cited by Prantl deserves closer analysis than it has received to date.
6. Occam was criticized in the fourteenth century on exactly the same point on which
Russell was criticized in the twentieth, namely that his strictures ruled out harmless selfreference ("Some propositions are true" or "All propositions can be discussed") along
with the paradoxical ones ("All propositions are false"). See Ashworth 1974, p. 105.
7. Note that the infeasibility of such a truth would also make for a negative answer to
the question: "Is truth-as-a-whole finitely axiomatizable?"
168
Chapter 9
supertruth would be a truth S such that: 'Itp[ Tp ~ (S ~ p)]. So the

question at hand comes down to the tenability of the thesis:
( 1) :3 q [q & 'ItP [p ~ (q ~ p)]
Adopting the idea of a supertruth implies that we should settle

this question via the definition:
(2) T= (tq)[q & 'ltp [p ~ (q ~ p)]
Now for this definition to satisfy VIR we need to establish both

uniqueness and existence. Here uniqueness is no problem. It is
readily shown that any two propositions q' and q" meeting the
condition at issue are relevantly identical (i.e., logically equivalent). But this existence claim at issue with (2 )'s VIR satisfaction
reverts to begging the very question at issue. That is, (2) could
only accomplish its intended missions if an affirmative answer to
the question of (1) 's tenability were already in hand.
Moreover, with (2) we are also plunged into problems in relation to SIP. For note that the formulation of what follows (tq) in
(2) involves a universal quantifier with respect to propositions, viz.,
'Itp. So T itself would here have to qualify as one of the values of
the relevant variable. And this means that SIP is also violated.
The principles with which we have been dealing thus rule out
the idea of a supertruth on grounds of its illegitimate self-encompassing nature. But is such an embargo actually necessary to avoid
paradox? Or are those principles blocking us off from something
that is actually unproblematic?
The paradoxical character of the supertruth idea emerges from
the apory:
(1) A supertruth entails all truths.
For every truth there is yet another dominator-truth, that is, a
truth that entails it but that it does not entail. The following two
illustrations demonstrate this fact-in each case the proposition at
issue dominates t:
t is a truth whose implications no one realizes in their

full detail
169
t is a truth of which someone may possibly not be

aware.
t is a truth that has consequences which some people

would probably deny.
(1) Now this being so, there will also be a dominator-truth

for that supposed supertruth S.
(2) But if a truth is dominated by another it cannot entail all
truths.
The best resolution of this Supertruth Paradox lies in seeing it as
a reductio ad absurdum of the very idea of a super truth S. And
the fact of S's violation of the meaningfulness requirements at
issue not only authorizes but enjoins this rejection.
It warrants note that the SIP identification principle applies to
items of all sorts and not just to objects or to propositions. For
example, arguments also come within its purview. Thus consider
the argument
(I)
Argument (I) is valid

Therefore: Argument (I) is sound
Recall that a valid argument is one whose premisses entail its conclusion, and that a sound argument is a valid argument with true
premisses. But now consider the following Valid At;gumentation
Paradox:
(1) A valid argument cannot have a conclusion that claims
more information than is provided by its premisses.
(2) By the thesis (1), (I) is not a valid argument.
(3) (I) is a valid argument, for there is no way for its premisses
to be true and its conclusion false (which is the definition
of argument validity).
Chapter 9
170
The exit from paradox here lies in noting the argument's failure
to distinguish between suppositional (or conditional) truths and
categorical truth. Moreover-and this is the presently crucial
point-the salient flaw of (1) lies in its blatant violation of the
Successful Identification Principle (SIP).
9.4
Russell's Paradox
One of the most notorious paradoxes encountered in twentiethcentury logic and mathematics is RusseWs Paradox, based on the
idea of "the set of all sets that do not contain themselves," symbolically {x: x ~ x}. (Here E is set membership and ~ its denial.)
We thus envision a certain set R supposedly defined through the
following specification of its membership:
By definition (of R): R is the set of all non-self-containing
sets, so that S E Riff S ~ S
This specification is supposed to provide a contextual definition of
R (in the specific context of membership attribution). And perplexity arises here when we ask whether or not R contains itself.
For if R e R, then by the definition of R we have R e R. And if
R ~ R, then by the definition of R we have R e R. Thus contradiction results either way.
Given these considerations, we confront the following aporetic
situation:
(1) R as defined by the preceding specification exists as a welldefined set-that is, one such that any existing item either
is or is not a member of it.
(2)
ReR~R~R
(3) ReR
(4)
R~R~RER
(5) ReR
by the definition of R
from (2) by standard logic
by the definition of R
from (4) by standard logic
171
The resolution of this paradox is, in a way, straightforward. Since

the whole course of argumentation pivots on (1), the only exit
from inconsistency has to proceed via its rejection.
When we ask a question like "Does the set R include itself or
not?" we presuppose that the set R exists. But what the argumentation of the Russell paradox actually shows is simply that it does
not-that this essential presupposition of the question is not satisfied. There is no alternative to taking the line that the Russell set
R is simply not a well-defined set. The only real question is why
this is so-that is, how it comes about that R's seemingly straightforward definition goes awry and is illegitimate. It is not R's
demise that is in question, but the result of the autopsy, the question of why it is that R cannot survive a proper scrutiny of its
meaningfulness.
The difficulty with the Russell set R is that it is introduced into
the arena of discussion by a definition of the form:
the set that includes all sets of a certain type (viz., non-selfinclusive sets).
We could conceivably use such a formula to describe a given set in
the range of sets if it indeed existed and answered to that description. But the SIP requirement means that we cannot introduce it
on this basis. For such an introduction to be appropriate we
would need to pre-establish the existence of a (unique) set x such
that
Vy(y E xiffye: y).
And the argumentation of the Russell Paradox shows that no such
set can exist. The fact is that the Russell set is an inappropriate
hypostatization (or an identificatory failure), because its introduction on the stage of discussion proceeds by a defining specification that is inappropriate exactly because there can be no such set.
It warrants note that from the formal point of view Russell's
paradox of the set that contains all but only sets that are non-selfbelongers strictly parallels the paradox of the barber who shaves
all but only persons who are non-self-shavers. Both paradoxes rest
on pretty much the same mis-presumption of the existence of
their pivotal item.
172
Chapter 9
The early Russell adopted an approach quite different from

that taken here. He took over from the French mathematician
Henri Poincare a principle he enunciated in 1906 and which
Russell following him characterized as the
Vicious Circle Principle (VCP): No collection (whole or totality) can contain members that are defined in terms of itself:
specifically, no existing collection can ever be a constitutive
part of itself. 8
As it stands, this principle is clearly a limitation upon the constitution of actual totalities-and a very strong limitation at that. For
Russell, saying that such a "collection" "has no total," is to say

that it does not exist as a collection. What we have here is a
restriction on the sorts of collections that can exist-that is, upon
how authentic collections can validly be constituted.
The present approach, by contrast, pivots matters on the SIP
Principle. It takes the line that only when something is properly
identified, can it serve as a subject of meaningful discussion. For
when we do not know what we are talking about, we do not
know what is or what is not to be said about it. With an inappropriately identified pseudo-object z it is every bit as plausible (or,
rather, implausible), to claim F(z) as not-F(z). The doorway to
8. See Bertrand Russell and A.N. Whitehead, Principia Mathematica, Vol. I

(Cambridge: Cambridge University Press, 1910 [re-issued in paperback, 1967]), pp. 31,
37. Elsewhere Russell puts it as follows: "If provided a certain collection has a total, it
would have members only definable in terms of that total, then the said collection has no
total." American Journal of Mathematics, vol. 30 (1908), pp. 222-262 (see p. 240).
Russell went on to explain that "When I say that a collection has no total, I mean that
statements about all its members are nonsense." (Ibid.; see also Principia MathematiclJ,
Vol I.) He appears to think, however, that the culprit here is "all" instead of "its"! For he
stresses that "Whatever involves all of a collection must not itself be one of the collection"
("Mathematical Logic as Based on the Theory of Types" (1908) reprinted. in J. van
Heijenoort, From Frege to Giidel [Cambridge, MA: Harvard University Press, 1967], pp.
153-182; see p. 155.) Elsewhere, Russell is more cautious "If, provided [we assume] a
certain collection has a total, it would have members only definable in terms of the total,
then the said collection has no total." (Bertrand Russell, "Mathematical Logic as Based on
the Theory of Types," American Journal of Mathematics, vol. 30 [1908], pp. 222-262;
see p. 227; cf. Principia Mathematica, Vol. I, pp. 31, 36.) The phrase "only definable in
terms of" helps matters but also introduces problems of its own, since what is to say that
something not definable in one way cannot possibly be defined-by hook or crook-in
some other way? All of Russell's formulations sound as though he thought that a problematic collection is there all right, but we must avoid making claims about all of it. This
seems quite wrong-headed.
173
contradiction and paradox is thrown wide open. Only that which

has been identified can be said to have a discernible descriptive
identity. Only when we know what the "it" at issue is can we
meaningfully attribute any descriptive features to it.
Accordingly, the line of argument contemplated here via the
Successful Introduction Principle (SIP) is quite different from
Russell's VCP even though it accomplishes much the same work.
For it does not address the ontological issue of what sorts of sets
or collections do or do not exist. Rather, it addresses itself to the
communicative issue of the conditions that must be met for a purportedly identifying description of an object to succeed in specifying a well-defined referent and thereby presenting a meaningful
object of discussion. All in all, SIP affords a more natural, less
restrictive and more economical way of sidelining the paradoxes
than the Russell-Poincare Vicious Circle Principle (VCP) puts at
our disposal. On this basis, the issue is at bottom not one of the
ontology of what does or does not exist, but one of the semantics
of what sort of statements can or cannot meaningfully be made.
9.S
Paradoxes Related to Russell's: Curry's

and Grelling's
Curry)s Paradox is best seen against the background of Russell's

Paradox. Instead of contemplating the Russellian "set of all sets
that are not members of themselves, {x: -(XE x)J, Curry contemplates the set of all sets such that if they are members of themselves then an arbitrary proposition obtains:
{x: (x E x)
P}
This enables Curry to recover Russell's paradoxical result on a

negation-free basis when p is self-contradictory. In view of the
absurdity of what follows the implication sign, this paradox is an
effective variant of Russell's, and is amenable to essentially the
same de constructive analysis.
In the original 1903 edition of his Paradoxes of Mathematics
(Cambridge University Press), Bertrand Russell introduced his
Impredicability Paradox. This is based on the idea of impredicable
predicates, namely those which, unlike "abstract" or "intelligi-
174
Chapter 9
ble," cannot be predicated of themselves. And he maintained that

the predicate "impredicable" cannot itself be classed as a predicate
on the basis of the following argumentation: "Let us assume that
'not-predicable of itself' is a predicate. Then to suppose either
that this [supposed] predicate is, or is not, predicable of itself, is
self-contradictory. The conclusion in this case seems obvious: 'not
predicable of oneself' is not a predicate ... " (p. 102). To allow it
to qualify as such would be to invite a contradiction. Russell's
analysis thus successfully averts the paradox.
The "Heterological" Paradox of the German mathematician
Kurt Grelling is an in-name-only variant of Russell's Impredicability
Paradox. 9 1t pivots on the idea of a heterological predicate. A predicate (or adjective) is autological if it applies to itself (for example,
the predicate "intelligible" is intelligible and the predicate "common" is common). Otherwise, a predicate is heterological; for
example, "green" is not green nor is "prehistoric" prehistoric.
Thus we have the defining specification: F is a heterological predicate iff it is not self-applicable:
(Het)
For any predicate F: H( F) iff - F( F)
Now Grelling)s Heterological Paradox pivots on the question: Is

"heterological" heterological or not? Letting the F of the preceding specification be H itself we arrive at
H(H) iff -H(H)
We arrive at a paradox. The predicate "heterological" is heterological (and thus self-applicable) iff it is not so.
The scope for resolution here is narrow, since there is nothing
to attack save (Het) itself. Our best bet here is to take the line
that it rests on an erroneous presupposition, namely the supposition that H as a predicate of predicates is eligible for inclusion
within the range of the variable F at issue in the preceding specifi-
9. On Grelling's Paradox see Kurt Grelling and Leonard Nelson, "Bemerkungen zu

den Paradoxien von Russell und Burali-Forti," Abhandlungenden der Fries'schen Schute, N.
5., vol. 2 (1908), pp. 301-334. See also Hermann Weyl, Das Kontinuum (Leipzig: Veit,
1918).
175
cations. For SIP precludes that H is a something that lies within

the range of the universal quantifier 'V that is at issue when we
stipulate: 'V F [H(F) == -F(F)]. The point is that in adopting Het
we have a "creative" definition that superadds something to a presumably pre-determined range of predicates:
the predicate of predicates which applies to a given predicate wherever that predicate does not apply to itself.
The flaw here is not self-involvement as such but self-involvement
in the context of an initiating-definition or item-specification. For
this straightaway leads to a violation of the Successful
Introduction Principle (SIP).
It warrants noting that Grelling's Heterological Paradox of the
"predicate that applies to all non-self-applicable predicates" is also
closely parallel to Russell's Paradox of "the set that contains all
non-self-inclusive sets." The former looks at the issue intentionally where the latter sees it extensionally. For consider any specifiable feature F of things. We will then obtain Ix: Fx}, the set of all
items that have F, and also ('A.x)Fx, the property that all F-bearing
things share in common. The Russell Paradox turns on the set of
all sets that are not self-inclusive, Ix: x j!: xl, while Grelling's
Heterological Paradox-as indeed Russell's own Impredicable
Paradox-turns on predicate applicable to all predicates that are
not self-characterizing, CAP) -F(F). So the difference is, in effect,
merely stylistic. In each case the same pattern obtains. And so,
both lend themselves to the same deconstructive analysis to the
effect that the paradoxical item at issue (supposedly a well defined
set in the one case and a meaningful predicate in the other) does
not and cannot qualify as appropriate because the Successful
Introduction Principle (SIP) is once again violated. In each case
the property F at issue is one whose formulation supposes the
availability of the very item in question. For in the Russell case it
is "the set of all sets, itself included, that have the predicate F"
and in Grelling's case it is "the predicate of all predicates, itself
included, that have the property F."
Grelling's paradox shifts the Russell paradox difficulty from
sets to predicates. And Russell himself extended his finding that
sets of sets can engender paradoxes to the idea that relations of
176
Chapter 9
relations can also do SO.10 RusselPs Relational Paradox projected

the idea of a special second-order relation T which obtains
between two relations (R and S) wherever it is not the case that R
obtains between Rand S.
(T) By definition (of T): T is the relation such that for any
two relations Rand Swe have T(R, S) iff -R(R, S):
T = (t V) VR VS [ V( R, S)
R( R, S)]
The question now arises: Given a relation S, does the relation T

hold between itself and S? Note that in virtue of definition (7) we
will have it that the substitution of Tfor R yields:
T(T, S) iff -T(T, S)
And this is of course paradoxical: it means that T( T, S) can neither obtain nor fail to do so.
There is but one way to resolve the paradox that arises here:
We have to abandon the idea that (T) affords the meaningful
specification of a relation. This being so, T will not belong to the
range of the variables at issue (R and S) and so the substitution of
T for R, which engenders the difficulty, becomes unraveled as a
meaningful inferential step. And so this paradox-like its predecessors-also falls to the ax of the Successful Introduction
Principle (SIP).
9.6
Berry's Paradox and Its Ramifications
An interesting paradox was credited to the Oxford University

librarian G.G. Berry by Bertrand Russell, who first published it in
1905Y Berry's Paradox emerges from the following narrative:
The integer 1 can be identified (in standard English) by a single word ("one"), whereas the number 10 exp.lO cannot.
10. Bertrand Russell, The Principles of Mathematics, revised ed. (New York: Norton,
1938), pp. 102-03, section 102.
11. Russell 1908, reprinted in Hiejenoort 1967.
177
What then of "the least integer that cannot be identified in

fewer than fourteen English words"? This clearly poses an
anomaly. For we have just identified it in thirteen words.
The paradox that arises here roots in the following aporetic
cluster:
(1) Some number or other is bound to answer to the identifying descriptive expression "the least integer that cannot be
identified in less then fourteen English words."
(2) In view of(1), the descriptive expression "the least integer
that cannot be identified in less then fourteen English
words" identifies this integer.
(3) But-contrary to this contention-the descriptive expression "the least integer that cannot be identified in less
then fourteen English words"-which has only thirteen
words-successfully identifies the integer.
The only practicable exit from inconsistency here lies in insisting
that the purportedly identificatory "least integer" expression fails.
And it does so because its use in the present paradox rests on a
mistaken existential presupposition.
And this seems plausible enough. For Berry's paradox could
be construed as a reductio ad absurdum of the idea that there is
such a number as "the least integer that cannot be identified in
fourteen English words." After all, suppose that there were such
an integer. Then let it be printed out in some (conceivably sizable) book to be entitled The Number Tome. Then the expression "the integer printed out in The Number Tome" will identify
this number, few though its words be. Only if the allowable ways
and means of number specification are specified clearly from the
outset will the theses that constitute Berry's Paradox be meaningful-and once this is done the contradiction at issue dissolves.
This way of addressing Berry'S Paradox was argued by
Alexandre KoynS, who emphasized that there are two very different ways of specifying an integer, first, via a naming-formula of
the standardized format for numbers ("five thousand six hundred
and ninety-two"), and second, via some identifying condition that
178
Chapter 9
this number alone satisfies (such as "the number of grains of sand

in the Sahara desert").12 The former (naming) would generally
require a long expression for number specification, but the latter
(identifying) might generally be achievable by compact means.
The paradox thus succumbs to a dilemma. With technically standardized number-naming, premisses (3 )-( 4) are true but (5 )-( 6)
false, while with unformalized number-identification, (5 )-( 6)
would be true but (3 )-( 4) false.
It is, however, possible to recast Berry's paradox into a more
challenging formulation. For consider the following line of
thought:
(1) Every finite text that can be presented in an alphabetized
natural language such as English is contained in the textseries which first presents in alphabetic order all possible
texts of one-letter length, and then all texts of alphabetic
length two, and then all texts of alphabetic length three,
and so on.
(2) Since any particular (positive) integer can be identified in
standard English by a finite text, it follows that any and

every particular integer will eventually be identified by
some text in this series.
(3) The number of integers that can be identified by texts of
length less than n is always finite because the number of
such texts is finite. This means that the (infinite) set Zn of
all integers that cannot be identified by a text of length
less than n will always exist.
(4) Like any other set of integers, finite or infinite, such a set
Zn will always have a smallest member Zn *. In the circumstances this will be the smallest integer that cannot be
identified in standard English by a text of alphabetic
length n or less.
12. Koyre 1946. However, instead of the present line that any integer can be identified very briefly in the second mode, Koyn:'s resolution takes a different line: "La question: 'peut-on ou ne peuton pas "nommer" un nombre donne en tent de syllables?'-ne
comportera plus de reponse, n'offrant plus aucun sens determine" (p. 17).
179
(5) Specifically, if n is 1,000 then z* 1,000 will-in virtue of its

construction-be "the smallest integer that cannot be
identified in English by a text of fewer than thousand
letters."
(6) However-and here lies the paradox-we have just identified the integer z* 1000 by far fewer than a thousand letters,
namely just exactly as "the smallest integer that cannot be
identified in standard English by fewer than a thousand
letters."
Here there is no easy escape from acknowledging the quantity at
issue. To do so, one must here tackle the most problematic premiss-number (3 )-by a frontal assault.
Perhaps the most plausible way to effect this break in the chain
of inconsistency is to introduce the distinction between direct and
indirect integer-identification. An integer is directly identified if it
is specified either by its standard name or by an identifying formula that it alone instantiates such as "the fourth power of 35."
It is indirectly identified when it is specified by some contingent
relationship such as "the integer written on the piece of paper at
such-and-such a location" or "the integer that John mentioned
on such-and-such an occasion" or "the integer identified in suchand-such a way in such-and-such a language." Clearly, any integer
whatsoever can in theory be identified by a brief text of the indirect type. So premiss (3) is tenable only with respect to direct but
not with respect to indirect identification. However the claim of
Zn *Js existence at issue in (4) - (6) rests upon a process of indirect
identification. The paradox can thus be resolved by noting the
crucial equivocation at issue with its key idea of integer-identification. That purportedly identifying expression of premiss (6) simply does not identify.
The point becomes even clearer when viewed from another
angle. Consider the sentence in the following box:
Now is the time for all good men to come

to the aid of the party.
180
Chapter 9
How many words must I use to instruct you to write this sentence?
There are various alternatives here:
(1) Write the sentence: "Now is the time for all good men to
come to the aid of the party."
(2) Write the standard typing practice sentence.
(3) Write the sentence in the box.
Short of specifying the eligible ways of presenting the instruction
at issue, that italicized question is simply not well-defined. It is
like asking you how many dollars must be spent in getting
from New York to Los Angeles without indicating whether this is
to be by boat, on foot, by train, by plane, by cab, and so forth.
The question as it stands is lost in a fog of meaning-obscuring
indefini teness.
One can also question premiss (3) from another angle, namely
that of the difference between the prospective and the retrospective standpoint. Consider "Every book has a title." This is true
(or at least presumably so) retrospectively: every book we have
dealt with so far has a title. But it is not true-or at least not true
yet-for the books of the future that have not yet been produced,
let alone titled. And similarly with "Every number has a name."
This is certainly true of any and all real numbers that we specifically consider. But since the real numbers are uncountably infinite
while the available names are at best countably infinite, there are
too few names to go around. Any and every number that we
might come to discuss can (ipso facto) be named and identified.
But this is not true for all numbers in toto. (If we have only ten
minutes, I can introduce you to any person in the ballroom, but
not to every person.)
The perplexity at issue is also akin to that of the puzzle: What
are we to call some thing that has no identifying name? WeHlet's adopt the convention: It shall be Nameless. But now-Io and
behold!-it is not this any more. But the fact is that there is an
equivocation going on here, as between the level of discourse at
the initial state of things antecedent to the renovation of terminology and the ultimate state of things that ensues thereafter. In
the post-designation circumstances, there indeed is something
that has no name; in the post-designation circumstances, there is
181
not. Contradiction is thus averted by drawing a suitable distinction (just as Aristotle said it has to be).
Accordingly, we need not look to mathematics for paradoxes
of this sort. For consider the purportedly identifying expression at
issue in the following narrative:
Some persons are mentioned in a written text of some sort-a
biography or a "wanted" bulletin or perhaps a birth certificate.
Other persons are never mentioned in any written text-like a
remote Eskimo fisherman or Anatolian shepherd, they live out
their existence in unmentioned obscurity. But now consider
the descriptive expression: The oldest living textually unmentioned individual. Note the anomaly that we just now haveto all appearances-managed to mention this individual.
This paradox is essentially of the same type as that of Berry. A
similar mistaken presupposition is at work, seeing that we once
more have a purportedly identifying expression that violates the
Successful Identification Principle (SIP) which effectively embargoes the problematic idea of defining an "unmentioned individual" as "one not adverted to in any text-that of the present
discussion included."
9.7
Richard's Paradox
The perplexity known as Richard's Paradox emerges from the following line of thought. 13
Let D be the set of decimals that can be specified in a finite
number of words. We can, of course, present this set by means
of an infinite list, by listing first (in alphabetical order) all those
decimals that can be specified by one word (such as zero =
.0000 ... ), then by two words, and so on. But now construct
a new decimal as follows: Begin with the first decimal in the
list and look at its first decimal place. Should this be n, then let
13. Richard's paper "Les principes de la mathematique et la probleme des ensembles"
was originally published in the RwuegenerllJe des sciences pures et appliquees in 1905. It is
translated in Heijenoort 1967, pp. 143-44.
182
Chapter 9
the first decimal place of our newly constructed decimal be

n + 1, unless n is 9 in which case use O. Then go to the second
decimal in the list and look at its second decimal place. Should
this be n, then let the second decimal place of our newly constructed decimal also be determined by the preceding rule.
And then on to the third decimal on the list with the same
process. What we are thus constructing is (1) a decimal that is
specified by a finite number of words (note the utility of "and
so on" here"), but (2) that nevertheless cannot be present on
that initial infinite list of all of the members of D.
One way of attacking this paradox is to deny its (tacit) premiss
that any given (finite) verbal formula can only specify a certain
number and not others, and in any case only a limited number of
numbers. There just is no such thing as "the maximum number
that an expression of length n can identify. Think, for example, of
the formula: "The number of numbers which I am thinking
now," or "The number written in the book on the table." (Recall
the preceding discussion of Berry's paradox on the equivocal
nature of "number identification.") Accordingly, the Successful
Introduction Principle (SIP) is again violated. Giuseppe Peano
put his finger on the crux of the problem in objecting: "Richard's
example does not belong to mathematics but to linguistics. For an
element that is fundamental in the definition of [the diagonalized
number] N cannot be defined in an exact way via mathematical
rules."14 After all, that initial starting list of "decimals specified by
a given number of words" is not-and cannot be-generated by
arithmetical means alone, but calls for settling mathematically
imponderable matters of linking verbal formulas to numbers
where these verbal formulas belong to unformalized rather than
rigidly formalized languages-in much the same way as with the
previous discussion of Berry's paradox.
However, Richard's Paradox has interesting larger ramifications. For consider the following contentions:
(1) With real numbers, to be is to be arithmetically identifiable.
It is inappropriate here to claim the existence of
14. Heijenoort 1967, p. 142.
183
something that lies outside the realm of arithmetic

tractability.
(2) In view of (1), formalized real number arithmetic is adequate to the real numbers.
(3) The identification-formulas that can be stated in any formalized language (that of real-number arithmetic
included) can be set out in toto in a suitably constructed
infinite list. Hence the formulas available for the possible
specification of real numbers are countable.
(4) It can be demonstrated (by a diagonal argument due to
Georg Cantor) that the real numbers themselves cannot
be fully enumerated and that their number is uncountable.
(5) In view of (3 )-( 4), formalized real number arithmetic is
not adequate to the real numbers
Here (1), (3), (4) constitute an inconsistent triad.
How should this inconsistency be addressed? Since demonstrable facts are at issue with (3 )-( 4) while (1) is at best a plausible
claim, we should-in this context!-be prepared to abandon (1).
The result would be to accept the idea that as far as the arithmetical systematization of real numbers is concerned explicit identification is one thing and actual existence another-and a greater.
Real numbers are uncountably many in quantity: the number
of reals is nondenumerably infinite. However, the number of number specifications--of number "names" that can be articulated in a
formalized system-is countable. The real numbers play musical
chairs with the number names-there just are not enough names
to go around. There is nothing wrong with the idea that even for
real numbers To be is to be nameable-to be arithmetically identifiable. It is clear that any real number could be named/identified.
But not every real number is ever named/identified. Ifwe said that
for real numbers, to be is to be named (instead of nameable) we
would severely underestimate the reals.
Painful or not, we are impelled towards the idea that there is
more to the manifold of real numbers-informally understood-
184
Chapter 9
than any axiomatically formalized systematization of the arithmetic involved is able to capture explicitly. IS
9.8
The paradox due to the Italian mathematician Cesare Burali-Forti

involves mathematical complications that make it too technical for
detailed consideration here. I6 The basic idea is that the series of
all ordinal numbers, duly arranged in order of magnitude, must
itself have an ordinal number, say Z. But it can be shown that the
series of all ordinals up to and including any given ordinal is
greater than that ordinal itself (specifically that it is that ordinal
plus one). And this would mean that Z has to be greater than
itself (by one). And the arithmetic of ordinals is such that there, as
elsewhere, this sort of self-surpassing is impossible.
The most readily practicable line of attack here is to abandon
the idea that every series of ordinals must itself have an ordinal
and specifically that this is the case with the series of all ordinals.
After all, "the ordinal of the set of all ordinals-itself included" is
a specification that once again violates the Successful Introduction
Principle (SIP). The upshot would be rejection of a totalization
ordinal for the set of all ordinals. (Compare the discussion of
illicit totalization in other contexts such as the Kantian antinomies
(above, pp 147-51).) But these issues are too technical for further
elaboration here.
9.9 Addressing the Mathematical Paradoxes

In the course of the twentieth century, mathematical logicians
have proposed a variety of devices for dealing with the paradoxes
arising in this domain. In each case their general strategy has been
the same: the articulation of a rationale for holding that one or
another of the propositions essential to the paradox is untenable.
15. For a more detailed discussion of this issue see Alonzo Church, "The Richard
Paradox," American Mathematical Monthly, vol. 61 (1934), pp. 356-361.
16. See Cesare Burali-Forti, "Una questione sui numeri transfinite," Rendiconti di
Palermo (1897), trans. Heijenoort 1967.
185
And in the search for such a rationale they have been guided by
the "many birds with one stone" guideline of seeking for a mechanism that can at one stroke accomplish the needed work with
various-and indeed ideally with all-of the paradoxes. This
search has not been entirely successful, but useful progress has
nevertheless been made. In particular, following Peano, F.P.
Ramsey distinguished two types of paradoxes: (1) the mathematical invoking the concepts of set, element, or number, and (2) the
semantical invoking the concepts of truth, reference, or identificationY The former, so Ramsey held, could all be handled by the
simple theory of types (without type ramification and the axiom
of reducibility). And latter could be handled by a variety of
restrictions upon proper linguistic usage. This point of view has
been widely accepted.
But another option can also be envisioned that dispenses with
the cumbersome machinery of type theory. For as regards the
mathematical paradoxes themselves, these can be divided into two
main groups:
The Cantor-Russell Style Paradoxes
This group includes Cantor's Paradox, Russell's Paradox,

Curry's Paradox, Grelling's Paradox, and Russell's
Relational Paradox. The crux of the problem here is the
vitiating self-involvement arising from the combination of
self-reference with totalization at issue in identificatory
specifications like "the set of all sets such that," "the property of all properties such that," "the relation of all relations such that," "the number of all numbers such that,"
etc.).
The Berry-Richard Paradoxes
These include Berry'S Paradox and Richard's Paradox.

The crux of the problem here is that there is a discrepancy
between the intra-systemic identification of numbers in a
formalized system and their externalized identification by
informal means.
17. See Quine'S account in Heijenoort 1967, p. 152.
186
Chapter 9
Now the paradoxes of the second group are not too alarming. For
one thing, they can be averted as actual contradictions by observing that they rest upon inappropriate presuppositions regarding
the relation between the formalized definitions of numbers and
their informal, extra-systematic specification. Moreover, these
paradoxes call into question not so much the consistency of formalized mathematics as its completeness in relation to looser, informal ideas. They indicate a G6delian discrepancy between what can
be said within a formalized system of arithmetic and what can be
said about it by externalized means. And they indicate that the
second will always press beyond the limits of the former: that no
formalization can capture the whole of what can truly be said
about arithmetic.
However, the paradoxes of the former group are threatening
to the very consistency of mathematics itself. And a variety of farreaching strategies have therefore been proposed for their elimination, preeminently the following four: 18
I.
Self-Reference Proscription (Early Russell)

No statement is to be accepted as meaningful and proper
that involves a reference to itself, however oblique.
II.
The Vicious Circle Principle (Poincare, Later Russell)

A supposed item of any particular sort ("the set of sets
such that ... ," "the predicate of predicates such that
... ," or the like) cannot exist if its specification refersdirectly or obliquely-("all sets belonging to it," "all
predicates bear it" etc.) to that item itself.
III. Theory of Type Levels (Russell-Whitehead)

All properly defined statements about objects possess a
type level. And a feature being attributed to an object can
make mention only of items which themselves are of a
18. For the relevant discussions see Heijenoort 1967.
187
lesser type then this object, so that every meaningful assertion about a totality is of a higher type than that totality
itself. Accordingly, no meaningful statement about a set
can attribute to it any members that are not of a lesser
type. (Note that this precludes the meaningfulness of
x EX, which attributes to an object (x) a feature (elementpossession) that mentions an object of equal type-level,
namely x itself.)
IV. Language Level Theory Language/Metalanguage
Distinction (Tarski)
Discourse always proceeds at a particular level in a hierarchy of languages. And talk about an object at some level
automatically shifts what is said to the next level. Hence
expressions like "All statements (or all predicates) such
that ... " must be construed as "All level i statements (or
predicates) such that ... " And this statement, being at
level i + 1, does not--cannot-refer to itself. Statements
of the format "This statement is ... " are automatically illformed-in effect non-statements.
In the context of this fourth approach, Tarski was able to show
that if the usual logical principles are accepted, then no language
is semantically self-encompassing in the sense of providing the
means for discussing the truth and falsity of its own expressions.
For with those conditions satisfied it will always be impossible to
construct an assertion of the Liar-Paradox type which attributes
falsity to itself. Given that a coherent language is inadequate to
self-characterization we are driven to a hierarchy of languages of
different levels where we can ascend to level n + 1 to discuss a
language of level n, but where the means of totalization to all levels at once is not and cannot be made available. In particular we
cannot even speak of generalization as being true or false flat-out,
but only as being so at this or that level of the hierarchy of languages and meta-languages. However, this paradox-averting
upshot is purchased at the great price of forgoing the unencumbered generality of discourse that we would clearly like to have in
mathematics as elsewhere.
188
Chapter 9
9.10 The Efficacy of the Successful Introduction

Principle (SIP)
The problem common to all of the standard avoidance-devices for
the mathematical paradoxes that we have been considering is that
they all invoke a large-scale and cumbersome restriction on the
meaningfulness of language. In this regard the present chapter's
focus on how items can meaningfully be brought upon the stage
of discussion (namely by identifications that do not violate the
Successful Introduction Principle) has a substantial advantage in
point of simplicity and naturalness.
Consider once more the item-specifications at issue in these
mathematical paradoxes:
the set of all sets-itself included (Cantor's Paradox)
the set of all non-self-containing sets (Russell's Paradox)
the set of all sets the claim of whose self-inclusiveness
allows an "articulating assertion to be made" (Curry's
Paradox)
the property characterizing all properties that are not selfcharacterizing (Grelling's "Heterological" Paradox)
the relation that obtains between all pairs of relators wherever the former of these does not hold between itself and
the second (Russell's Relational Paradox)
the ordinal of all ordinals-itself included-that satisfy a
certain condition (Burali-Forti's Paradox)
the number relevant in a certain way to any number-itself
included (Berry's and Richard's Paradoxes)
The critical fact to note here is that all of these paradoxes pivot on
an item-introducing descriptive specification of the format "the
item meeting such-and-such a condition" (tx)Cx. But the identification operator at issue here so functions that this so-called item
is specified in a deeply problematic way. For what is at issue
throughout is a violation of the Successful Introduction Principle
(SIP).
189
Two modes of communicative anomaly can be distinguished.

One is a matter of propositional meaninglessnes~be it hermeneutically concept-based or semantically truth-based. But there is also
the referential vacuity by way of flawed identification or introduction of terms that leaves (tx)Cx as not properly specified (defined
or identified). And when we treat as meaningfully specified an
item that is not actually so, then the result is that the claims that
we make are bound to be semantically meaningless. We commit
the semantic equivalent of dividing by zero. Our contentions are
meaningless on the basis of identificatory presuppositions that go
unsatisfied.
It's clear that referential vacuity lies at the root of the mathematical paradoxes of the Cantor-Russell variety. All of them result
from a common cause-the inappropriate presupposition that
something exists which in fact does not. Now this may seem odd
on first view because in mathematics one can define things
into existence-we can bring things into being by assumption or
supposition. Or so it seems-but the seeming is not quite right.
For there is one very crucial limit to defining things into existence-namely the preservation of consistency. Even in the realm
of mere possibility you must at least keep your stipulations compatible with your own commitments. And the very fact of paradox here indicates that something has gone awry-that some
condition crucial to the maintenance of consistency has been
violated.
Another instructive lesson also emerges from these considerations. Paradox management calls for noting that, where, and how
certain premisses go awry. All this is a matter of diagnosis and is in
general relatively straightforward. But therapy (that is, treatment)
is something else again. Determining exactly how to fix thingsthat is, positing a cogent rationale for the specific steps required
to eliminate the difficulty-is all too often a large and difficult
challenge. For the medicaments needed to put things right should
be:
natural-by way of being inherently plausible and not arbitrary or ad hoc to the particular case in hand.
not too weak-by way of failing to deal with other versions
or reformulations of the paradox at issue.
190
Chapter 9
not too strong-by way of forcing us to forgo propositions

we want for the sake of eliminating those we do not.
The avoidance of referential vacuity through honoring the conditions of proper identification and introduction seems a promising
prospect in this regard.
To be sure, it is easy enough to offer the paradox-avoiding recommendation: "Do not make unwarranted presuppositions!" Yet
this tautology is of no real help. The devil is in the details, which
is why we need to lay all our cards explicitly on the table in paradox analysis. The problem in each case is to pinpoint that putatively unwarranted presupposition, and then, having done so, to
establish the rationale of its untenability. This aspect of paradox
analysis is not something that is altogether automatic, and the
mathematical paradoxes examined here show only too clearly that
complex considerations are at work in providing a rationale for
the thesis abandonments which alone can resolve such paradoxes.
Prescribing the appropriate remedy that will restore health to our
cognitive organisms in the wake of a paradox is-or can be-the
single most challenging part of aporetics.
The resolution of paradoxes is an invitation to innovation
exactly because it requires the adoption of novel mechanisms by
way of distinctions, prioritizations, and presuppositions. The
mathematical paradoxes invite close attention to the identification
of objects and their referential introduction. The Sorites Paradox
provides a pathway to plausibility theory and the sort of vagueness manipulation afforded by "fuzzy logic." The Liar Paradox
encourages the exploration of pathways to lack of meaning as per
type theory and the doctrine of language levels. The paradoxes of
equivocation calls for the development of a diversified manifold of
conceptual distinctions. In this way we encounter the bracing
atmosphere of theoretical innovation throughout the realm of
paradox.

Self-Counterexampling Paradoxes
Self-Falsification Paradoxes
The Liar Paradox of Epimendes the Cretan
(and its cousins)
Self-Falsification Paradoxes (single and
dualized)
Self-Negation Paradox
The Dualized Liar Paradox
The Liar Chain Paradox
Box Paradoxes
Truth-Status Assignment Paradoxes
The Instantiation Paradox
The Preface Paradox
Semantical Paradoxes that

Involve Conflicting Claims
Regarding Truth
10.1
Self-Falsification
Any contradiction-in-terms (contradictio in adjecto) is automatically paradoxical. To speak of "learned ignorance" is at once to
attribute and to deny sagacity. To locate something in a utopia
(literally a "noplace") is to gainsay the very locatedness that is
being claimed. But such "paradoxes" are mere rhetorical tropes,
and sensible interpreters at once recognize that caution is needed
with such literary flourishes because one must not take them too
literally, even as is the case with metaphor or analogy.
However discourse that is blatantly self-falsifying is something
else again. There are different forms of self-negation: A statement
can occur in a context such that it
is incompatible with its own claim to meaningfulness (by
somehow proclaiming the claim's own meaninglessness).
(Example: "This statement is nonsense.")
is incompatible with its own claim to truth (by maintaining
the claim's own falsity or indeterminacy). (Example: "This
statement is false.")
is at odds with asserting the claim itself. (Example: "No
claim made by statements on this page is true.")
193
194
Chapter 10
contradicts the substance of the claim itself. (Example: "No

statement at all is made on this page." or "This sentence has
six words.")
This last case is a particularly striking form of anomaly. Thus consider: "No proposition is negative, " "All propositions are negative," "Everything said in print is true," "I never say anything
about what I say," "This statement is interesting." Such statements are not just false but themselves present an instance of their
falsity. They instantiate the idea of self-counterexemplifying paradoxes. Medieval logicians examined such statements but were
reluctant to characterize them as insolubilia. 1 A real insoluble, so
they held, is such that it will be false if viewed as true, but the
reverse must hold as well-the statement's truth must also somehow follow from the assumption of its falsity.
The indicated examples show that single statements can
engender apories. However, they rarely pull off the rather difficult
feat of having some semblance of plausibility while nevertheless
being self-inconsistent. In this regard they differ from:
(G) All global generalizations are false.
Given our vast experience with facile generalizations that go awry,
this thesis has an air of plausibility about it. Nevertheless, this
statement is self-falsifying in the same manner in which a thesis
like "Some statements are true" is self-verifying. Resolving the
paradox created by such single-statement examples is easy: we
have to abandon the statement itself. Thus in this present case we
can take (G) at face value and dismiss it as false in line with its
own contention. Here too a conflict of truth claims stands before
us.
A statement will clearly be paradoxical if it does not merely
manifest its own falsity but proclaims it outright. Thus consider
the statements:
This statement is false.
Every statement I make today is false.
1. See Ashwonh 1974, pp. 102-03.
Semantical Paradoxes
195
Everything said on this page is false.

If this statement is true, then I'm a monkey's uncle.
This line of thought leads us back to John Buridan's insolubilium
"This statement [the n-th on the present page] is false," or that
engendered when Socrates supposedly says "Everything that I say
today is false," or again "What I am (now) saying is false."2
Contentions of this sort are not semantically viable. They
straightaway give rise to a Self-Falsification Paradox along such
lines as:
(1) By definition, all (semantically) meaningful statements are
either true or false but not both.
(2) The statements in question are semantically meaningful.
(3) By (1) and (2), the statements in question are either true
or false but not both.
(4) Thanks to their content, both statements are false if true
and true if false. Hence we must either class them both as
true or both as false.
(5) By (1) and (4), the statements in question are not semantically meaningful.
Here I(1), (2), (4) I constitutes an inconsistent triad. But save for
intuitionist logicians (1) constitutes a truth embodied in our definitional understanding of what is at issue with semantical meaningfulness. And (4) is a fact-of-life in the situation at hand.
2. See Prantl, Geschichte, Vol. IV, p. 37, pp. 145 and 146; and G.E. Hughes, John
Buridan on Self-Reference (Cambridge: Cambridge University Press, 1982), pp. 51, 58.
See also Albert of Saxony as discussed in Norman Kretzmans and Eleonore Stump's, The
Cambridge Translations of Medinal Philosophical Texts, Vol. I, Logic and Philosophy of
LIInguage (Cambridge: Cambridge University Press, 1988), pp. 342-43. As the medievals
saw it as insoluble (insolubilium) it is not just a sophism, but specifically a proposition that
poses a truth-value amphiboly, that is, a proposition such that there are structurally equivalent arguments for classing it as true and as false.
Precisely because insolubilia were not really outside the range of the solublc-and
indeed the authors who discussed them all proposed solutions-renaissance writers preferred the Ciceronian term inexplicabilia, which, of course, has problems of its own. (See
Ashworth 1974, p. 114_)
196
Chapter 10
However, (2) is no more than a plausible supposition. Hence we

obtain the priority order (1) > (4) > (2) in virtue of which (2)
must be abandoned.
The general contentions "Every statement I make today is
false" and "Every statement on this page are false," although selfnegating and indeed paradoxical, nevertheless have perfectly
unproblematic negations, namely "Some statement I make today
is true" and "Some statement on this page is true" respectively.
However, this is not the case with "This statement is false." For
its negation is "That statement-namely, 'This statement is
false'-is true." And this continues to be every bit as paradoxical
as the original was. Such a self-falsifYing statement has the feature
that one can neither class it as true nor as false: there is no alternative but to abandon it as semantically deficient and meaningless.
Things can get worse yet. Consider the following statement:
(S) This sentence is (semantically) meaningless [that is, it
lacks a truth value in being neither true nor false].
We ourselves can assign no truth value to (S). For consider the

following:
(1)
Truth-value
assigned to S
(2)
Truth-value of S if
the assignment of
(1) is appropriate
(3)
Agreement between
(2) and the sentence
itself?
No
No
Only by refusing to give S a truth-value can we bring what it

claims into consonance with the prevailing situation, and even
then we do not render the sentence true seeing that in doing so
we would at once falsify it. The only way to treat S is to class it
not just as semantically meaningless but as hermeneutically meaningless. There is no viable way to make sense of it.
However, as we saw above meaningfulness involves yet another
more subtle and complex issue, namely that of referential adequacy. At this point we revert to the Successful Introduction
197
Principle (SIP) of the preceding chapter. It is clear that in the

light of this principle the purported specification:
(L) L is false.
is an inappropriate and unavailing introduction of the statement

L. For its identification here as "the statement which says of itself
that it is false" [or for that matter true!] clearly requires that "it"
have a prior specification if that anaphoric back-reference is to
take hold of anything. Accordingly the trouble with (L) or with
"This sentence is false" is that this crucial principle is violated. To
be meaningful, the claims "L is false" or "It is false that L" presuppose that L is already available for consideration, and we cannot embark upon a purportedly identifying introduction that
presupposes this.
And generally, use of the anaphoric phrase "this sentence" presupposes that the sentence at issue has already been identified in a
prior specification. Indeed any self-referential statement introduction of the format:
(L) Lis such-and-such ["L is the sentence of which it holds

that this sentence is such-and-such."]'
is inappropriate, seeing that the standard process of substituting
their definitions for defined terms here embarks us on an infinite
regress as per:
(L is such and such) is such and such.
( (L is such and such) is such and such) is such and such.
and so on. That supposed definition does not enable us to take

anything stably meaningful into view. Here a thesis that is (putatively) being introduced as a (purportedly) meaningful assertion
simply does not quality as such. No such meaningful statement
exists. Thus a Self-Falsification Paradox and its cognates are
semantically meaningless because they resist the stable assignment
of any truth value through their plunge into paradox.
And it should be emphasized that this is equally true of the
harmless seeming
198
Chapter 10
(L) L is true.
For the real problem in this context is not self-falsification as such

but self-characterization in a way that vitiates the sort of identification required for access to a meaningful referent.
To be sure, the contention that "This statement is such-andsuch" or equivalently:
(L) Lis such-and-such.
could be reformulated free of explicit self-reference as
vp [p = L iff P asserts that Lis such-and-such].

But in this formulation the thesis does not purport to define or
introduce L. And the problem can now be sidelined by observing
that its paradoxical nature precludes L from lying within the range
of the propositional variable p, namely the set of semantically
meaningful presuppositions. That "reformulated" thesis is not a
well-formed statement.
The study of these insolubilia formed a key component of the
arts curriculum of the major medieval universities-usually in the
third or fourth year which focussed upon public disputation.
Some of the medieval school men took a line that actually makes
very good sense. The term "statement,"they in effect maintained,
is equivocal. For we must draw the important distinction between
a sentence or assertion (oratio) which is something merely verbal,
and an actual proposition (propositio) which is a communicatively
meaningful item of information. Normally and when all goes well,
sentences convey propositions. But things do not always go
well-and in particular they do not go well with sophisms and
insolubilia. 3
Specifically they held that this happens with what we would
characterize as semantically meaningless propositions. Thus
Buridan maintained that statements which cannot in the nature of
things be classed as true or false are thereby inherently
3. An insolubilium, says Paul of Venice, is only a verbal proposition (propositio vocalis)

but one that is not authentically communicative (propositio scripta vel mentalis improprie
dicta); Prantl, Geschichte Vol. IV, p. 139 nS69).
Semantieat Paradoxes
199
untenable. 4 And Paul of Venice was even more explicit. He held

that such a statement as "What I now say is false" is not a proper
statement at all. s It is neither true not false but stands in an indifferently untenable vacillation between the two (non est 7Jerum nee
fatsum, sed medium indifferens ad utrumque; ibid., pp. 138-39).
It is, strictly speaking, no proposition at all. (Nuttum insotubite est
7Jerum et fatsum, quia nuttum tate es propositio; ibid, p. 139,
n539.)
10.2 The Liar and His Cousins

Concern for problems of self-reference originated with the Liar
Riddle (pseudomenos) of Eubulides: "Does the man who says 'I
am lying' lie?" (Also "Does the witness who declares 'I am perjuring myself' perjure himself.")6 The problem that arises here was
posed via the following dilemma:
The declaration that I lie will be either true or false. But if this
declaration is true, then I lie, and my declaration will be false.
But if that declaration is false, then what it says-namely that
I lie-is not the case and I must be speaking the truth. Thus
either way the truth status being assigned is inappropriate.
Eubulides's riddle was immensely popular in classical
antiquity.7 And it gave rise to the problem encapsulated in the
ancient story of Epimenides the Cretan, who is supposed to have
said that" All Cretans are liars"-with "liar" being understood in
4. Prand, Geschichte, Vo!' IV, p. 37 n146.

5. Socrates dicens, se ispum dicere falsum, nihil dicit. (Prantl, Geschichte, Vol. IV, p. 139
n569.) It became a commonly endorsed doctrine in late medieval times that paradoxical
statements are not proper propositions and for this reason cannot be classed as true or
false. (See Ashworth 1974, p. ll5 for later endorsements of this approach.) Thus later
writers dismissed insolubles as not being propositions at all, but "imperfect assenions"
(orationes imperfeaae). (See Ashworth 1974, p. 116.)
6. Aristode, Soph. Elen., 180a35; Nicomachaean Ethics, 1146a71. Prand, Geschichte,
Vol. I, pp. 50-51. Si dicis te mentiri perumque dicis, mentirs. (Cicero, Academica priora,
11,30,95-96; and compare, De dipinatione, II, 11).
7. It was discussed not only by Aristotle and Cicero (see the preceding note) but by
the Stoics (Prantl, Geschischte, Vol. I, p. 490). In medieval times it was a staple in the
extensive: discussions ofinsolubilia. See Prand, Geschichte, Vol. IV, pp. 19,41.
200
Chapter 10
the sense of a "congenital liar," someone incapable of telling the

truth. 8 What we have here is a self-falsifying statement that
involves a conflict of truth claims. For if we accept the claim of
Epimendes as true, then it itself will have to be false so that the
statement implies its own negation. However, nothing untoward
results from classing the statement of Epimenides as false-there
is no problem about the idea that some Cretans sometimes do not
lie. Self-falsifying statements ("I am lying,"9 "This statement is
too complex to formulate," "All universal statements are false")
are self-contradictory and therefore false, seeing that a conflict
arises between the statement and its own consequences. We have
a paradox alright, but one that is readily resolved-both as
regards culprit identification and as regards the diagnosis of
involving a referential failure.
The Epimenides problem is thus not as grave as the classic Liar
Paradox, which poses larger and deeper issues. And there is nothing novel about this claim. Riistow enumerates sixteen approaches
to the resolution of the Liar Paradox that were offered in ancient
and medieval times. (Paul of Venice already listed fifteen.) These
fall into four groups: (1) charges that a standard fallacy is committed (for example, secundum quid, equivocation per accidens, non
causam ut causam); (2) self-destruction through self-negation;
(3) semantic meaninglessness through failing to be either true or
false; and (4) referential failure (non potest supponere pro tota ilia
propositione cuius est pars." 10
Semantical paradoxes arise when matters of truth, falsity, and
reference are explicitly at issue. And here, as elsewhere, appearances can be deceiving. In this regard consider the variant version
of Eubulides's Liar paradox, posed by the riddle: "Does the per-
8. Several Greek philosophers, preeminently including the Aristotelian Theophrastus

and the Stoic Chrysippus, wrote treatises about the Liar Paradox. See Diogenes Laertius,
Lives of Eminent Philosophers, V 49 and VII 196. The poet Phietas of Cos is said to have
worried himself into an early grave by fretting over this perplex, and its notoriety was such
that even St. Paul adverted lO it in Titus I: 12-13. The history of the Liar Paradox is discussed in substantial detail in Alexander Riistow's Der Lugner: Theorie, Geschichte, und
Auflosung (Leipzig: B.G. Treubner, 1910; reprinted, New York: Garland, 1987).
9. This anomaly gives rise to the comic variation of the "King of Beasts" who could
not get his subjects to believe him when he said "I'm lion." See also Martin Gardner,
Aha! Gotcha (San Francisco: Freeman, 1982), pp. 4-7.
10. Riistow 1908, pp. 114-16. For a contemporary analysis of the difficulties involved
see Barwise and Etchemendy 1987 as well as Martin 1990 and Martin 1993.
201
son who says that he is now lying speak truly?" This SelfFalsification Paradox now at issue roots in the thesis:
(5) This statement (that is, this very statement now being
enunciated) is false.
This statement confronts us with the perplex:
We class S as
The truth-value of S according to

itself, in consequence of the preceding
status of what it claims
In no case can we bring these two into alignment. Nor will it

help to introduce a third truth value ("indeterminate" or "undecided."). For consider:
(S') This statement is false or undecided.
The situation now becomes:
We class S as
Truth-value of S' in
consequence of the preceding
status of what if claims
Again there is no way of bringing these two columns into accord.

The only viable alternative is to view such statements as Sand S'
as semantically meaningless, that is, as lacking any stable truthvalue status whatsoever-even "indeterminateness" insofar as this
is seen as a truth-value.
The explicitly self-falsifying claim "This statement is false" is
dearly paradoxical. Yet we could not simply class it as false
Chapter 10
202
because in doing so we would in effect verify it. Instead we canand indeed should-proceed to view this statement (S) in a wider
context:
(1) S is a semantically meaningful statement-that is, it is
either true or false and not both.
(2) Given what it asserts, if S is true-that is, if it is true that S
is false-then it is false. And(3) Given what is asserts, if S is false, then it is true.
(4) No semantically meaningful statement can be both true

and false.
(5) If S has one truth value (if S is true or false) then it also
has the other. By (2), (3).
(6) S is semantically meaningless-it is neither true nor false.
(From (4), (5).)
Since (4) is fixed in place as the consequence of a definition,
the inconsistency of the quartet {(I), (2), (3), (4)} means that
we are driven to a choice between (1) and the coordinated pair
(2), (3). But (2) and (3) are parts of the issue-defining
hypothesis, while (1) is no more than a plausible supposition,
and one to which an intuitionist logician might well take
exception. Thus the plausibility priority situation is:
(4) > [(2), (3)] > (1)
Accordingly (2), (3), (4)/(1) with its retention profile of {I, 1,

O} represents the plausibility-optimal resolution and S must
therefore be dismissed as a semantically meaningless statement.
The paradox thus admits of this decisive resolution which
uniquely averts abandoning a proposition of the first priority
level. S is the culprit and has to be abandoned. But note that S
alone does not constitute the paradox here all by itself; rather, it
opens the door to paradox via (1)-(5). Moreover, (1)-dismissalthough certainly in order-is not the end of the matter. We need
to determine of validating rationale for this step that principles
like SIP provide.
203
There is also the Self-Negation Paradox where truth and falsity are now not at mention. It pivots on the proposition P which
asserts: "This proposition-viz. P itself-does not obtain." We
thus have:
(P) not-P
This directly gives rise to a paradox, as follows:
(1) PH not-P
From (P) because that is

what this affirms
not-P
From (1)
From (1)
(2)
P~
(3)
not-P~
(4) not-P
From (2) by standard logic
(5) P
From (3) by standard logic
Here (1) is the source of the difficulty all by itself. But what
exactly is wrong with it? It is simply untenable because it concurrently claims and disclaims P. In consequence, (1) does not specifY a hermeneutically meaningful statement-we can make no
sense of it. As we have seen time and again, such statements that
violate the SIP principle are a fertile source of paradox.
10.3
Problems of Self-Reference
When we assume too incautiously that various items are already

available for use in their own specification we may well run into
trouble. Thus consider the purportedly self-denying statement
introduced by the following specification:
(S) not-S
It is clear on the surface of it that this is something paradoxical.
The question, "Is it or is it not the case that S?" at once plunges
us into a contradiction. The SIP-prohibited self-involvement at
issue here leads us ad absurdum.
204
Chapter 10
And again, consider a statement that is specified as the conjunction of itself with some claim:
(S) S & c
For example, if c is 2 + 2 = 4, the question "How many times is

the number two mentioned in S?" becomes problematic. The very
idea of counting all of this statement's twos leads to difficulties.
(Only for infinite N can we have it that N = N +1, but any actualizable statement must be one of finite length.) In this context we
are dealing with specifications whose meaningfulness and viability
are erroneously presupposed.
Consider next the regress generated by the following question-answer cycle:
Statement: (S) This statement has feature F.
Question: (.0 But what statement are you talking about?
What is the referent of your own statement?
Response: (R) Go back to that initial statement. The referent
is S itself.
The cycle never closes. The question of a referent is never fully
resolved. Self-involving identification is effectively toothless: it has
no identifactory bite. The situation is akin to that of the
exchange: Q: "Is this a proper question?" A: "Only if this is a
proper answer to it." In its own bizarre way, this exchange actually works out. For the question is not a proper one exactly
because the "answer" to it is No. And the answer is not a proper
one either, so that its negative standing provides what is, in the
circumstances, an appropriate sort of response.
The difficulty in all these cases is again one of unsuccessful
identification. For in each case we have a thesis identified via a
format that violates the SIP principle of successful identification.
To be sure, in addressing problems of this sort, an accusing finger
is sometimes pointed at self-reference as the source of the difficulty, and some writers-Russell for example-accordingly suggest a general embargo on all self-referential statements. But this
is going too far. What is problematic is not self-referential asser-
205
tion as such, but self-referential statement introductions or specifications. Self-referential statements are often not only unproblematic in meaning but even true. 11 There is nothing anomalous
about "This sentence contains a verb." and "This sentence occurs
on p. 205 of this book" or "This sentence is written in English."
There is nothing wrong with the self-referential contention "This
statement is an affirmative proposition formulated in English" or
with "This denial that cats are crustaceans is a true statement
about felines." What does go amiss is the introduction of an item
on the stage of discussion in a way that presupposes it as already
available as a term of the kind at issue. For example, to write
(L) L is a truth.
is to indicate on the one hand that we are introducing (specifying)

the statement indicated by L, and on the other that L is already
available for a reflective claim to be made about it (viz. that it is
true). And this just will not work.
Against those who want to proscribe overt self-reference altogether in the interests of avoiding paradox, one can urge two
points (among others) as prime reasons not to adopt this policy.
(i) It throws out the baby with the bath water. Self-reference
of and by itself cannot be the problem. For there are many
informative and unproblematic self-referential statements
such as "This statement is made by means of an English
sentence." And many instructive generalizations have such
statements as consequences (for example): "All statements
made in this chapter are relevant to paradoxes").
(ii) Various useful and important logical and mathematical
theses have to be formulated by means of self-referential
or self-inclusive statements. (For example, "Semantically
meaningful statements have some truth-value or other.")
And some statements in metamathematics (including
assertions pivotal to Godel's proof of the incompleteness
of arithmetic) have to be formulated self-referentially.
11. This was emphasized in Kurt Godel, op. cit.
206
Chapter 10
Only a very few assertoric processes will engender paradox automatically and invariably. Direct or indirect self-negation of the following sort is an obvious example:
The statement preceding the next statement is meaningless.
All generalizations have exceptions. 12
However, the difficulty is more subtle with statements which
themselves counter-exemplify that which they overly assert:
No proposition is negative.
All generalizations are true.
Here the first statement is a negative proposition (something
whose existence it itself denies) and the second is a false generalization (again something whose existence it itself denies). Since
the mere existence of these statements establishes their own falsity, there is no way to establish harmony as between what statement T is and what it claims.
In the end, then, it is best to take a discerning and differentiated view of self-reference, recognizing that it can take both lawful
and illicit forms. A general point is at issue here, namely that there
is a great difference between saying that something is a process
that can engender paradox and that it is a process that must engender paradox. The sorts of paradox-inviting assertoric processes that
we have considered here--equivocation, self-reference, questionable totalization and problematic presupposition, for example-are
merely paradox facilitating and not paradox necessitating. And one
must accordingly always distinguish between illicit and innocuous
uses of such a process. The sensible policy is to indicate more
specifically the conditions under which such processes are likely to
produce inconsistencies. And this means that wholesale measures
of paradox avoidance-an embargo on self-reference, for example-are generally involved in overkill by eliminating much that is
harmless along with the inappropriate.
12. This statement is obviously a generalization. According to its own assertion it thus
has exceptions. But generalizations with exceptions are false. Ergo if true, then it is falseand false on the basis of its own explicit content.
207
10.4 Paradoxes of Truth-Claim Dissonance

Seman tical paradoxes generally involve a conflict between what
we are saying in the (internal) content of a statement and the
(external) claim involved by making it. Thus with the self-falsifying statement:
(P) not-P
we are claiming that P by making the statement, but saying that

not- P in the statement itself Or again with
All propositions are negative.
we are stating an affirmative claim by making the statement but
denying this prospect in the statement itself.
An analogous clash can also occur within a plurality of statements. Thus consider a dualization of the self-falsifying statement
Pabove:
Irrespective of how we assign truth-values here, there is no way to

bring our truth-classification of those statements into alignment
with what the statements themselves say. The statements at issue
have to be dismissed as semantically meaningless in the circumstances.
But note that what has gone wrong here is a matter of the
identiftcatory self-reference that is proscribed by the Successful
Identification Principle (SIP). For when we return to (PI) and
(P ) and replace defined terms by their definitions we get:
2
(PI) not-PI
(P2 ) not-P2
Both involve the same sort of self-negation that is at issue with P
above.
208
Chapter 10
Again, consider the following pair of statements:

(A) B is true.
(B) A is false.
These statements issue in the Dualized Liar Paradox:
(1) Statement A declares " B is true."
(2) Statement B declares" A is false."
( 3) A and B are semantically meaningful statements (ones that
are true or false).
(4) Thanks to their content, neither A nor B can stably be categorized as true or as false.
Here statements (1)-(4) are collectively inconsistent and constitute an aporetic cluster, as we have already seen in detail. Now the
only candidate for rejection here is (3), since (1) and (2) obtain
by the defining stipulation of the problem and (4) is part of the
situational realities of the case. With the dismissal of (3) we
achieve a decisive resolution.
Or consider the closely related paradoxical pair:
The immediately succeeding statement is true.
The immediately preceding statement is false.
Note that unlike the single self-contradictory statement "This
statement is false" the paradoxical character of this pair is contingent. It is a matter of just what those succeeding or preceding
statements happen to say. Each of these two statements is self-referential-but only conditionally and contingently so, not directly
and by nature. What we have here is merely circumstantial selfreference and merely contingent paradoxicality. The crux lies not
in the meaning of what is said but on the circumstance of the saying of it, exactly as with other ostensive statements such as "That
object is a knife." Here Plato effectively maintains: "'What Plato
says is false' [= what Socrates says] is true." And Socrates effec-
Seman tical Paradoxes
209
tively maintains" "'What Socrates says is true' [= what Plato says]

is false." Both enmesh themselves in self-annihilating contentions.
The Dualized Liar Paradox also gives truth to the longer (sorital) version of the Liar Chain Paradox as per:
A: What B says is false.
B: What C says is false.
C: What A says is false.
Again, no matter what truth-status we propose to assign to these
statements we run into contradiction. 13
Medieval logicians were very devoted to such variations on the
Liar paradox. Thus one twelfth-century theorist posed the conundrum: "Plato says 'What Socrates says is trUe' and Socrates says
'What Plato says is false.' Neither says anything else. Is what
Socrates says true or false?,,14 And this sort of thing can occur in
innumerable other ways. Consider the pair of truth-value attribu. The preceding sentive sentences: "The next sentence is
tence is
." Of the four possible ways of filling in the
blanks-namely TT, TF, FT, and FF---only the first and last are
nonparadoxical. And this means that no paragraph of the format
. The next sentence
"Just one of the next two sentences is
is
. The preceding sentence is
." can possibly be
rendered consistent. Such a cluster is bound to be paradoxical.
Thus consider the Truth-Status-Assignment Paradox posed by
the two statement in the following display:
(A) The statements in this display differ in truth-value.

(B) George Washington was the first president of the U.S.A.
13. On this paradox, due to Albert of Saxony, see Kretzmann and Stump 1988, p.
353.
14. See William and Martha Kneale, The Development of Logic (Oxford: Oxford
University Press, 1985), p. 228. Another version of this paradox envisions a card with one
sentence written on each side. The one reads "The sentence on the other side of this card
is true" and the other reads "The sentence on the other side of this card is false."
210
Chapter 10
The paradox at issue here runs as follows:

(1) (B) is true. (A well-known fact.)
(2) (A) is false. From (A), (1).
(3) The statements in the box agree in truth-value. From
(2 )'s stipulation of (A)'s falisty.
(4) Both statements are true. From (1), (3).
(5) Both statements are false. From (2), (3).
The only viable way out here is to dismiss A as (semantically)
meaningless in the prevailing context.
Box paradoxes of this kind were initiated by Henry Aldrich
(1647-1710), who offered the example of the box that contains
only a sentence saying that every sentence in this box is false: 15
Omne enunciatum intra hoc quadratum scriptum est falsum.
Again consider:
It is - - - that a - - - statement is
presented in this box.
Here let the blanks be filled in by true (1) or false (P). Note now
the following survey of possibilities.
15. HenlY Aldrich, Artis logicae compendium (London, 1691). See Ashworth 1974,
pp. 18,114.
211
The truth-values
we supply are
On this basis the

boxed statement is
IT
TF
FT
FF
Consider now the pair of assertions.

(1) The boxed statement reads as indicated.
(2) The two blanks of the boxed statement are filled in by
opposite truth-values.
This pair of statements is dearly aporetic: there is no way to get
them to be true together. The only way out is to deny the (tacit)
premiss of the paradox that within the parameters of the problem
the boxed statement qualifies as (semantically) meaningful. (Note,
however, that this upshot is entirely contextual. If assertion (2)
read "identical" for "opposite" the situation would be radically
altered.)
Again, consider the Box Paradox engendered by the following
display:
(1) Every sentence in this box is false.
(2) Isaac Newton wrote the plays attributed to Shakespeare.
Since (1) is self-contradictory (paradoxical), it must be false. But
(l)'s falsity means that its negation must be true. In consequence
some sentence in the box must be true. Since (1) is out of bounds
in this regard, this means that (2) must be true. 16 Clearly any
absurdity can be "demonstrated" on this basis. The paradoxicality
of (1) means that anything we put into (2)'s place will be "innocent by association." And the appropriate exit from paradox here
16. This paradox is due in essence to Popov and Popov, Dodkalny Mathenalizy, 1996.
lowe this reference to Alexander Pruss.
Chapter 10
212
again lies in rejecting the culprit statement (here (1)) as semantically meaningless.
Another case that merits consideration is the variant TruthStatus Misallocation Paradox presented by the following two
groups of statements:
Group A
1. Most statements of Group B are true
2. 2 + 2 = 4
3. 2 + 2
GroupB
1. Most statements of Group A are false
2. 2 + 2 = 4
3. 2 + 2
Here the only way to make Al true is to make BI true, yet the
only way to make BI true is to make Al false. Again, in reverse,
the only way to make BI true is to make Al false, yet the only
way to do this is to make BI false. Thus neither Al nor BI can be
true. But, by analogous reasoning, neither Al nor BI can be false.
The only way out is again to dismiss some of the key statements
involved-specifically Al and BI-as (semantically) meaningless
in the present context. As in the previous examples there is here a
clash between the external (semantical) status of the problematic
statements and their (internal) meaning-content. Such explicitly
or obliquely (contextually) self-falsifying statements are also a fertile source of paradox. Collectively unsatisfiable statement groups
of the sort at issue here are the verbal analogue of the geometric
anomalies of Escher drawings.
A generic contention regarding an otherwise unspecified item,
"X is a Siamese cat" or "X is an odd integer" or "X is an affirmative statement"-a propositional function, as it is called-will be
true of some things and not true of others. In particular some
propositional functions will be true of themselves, for example "X
213
is a propositional function" or "X is a statement schema that can

be rendered true by a suitable instantiation." Such propositional
functions are self-instantiating. But now consider the generic thesis
(1) X is a non-self-instantiating generic thesis.
Is this statement schema self-instantiating or not? Suppose we
accept that it is. Then we have
"X is a non-self-instantiating generic thesis" is a non-selfinstantiating generic thesis.
But this immediately falsifies our supposition. On the other hand,
suppose it is not. Then we have:
"X is a non-self-instantiating generic thesis" is a self-instantiating generic thesis.
And this contradicts the supposition. And so neither alternative is
viable.
With this Instantiation Paradox the only escape from inconsistency lies in rejecting (1) itself. The idea that a meaningful concept is at hand with the specification of a "non-self-instantiating
generic thesis" must be abandoned.
This paradox closely parallels Russell's previously discussed
Impredicability Paradox regarding predicates (see above, pp.
173-74). However, via the pivotal role of truth in the ruling conception of "is true of" at issue with generic theses (rather than "is
predicable of" with predicates), this propositionalized version of
the paradox makes clear its status as a semantical paradox involving conflicting claims to truth.
10.5
The Preface Paradox
Consider the following situation that defines what has become

known as The Preface Paradox. An author's preface reads in part
as follow: "I realize that, because of the complex nature of the
issues involved, the text of the book is bound to contain some
errors. For these I now apologize in advance." There is clearly
214
Chapter 10
something paradoxical going on with this otherwise far from outlandish disclaimer because the overall set of implicit truth-claims
at issue here is, in aggregate, incoherent. The statements of the
main text are flatly asserted and thereby claimed as truths while
the preface statement affirms that some of them are false. Despite
an acknowledgement of a collective error there is a claim to distributive correctness. Our author obviously cannot have it both
ways.I7
To clarify the situation it is helpful to let T serve as abbreviation for the conjunction of statements of the text and P analogously for the preface. On this basis we straightaway establish the
aporetic nature of our author's declarations. For we have both of
the following:
(1) T, that is, "Tis true," which is tantamount to "All components of Tare true."
(2) P, that is, "Some few of the (otherwise unspecified components of 7) are false."
Something has to give way here. But how are we to resolve this
perplex?
Perhaps the most sensible way to look at the matter is something like this. The prominence our author gives to P through its
synoptic role in his preface prioritizes it over T. Thus P must be
accepted while Tis abandoned in its composite totality. (However
the plausibility of T gives us an incentive to salvage as much as we
can, so that we would presumably settle for prefixing to the text
T the (otherwise tacit) preliminary statement "All but some few
of the statements that follow are true.") The course of subsequent
events can then be allowed to determine just which of the otherwise unidentified statements turn out to be falsehoods.
The idea of plausible claims yields a prospect for authorship
that is seldom exploited. For authors who advance their claims in
the mode of plausibility can proceed on a tentative basis and need
not present their assertions as categorical claims to truth. In con-
17. On the Preface Paradox se A.N. Prior, "On a Family of Paradoxes," Notre Dame
Journal of Formal Logic, Vol. 2 (1961), pp. 26-32. See also D.C. Makinson, "The Paradox
of the Preface," Analysis, vol. 25 (1965), pp. 205-{)7.
Seman tical Paradoxes
215
sequence, they need not be intensely concerned to maintain the

compatibility of their statements. The prospect of an occasional
inconsistency will not distress them. They need not view consistency as the hobgoblin of small minds, but rather as that of minds
that are fearful and overcautious, somewhat neurotically reluctant
to run the risk of error in the pursuit of truth and information.
Their perhaps overliberal policy of information management
leads such authors to be prepared to risk maintaining too much in
the interests of assuring that they maintain enough. Given that we
are finite creatures, the pursuit of truth is impracticable for us if
we are unwilling to run the risk of error. And as these authors see
it, the prospect of an occasional and limited inconsistency constitutes no additional and thereby no graver and more portentous
threat. In their view neither epistemic self-respect nor cognitive
credibility requires a commitment to consistency at all costs. In
the pursuit of truth, consistency is a great asset but not an
absolute and non-negotiable demand-a categorical requirement.
There are but a few authors who take this line openly and
explicitly and flatly put their readers on warning to be prepared to
take an occasional inconsistency in their stride. But there is certainly no tellingly cogent reason why there should not be more of
them. To be sure they should, in due candor, avow their policy
position and inscribe caveat emptor on their title pages.
The seman tical paradoxes all root in one or the other of two
sorts of error. On the one hand, with particular statements there
is identificatory self-reference of the sort proscribed by the
Successful Introduction Principle (SIP). On the other hand, with
general statements there is auto-inclusion in a group that the
aporetic premisses at issue reject. "This statement is false" is a paradigm example of the former problem. And "All statements made
in this book are false" is a paradigm example of the latter. The
invalid introduction of items (in the case of particulars) and autodestruction by self-impugnment (in the case of generalizations)
are the prime access-roads to semantical paradox.

Equivocal Evidence Paradoxes
The Of-Two-Minds Paradox
The Lottery Paradox
Hempel's Raven Paradox
Inductive Paradoxes
(Conflicts of Probability
and Evidence)
11.1
Evidential Indecisiveness
An important source of paradox is constituted by evidential situations where the epistemic issues are so complicated and knotty
that their upshot is indecisive. We find ourselves confronted with
a series of claims for each of which there is much to be said by
way of support and substantiation, but which are nevertheless collectively inconsistent. Just what to conclude now becomes a problematic issue.
Such a situation of conflicting evidence arises, for example,
when one credible witness or source of information declares p and
another not-po Or again when one evidential datum speaks for p
and another for not-p, as when we want to determine somebody's
height and are informed both that she is expert at gymnastics
(most good gymnasts being rather short) and at basketball (most
basketball adepts being very tall).
It is a plausible principle of inductive reasoning that where the
available evidence (E) gives strong inductive support (=> ) to a certain conclusion (C) then this conclusion can be inferred as per the
inferential pattern of modus ponens:
(MP)
E~C
:. C
219
220
Chapter 11
However ambivalence can arise when the evidence at our disposal

is indecisive in such a way that either of two incompatible conclusions can reasonably be drawn from it-when, as in a detective
story, there is good reason for suspecting both the ne'er-do-well
nephew and the unexpected guest of the murder. Such cases yield
Equivocal Evidence Paradoxes of the following format:
(1) Principle MP as stated above
(2) El &E2
The aggregate body of equivocal

evidence
(3) El ~Cl
By hypothesis
(4) E2~ C2
(5) C 1
By hypothesis
(6) C 2
From (1), (2), (3)

From (1), (2), (4)
A fact of logic in the prevailing
circumstances
Here 1(1), (2), (3), (4)} are an inconsistent quartet. Now (2) is
(by hypothesis) a given fact, and (3)-(4) are also fixed in place the
defining stipulation of the problem. We thus have the priority
ranking: [( 3), (4)] > (2) > (1). Accordingly, the proper link at
which to break the chain of inconsistency is with the principle (P).
The fact is that this is not a universally valid principle that
deserves flat-out acceptance, but only a plausible rule of inductive
inference that can only be employed in circumstances devoid of
counter-indications, unlike those of the situation that we presently
confront.
This example also conveys another lesson. It is sometimes said
that paradox occurs when apparently acceptable premisses yield
apparently unacceptable conclusions by apparently cogent arguments. 1 But the present example shows that we can speak simply
of valid arguments here. For that "seemingly cogent argumenta-
1. See, for example, R.M. Sainsbury, Paradoxes, 2nd ed. (Cambridge: Cambridge
Press, 1995), p. 1.
Unive~ity
Inductive Paradoxes
221
tion" can itself always be transposed into just another of those

explicit "seemingly acceptable premiss"-exactly as was done at
step (1) here.
11.2
Self-Deception
Not only are people sometimes deceived, but they can also
deceive or delude themselves about all sorts of things-about
being a good driver for example. (Paradoxically, some 80 percent
of drivers think they drive better than the average driver.) To be
sure, there are, strange though it may seem, some issues about
which people cannot possibly deceive themselves. For example,
one cannot deceive oneself about:
being liable to self-deception.
being sometimes mistaken.
Here "deceiving oneself" would simply substantiate the point at
issue.
Self-deception can occur when it is we ourselves rather than
the evidence that are indecisive. It involves accepting in one frame
of one's mind or in one phase of one's thinking something that
one rejects in another without acknowledging the incompatibility.2 It is effectively a matter of "being of two minds" about something, so that here both "X is minded to reject p" and "X is
minded to accept p" come to figure as concurrently applicable
contentions. The Of Two Minds Paradox that results stands
roughly as follows:
( 1) X accepts p (on the basis of certain considerations).
(2) X rejects p (on the basis of others).
( 3) X qualifies as a fully rational person who keeps his beliefs
compatible and consistent.
2. On paradoxical aspects of self-deception see T.S. Champlin, ReflexiTJe Paradoxes

(London: Routledge, 1988), pp. 10-23.
222
Chapter 11
Here the only available exits from inconsistency are:

(1) - or-( 2 )-abandonment. Holding that someone who can
appropriately be said to "reject" a proposition p cannot
with yet another breath appropriately be said to "accept"
it-and conversely.
(3) - abandonment. Rejecting the idea that someone caught up
in such a situation of ambivalence can qualify as a fully
rational being.
Accordingly there are two avenues to paradox resolution: One can
abandon the idea that the "accept" /"reject" demarcation has to
be drawn in a way that renders these two stances mutually exclusive. Alternatively, one can abandon the idea that rationality must
be so understood that no one who wittingly "accepts" a contradiction can possibly qualify as rational. In introducing the contrast
between acceptance as true and acceptance as plausible we have
sought to negotiate this paradox through a guarded and qualified
endorsement of the second alternative.
11.3
The Lottery Paradox
Probabilities afford one of the possible routes to plausibility

appraisal. But here too we must acknowledge and accommodate
the prospect of possible entanglement in aporetic situations. After
all, the nonoccurrence of any and every particular outcome of a
die toss is rather probable. But nevertheless one of them must
occur. This clearly means that we must not use probabilities as a
basis for acceptance-as-true since truth admits no contradictions
and is therefore bound to be paradox-free. However, the shift to
plausibility averts such problems, seeing that plausibilities, unlike
truths, need not be mutually consistent.
Let us consider this situation in the light of the so-called
Lottery Paradox. 3 Suppose a lottery with 100 markers inscribed 1,
3. On this paradox see Henry Kyburg, Probability and the Logic of Rational Belief
(Middletown, CT: Wesleyan University Press, 1961), L.J. Cohen, The Probable and the
Provable (Oxford: Clarendon, 1977), and Robert Stalnaker, Inquiry (Cambridge, MA:
MIT Press, 1984). Compare also pp. 52-53 above.
Inductive Paradoxes
223
2, 3, ... , 100. Here each outcome one is only an "one-in-a-hundred shot." each is very unlikely. So if we have the following manifold of outcomes:
O( i)
the outcome of the lottery drawing is i.
then any and every particular outcome-claim is distinctly improbable. (And we could make it even more so by letting the lottery
have a thousand or even a million entries.)
Now if we were to let high probability be our guide to outright acceptance, then clearly we would have to accept "Not
O( i)" systematically, for any particular i. This leads straightaway
to paradox. For on the one hand we now accept:
X(i)
Not-O(i)
for every value of i in the range 1, 2, ... , 100. But on the other
hand we must also accept:
x = 0(1) v 0(2) v ... v 0(100)

which reflects the limits of the pertinent outcome range at 100.
After all, some outcome must result, so that this disjunction is a
flat-out truth. But now the overall collection of all 101 of these
claims-X together with all of the X(i)-is logically inconsistent.
So we certainly cannot accept all these theses as true. But matters
stand quite differently if our "acceptance" of the X( i) is merely an
acceptance as plausible (unlike that of X itself) since plausibilities
as such need not be constant.
The lesson is that we can accept sufficiently probable propositions as plausible but not as true.
To be sure, an aporetic situation still remains. But we can now
approach it from the angle of our general methodology. Thus
consider:
1. Aporetic cluster: {X(l), X(2), ... ,X(lOO), X}

2. Maximal consistent subsets:
{X(l), X(2), ... ,X(lOO)} plus 100 sets of the format
{X, X( 1 )- X( 100) with exactly one omission among the last
100.
224
Chapter 11
3. Acceptance alternatives:
All 100 posibilities for the group X, X(l)-X(lOO) with
one omission among the X(i) and also X(l)-X(lOO)
5. Priority ranking:
X> [X(l), X(2), ... ,X(lOO)]
6. Optimal resolution:
A disjunction of all the R/A alternatives that exclude
exactly one of the X( i).
Here we would, as ever, want to break through the "chain of
inconsistency" by seeking out the weak links of the chain. But
unfortunately, the circumstances of the example do not permit us
to differentiate among those plausible theses X( i), all of which lie
on the same plane-to all visible appearances. Apart from recognizing that at least one of the XU) must be false, any more definite resolution of the paradox is impracticable. The specifics as to
the choice between one alternative and another remain undecided, so that this resolution of the apory again represents an
inconclusive disjunction of uneliminable alternatives.
11.4
Hempel's Paradox of the Ravens
A substantial literature has grown up over a problem of inductive

reasoning first posed by C.G. Hempel in 1946, the so-called
"Paradox of the Ravens."4 It is rooted in the fact of deductive
logic that "All X is T' is interdeducibly equivalent by contraposition to "All non- Y is non-x''' Accordingly, "All ravens (R) are
black (B)" is deductively equivalent to "All non-black-objects
(non-B) are non-ravens (non-R)." Now the problem is thus:
given this equivalence, why should it be that in inductive contexts
one is inclined to accept black ravens as confirming instances of
that initial lawful generalization but not white tennis shoes?
4. C.G. Hempel, "A Note on the Paradoxes of Confirmation," Mind 55 (1946), pp.
79-82. See also Rudolf Carnap, Logical Foundations of Probability, pp. 223-24.
Inductive Paradoxes
225
The situation at issue here is such that we confront the following aporetic cluster:
( 1) The theses
(HI) All ravens are black
(H2) All non-black objects are non-ravens
state logically equivalent hypotheses.
(2) If two hypotheses are logically equivalent, then any datum
that confirms the one must also confirm the other-and
will do so to the same extent.
(3) A black raven will confirm HI (to some nontrivial extent),
and analogously a non-black non-raven (such as a white
shoe) will similarly confirm H2 (to some nontrivial
extent).
(4) However, a white shoe will not confirm HI (or at least
will not do so to any extent worth mentioning).
It is clear that these theses are inconsistent, seeing that (1 )-( 3)
entail the denial of(4).
Since (1) is an unproblematic fact of logic, and (3 )-( 4) represent rock-bottom intuitions of inductive thought, it is thesis (2)
that attracts our suspicions. The priority situation stands at (1) >
[(3), (4)] > (2). We are thus led to the R/A option
(1),(3),(4)/(2) as the appropriate resolution, exhibiting as it does
the resolution optimal priority profile {1, 1, O}.
But what reason is there for thinking (2) to be the weak link in
the chain here? Let us consider this question more closely.
It helps to clarify matters to consider the situation as represented in a Venn diagram of the following sort:
R
[1]
226
Chapter 11
Here the regions represent different classes of objects. Specifically

the R are the Ravens and the B the black objects. Now clearly the
emptiness of [2]-which is tantamount to "All R is B"-is quite
equivalently verified by going to the R's = [2] + [3] and seeing
that only [3]'s are encountered here as by going to the non-B's =
[1] + [2] and seeing that only [1]'s are encountered here. Either
way, we are simply monitoring the emptiness of compartment [2].
There seems to be no logico-theoretical reason for granting one
of these approaches a preferred status over the other.
But appearances are misleading here. For there is a crucial
informative difference between the two approaches to verifying
our hypothesis HI. To check the emptiness of [2] by the "natural" approach (through the R's) means going to the R's = [2] +
[3] and checking their color. To establish it by the "unnatural"
approach (through the non-B) means going to the non-B's = [1]
+ [2] and checking their type.
But note that the size of these compartments is critically different. The R's of [2] + [3] may run in the millions but the nonB's of [1] + [2] will run in the zillions. Hence a satisfactory
member of [2] + [3] (viz. one that is in [3] like that black raven)
will make a contribution of one-in-some-millions towards assuring that "All ravens are black." But a satisfactory member of [1]
+ [2] (viz. one that is in [2] like that white shoe) will only make
a one-in-many-zillions contribution to ascertaining that "All
ravens are black."
There is thus a striking evidential disparity between these two
strategies of verification-one that puts black ravens and white
tennis shoes on an altogether different plane in inductive contexts. Someone who hands us a verified black raven, has made a
modest but nontrivial contribution to the whole project ofverification. But the contribution of a white tennis shoe, albeit
nonzero, is vanishingly small.
As these considerations indicate, thesis (2) is the weak link in
this chain of inconsistency. It is in fact simply false that the evidentially confirmatory contribution of black ravens and white
shoes is the same. That "to the same extent" clause of thesis (2) is
simply false. Logically equivalent general propositions are indeed
inductively equivalent, but does not means that their instances will
bear equal evidential weight.
Inductive Paradoxes
11.5
227
In an influential essay of 1953, Nelson Goodman presented a

problem for the theory of inductive inference that has occasioned
a substantial literature over the ensuing years. 5 This problem is
based on a distinctly unorthodox pair of color concepts:
grue
green if examined before the temporal referencepoint to and if examined thereafter blue. (to is an
arbitrary moment of future time.)
bleen = as above but with green and blue reversed.

Goodman noted that if we do our inductive reasoning on the
basis of this color (or "grulor") taxonomy, we shall seemingly
obtain excellent inductive support for the thesis that all emeralds
will eventually have the appearance we have standardly indicated
by the description "blue" (after tQ)-since all that have been
examined to date are grue. On this basis, our "normal" inductive
expectations would be totally baffled, and we arrive at a Humelike result from a rather un-Humean point of departure. 6 Nature
becomes predictively inscrutable.
Various considerations mark this issue as an inductive paradox
that cuts deeper than one at first might think:
1. No amount of empirical evidence can differentiate the
"abnormal" grue/bleen color taxonomy and the normal
green/blue one. Empirical evidence relates in the nature
of things to past-or-present and there is (ex hypothesi) no
difference here.
2. We cannot object that grue and bleen make explicit reference to time (to to)' because this only seems so from our
5. See Goodman 1955 and especially Stalker 1999.

6. The situation would be quite otherwise if the "big change" came about in a gradual
and unnoticed way, as a slow shift with no rude shocks to our memory of how things used
to look, or else with a gradual readjustment, so that the recollection of discrepancies would
fade away in the manner of an image-inversion experiment. For then we would continue to
make our inductive projections on the old basis; for example, grass would still be described
as green, even thought it "really" looks blue.
228
Chapter 11
own parochial standpoint. From the perspective of the

grue-bleeners, it is our color taxonomy that is then timedependent:
green
examined before to and grue or not examined

before to and bleen.
blue
examined before to and bleen or not examined

before to and grue.
Neither perspective is evidentially privileged. The situation is

entirely symmetric from their perspective, and what's sauce for the
goose is also sauce for the gander as far as any theoretical objections on the basis of general principles go. The case thus looks to
be one of total parity as between our color talk and that of the
grue-bleeners, with no theoretical advantage available to decide
the case between us.
In the face of this parallelism Goodman himself effectively gave
up on finding a preferential rationale for choice on a basis of theoretical general principles. Instead he relied on an appeal to the
essentially practical factor of "linguistic entrenchment." The pivotal consideration is accordingly seen to lie in the fact that custom
and habit has settled the issue preemptively in favor of our familiar
color discourse. The descriptors blue/green unlike grue/bleen
have a securely established place in our actual communicative
practice. However, few commentators have found this resolution
by an appeal to the mere custom of things to be convincing?
But let us reexamine the paradox from the standpoint of the
present approach. In effect it invites us to contemplate two conflicting general claims:
(A) All emeralds are green.
(B) All emeralds are grue.
7. For a survey of objections and positions see Henry E. Kyburg, Jr., "Recent Work
on Inductive Logic," (American Philosophical {bIarterly, vol. 1 [1964]), who summarizes
his discussion with the observation "The problem of finding some way of distinguishing
between sensible predicates like 'blue' and 'green' and the outlandish ones suggested by
Goodman, Barker, and others, is surely one of the most important problems to come out
of recent discussions of inductive logic" (p. 266).
Inductive Paradoxes
229
And now consider the following aporetic cluster:

(1) The available evidence speaks equally for A and for B (seeing that as far as our observational evidence goes each is
confirmed-and confirmed equally-by exactly the same
supporting instances).
(2) Claims for which the available evidence speaks equally are
equally credible.
(3) A is highly credible.
(4) B is not credible.
(5) (4) and (5) are incompatible.
Goodman himself effectively resolves this perplex by holding that
(1) is the weak spot here. For, as just indicated, he saw a generalization's evidentiation as hinging on the linguistic "entrenchment" of the predicates in whose terms the generalization is
formulated. On this basis it becomes a pivotal consideration of
Goodman that "green" is a widely used familiar and customary
conception. For Goodman, principles of inductive reasoning are
"justified by their conformity with our inductive practice"-as
indeed holds analogously with principles of deductive reasoning
also. s Goodman characterizes this conformity with practice as
"entrenchment," and in this regard "green" wins out easily over
"grue" because, plainly, "green," as the subject of innumerably
more prior projections (predictions), has a more impressive biography. The predicate "green," we may say, is much better
entrenched then the predicate "grue."9
However, the basic problem with this approach-and the reason why most theorists have found it unsatisfying-is that accustomed "entrenchment" as such does not seem able to bear the
weight of the structure that is being built upon it. For the crux is
clearly a matter not simply of the fact that a predicate is
entrenched but of the underlying issue of why it is entrenched. It
is a matter not just of custom but of the greater utility and effi8. See Goodman 1955(3), p. 63.
9. Ibid., p. 94.
230
Chapter 11
cacy in application on the basis of which the custom of a rational

community is ultimately shaped.
Yet notwithstanding whatever problems it may have,
Goodman's overall analysis surely moves in the right direction.
For consider our aporetic cluster {(1), (2), (3), (4)}. Here (2) is
to all intents and purposes a non-negotiable principle of inductive
rationality. And (3) and (4) are "facts of life" that we have litde
choice but to accept at face value. We thus arrive at the priority
ranking: (2) > [(3), (4)] > (1). Thesis (1), with its problematic
supposition that evidentiation turns only on the amount (and not
the quality!) of available instantiation, is thus the most vulnerable
contention here. IO And so, irrespective of whether one agrees
with Goodman's diagnosis of the rationale of (1)'s vulnerability
and remains sceptical about his reason-why analysis of the situation, the fact remains that (1) is the weakest chink in the armor of
argumentation here. I I
As this survey of examples indicates, the same general methodology of paradox analysis that operates elsewhere applies in these
evidential and inductive situations as well. Here too we encounter
an aporetic overcommitment that can be resolved-insofar as such
resolution is possible-by looking to plausibilistic considerations
to identify the vulnerable members of the circle of inconsistency
through which the aporetic perplexity at issue arises. Careful heed
of the considerations that support comparative assessments of
plausibility can prove critically useful here.
An important point deserves emphasis here. The various flaws
we have been considering-vagueness, equivocation, unjustified
presupposition, problematic self-reference, counterfactual ambiguity, evidential fragility, and so forth-are not guarantors of paradox
any more than crossing a street blindfolded guarantees an accident.
But all of these are risk factors: they issue an invitation to paradox
that is all too readily accepted. We are well advised to post a warning "Thin ice!" wherever these factors come upon the scene.
10. Note that (1) suffers from the frailty that its invocation of equally weighty support
it exhibits exactly the same flaw encountered in Hempel's Paradox of the Ravens. (See pp.
224-26.)
11. See Stalker 1999.

Counterfactual Hypothesis Paradoxes (various)
Reductio Style Paradoxes
The Incommensurability of the Diagonal

Thompson's Lamp Paradox
Per-Impossibile Style Paradoxes
Paradoxes of Hypothetical
Reasoning (The Problem
of Assumptions that Conflict
12.1
Hypotheses and Burley's Principle
To this point the focus has been upon paradoxes that arise in contexts where the choice between retention and abandonment is a
matter of how loudly the probative considerations of evidence and
conceptual cogency speak for the truth or plausibility of the relevant contentions. But we shall now deal with contexts of supposition-of postulation, assumption, or hypothesis where claims of
actual truth are put aside and disbelief suspended. When a statement is stipulated in this way it thereby acquires a priority status
that is independent of and indifferent to whatever supportive
indication for its truth there mayor may not be in other regards.
In virtue of their status as such, hypotheses acquire an automatic
priority in our reasoning. Like it or not we have to take them in
stride and make the best and most of things on that basis-at least
for the time being. l
A "counterfactual conditional" along the lines of "If Napoleon
had stayed on Elba, the battle of Waterloo would never have been
1. The treatment of suppositions presented in this chapter was initially set out in the
author's "Belief-Contravening Suppositions," The Philosophical Rc"piew, vol. 70 (1961), pp.
176-196. It was subsequently developed in Hypothetical Reasoning (Amsterdam: North
Holland, 1964).
233
234
Chapter 12
fought" is, in effect, a conditional that elicits a consequence from

an antecedent which represents a belief-contravening hypothesis. 2
Any such conditional will accordingly exhibit the same problems
and difficulties found to be generally present in aporetic situations. For the reality of it is that in the context of prevailing
beliefs counterfactual hypotheses are always paradoxical. 3
The interlinkage of our beliefs is such that belief-contravening
suppositions always function within a wider setting of accepted
beliefs B1 , B2 , . . . , Bn of such a sort that when one of them, for
simplicity say B 1 , must be abandoned owing to a hypothetical
endorsement of its negation, nevertheless the resulting group
-BI' B2 , , Bn still remains collectively inconsistent. The reason
for this lies in the logical principle of the systemic integrity offact.
For suppose that we accept B1. Then let B2 be some other claim
that we flatly reject-one that is such that we accept -B 2. Now
since we accept Bl> we will certainly also accept BI v B2. But now
consider the group of accepted theses: B 1, BI v B2, -B 2. When we
drop BI here and insert -Bl in its place we obtain -Bl' Bl v B2,
-B 2. And this group is still inconsistent.
Facts engender a dense structure, to employ the mathematicians' use of this term figuratively. Every determinable fact is so
drastically hemmed in by others that even when we erase it, it can
always be restored on the basis of what re~ains. The fabric of fact
is woven tight. Then suppose that we make only a very small
alteration in the descriptive composition of the real, say by adding
one pebble to the river bank. But which pebble? Where are we to
get it and what are we to put in its place? And where are we to
put the air or the water that this new pebble displaces? And when
we put that material in a new spot, just how are we to make room
for it? And how are we to make room for the so-displaced material? Moreover, the region within six inches of the new pebble used
to hold N pebbles. It now holds N + 1. Of which region are we to
2. Sometimes what looks like a counterfactual conditional is only so in appearance.

Thus consider "If Napoleon and Alexander the Great were fused into a single individual,
what a great general that would be!" What is at issue here is not really a counterfactual
based on the weird hypothesis ofa fusion of two people into one. Rather, what we have is
merely a rhelOricaily striking reformulation of the truism that "Anybody with all of the
military talents of Napoleon and of Alexander combined, is cenainly a great general."
3. Compare Roderick M. Chisholm, "Law Statements and Counterfactual
Inferences," Analysis, vol. 15 (1955), pp. 97-105 (see especially pp. 102--05).
235
say that it holds N - 1. If it is that region yonder, then how did

the pebble get here from there? By a miraculous instantaneous
transport? By a little boy picking it up and throwing it? But then
which little boy? And how did he get there? And if he threw it,
then what happened to the air that his throw displaced which
would otherwise have gone undisturbed? Here problems arise
without end.
Every hypothetical change in the physical make-up of the real
sets in motion a vast cascade of physical changes either in the
physical makeup of the real or in the laws of nature. For what
about the structure of the envisioning electromagnetic, thermal,
and gravitational fields? Just how are these to be preserved as they
were, given the removal and/or shift of the pebbles? How is matter to be readjusted to preserve consistency here? Or are we to do
so by changing the fundamental laws of physics?
Such deliberations indicate that we cannot make hypothetical
redistributions in the make-up of the real without thereby raising
an unending series of questions. And not only do redistributions
raise problems but even mere erasures, mere cancellations do so
as well because reality being as is they require redistributions to
follow in their wake. If by hypothesis we zap that book out of
existence on the shelf, then what is it that supports the others?
And at what stage of its production did it first disappear? And if
it just vanished a moment ago then what of the law of the conservation of matter? And whence the material that is now in that
book-denuded space? Once more we embark upon an endless
journey.
Their density means that facts are so closely intermeshed with
each other as to form a connected network. Any change anywhere
has reverberations everywhere. This condition of things is old
news. Already in his influential Treatise on Obligations 4 the
medieval scholastic philosopher Walter Burley (ca. I27S-ca.
1345) laid down the rule: when a false contingent proposition is
posited, one can prove any false proposition that is compatible with it.
His reasoning was as follows. Let the facts be that:
4. Translated in part in N. Kretzman and E. Srump, The Cambridge Translation of

Medief1lJl Philosophical Texts, Vol. I: Logic and Philosophy of Language (Cambridge:
Cambridge University Press, 1988), see pp. 389--412.
236
Chapter 12
(P) You are not in Rome.

(Q) You are not a bishop.
And now, of course, also:

(R) You are not in Rome or you are a bishop. (Por not-Q)
All of these, so we suppose, are true. Let us now posit by way of a

(false) supposition that:
Not-(P) You are in Rome.
Obviously (P) must now be abandoned-"by hypothesis." But
nevertheless from (R) and not-(P) we obtain:
You are a bishop. (Not- Q)
And in view of thesis (Q) this is, of course, false. We have thus
obtained not- Q where Qis an arbitrary true proposition.
It is clear that this situation obtains in general. For let p and q
be any two (arbitrary but nonequivalent) facts. Then all of the following facts will also of course obtain: -( -p), p & q, p v q, p v -q v
r, -p v q, -( -p & q), and so forth. Let us focus upon just three of
these available facts:
(1) P
(2) q
(3) -( -p & q), or equivalently p v -q
Now let it be that you are going to suppose not-po Then of
course you must remove (1) from the list of accepted facts and
substitute:
(I') -p
But there is now no stopping. For together with (3) this new item
at once yields -q, contrary to (2). Thus that supposition of ours
that runs contrary to accepted fact (viz., not-p) has the direct consequence that any other arbitrary truth must also be abandoned.
237
On this basis Burley's Principle has far-reaching implications.

As far as the logic of the situation is concerned, you cannot
change anything in the domain of fact without endangering
everything. Once you embark on a contrary-to-fact assumption,
then as far as pure logic is concerned all bets are off. Nothing is
safe any more. To maintain consistency you must revamp the
entire fabric of fact, which is to say that you confront quite a sizable task. (This is something that people who speculate about
other possible worlds all too easily forget.)
The reality of it is that all counterfactual propositions are contextually ambiguous. For once we assume that not-p in a context
where p is already in place, we embark on an endless succession of
yet-unresolved choices. Let it be that we accept p. Then for any
arbitrary q we are also committed to p v q. And let it be that we
accept not-q. We thus (by hypothesis) stand committed to:
p
pvq
not-q
But now when p is replaced by not-p so that the first thesis is
negated, we shall have to reject at least one of the other two. But
logic alone does not tell us which way to go. The situation is
totally ambiguous in this regard. When an inconsistency arises, all
that logic can do is to insist that consistency must be restored. It
will not tell us how to do this. This requires some altogether
extra-logical resource.
And this is always the case. If we assume, say "Napoleon died
in 1921" we have to choose between keeping his age (and thus
having him born in 1769) and keeping his birthday (and thus
having him live to age 152). And nothing about the counterfactual itself provides any guidance here. The situation in this regard
is totally ambiguous and indeterminate: we are simply left in the
dark about implementing the counterfactual assumption at issue.
To decide matters in one way or the other and remove the ambiguity at issue we need to have additional, hypothesis-transcending
information-a mechanism of precedence and priority to enable
us to chose among the alternatives that confront us. Let us see
238
Chapter 12
how this sort of program is to be implemented in the context of

counterfactual conditionals.
12.2
Counterfactual Conditionals
Consider the counterfactual "If the letters A and B were indistinguishable-say that both alike looked like X-then the random
series ABBABAAAB. . . would become the uniform series
XXXX. ... " The situation issues in the following paradox:
(1) The given A-B series is not uniform. (A known fact.)
(2) A is indistinguishable from B. (The issue-definitive supposition, which contradicts our realization that A can in fact
be distinguished from B.)
(3) The series is uniform. (Given (2) every A-B series is now
uniform.)
Since (2)'s consequence (3) contradicts (1), the pair I( 1), (2) I
represents an aporetic duo. But owing to (2)'s priority status as an
issue-definitive hypothesis, (1) must give way to it. And so we
arrive at the counterfactual conditional under consideration.
Again, consider the more complex situation of counterfactual
reasoning in the context of the conditional: "if this stick were
made of copper, it would conduct electricity." The situation is as
follows:
(1) This stick is made of wood.
(2) This stick is not made of copper.
(3) Wood does not conduct electricity.
(4) Copper does conduct electricity.
(5) This stick does not conduct electricity.
And now let us introduce the (1 )-modifying assumption:
( 6) This stick is made of copper.
How is consistency now to be restored?
239
Note that our initial givens fall into two groups: general laws
(3), (4), and particular facts (1), (2), (5). Now when (6) is introduced as an issue-definitive hypothesis we of course have to abandon (1) and (2) in the wake of this assumption. But that still does
not restore consistency since (6) and (4) still yield not-( 5).
However, the aforementioned standard policy of prioritizing more
general principles such as laws over particular facts in counterfactual contexts leads to the priority schedule: [(3), (4)] > (5). And
this means that it is (5) rather than (4) that should now be abandoned. We thus arrive at the natural (5 )-rejecting counterfactual
conditional:
If this stick were made of copper, then it would conduct
electricity (since copper conducts electricity).
in place of the "unnatural," (4 )-rejecting counterfactual conditional:
If this stick were made of copper, then copper would not
conduct electricity (since this stick does not conduct electricity).
Suppose, however, that for the sake of contrast we take the radical
step of altering the fabric of natural law by contemplating the
assumption:
(6') Wood conducts electricity.
We would now of course have to abandon (3) in the wake of this
issue-definitive hypothesis. But again this is not enough, seeing
that (6') and (1) will again yield not-(5). At this point, however,
we must resort to the principle: in counterfactual contexts, particular mode-oj-composition statements take priority over particular
mode-of-behavior-statements.5 And this means that the priority
schedule will be: (6') > (1) > (5). Accordingly, it is once more
5. But why should thing-type characterizations be taken as more fundamental than

mode-of-behavior characterizations? The answer lies in the processual perspective that to
be an X is to behave as X's do: that to be wooden is to behave as wooden things do. Since
such systemic comportment encompasses the whole range of relevant lawful behavior, it is
240
Chapter 12
(5) that should be abandoned. We thus arrive at the "natural"

counterfactual :
Ifwood conducted electricity, then this stick would conduct electricity (since it is made of wood)
in place of the "unnatural" counterfactual:
Ifwood conducted electricity, then this stick would not be
made of wood (since it does not conduct electricity).
Since alternative outcomes are always possible in such cases, a
logical analysis of the situation will not of itself be sufficient to
eliminate the basic indeterminacy inherent in counterfactual situations. We once again require a principle of precedence and priority to spell out what has to give way in case of a clash in conflict.
Observe that the choice between (A) and (B) above offers us
an option between abandoning a law of nature ("Copper conducts electricity") and readjusting the features of a particular
object, that stick. Following the policy of minimizing the scope of
environmental changes in the wake of hypotheses we will obviously opt for the second alternative.
Now as Chapter 3 has indicated, the general policy of prioritization in factual contexts is to give established facts precedence
over otherwise well-confirmed generalizations. A kind of specificity favoritism is at work here. But in counterfactual contexts
this prioritization is reversed. When we play fast and loose with
the world's facts we at least need the security of keeping its laws
in place. Accordingly it is standard policy that in counterfactual
contexts well-confirmed generalizations take priority over particular
facts.
To deal effectively with counterfactual conditionals we must be
in a position to distinguish, within the group of logically eligible
alternatives, between more and less "natural" ways of reconciling
a belief-contravening hypothesis with the entire set of residual
beliefs which continue to be collectively inconsistent with it. And
inherendy more general than a specific item-and one particular mode of behavior. The
prioritization at issue is thus provided for by the standard policy of giving precedence to
what is more general, pervasive, and fundamental.
241
once the problem of counterfactuals is seen in this light, its assimilation to the broader issue of aporetic reasoning becomes
straightforward, seeing that counterfactual conditionals cry out
for analysis in terms of the priority-geared machinery of paradoxmanagement with which we have been operating throughout. (It
is just that the principles of priority determination are distinctively
different in this domain.)
For the sake of another illustration, consider the counterfactual "If Puerto Rico were a state of the Union, the U.s. would
have 51 states." Here the following three propositions may be
taken as known givens:
( 1) Puerto Rico is not a state of the Union.
(2) There are 50 states.
(3) The list of the states includes all the following: Alabama,
Arizona, and so on.
(4) The list of the states includes only th~ following: Alabama,
Arizona, and so on.
Now consider introducing the fact-contravening supposition: suppose that not-(1), that is, suppose Puerto Rico to be a state.
In embarking on this assumption we obviously now have to
jettison (1) and (4) and replace them by their negation. But this is
clearly not enough. The set {not-(l), (2), (3)} still constitutes an
aporetic trio. To keep within the limits of consistency we must
also abandon either (2) or (3). Of course we have a choice
between these two. And with it we arrive at a choice between two
alternative counterfactuals.
(A) If Puerto Rico were a state, then there would be 51 states
since all of the following are states: Alabama, Arizona, and
so on. (Here we retain (3) and abandon (2).)
(B) If Puerto Rico were a state, then one of the present states
would have departed, since there are just 50 states. (Here
we retain (2) and abandon (3).)
Now it is clear that the first of these is plausible and the second
distinctly less so. The question comes down to a matter of priority
242
Chapter 12
and precedence. And the answer is (3) > (2), its justifYing rationale being as follows: The number of states has repeatedly
changed throughout U.S. history as new states were added to the
Union. But no state has ever dropped out-and the attempt on
the part of some to do so proved a disastrous failure. The generalization "States can enter the union" is a more tenable contention
than "States can leave the union." The crux is that in these counterfactual contexts (unlike the fictional case), well-established
generalizations take precedence over particular facts. Hence (3) is
distinctly more tenable than (2) and the first of the indicated
counterfactuals accordingly is more eligible then the second.
The example accordingly illustrates a general situation.
Counterfactuals arise when assumptions that conflict with some
of the factual beliefs at our disposal are introduced into the wider
setting of other accepted beliefs. And the process of consistency
restoration via plausibility that is effective with other paradoxes
also applies straightforwardly and constructively in this present
case-subject, however, to the crucial considerations (i) that
hypotheses and their logico-conceptual consequences now
assume a position of prime priority and (ii) that the usual principle of specificity prioritization familiar from factual contexts is
now reversed seeing that well-established generalizations will
take precedence here.
12.3
Further Examples
Consider the counterfactual contention: "if the Eiffel Tower were

in Manhattan, then it would be in New York State." This conditional introduces a fact-contravening hypothesis
The Eiffel Tower is in Manhattan.
into a context where the following statements are in place as
accepted truths:
(1) The Eiffel Tower is in Paris, France.
(2) The Eiffel Tower is not in Manhattan.
(3) Manhattan is in New York State.
243
As usual, the result of replacing a thesis by its counterfactual

negation still leaves an obviously inconsistent situation.
How is this inconsistency to be overcome? We are now confronted by the need to choose between (1) and (3). Here (3) will
have the upper hand as long as general geographic facts such as
(3) are granted priority over the location of specific structures
such of the sort at issue with (1). Accordingly, (1) would have to
be abandoned as well. And we would thus arrive at the conditional: "If the Eiffel Tower were in Manhattan, then it would not
be in France but in New York State."
In theory, to be sure, our counterfactual hypothesis leaves two
alternatives open: namely the more natural "If the Eiffel Tower
were in Manhattan then it would be in New York State" and its
anomalous reverse: "If the Eiffel Tower were in Manhattan then
Manhattan Island would be in Paris (just like the Isle de la Cite)."
In the former case we would leave Manhattan Island in New York
State, while in the latter we would have to shift it to Paris. We
have a choice between moving an individual structure and moving
a whole island. And subject to the policy of minimizing change by
granting precedence to which is more general and fundamental
we keep that island in place.
The preferred resolution here is underwritten by a definite priority order: Hypotheses take precedence, and beyond this systemic fundamentality is our guide. And this policy enables us both
to validate counterfactual conditionals and to explain how it is
that some counterfactuals are natural and acceptable and others
unnatural and unacceptable.
This distinction between "natural" and "unnatural" counterfactuals is crucial. For the sake of a further illustration consider
the following example due to David Lewis. The case at issue is by
stipulation one where we are taken to know:
(1)
J.F. Kennedy was assassinated.
(2) L.H. Oswald assassinated Kennedy.

(3) No one other than Oswald assassinated Kennedy.
Suppose now that we are instructed to suppose not- (2), and
assume that Kennedy was not killed by Oswald. Then we clearly
cannot retain both (1) and (3), since in the presence of not-(2),
244
Chapter 12
(3) entails that no one assassinated Kennedy which contradicts

(1). Either (1) or (3) must go--one must be subordinated to the
other. And now the very way in which a counterfactual is formu1ated instructs us as to the appropriate resolutions:
(A) If Oswald did not assassinate Kennedy, then someone else
did. (Subordinates (3) to (1).)
(B) If Oswald had not assassinated Kennedy, then Kennedy

would not have been assassinated at all. (Subordinates (1)
to (3).)
However, if we were to supplement our beliefs (1)-(3) with a
conspiracy theory by way of adopting
(4) Kennedy was the assassination victim of a successful
conspiracy.
then we would also arrive at
(C) If Oswald had not assassinated Kennedy, then someone
else would have. (Subordinates (3) to (4).)
The very way in which these conditionals are formulated informs
us about (and corresponds to) the sorts of subordination relationships that are at work among those "factual" items that we take
ourselves to know within the information-context of the counterfactual at issue.
12.4 The Crucial Difference between Factual and

Hypothetical Contexts
It's important for the proper understanding of counterfactual reasoning to note the difference between reasoning from purely
hypothetical and reasoning from putatively actual counterinstances. For these lead to very different results in the face of real
or supposed laws of nature. Thus assume as given a law of the
form: All X's are Ys (say "All copper bodies are electricity conductors" or "Copper conducts electricity" for short). If in the
245
course of empirical inquiry into their status we were to discover a

copper object that did not conduct electricity, then of course we
would have to withdraw and revise that universal generalization,
thereby making it yield way to the newly discovered observational
fact. (Here we have the situation noted in Herbert Spencer's quip
that Henry Buckle's idea of a tragedy is a promising theory
destroyed by a recalcitrant fact.) In counterfactual contexts, however, a different rule of prioritization prevails. The priority situation of empirical inquiry ("Laws give way to facts") is thus
radically altered. For while putative laws viewed as mere theories
do and must give way to real facts, accepted laws need not and
will not give way to merely hypothetical or suppositional facts.
The preeminent status enjoyed by suppositions in hypothetical
contexts has other ramifications. When we encounter a disjunctively underdetermined paradox in matters of factual inquiry, we
see this as a sign of incomplete information and an invitation to
look for additional considerations to effect a further definite resolution. But with hypothetical contexts we have to take such
underdetermination at face value: we have no choice but to see it
as final.
12.5
Reductio-ad-Absurdum Reasoning
Reductio-ad-absurdum reasoning is also a version of reasoning

from a belief-contravening supposition-but now with a very particular sort of end in view. For the aim of the enterprise is to
establish a certain thesis T. To do so we proceed as follows. We
begin by assuming not- T by way of a hypothetical supposition.
From not- T, together with certain pre-established facts or principles PI' P2 , . , Pn that are already at our disposal, we then derive
a contradiction:
not- T, Pi' P2 ,
... ,
Pn ~ contradiction
To restore consistency to the group of theses to the left of the ~

symbol we must abandon (and thereby endorse the denial of) at
least one of them. But since all of the theses PI' P2 , , Pn are (by
hypothesis) established principles, while not- T is no more than a
tentatively adopted provisional assumption, we grant priority to
246
Chapter 12
those preestablished principles, and must accordingly abandon

not- T. On this basis we can now class T itself as an established
fact.
Here once again we have an aporetic cluster {not- T, PI> P2 ,
... , Pn } and resolve the inconsistency by the standard process of
"breaking the chain of inconsistency at its weakest link." On this
basis, not- T emerges as an inexorable result of the only acceptable
option for restoring consistency.
An example will help to clarify the issue. A classic instance of
reductio reasoning in Greek mathematics relates to the Pythagorean
discovery-revealed to the diagram of his associates by Hippasus
of Metapontum in the fifth century B.c.--of the incommensurability of the diagonal of a square with its sides. The reasoning at
issue runs as follows:
Let d be the length of the diagonal of a square and s the length
of its sides. Then by the Pythagorean theorem we have it that
d 2 = 2s2. Now suppose (by way of a reductio assumption) that
d and s were commensurable in terms of a common unit n, so
that d = n x u and s = m x u, where m and n are whole numbers (integers) that have no common divisor. (If common divisor there were, we could simply shift it into u.) Now we know
that
( n x U)2 = 2 ( m x U)2
We then have it that n 2 = 2m2 . This means that n must be

even, since only even integers have even squares. So n = 2k.
But now n 2 = (2k)2 = 4k2 = 2m2, so that 2k2 = m 2. But this
means that m must be even (by the same reasoning as before).
And this means that m and n, both being even, will have common divisors (namely 2), contrary to the hypothesis that they
do not. Accordingly, since that initial commensurability
assumption engendered a contradiction, we have no alternative
but to reject it. The incommensurability thesis is accordingly
proven. 6
6. On the historic background see T.L. Heath, A History of Greek Mathematics

(Oxford: Clarendon, 1921).
247
In mathematics this sort of proof of a fact by deriving a contribution from its negation is characterized as an indirect proof.
The contrast between hypothetical and reductio reasoning is
instructive. In hypothetical reasoning the assumptions we make
are issue definitive stipulations and consequently allowed by fiat to
prevail come what may. In reductio reasoning, however, our
assumptions are merely provisional and must in the end give way
in cases of a conflict with established facts. The situation is thus
altered radically from that of counterfactual reasoning. There
assumptions were seen as fixed points around which everything
else had to revolve. Their priority was absolute. With reductio ad
absurdum reasoning, however, the matter is reversed. Here
assumptions are viewed as merely provisional hypotheses and
accordingly become frail and vulnerable: they stand at the foot of
the precedence-priority scale. And the circumstance that established propositions prevail over mere hypotheses in this context
makes reductio-ad-absurdum reasoning a relatively straightforward business.
12.6 Thompson's Lamp as an Illustration of Reductio

Reasoning
The Thompson)s Lamp Paradox was suggested by the English
philosopher James Thompson,? who posed the following
question:
A lamp has two settings: ON and OFF. Initially it is ON.
During the next 1/2 second it is switched OFF. During the
subsequent 1/4 second it is switched ON. And so on with a
change of switch setting over every interval half as long as the
preceding, alternating ON and OFF. Question: What is its setting at exactly one second after the start?
This situation gives rise to the following apory:
7. See James F. Thompson, "Tasks and Super-Tasks," Analysis, vol. 15 (1954), pp.
1-13; reprinted in R-M. Gale, The Philosophy of Time (London: Macmillan, 1968).
248
Chapter 12
(1) A lamp of the hypothesized kind is possible.

(2) At any given time, the lamp is ON or OFF, but not both.
Moreover,
(3) Physical processes are continuous. A physical condition
that prevails at some time within every E -sized interval
prior to t, no matter how small E may be, will prevail at t
as well.
(4) Within every E -sized interval prior to t = 1 second the
lamp is frequently ON.
(5) Within every E -sized interval prior to t= 1 second the
lamp is frequently OFF.
(6) At
t=
1, the lamp is ON. (By (3) and (4))
(7) At t = 1 the lamp is OFF. (By (3) and (5))

(8) (5) and (6) contradict one another.
Here {(1), (2), (3), (4), (5)} is an inconsistent cluster. Now in the
context of reductio argumentation, the basic hypothesis (1) has to
be seen as no better than the provisional assumption of a merely
plausible supposition, while (2), (4), (5) are defining stipulations
of the problem. (3) is a fundamental physical principle. The
resulting priority profile is [(2), (4), (5)] > (3) > (1). Clearly (1)
has to go here and we can interpret the apory in question as a
reductio ad absurdum of that conjectural lamp.
A variant prospective also yields the same result. For consider:
(1) A lamp of the hypothesized kind is possible.
(2) At any given time the lamp is ON or OFF but not both.
(3) The lamp situation is entirely uniform (symmetric) as
between ON and OFF. (It is only a matter of which interval we choose to call "the first.")
(4) Symmetric conditions yield symmetric results.
(5) Therefore: the final, t = 1 setting of the lamp must be
indifferent as between ON and OFF.
249
Here {(I), (2), (3), (4)} is an inconsistent quartet. Moreover (2)

and (3) are defining stipulations of the problem, (4) is a fundamental principle of physics, and (1) is a plausible supposition. We
thus have the priority profile: [(2), (3)] >(4) > (1). Once more
( 1) has to be abandoned. The lamp is again abolished through a
reductio. And of course once the lamp is abolished so is the issue
of its supposed comportment.
And the physical situation itself lends further credence to this
result. For one thing, it would ultimately require unendingly
swifter velocities to accomplish that hypothetical switching-in
clear violation of the strictures of special relativity theory. For
another, the supposition that the lamp is ON or OFF at any given
time unrealistically sidelines the prospect that it might simply
blow up-that is, cease to exist as a lamp. Underlying the analysis
is thus the reality that one would not be prepared to sacrifice
physical principles fundamental here to a mere "thought experiment" such as Thompson's Lamp.
The lesson here is that when a hypothesis-engendered paradoxical situation becomes too paradoxical, the appropriate course
may be to dissolve the paradox by concluding that the underlying
suppositions on which it rests are simply inappropriate. Here the
paradox at issue effectively self-destructs by constituting its own
reductio ad absurdum.
One final-and important-point. When p yields a contradiction upon being superadded to a family of true propositions we
can infer not-po (That is exactly how reductio argumentation
works.) However, when p yields a contradiction upon being
superadded to a family of plausible propositions, the situation is
different. For everything now depends on the priority ranking of
the proposition at issue. Only when p is of minimum priority can
not-p now be inferred: otherwise the onus of untenability falls
elsewhere. Truths cannot be destabilized by incompatible additions. But with merely plausible propositions everything will
depend on considerations of precedence and priority.
12.6
Per-Impossibile Reasoning
Impossible suppositions are not just false but necessarily false, that
is, in logical conflict with some necessary truths, be the necessity
250
Chapter 12
at issue logical or conceptual or mathematical or physical. Thus a

counterfactual's antecedent may negate:
a matter of (logico-conceptual) necessity ("There are infinitely many prime numbers").
a law of nature ("Water freezes at low temperatures").
Suppositions of this sort give rise to per impossibile counterfactuals
(in the physical and conceptual modes, respectively). "If (per
impossibile) water did not freeze, then ice could not exist" and "If
(per impossible) there were only finitely many primes, then there
would be a largest prime number" afford examples. 8 Nevertheless,
in validating such counterfactuals we proceed-as usual-by principles of priority that keep the more fundamental principles of the
domain intact.
Thus consider the counterfactual, "If 2 + 1 were even, then (2
+ 1) + 1 would be odd. Here we have the aporetic cluster:
(1) 2 + 1 is odd, not even.
(2) (2 + 1) + 1 is even, not odd.
(3) The successor of an integer N is N + 1.
(4) Whenever an integer is even, its successor is odd.
(5) Whenever an integer is even, its successor is odd.
(6) By supposition: 2 + 1 is even.
In view of (6), we must abandon (1). But the remaining group
(2 )-( 6) is still inconsistent since the trio (6), (3), (4) yields that
(2 + 1) + 1 is odd, contrary to (2). Now here (6) is a prob1emdefinitive supposition, (3) a definition, (1) a particular arithmetic
fact, and (4 )-( 5) general principles (arithmetical laws). The resulting priority situation is (6) > (3) > [(4), (5)] > (1) with the result
that (2) must be abandoned. And this validates the counterfactual
8. A somewhat more interesting mathematical example is as follows:

If, per impossibile, there were a counterexample to Fermat's Last Theorem, there
would be infinitely many counterexamples, because if x' + y' ~ z', then (nx)' + (ny)' (nz)', for any k.
251
at issue: "If (per impossible) 2 + 1 were even, then (2 + 1) + 1

would be odd."
With both reductio and with per impossibile reasoning we introduce an "absurd" or "logically untenable" hypothesis. But the
aim of the enterprise is different. With reductio argumentation the
aim is to make this absurdity patent through the detection of
impossible consequences. With per impossible reason the aim is to
indicate interesting consequences that bring instructive implications of the "impossible" supposition to light. With reductio the
derivation of a contradiction is essential to the project, with per
impossible reasoning it defeats the aims of the enterprise and frustrates its very reason for being.
Consider such counterfactuals as:
If (per impossible) 9 were divisible by 4 without a remainder, then it would be an even number.
If (per impossible) Napoleon were still alive today, he would
be surprised at the state of international politics in Europe.
Our real interest in such cases is not in the impossible antecedent
but in the general status of the consequence. A virtually equivalent formulation of the very point at issue with these two contentions is:
Any number divisible by 4 without remainders is even.
By the standards of Napoleonic France the present state of
international politics in Europe is amazing.
Per impossibile suppositions pose special challenges when the

impossibility at issue is not merely one that runs afoul of laws of
nature ("Suppose one could travel at a speed faster than the speed
of light") but violates matters of logico-conceptual necessity
("Suppose there were a circle with zero radius"). For in the latter
case we need to distinguish between more and less fundamental
principles-as we need generally not do with matters of logicoconceptual necessity.
An interesting lesson emerges here. Both with reductio ad
absurdum and with per impossibile reasoning we set out from an
"impossible" or "absurd" assumption that stands in contradiction
252
Chapter 12
with what we know. But there is a crucial purposive difference

between the two modes of thought. (Whoever said that purposes
do not enter into logic?) In the case of reductio the aim is to
establish the absurdity at issue by indicating that and how the contradiction in question follows. In the case of per impossibile reasoning, the impossibility is conceded but waived. We simply want
to show that a certain consequence follows. Our interest is not
really in the impossible antecedent, but in the larger import of the
consequent.

The Voting Paradox
The Demands of Reason Paradox
The "Chicken" Paradox
The Unachievable Wisdom Paradox
The Prisoner's Dilemma Paradox
The Predictive Competition Paradox
Allais's Paradox
The St. Petersburg Paradox
The Buridan's Ass Paradox
Paradoxes of Choice
and Decision
(Conflicting Reasons
for Action)
13.1
A Sample Paradox of Decision
Matters of decision and action will have to be subject to the same

consistency-preserving safeguards that obtain with matters of
assertion and denial, and the same general processes are operative
in both cases alike. The so-called Voting Paradox is a prime example of a paradox of decision. l It roots in the fact that an otherwise
natural use of the principle of majority rule can lead to a contradictory result. For example:
Three individuals agree (unanimously) in the abstract that one
of them should enjoy a certain privilege. Yet in no case are the
other two willing to let the third enjoy this benefit.
The paradox runs as follows:
(1) Whatever the majority agrees upon is to be done.
(2) Everyone agrees that one of the trio should do X.
(3) By (1) and (2): One of the trio is to do X.
1. On paradoxes of voting, decision, and political processes in general see Steven
Brarns, Paradoxes in Politics (New York: The Free Press, 1976).
255
J.
256
Chapter 13
( 4) A majority is against having trio member # 1 do X.

(5) By (1) and (4), trio member #1 is not to do X.
(6) As with #1 in theses (4)-(5) so also with #2 and #3. By
the "majority rule" principle of (1), they too are not to do
X.
(7) By (5), (6), none of the three is to do X.
Here {( 1), (2), (4) plus its two analogues} constitutes an inconsistent quintet. But since (2), (4), and the (4 )-analogues are simply
given facts, while (1) is no more than a plausible principle, it is
clear that (1) should be sacrificed here. Evidently the principle of
majority-rule cannot be maintained flat-out, but only qualifiedly,
subject to its restriction to circumstances where problems of
implementational infeasibility do not ensue.
Decision paradoxes often root in conflicting evaluations. They
frequently relate to conflicts of advantage arising when there are
conflicting pro's and con's regarding competing alternatives, so
that different ones win out from alternative points of view, while
nevertheless the prospect of a combined or overall point of view
is impractical. In these circumstances one of those conflicting
assessments must simply be subordinated or sacrificed to the
other, unless (more drastically) both of them are abandoned
together.
Paradoxes of decision have played a particularly prominent
part in twentieth-century philosophy because of their instructive
bearing upon one of its central topics: rationality. The determination of the rational thing to do in various sorts of problematic circumstances and conditions is, after all, one of the best ways of
approaching this topic. It is here instructive as a matter of stagesetting to begin with what might be called The Demands of
Reason Paradox which arises from the potential conflict between
real and apparent advantage:
(1) Reason requires us to choose the (really and actually) best
available alternative.
Paradoxes of Choice and Decision
257
(2) We can do no more toward determining what is best than

to determine what looks best: that is, we can get no closer
to the real optimum that to determine the apparent
optimum.
(3) In letting the apparent optimum stand surrogate for the
real optimum, we might very well be off the mark: the
apparent best may well not be anything like the actual
best.
(4) Reason can appropriately demand no more of us than the
best that can be done in the circumstances.
(5) From (2) and (4) it follows that reason can do no more
than to require of us that we choose the best apparent
alternative among the discernibly available options. And
(3) indicates this apparent best might well not be the
actual best.
(6) (5) conflicts with (1)
Here {(I), (2), (3), (4)} constitutes an inconsistent quartet. Now
(2) and (3) are unquestionable facts of life, while (4) and (3) are
reasonable-looking general principles, with the latter rather more
securely in place than the former. We thus obtain the priority
ranking [(2), (3)] > (4) > (1) as a guide for resolving the contradiction at issue. The most promising way to restore consistency is
accordingly to abandon (1) and rest content with (4) because (1),
despite its seeming plausibility, is less plausible than its competitors. What reason demands of us is not the best as such, but the
best that we can effectively realize. (Note, however, that premiss
(1) would be in better shape were it to distinguish between what
it requires us to do and what it requires us to try to do.)
Decision theorists also deliberate about "Chicken," the rather
idiotic "game" that seems to have originated in the depression era
and was later popularized by the film Rebel Without a Cause set
among Californian teenagers in the 1950s. As two drivers
approach each other head on at high speed on a narrow road,
each has the choice of swerving to the right or continuing on a
collision course. The resulting outcome possibilities are:
258
Chapter 13
#1
#2
Result
swerves
swerves
persists
persists
swerves
persists
swerves
persists
a draw
#2 wins
#1 wins
a collision
The resulting Chicken Paradox runs as follows:

(1) In playing a game a player should adopt a strategy that
affords some chance of winning.
(2) Only by persisting will a player have a chance of winning
in this instance.
(3) By (1) and (2), the players will both persist.
(4) No decision maker will risk actual disaster when the best
possible outcome is a trivial gain such as "winning" in a
silly game.
( 5) Only by swerving can a player insure against disaster.
(6) Therefore the players will both swerve.
The pathway to resolution lies in noting that in deliberating about
what "a player" would or would not have to do, this has to be
understood as referring to a rational or sensible player. But of
course a rational or sensible person will not play this game at alP
The problem dissolves over this consideration that the crucial
rationality presupposition which underlies the paradox at issue is
just not satisfied.
A challenging paradox of rational decision arises from the situation at issue in the Unachievable Wisdom Paradox which emerges
from the following narrative: 3
2. To be sure, ifforced to play the "game" in circumstances where some great disaster
can only be averted by my "winning," the rational upshot would be not simply swerpe, but
swerve last. And if my opponent was in the same position, a collision would become
inevitable.
3. The example is adapted from H. Gaifman, "Infinity and Self-Applications, I,"
Erkenntnis, vol. 20 (1983), pp. 131-155 (see pp. 150-52). On this paradox see also
259
The Arranger offers you a choice between two bills of money,

one worth $1 and the other worth $10. And he says, "Let me
offer you some reassurance. You are not to worry. For if you
fail to choose wisely, I will compensate you with an additional
$10." Which bill do you choose?
The following line of reasoning looms before you:
IfI choose $1 rather than $10, that would dearly be
unwise. So it will earn me that extra $10 for a total of$l1.
Let us take the preceding reasoning at face value. Then
obviously taking that $1 will be the sensible choice-the
wise thing to do. So to act unwisely I must choose that $10
and will thereby realize a total of $20.
But let us now take step 2 reasoning at face value. This
would mean that opting for $10 is the wise choice. So for a
maximum gain I must return to step I and choose that $1
so as to settle for $11.
And this cyde goes back and forth without end. In these circumstances, it would accordingly seem that there is nothing that one
can sensibly do. But in doing nothing one would forgo the benefit of a sure gain of $10. So that does not offer a good prospect
either.
How is this perplex to be resolved? Note first of all that the
paradox that confronts us here is as follows:
(I) By the stipulations of the problem situation, proceeding
unwisely assures one of a gain greater than what one
would otherwise obtain.
(2) Maximizing one's overall gain is the wise thing to do.
(3) A wise course of action is available in the specified circumstances.
Roben C. Koons, Paradoxes of Belief and Strategic Rationality (Cambridge: Cambridge

University Press, 1992), pp. 17-19. The paradox is sometimes called "The Two Envelopes
Paradox."
260
Chapter 13
This trio is collectively inconsistent since (1) assures that the

course of unwisdom will-if available as per (3 )-be the more
rewarding, while (2) indicates that this cannot be so. Where is this
chain of contradiction to be broken?
Here (1) is fixed as a defining condition of the problem. And
(2) is more or less axiomatic as a basic principle of economic
rationality. Of the theses at issue (3) is therefore the most questionable. We thus have the priority ranking: (1) > (2) > (3). In
consequence, (3) has to give way here. We have to acknowledge
that in the circumstances there is really no such thing as "the wise
thing to do" and the arranger should come up with that bonus
come what may. (Here one might as well take the $10 and run. 4 )
So here we have another paradox of decision, one whose seeming conflict of advantage is straightforwardly resolvable on the
basis of plausibilistic considerations.
13.2 The Prisoner's Dilemma

The Prisoner)s Dilemma Paradox has been debated for over half a
century, having initially been propounded by Melvin Dresher and
Merril M. Flood of the RAND Corporation around 1950. 5 It
emerges from a story that runs as follows:
You and your accomplice have committed some crime. The two
of you are eventually arrested and charged. The public prosecutor offers you a plea bargain: Confess and turn state's evidence
4. The reasoning here is based on the following line of consideration:
I CHOOSE
$1
$10
I GET IF
THE WISE CHOICE
IS ACTUALLY $1
I GET IF
THE WISE CHOICE
IS ACTUALLY $10
I GET IF
THERE ISNO
WISE CHOICE
$1
$20
$11
$10
$10
$I
By choosing $10 I sustain a trivial comparative loss if worst comes to worst and am substantially better off in any other case.
5. For a many-sided discussion of the problem see Richmond Campbell and Lawring
Sowden (eds.), Paradoxes of Rationality and Cooperation: Prisoner's Dilemma and
Newcomb's Problem (Vancouver: University of British Columbia Press, 1985). See also
Brams 1976.
261
against your accomplice and she will ensure that the courts will
treat you leniently-provided that your confession turns out to
be useful for her case. Note further: (1) The utility of your confession depends on whether or not your accomplice keeps
silent. (Ifhe does, your confession is valuable, while ifhe too
confesses, its value is substantially diminished.) (2) If neither of
you confesses, then the prosecutor's case is in fact so weak that
both of you will almost certainly receive no more than a rather
light penalty. Accordingly we may suppose the following punishment schedule, measured in years lost in incarceration:
Action
Years of Incarceration for
You
Accomplice
You
Accomplice
confess
confess
not confess
not confess
confess
not confess
confess
not confess
5
1
10
2
5
10
1
2
It is to be assumed here that both of you are rational agents who

make choices in line with the usual principles of prudential rationality. And the paradox is that by doing the decision-theoretically
rational thing-namely opting for that alternative where you
might gain but cannot lose, no matter which way the opponent
acts-you and your similarly circumstanced opponent are condemned to confession, thereby foregoing the prospect of arriving
at an option that is dearly preferable-namely mutual nonconfession with the result of a much diminished penalty for each of you.
The following contentions make explicit the paradox that
arises here:
( 1) Other things equal, the rational thing is to prefer an
option that can only leave one better off. (A fundamental
principle of economic racionality.)
(2) In the sort of situation at issue here, it is rational to select
that "casewise dominant" option by which one comes out
ahead regardless of how the opponent chooses. (A reasonable principle of decision-theoretic procedure.)
262
Chapter 13
(3) In view of the defining features of the problem this means

that: both agents will choose C, and thereby arrive at
mutual confession with the result -5/-5. (From (1) and
(2).)
(4) Each agent would actually fare better if both kept silent
and neither one confessed. (A defining feature of the
case.)
( 5) Both agents-being rational themselves and presuming

the other to be so as well-will therefore choose nonconfessions and thereby arrive at (-2, -2).
Here the set {(I), (2), (4)) presents an aporetic cluster that needs
to be reduced to consistency. And the appropriate priority ranking
is (4) > (1) > (2). For this situation casts a shadow of doubt over
thesis (2) with its assertion of the decision-theoretic casewise
domain standard. It is thus by abandoning (2) as its weakest link
that the chain of inconsistency is optimally broken.
Yet another way of addressing the paradox at issue here is to
proceed as follows:
(1)
Other things equal, the rational thing is to prefer an

option that can only leave one better off. (A fundamental
principle of economic rationality.) [Just as before.]
(2)
In the sort of situation at issue here, it is rational to select

that "casewise dominant" option by which one comes
out ahead regardless of how the opponent chooses. (A
reasonable principle of decision-theoretic procedure.)
[Just as before.]
( 3 ) In view of the defining features of the problem this

means that: both agents will choose C, and thereby
arrive at mutual confession with the result -5/-5. (From
(1) and (2).) [Just as before.]
(4') It is inherent in the description of the situation that both
agents are in exactly the same boat: the situation is
263
entirely symmetric as between the two. (A defining feature of the problem.)

(5') Rationality is impersonal and uniform: what is rational
for one person is rational for any other who is in exactly
the same circumstances. (A general principle.)
Therefore( 6') The agents must both choose alike (that is either both C
or both not-C): these are the only "realistically available"
alternatives. (From (5).)
(7') The agents will both choose not-C and arrive at -2/-2.
(From (1), (6').)
(8') (7') contradicts (3).
This second dilemma roots in the inconsistency of {(1), (2), (4')
(5') }, and has to be viewed in the light of the following priority
schedule:
(4') > [(1), (5')] > (2)
So here once more the abandonment of the minimally plausible

(2) serves to break the chain of inconsistency.
It is clear that the problem here turns on pitting the decisiontheoretic construction of rationality via casewise dominance
against other, more elemental and common-sensical construals of
this conception.
What is noteworthy in this case is that we here have a compound paradox with two interlocking aporetic clusters, namely
the inconsistent trio {( 1), (2), (4)} and the inconsistent quartet
{(1), (2), (4'), (5')}. And while our dismissal of(2) is in each case
dictated by plausibility considerations, it is interesting-and most
convenient-that in this case one stone kills both of these aporetic
birds together (which is by no means necessarily the case).
264
13.3
Chapter 13
In 1960 the American physicist William A. Newcomb propounded a paradox that has beguiled theorists in recent decades. 6
NewcombJs Paradox is presented by the problem situations of the
following narrative:
The Predictor says: "Here are two boxes A and B. You may
pick one or both. You see that I am now putting this $1,000
into box A. And as regards box B, be informed that I have
already put in a million dollars if I predicted that you will
choose just one box, but nothing if I predicted that you would
choose both boxes. Knowing my extensive track record as a
good predictor, you would be well advised to heed that prediction." How do you choose?
This problem straightaway engenders a paradox because the following theses are all plausible in this situation.
(1) The reasonable thing to do is to pick both boxes. (After
all you will then get whatever there is to be gotten.)
(2) The reasonable thing is to be guided by expected value
calculations.
(3) An appropriate expected-value assessment can be carried
out as follows:
Let P be the probability that the Predictor has forecast your
picking both boxes. Then your gain will be as follows:
You choose A only:
$1,000
You choose B only:
p(O) + (l-p)(I,OOO,OOO) =
(l-p)( 1,000,000)
6. On Newcomb's Paradox see Robert Nozick, "Newcomb's Problem and Two

Principles of Choice" in N. Rescher (ed.), Essays in Honor of Carl G. Hempel (Dordrecht:
Reidel, 1969), pp. 114-146. See also: R. Campbell and L. Sowden, Paradoxes of
Rationality and Cooperation (op. cit.); Martin Gardner, "Mathematical Games," Scientific
American, July 1973, pp. 102-08; Isaac Levi, "Newcomb's Many Problems," Theory and
Decision, vol. 6 [1975], pp. 161-175, as well as Brams 1976. Michael D. Resnick, Choices:
An Introduction to Decision Theory (Minneapolis University Press, 1987), pp. 109-112.
You choose both:
265
the sum of the two preceding

amounts.
Here choosing both wins out-automatically so because

its expected value is greater than that of either of the alternatives.
(4) An appropriate expected -value assessment can be carried
out as follows:
Let P be the probability that the Predictor has forecast
your choice correctly. Then your expected return will be as
follows:
You choose A only:
1,000
You choose B only:
p (1,000,000) + (1 - p)[~ (0) +

~ (1,000,000)] = 500,000(1 + p)
(Note: This assumes a 50:50 division of how the predictor

goes wrong as between choosing "Both," and "A only.")
You choose both:
1,000 + p(O) + (1- p)

(1,000,000) = 1,001,0001,000,000p
Since p cannot be negative, B-only automatically wins out

over A-only. Moreover B-only will win out over Both
whenever:
500,000 (1 + p) > 1,001,000 - 1000,000 P
500,000 + 500,000 P > 1,001,00 -1000,000 P
1,500,000 P > 501,000
>
roughly one-third
On this basis it transpires that you are well advised to

choose B-only as long as you are convinced that the
Predictor has a better than roughly one-third chance of
predicting your choice correctly.
266
Chapter 13
And so, the indicated cluster of contentions again puts us into an

aporetic situation. For (2) and (3) agree with (1) in ruling
"Both," while (2) and (4) combine in ruling "B-only" in suitable
circumstances. And we clearly cannot have it both ways. Either
(4) must be abandoned or (1) and (3) must go since there are but
two A/R-alternatives: (1), (2), (3)/(4) and (2), (4)/(1), (3).
(Here (2) is inevitable, given that we are talking about economic
rationality. )
How to choose? Theses (1) and (2) to all appearances represent equiplausible construals of the demands of choice-theoretic
rationality, while (3) and (4) offer competing means for an implementing process of precedence determination. So we arrive at the
priority ranking [(1), (2)] > [(3), (4)]. On this basis (1), (2),
(3)/(4) with retention profile { h I} wins out over (2), (4)/(1),
(3) with retention profile { hi}. In effect we see (2) + (3) and
(2) + (4) as canceling each other out and let (1) carry the day.
The rationale that undergirds this analysis goes something like
this: "the probabilities at issue in (3) and (4) are problematic;
they may very possibly fail to be well-defined, meaningful quantities. (After all, both presuppose that there are correct probabilityevaluations regarding otherwise imponderable human choices.)
So let us refrain from commitment either way and let (1) be
determinative.
As we saw above, Prisoner's Dilemma casts a shadow of doubt
across the standard decision-theoretic criterion of "casewise dominance" as a universal instrument of rational choice. By contrast,
Newcomb's Problem highlights the potential shortcomings that
expected value calculations encounter in the presence of problematic probabilities.
To be sure, one can also view this problem in a very different
light. This would be to view it as a reductio ad absurdum of a predictor with the capability at issue. For if there can be one such
Predictor-short of God himself-then there could also be two of
them. And these two can then be pitted against each other in a
matching game, as follows. Each gets to circle A or B on a card. If
the cards match, Predictor No. 1 wins, if they do not, then No.2
wins. Now if No. 1 were to write A, then No.2, foreseeing this,
would write B. But then A, foreseeing this, would have to write
B, and so on. This Predictive Competition Paradox indicates that
an assumption of predictive competence with respect to other free
agents can be deeply problematic.
267
13.4 Allais's Paradox and the St. Petersburg Paradox

The inner stresses of the standard theory of rational decision are
further illustrated by Altais)s Paradox, which emerges from the
following considerations:
(1) Rational people make their choices in line with expectedvalue contributions.
(2) Consider a choice between two alternatives. If you pick
alternative one you get a million dollars minus a couple of
dollars. If you pick alternative two, we'll toss a coin and
you get a 50:50 chance of getting two million dollars or
nothing?
(3) In view of (1), rational people confronted with this situation must pick alternative (2) since this has the greater
expected-value. They should thus chance everything on
the toss of a coin.
(4) But in real life, most people-and we must presume that
the bulk of them is rational-would actually opt for alternative one and would setde for that almost-a-million.
(5) Thus contrary to (3), it seems that rational people could,
would, and should prefer that sure-thing alternative,
expected values to the contrary notwithstanding.
Paradoxes of this sort, which cast a shadow of doubt across the
claims of expected value calculations to serve as principles of rationality, were investigated by the French economist Maurice Allais.
They strongly suggest once more that in a broad range of cases it
is well advised to acknowledge the rationality of sure-thing preferences over against expected-value juggling. However, taking this
line means that one must abandon the idea that expected-value
comparison suffices as a rationally failproof guide in matters of
probabilistic choice. 8
7. Strictly speaking the units of reward should be measured in "utility" rather than in
money.
8. For a discussion of A11ais's paradox situations see R. Duncan Luce and Howard
Raiffa, GlnneSIJnli Decisions (New York: WLley, 1957).
268
Chapter 13
Such an approach is supported by yet another obstacle to the

standard theory of decision based on expected value analysis. This
is the so-called St. Petersbu'IlJ Paradox which is based on the following hypothetical game:
A fair coin is to be tossed until a "heads" appears. Now if this
should happen on the n-th toss, then the player is to get a payoff of 2 n 1 units (which again should be units of "utility" as
economists conceive of this). Question: How much should a
rational person be willing to pay this game?
The expected value here is going to be the infinite sum (over n)
of the product:
prob(heads appears first on the n-th toss) x 2 n l
Since the probability at issue is ( )n it transpires that this product

is uniformly h and the sum in question is therefore + t + t +
... , which is to say that it is infinite (that is, larger than any finite
quantity). The upshot would be that no price would be too large
for the opportunity to play this game. And yet this gambling scenario seems distinctly counterintuitive-and not just because it
requires a bank with infinite patience and infinite resources.
We arrive on this basis at the following paradox:
(1) The St. Petersburg scenario describes a practicable game

that defines a real choice.
(2) The expected utility-value of this game is a meaningful,
well-defined quantity, albeit an infinite one.
(3) Being guided by expected-value considerations, a rational
person could deem no price too large to play the St.
Petersburg game.
(4) But nevertheless it is clear on general principles that no
sensible person would play this game if the price were
truly enormous.
Given that (4) seems intuitively unavoidable and that (1) is to be
accepted as a formative hypothesis of the problem situation, there
269
are only two exits from inconsistency here:

(2) - abandonment. Rejecting the idea that infinite utilities are
meaningful quantities.
(3) - abandonment. Rejecting the idea that expected-value
calculations provide an appropriate guide to rational
choice (at least in exotic cases such as this one).
Decision theorists generally opt for the first of these alternatives,
while people less ardently committed to the decision-theoretic
view of rationality are inclined to opt for the second. 9
13.5
The Dr. Psycho Paradox affords yet another instructive example,

which involves a clash of two different but equally plausible ways
of implementing the process of expected-value maximization. 1o
Consider the problem posed by a somewhat eccentric friend of
yours, Dr. Psychic Psycho, an otherwise intelligent, serious, reliable, and generally sagacious and self-assured biochemist who fancies himself a clairvoyant psychic and indeed has a good track
record for oddball predictions. After you have just eaten apples
together, he proceeds to astonish you with the following
announcement:
I have interesting news for you. You must seriously consider
taking this pill. As you know (since we have recently determined it together) it contains substance X which (as you also
know-but consult this pharmacopoea ifin doubt) is fatally
poisonous by itself, while nevertheless furnishing unfailing
antidote to poison Z-though it does have some minorly
unpleasant side effects. Now the apple I gave to you, which
9. On the origin of the problem in the work of Nicholas and Daniel Bernoulli see
Isaac Todhunter, A History of the Mathematical Theory of Probability (New York: G.E.
Stechert, 1931), pp. 134 and 220-22.
10. On this problem see the author's "Predictive Incapacity and Rational Decision" in
The European ReTliew, vol. 3 (1995), pp. 325-330.
Chapter 13
270
you have just finished eating, was poisoned by me with Z--or

not-in line with my prediction as to your taking or not taking
the antidote pill. Benign old me of course only poisoned the
apple if! foresaw that you were indeed going to take the antidote. And not to worry-I'm a very good predictor.
At this point your strange friend rushes off and vanishes from the
scene. Any prospect of beating the truth out of him disappears
with his departure. And an awful feeling comes over you-you
cannot but believe him. In fact, you strongly suspect that he went
through the whole rigamarole to get you to take that problem
pill. What do you do? Your very life seems to depend on predicting the result of taking that pill.
Of course you proceed to do a quick bit of decision theoretic
calculation. For starters, you map out the spectrum of available
possibilities:
According Your
to
System
Contains Result?
Him/You
He
You
take
take
not take
not take
predicts correctly
predicts incorrectly
predicts correctly
predicts incorrectly
take
not take
not take
take
Z,X
X
neither
Z
survive
die
surVIve
die
Not a pretty picture. After all, you stand a chance of dying

whether or not you take that accursed pill. What to do?
Two lines of analysis are available here:
[B] Analysis No.1
Let P = probability that he correctly predicts your action. In

terms of an expectation-of-life measure we then have:
EV (take)
p (+ 1-) + (I - p) (- I)
EV (not take) = p (+ I) + (I - p) (- I)
Accordingly EV (take) < EV (not take), irrespective of the
value of p. (Note that the small superscripted minus sign -
271
means "a smidgeon less" and comes into play owing to the
minor negative side effect of taking the pill.)
On this basis, the expected-value comparison envisioned in
orthodox decision theory appears to rule against taking that
pill, quite independently of any estimate of how good a predictor Dr. Psycho is or isn't.
[B] Analysis No.2

Let P = probability that he actually poisoned the apple.
Calculating the expected value in lives gained or lost, we now
have:
EV (take) = p (+ 1 -) + (1 - p) (- 1) = 2p - - 1
EV (not take) = p (-1) + (1-p) (+ 1) = -2p + 1
Note that:
EV (take) > EV (not take) whenever:
2p--1 >-2p+1
4p- > 2
Now let it be (ex hypothesi) that, additionally to the given

information, you strongly suspect that he actually went
through this bizarre exercise to induce you to take the pill
(which, as he presumably sees it, you will do iffhe poisons the
apple). Then clearly the preceding condition on p-namely its
being non trivially greater than -is satisfied, and you are
decision-theoretically well advised to take the pill.
Overall, we here face the aporetic puzzle created by the following

theses:
(1) The rational and sensible thing to do is to follow the guidance of a decision theoretic analysis.
272
Chapter 13
(2) The argumentation of Analysis No.1 is decision-theoretically cogent.

(3) The argumentation of Analysis No.2 is decision-theoretically cogent.
(4) The two rationally cogent analyses in question lead to discordant resolutions.
(5) Rationality is coherent: rationally cogent problem -resolutions do not lead to conflicting results.
Now (5) is a fundamental principle of rationality: there is no sensible alternative to accepting it. And (4) is simply a fact of the situation. (1) is a plausible but nevertheless problematic supposition.
We thus arrive at the priority ranking (5) > (4) > (1) > [(2), (3)].
What we have here is an aporetically inconsistent family of theses,
whose m. c. s. are the inconsistent quartets {(I), (2), (4), (5)} and
{( 1), (3), (4), (5)}. This leaves us so circumstanced that one of
(2) and (3) must be jettisoned and one of those two expected
value analyses has to be abandoned in consequence.
But which of these two discordant analyses is right? That is difficult to say-indeed effectively impossible. On all the available
indications both available alternatives seem equally cogent. The
abstract case that can be made for the one is every bit as good as
that which can be made from the other. You pay your money and
take your choice-or toss your coin. 11
The upshot is that we here again encounter a situation that
reinforces the indication afforded by Newcomb's problem that in
various contexts of decision the standard machinery of expectedvalue analysis may leave us in the lurch because the probabilities at
issue may simply fail to be well-defined quantities.
In the present aporetic situation it transpires that plausibilistic
considerations, though generally helpful, cannot dictate a single
uniquely definitive resolution but still leave matters in a state of
total bafflement. And this should not be altogether surprising.
For as is clear from games like "Rock, Paper, Scissors"-in con-
11. To be sure, if you knew for certain that one of the two analyses had
you might be able to effect a second-order expected value comparison_
to
be correct,
273
trast to solvable games like Tic-Tac-Toe-there sometimes is no

effective strategy of play, no definite way of making "the right
choice." In such uncooperative circumstances the resources of
rationality have no advice for us. No analysis of the range of
options indicates one option to be superior to the rest: every
alternative that confronts one has as much to be said for (or
against) it as any other.
All of these paradoxes of choice and decision arise in a common
way in that in each instance there is a seemingly equi-meritorious
case for resolving a decision issue in either of two incompatible
ways. And all these paradoxes admit of the same generic response.
For since the case is equally good for those incompatible resolutions, we might as well "do what comes natural" whatever seems
to be the most favorable, least burdensome of the choice alternatives. Rational decision in matters of truth cannot but await the
preponderance of evidence. But in matters of practice we can let
equivalent opposites balance each other out and do as the circumstances of the case might appeal to our fancy.12
The fact of it is that sometimes rational analysis underdetermines the choice of an advisable resolution: there are no good
arguments for selecting one possibility over against some other
alternative. And the present situation is, as it were, a cousin to
that one. In both cases alike it accordingly becomes impossible to
say with warranted confidence how even an ideally rational agent
would proceed. Our only recourse here is brute random selection.
And this points in the direction of a notorious paradox.
Another closely related example is provided by the Buridan's
Ars Paradox, attributed to the medieval French philosopher Jean
Buridan (ca. 1295-ca. 1358). This paradox is based on the story
of a donkey, starving between two equally eligible bales of hay,
unable to decide which one to eat. The moral of the story seems
to be that absent a free will to force a choice between equally
powerful motivations absurd consequences will ensue. 13
12. On the different groundrules of decision in factual and practical contexts see the
author's "Ueber einen zentralen Unterschied zwischen Theorie und Praxis," Deutsche
Zeitschriftfor Philosophie, vol. 47 (1999), pp. 171-182.
13. On the historical background of the Buridan's Ass perplex see the author's
"Choice Without Preference" in his Essays in Philosophical Analysis (Pittsburgh: University
of Pittsburgh Press, 1969), pp. I11-lS7.
274
Chapter 13
The paradox at issue stands roughly as follows:

(1) A rational being would not starve in the presence of available food.
(2) A rational creature must decide between alternatives on
the basis of preponderating reasons.
(3) Where there is an equilibrium of reasons-a balance where
an equally good case exists either way-there is, ex
hypothesi, no preponderating reason for going one way
rather than the other.
(4) In consequence of(2) and (3), a rational creature cannot
choose between equivalent alternatives.
( 5) In view of (4), our donkey would starve-even though
food is available.
Here {(I), (2), (3)} constitutes an inconsistent triad. And (2)
seems to be the most vulnerable of these three plausible principles
since both (1) and (3) are beyond reasonable cavil. For it seems
sensible to argue that while it is indeed the case that where there
are preponderating reasons a rational creature can and should
heed them, it would nevertheless be totally irrational to let immobilization occur if-as Buridan was presumably trying to showstalemates can be broken by the free agency of the will.
Accordingly, (2) - abandonment is our best prospect for breaking
through the chain of inconsistency here. We can however again
save something of (2)'s message by a suitable distinction-this
time between a narrower and a broader sense of "rationality." For
while there indeed are no immediate reasons for preferring one
bale of hay over the other, because of the symmetry that (ex
hypothesi) obtains, there is excellent reason to prefer one to none,
and thus for preferring an arbitrary choice to immobilization, an
arbitrary choice which, even if beyond the capacity of the will
itself, could be delegated by it to a random device.
There is something particularly problematic-particularly paradoxical-about paradoxes that arise in matters of decision and
choice. In matters of theory--of question-answering and beliefacceptance-we can, as it were, temporize by suspending judg-
275
ment and deferring the matter for another day when ampler information becomes available. Or we can haver between the unresolvable alternatives P and Q by the simple device of disjunction, that
is by adopting P-or- Q and being no more committal than that.
But in practical matters of decision and action the situation wears
a different aspect. Here inaction is itself a mode of action.
Confronted with a choice between doing A and doing B we face
the choice system:
do nothing
doA
do B
And here it may well be that doing nothing is the very worst of
available alternatives-as the example of the Buridan's Ass paradox brings vividly to realization.
This aspect of the matter give a special bite to the paradoxes of
practice.
In solving theoretical and practical paradoxes alike we want to
minimize the damage. But damage is measured very differently in
these two domains. Theoretical damage consists in the epistemic
error of accepting what is false or implausible. But practical damage can be something even more troubling of a more grave and
painfully affecting sort. And this is why those paradoxes like the
Prisoner's Dilemma or the Dr. Psycho Paradox, whose difficulty
lies in the fact that here the very principles of proper damage assessment are in question, are peculiarly troublesome.
13.6
Retrospect: The Pragmatic Dimension
When paradoxes arise, logico-inferential rationality alone avails

but little. For when the data are inconsistent, inferential reason
can only say that consistency should be restored, but cannot tell
us how this must be achieved. Rational analysis can define the
alternatives but cannot resolve them. In this regard paradox solution is not in general a matter of mechanical rules and automatic
routine. Logic-transcending resources are required.
276
Chapter 13
In the course of these deliberations we have met with a considerable array of "tools of the trade" in matters of paradox resolution. Prominent among these are:
dis-ambiguation
dissolution through equivocation removal
presupposition rejection or invalidation
meaningfulness denial and concept invalidation
implausibility dismissal
item abolition (on grounds of inappropriate hypostatization)
prioritization of "deeper" principles and "superior"
alternatives
hypothesis invalidation (through reductio, for example)
choice resolution through preferentiability validation
All of these processes can provide a means for reestablishing
consistency when confronted with a family of inconsistent
premisses.
The striking feature of all of these resources is that in terms of
their thematic locus in the cognitive scheme of things they all
belong rather to the domain of rhetoric than to that of theoretical
logic proper. In the sections devoted to sophisms and insolubilia
in Prantl's classic Geschichte der Logik im Abendlande one encounters time and again a complaint to the effect that "this sort of
thing belongs rather to rhetoric than to logic." And Prantl's sensibilities in this matter are entirely correct. (What one can object to
is his reaction to this circumstance-based on the sentiment that
if something does not belong to logic proper, then it is not all
that important.)
Be this as it may, the problem of resolving aporetic conflicts
recurs in a diversified variety of contexts, preeminently including
the purely hypothetical, the proof-theoretic (reductio ad absurdum), the evidential, and the philosophical. But throughout,
277
one selfsame master principle is determinative for the process

of resolving aporetic inconsistency: "restore consistency with
minimal disruption, disruptiveness in terms of the purposive
nature of the enterprise.)) And the nature of the context will play
an important role here. In factual contexts, in particular, plausibilistic priorities are determinative. However, there are also hypothetical contexts of supposition and reductio argumentation. And
matters are different here. Hypotheses and their implications are
now paramount and other plausibilistic considerations are displaced and readjusted. It is all a matter of practical policies determined by the purposive considerations operative in the context at
issue.
It must accordingly be emphasized that the priority or precedence at issue with propositions that conflict in these aporetic situations need not and generally will not be absolute or categorical;
rather it is something that is variable and context-dependent. We
proceed differently in different sorts of cases. Thus, for example:
In reductio ad absurdum contexts we sacrifice hypotheses to
givens; in purely hypothetical contexts we do the reverse.
In evidential contexts we sacrifice generalities to specificities; in purely hypothetical contexts we do the reverse.
In philosophical contexts we must take evidential considerations into account; in purely hypothetical contexts we do
not worry ourselves about them.
What qualifies such rules as the natural or appropriate way of
proceeding in these different cases is a matter of the purposive
nature of the relevant domain at issue-its governing aims. In
each case we proceed on the basis of guiding principles determined by the specific objectives of the particular endeavor at
hand. The contextuality of the principles of propositional prioritization must therefore be noted and acknowledged. And it
should accordingly be stressed that the priority among conflicting propositions in aporetic settings is not a matter of the personal preferences and predilections of individuals, but is
determined objectively on the basis of the purposive orientation
of different contexts of discussion. 14
278
Chapter 13
And here, as we have seen, one must distinguish the following

four cases:
1. Purely hypothetical contexts. In restoring consistency, salvage as much information as you can. Hence give priority
to the comparatively more informative-contextually
stronger-statements. Keep what is more general and
fundamental in place. The determinative issue is: "What
are the informatively weakest links in the chain of
inconsistency? "
2. Reductio ad absurdum contexts. In maintaining consistency
give priority to what has already been established.
Therefore, it will be those conflict-engendering assumptions themselves that must give way. The determinative
issue is: "Which claims are minimally at odds with what is
already established?"
3. Evidential-inductive contexts. In restoring consistency, preserve as much as you can of the probative/evidential fabric.
Always make more weakly evidentiated claims give way to
those that are more strongly evidentiated. The determinative issue is: "Where are the evidentially weakest, probatively most vulnerable links in the chain of inconsistency?"
4. Counter/actual contexts. In restoring consistency give
prime precedence to issue-definitive assumptions. And from
there on in opt for the least amount of factual disruption
that can be managed, thus proceeding as in case 2 above.
5. Philosophical contexts. In restoring consistency maintain
overall credibility. Give priority to those propositions that
provide for the systematically optimal combination of plausibility (evidentiation) and problem resolving (informative-
14. The context-dependent nature of the project of conflict resolution means the
aporetic approach is able to unify imponant aspects of the theory of reasoning in very different domains (proof theory, empirical inquiry, hypothetical reasoning, philosophical reasoning) within a single overarching integrating perspective. This unification patently
integrates the author's approach to these various issues in such books as Hypotheti&a/
Reasoning (1967), Plausible Reasoning (1974), Empiri&al Inquiry (1982), and The Strife
of Systems (1985), and thereby unifies in a synoptic perspective the pragmatic tendency of
my overall position.
279
ness). (The process here is accordingly a hybrid mix-a balanced combination or fusion of the evidential and the hypothetical approaches.) The determinative issue is: "What are
the least plausible links in the chain of inconsistency-those
whose abandonment would minimally impede the construction of a coherent system of understanding?"
As this survey indicates, different purposes are at work in different
contexts. The purposive/teleological character of the particular
enterprise at issue provides the guiding basis for the different
ground-rules of precedence-determination that are operative in
these different contexts.
The salient point with respect to inconsistency-resolution is
one of pragmatics, of rational practice and procedure. Different
contexts of deliberation reflect different ranges of purpose. And it
is this teleological issue of the purposive orientation of the relevant context that determines what sorts of principle of priority
will be operative within it. For it is their capacity to facilitate the
realization of the relevant purposes that determines the appropriateness of the requisite principles of prioritization.
Bibliography
NOTE: Regrettably there is as yet no general history of paradoxes.
However, the thinkers who dealt with them in classical antiquity
are all discussed in Eduard Zeller's magisterial Philosophie der
Griechen in ihrer Geschichtlichen Entwicklung published in Leipzig
by O.R. Reisland in three massive double volumes, 6th ed., 1919.
The paradoxes themselves are discussed in greater detail, albeit in
general rather dismissively, in Carl Prantl's Geschichte der Logik
ime Abendlande, published in Leipzig in three volumes in 1855
by S. Hirzel, which offers much information on ancient and
medieval contributions to the subject. Salmon 1970 gives a
comprehensive bibliography on Zeno's paradoxes. Riistow 1910
gives a synoptic account of the history of the Liar Paradox and its
cognates.
On paradoxes among the medieval schoolmen see the extensive bibliography presented in Ivan Boh, Epistemic Logic in the
Later Middle Ages (London: Routledge, 1993). Medieval texts on
insolubles and sophisms are available in Buridan 1977, Grabmann
1940, Heytesbury 1979, Hughes 1982, Kretzmann and
Kretzmann 1990, Nicholas of Cusa 1954, Perreiah 1978, Rijk
1962-67, and Wyclif 1986. And for analyses on these topics see
Biard 1989, Bottin 1976, Kretzmann et al. 1982, Read 1993, and
Weidemann 1980. On semantic paradoxes in the sixteenth century see the section of that title in Ashworth 1974. And for the
Renaissance see Colie 1966.
Ameriks, Karl. 1982. Kant's Theory of Mind: An Analysis of the Paralogisms of

Pure Reason. Oxford: Clarendon.
Aristotle. 1908-31. Oxford Translation of the Words of Aristotle, ed. by J.A.
Smith and W.D. Ross. Oxford: Clarendon.
Ashworth, E. J. 1974. Language and Logic in the Post-MedieTJaI Period.
Dordrecht: Reidel. [See Chapter N on Semantic Paradoxes, pp. 101-117.]
Axelrod, Roben M. 1985. The Evolution of Cooperation. New York: Basic Books.
Bandmann, H. 1992. Die Unendlichkeit des Seins: Cantors transfinite
Mengenlehre und ihre metaphysischen Wurzeln. Frankfun: Peter Lang.
281
282
Bibliography
Bar-Hillel, Yehoshua. 1939. Le probleme des antinomies et ses developpements

rI:cents. Rwenue de Metaphysique et de Morale, vol. 36, pp. 225-242.
Bartlett, Steven J., and Peter Suber, eds. 1987. Self-Reference. Dordrecht:
Nijhoff.
Barwise, Jon, and John Etchemendy. 1987. The Liar. New York: Oxford
University Press.
Behmann, Heinrich. 1937. The Paradoxes of Logic. Mind, vol. 46, pp.
218-221.
Bemardete, Jose A. 1964. Infinity: An Essay in Metaphysics. Oxford: Clarendon.
Biard, Joel. 1989. Les sophismes du savoir: Albert de Saxe entre Jean Buridan et
Guillaume Heytesbury. Vivarium, vol. 27, pp. 36-50.
Bolzano, Bernard. 1955 [1851]. Paradoxien des Unendlichen. ed. F. Prihonsky.
Leipzig: Publisher, 1851; reprinted Hamburg: F. Meiner, 1955. English version: Paradoxes of the Infinite. London: Routledge, 1950.
Borel, Emile. 1946. Les Paradoxes de l'infini. L'avenir de la science, No. 25, Part
3.
Bottin, Francesco. 1976. Le antinomie semantiche nella logica mediwale. Padova:
Antenore.
- - - . 1980. Gli 'insolubilia' nel 'Curriculum' scolastico tardomedevale.
Logica, Epistmologia, Storia della Storiografia (Padova: Editrice Anternore,
pp.13-32).
Brams, Steven, J. 1976. Paradoxes in Politics: An Introduction to the Nonobvious
in Political Science. New York: Free Press.
Burali-Forti, Cesare. Una questione sui numeri transfiniti. Rendiconti del Circulo
Mathematico di Palermo, vol. 11, pp. 154-164.
Buridan, Jean. 1977. Sophismata, ed. T.K. Scott. Stuttgart-Bad Cannstatt:
Frommann-Holzboog. English version: Sophisms on Meaning and Truth,
trans. T.K Scott. New York: Appleton-Century-Crofts, 1966.
Burgess. Theodore C. 1902. Epideictic Literature. Chicago: University of
Chicago Press.
Burns, L.C. 1991. Vagueness: An Investigation into Natural Languages and the
Sorites Paradox. Dordrecht: Kluwer.
Campbell, Richard, and Lanning Sowden, eds. 1985. Paradoxes of Rationality
and Cooperation: Prisoner's Dilemma and Newcomb's Problem. Vancouver:
University of British Columbia Press.
Cargile, James. 1979. Paradoxes: A Study in Form and Predication. Cambridge:
Cambridge University Press.
Carnap, It- 1934. Die Antinomien und ide Unvollstandigkeit der Mathematik.
Monatshefte der Mathematik und Physik, vol. 41, pp. 263-284.
Champlin, T. S. 1988. ReflexiTle Paradoxes. London: Routledge.
Chihara, Charles. 1979. The Semantic Paradoxes: A Diagnostic Investigation.
Philosophical Rwiew, vol. 88, pp. 590--618.
Church, Alonzo. 1934. The Richard Paradox. American Mathematical Monthly,
vol. 41, pp. 356-361.
Colie, Rosalie L. 1966. Paradoxica Epidemica: The Renaissance Tradition of
Paradox. Princeton: Princeton University Press, 1966.
Bibliography
283
Davis, Martin. 1965. The Undecidable: Basic Papers on Undecidable Propositions,

Unsolvable Problems, and Computable Functions. Hewlett, NY: Raven
Press.
De Laguna, Theodore. 1916. On Certain Logical Paradoxes. Philosophical
ReTliew, vol. 25, pp. 16-27.
de Morgan, Augustus. 1872. A Budget of Paradoxes. London: Longmans,
Green; reprinted Freeport, NY: Books for Libraries Press, 1969.
Dumitriu, Anton. 1969. The Solution of Logico-Mathematical Paradoxes.
International Philosophical QJt,arterly, vol. 9, pp. 63-100.
- - - . 1974. The Logico-Mathematical Antinomies: Contemporary and
Scholistic Solutions. International Philosophical Quarterly, vol. 14, pp.
309-328.
Escher, Maurits Comelis. 1967. Graphic Work. Revised ed. London: Dutton.
Ferber, Rafael. 1981. Zenons Paradoxien der Bewegung und die Struktur von
Raum und Zeit. Munchen: C.H. Beck.
Finsler, Paul. 1944. Gibt es unentscheidbar Satze? Commentarii Mathematici
Helvetici, vol. 16, pp. 310-320.
- - - . 1926. Gibt es Widerspliiche in der Mathematik? Deutsher Mathver., vol.
34, pp. 143-155; reprinted G. Unger, ed., Aufsiitze sur Mengenlehre.
Darmstadt: Wissenschaftliche Buchgemeinschaft, 1975.
- - - . 1927. Uber die LOsung von Paradoxien. Philosophischer Anzeiger, vol. 2,
pp. 183-192.
Fitch, F.B. 1946. Self-reference in Philosophy. Mind, vol. 55, pp. 64-73.
Franck, Sebastian. 1909. Paradoxa. Jena: H. Ziegler. English trans. E.J. Furcha:
Paradoxa: Two Hundred Eighty Paradoxes or Wondrous Sayings. Lewiston,
NY: Edward Mellon Press, 1986.
Frontera, G. 1891. Etude sur les arguments de zenon d'Etee contre Ie mouvement.
Paris: Hachette.
Gaifman, H. 1983. Infinity and Self-Applications: I. Erkenntnis, vol. 20, pp.
131-155.
Gale, Richard. 1993. The Nature and Existence of God. Cambridge: Cambridge
University Press.
Garciadiego, A.R. 1992. Bertrand Russell and the Origins of the Set-Theoretic
cParadoxel. Basel: Birkhauser.
Gardner, Martin. 1982. Aha! Gotcha: Paradoxes to Puzzle and Delight New
York: W.H. Freeman.
- - - . 1971. Infinite Regress. In Sixth Book of Mathematical Games from
Scientific American. San Francisco: W.H. Freeman.
- - - . 1968. The Paradox of the Unexpected Hanging. In The Unexpected
Hanging and Other Mathematical Diversions. New York: Simon and
Schuster.
Geyer, P., and R. HagenbucWe, eds. 1992. Das Paradox: Eine Herausforderung
des abendliindischen Denkens. Tubingen: Stauffenburg, pp. 159-189.
Good, I.J. 1966. A Note on Richard's Paradox. Mind, vol. 75, p. 431.
Goodman, Nelson. 1955. Fact, Fiction, and Forecast. Cambridge, MA:
Harvard University Press; 2nd ed. Indianapolis: Bobbs-Merrill, 1965; 3rd
ed. 1973.
284
Bibliography
Grabmann, Martin. 1940. Die Sophismataliteratur des 12. und 13. Jahrhunderts,
mit Textausgabe eines Sophisma des Boetius pon Dacien. Munster:
Aschendorff.
Grelling, Kurt. 1936. The Logical Paradoxes. Mind, vol. 45, pp. 481-86.
- - - . 1937. Der Einfluss der Antinomien auf die Entwicklung der Logik im
20 Jahrhunders, TraMux du Ixieme congres international de philosophie,fasc,
vol. VI, pp. 8-17.
Grelling, Kurt, and Leonhard Nelson. 1907-1908. Bemerkungen zu den
Paradoxien von Russell und Burali-Forti. Abhandlungen der Fries]schen
Schule n.s., Vol. 2, pp. 301-334.
Hallden, Soren. 1949. The Logic of Nonsense. Uppsala: Lundequistska
Bokhandeln.
Hamblin, c.L. 1986. Fallacies. Revised ed. Newport News: Vale Press.
van Heijenoort, Jean, ed. 1967. From Frege to Godel: A Source Book
in Mathematical Logic] 1879-1931. Cambridge, MA: Harvard University
Press.
- - - . 1967. Logical Paradoxes. In The Encyclopedia of Philosophy. Paul
Edwards ed. New York: Macmillan.
Heiss, Robert. 1928. Der Mechanisumus der Paradoxien und das Gesetz der
Paradoxienbildung. Philosophischer Anzeiger, vol. 3, pp. 403-432.
Helmer, Olaf. 1934. Remarques sur Ie probleme des antinomies. Philosophisches
Jahrbuch der Gorresgesellschaft, vol. 47, pp. 421-24.
Helrnont, J. B. van. 1650. A Ternary of Paradoxes. Ed. and translated by Walter
Charleton. London: James Flesher.
Heytesbury, William. 1979. On Insoluble Sentences: Chapter One of His Rulesfor
Solping Sophisms. Translated with an introduction and a study by P.V. Spade.
Toronto: Pontifical Institute of Mediaeval Studies. Mediaeval Sources in
Translation, No. 21.
Hinske, Norbert. 1965. Kant's Begriff der Antinomie und die Etappen seiner
Ausarbeitung, Kant-Studien, vol. 56, pp. 485-496.
Hintikka, Jaakko. 1957. Vicious Circle Principle and the Paradoxes. Journal of
Symbolic Logic, vol. 22, pp. 245-49.
Hofstadter, Douglas R.. 1980. Godel, Escher, Bach: An Eternal Golden Braid.
New York: Random House.
---.1986. Metamagical Themes New York: Basic Books.
Hofstadter, Douglas R.., and Daniel C. Dennett. 1981. The Mind]s I: Fantasies
and Reflections on the Self and Soul. New York: Basic Books.
Hughes, George Edward. 1982. John Buridan on Self-Reference: Chapter Eight
of Buridan]s Sophismata. Cambridge: Cambridge University Press.
Hughes, Patrick, and George Brecht. 1975. Vicious Circles and Infinity: A
Panoply of Paradoxes. Garden City, NY: Doubleday.
Hyde, M. J. 1979. Paradox: The Evolution of a Figure of Rhetoric. In. R..L.
Brown Jr. and M. Steinman Jr., eds., Rhetoric 78: Proceeding of Theory of
Rhetoric: An Interdisciplinary Conference. Minneapolis: University of
Minnesota Center for Advanced Studies in Language, Style, and Literary
Theory, pp. 201-225.
Bibliography
285
Intisar-Ul-Haque. 1969. Critical Study of Logical Paradoxes. Pakistan: Peshawar

University Press.
Jourdain, P.E.B. 1913. Tales with Philosophical Morals. In The Open Court, vol.
27,pp.310-315.
Kainz, Howard P. 1988. Paradox, Dialectic, and System: A Contemporary
Reconstruction of the Hegelian Problematic. University Park: Pennsylvania
State University Press.
Kneale, William, and Martha Kneale. 1962. The Development of Logic. Oxford:
Clarendon.
Koons, Robert C. 1992. Paradoxes of Belief and Strategic Rationality.
Cambridge: Cambridge University Press.
Koyre, Alexandre. 1946. Epimenide Ie menteur. Paris: Ensemble et categorie,
Actualites scientifiques et industrielles.
Kretzmann, Norman, and B.E. Kretzmann, eds. 1990. The Sophismata of
Richard Kilvington. Cambridge: Cambridge University Press.
Kretzmann, Norman, and Eleanore Stump. 1988. The Cambridge Translations
of Medieval Philosophical Texts, Vol. I. Contains an English translation of
Albert of Saxony's Insolubilia. Cambridge: Cambridge University Press.
Kretzmann, Norman, Anthony John Patrick Kenny, and Jan Pinborg, eds. 1982.
The Cambridge History of Later Medieval Philosophy: From the Rediscovery of
Aristotle to the Disintegration of Scholasticism, 1100-1600. Cambridge:
Cambridge University Press.
Kutschera, Franz von. 1964. Die Antinomien der Logik. Freiburg: K. Alber.
Lando, Ortensio. 1593. The Defense of Contraries. Translated by Anthony
Munday. London: by John Windet for Simon Waterson.
Langford, C.H. and M. Langford. 1959. The Logical Paradoxes. Philosophy and
Phenomenological Research, vol. 21, pp. 1l0-l3.
Lubac, Henri de. 1946. Paradoxes. Paris: Editions du Livre fran"ais.
---.1958. Further Paradoxes. London: Longmans Green.
Mackie, J.L. 1973. Truth, Probability and Paradox: Studies in Philosophical Logic.
Oxford: Clarendon.
Makinson, D.C. 1965. The Paradox of the Preface. Analysis, vol. 25, pp.
205-07.
Margalit, Avishai, and Maya Bar-Hilled. 1983. Expecting the Unexpected.
Philosophia, vol. 13, pp. 263-288.
Martin, Robert L. 1968. On Grelling's Paradox. The Philosophical Review, vol.
77, pp. 321-331.
- - - , ed. 1970. The Paradox of the Liar. New Haven: Yale University Press;
2nd ed., Resada, Ca: Ridgeview, 1978. [An anthology of some dozen
papers plus a very extensive bibliography.]
- - - . 1967. Toward a Solution to the Liar Paradox. Philosophical Review, vol.
76, pp. 279-311.
- - - , ed. 1984. Recent Essays on Truth and the Liar Paradox. New York:
Oxford University Press.
McGee, Vann. 1991. Truth, Vagueness, and Paradox: An Essay on the Logic of
Truth Indianapolis: Hackett.
286
Bibliography
Melhuish, George. 1973. The Paradoxical Nature of Reality. Bristol: St.

Vincent's Press.
Mirimanoff, D. 1917. Les antinomies de Russell et de Burali-Forti et Ie probIerne fondamental de la theorie des ensembles. VEnseignement Math., vol.
19,pp.37-52.
Nicholas ofCusa (Nicolas Cusanus). 1954. Of Learned Ignorance. Translated by
Fr. Germaine Heron. New Haven: Yale University Press, and London:
Routledge. Reprinted Westport, CT: Hyperion, 1991.
Nordau, Max. 1886. Paradoxes. Chicago: Laurd and Lee.
Northrop, Eugene Purdy. 1958 [1944]. Riddles in Mathematics. London:
English Universities Press. Reprinted New York: Van Nostrand, 1966.
Nozick, Robert. 1974. Reflections on Newcomb's Problem. Scientific American
(March), pp. 228-241.
O'Carrol, M. J. 1967. Improper Self-Reference in Classical Logic and the
Prediction Paradox. Logique et analyse, vol. 10, pp. 167-172.
O'Connor, D.J. 1948. Pragmatic Paradoxes. Mind, vol. 57, pp. 316-329.
Pap, Arthur. 1954. The Linguistic Hierarchy and the Vicious-Circle Principle.
Philosophical Studies, vol. 5, pp. 49-53.
Perelman, C.H. 1936. Les Paradoxes de la Logique. Mind, vol. 45, pp. 204-08.
Perreiah, Alan. 1978. Insolubilia in Paul of Venice's Logica Parra. Medioevo, vol.
4, pp. 145-171.
Prior, A.N. 1958. Epimenides the Cretan. Journal of Symbolic Logic, vol. 23, pp.
261-66.
- - - . 1961. On a Family of Paradoxes. Notre Dame Journal of Formal Logic,
vol. 2, pp. 16-32.
Probst, P., H. Schroer, and F. von Kutschera. 1989. Das Paradoxe, Paradoxie.
In Historisches Workerbuch der Philosophic, Vol. VII, pp. 81-97.
Quine, W.V.O. 1962. Paradox. Scientific American, vol. 206 (April), pp. 84-96.
- - - . 1966. The Ways of Paradox and Other Essays. New York: Random
House.
Ramsey, F.P. 1925. The Foundations of Mathematics. Proceedings of the London
Mathematical Society, series 2, vol. 25, part 5 (read 12 November), pp. 338384. Reprinted in D.H. Millar, ed., F.P. Ramsey: Philosophical Papers
(Cambridge: Cambridge University Press, 1990).
Rapaport, Anatol, and Albert M. Chammah. 1965. The Prisoner's Dilemma. Ann
Arbor: University of Michigan Press.
Read, Stephan. 1993. Sophisms in Medieval Logic and Grammar. Acts of the
Ninth European Symposium for Medieval Logic and Semantics, St. Andrews,
June 1990. Dordrecht: Kluwer.
Rescher, Nicholas. 1961. Belief-Contravening Suppositions. Philosophical
Review, vol. 70, pp. 176-195.
---.1964. Hypothetical Reasoning. Amsterdam: North Holland.
Rescher, Nicholas, and Robert Brandom. 1980. The Logic of Inconsistency.
Oxford: Oxford University Press.
Richards, T.J. 1967. Self-Referential Paradoxes. Mind, vol. 76, pp. 387-403.
Rijk, Lambertus Marie de. 1962-67. Logica modernorum: A Contribution to the
History of Early Terminist Logic. 2 vols. Assen: Van Gorcum.
Bibliography
287
Riverso, Emanuele. 1960. II Pardosso del Mentitore. Rassengna di Scienze

Filosoftche, vol. 13, pp. 296--325.
Rivetti, BarbO. 1959. Vorigine dei paradossi ed il regresso all'infinito. RiTJista di
ftlosophia neo-scolastica, vol. 51, pp. 27-60.
- - - . 1961. L'antinomia del mentitote nel pensiero contemporaneo, da Peirce a
Tarski, Studi-testi-bibliografta. Milano: Societa Editrice Vita e Pensiero.
Rozeboom, W.W. 1957-58. Is Epimenides Still Lying? Analysis, vol. 18, pp.
105-113.
Rudavsky, Tamar, ed. 1985. Divine Omniscience and Omnipotence in Medieval
Philosophy. Dordrecht: Reidel.
Russell, Bertrand. 1903. The Principles of Mathematics. Cambridge: Cambridge
University Press. Revised ed., London: XYZ, 1937 and New York: Norton,
1938.
- - - . 1906. Les paradoxes de la logique. Rev. Met mor., vol. 14, pp. 627-650.
English: On Insolubilia and the Solution by Symbolic Logic. In Russell,
Essays in Analysis. D. Lackey, ed., London: Allen and Unwin, 1973, pp.
190-214.
- - - . 1908. Mathematical Logic as Based on the Theory of Types. American
Journal of Mathematics, vol. 30, pp. 222-672. Reprinted in Heijenoort
1967.
- - - . 1918. Introduction to Mathematical Philosophy. London: Macmillan.
Russell, Bertrand, and A.N. Whitehead. 1910-13. Principia Mathematica. 3
vols. Cambridge: Cambridge University Press, 1910, 1912, 1913.
Riistow, Alexander. 1910. Der Lugner: Theorie, Geschichte, und Auflosung.
Leipzig: B.G. Teubner.
Ryle, Gilbert. 1961. Dilemmas. Cambridge: Cambridge University Press.
Sainsbury, Richard Mark. 1988. Paradoxes Cambridge: Cambridge University
Press.
Salmon, Wesley C. 1970. Zen's Paradoxes. Indianapolis: Bobbes-Menill.
Saperstein, Milton R. 1966. Paradoxes of Everyday Life. New Haven, CT:
Fawcett.
Schilder, Klaas. 1933. Zur Begrifftgeschichte des Paradoxons: mist besonderer
Berucksichtung Calvins und des nach-kierkegaardschen Paradoxon. Kampen:
J.H. Kok.
Schroer, Henning. 1960. Die Denkform der Paradoxalit&t als theologisches
Problem. Gottingen: Vandenhoeck and Ruprecht.
Scriven, Michael. 1951. Paradoxical Announcements. Mind, vol. 60, pp.
403-07.
Skyrms, Brian. 1970. Return of the Liar: Three-Valued Logic and the Concept
of Truth. American Philosophical Q;tarterly, vol. 7, pp. 153-161.
Slaatte, Howard A. 1968. The Pertinence of the Paradox: The Dialectics of
Reason-in-Existence. New York: Humanities.
Smullyan, Raymond M. 1978. This Book Needs No Title. Englewood Cliffs:
Prentice-Hall.
- - - . 1978. What Is the Name of This Book? Englewood Cliffs: Prentice-Hall.
Stalker, Douglas, ed. 1999. Grue!: The New Riddle of Induction. Chicago: Open
Court.
288
Bibliography
Stenius, Eric. 1949. Das Problem der Logischen Antinomien. Societas

Scientarium Fennica, Commentationes Physico-Mathematicae, vol. 14.
Strawson, P.F. 1967. Paradoxes, Posits, and Preposition. Philosophical RClIiew,
vol. 76, pp. 214-229.
Stroll, Avrum. 1954. Is Everyday Language Inconsistent? Mind, vol. 63, pp.
219-225.
Tarski, Alfred. 1956. Logic, Semantiu, Metamathematics. Translated by J.H.
Woodger. Oxford: Oxford University Press.
Thines, Georges. 1974. L'aporie. Brussels: Leclerc.
Thompson, M.H., Jr. 1949. The Logical Paradoxes and Peirce's Semiotic.
Journal of Philosophy, vol. 46, pp. 513-536.
Toms, E. 1952. The Reflexive Paradoxes. Philosophical Review, vol. 61, pp.
557-567.
Urbach, B. 1910. Ober das Wesen der logischen Paradoxa. Zeitschrift fur
Philosophie und Philosophische Kritik, vol. 140, pp. 81-108.
- - - . 1927-28. Das Logische Paradoxon. Annalen der Philosophie und
philosophische Kritik, vol. 6, pp. 161-176 and 265-273.
Urquhart, W.S. 1942. Paradox in Religious Thought. The Contemporary RClIiew,
vol. 161 (April).
Vailati, G. 1904. Sur une classe remarquable de raisonnements par reduction it
l'absurde. RClIue de Metaphysique, vol. 12, pp. 799-809.
van Fraassen, B.C. 1968. Presupposition, Implication, and Self-Reference.
Journal of Philosophy, vol. 65, pp. 136-152.
Venning, Ralph. 1657. Orthodox Paradoxes: Theoreticall and Experimentall.
London: J. Rothwell.
Visser, Albert. 1989. Semantics and the Liar Paradox. Handbook of Philosophical
Logic, vol. IV. Dordrecht: Reidel, pp. 617-706. [Has an extensive bibliography.]
von Wright, G.H., 1974. The Heterological Paradox. Societas Scientarium
Fennica, Commentationes physico-mathematicae, vol. 24.
Vredenduin, P.G.J. 1937-38. De Paradoxen. Algemeen Nederlands tijdschrift
voor wijsbegeerte en psychologie, vol. 31, pp. 191-200.
Walton, Douglas N. 1991. Begging the QJt.estion. Westpoint CT: Greenwood.
Weidemann, Hermann. 1980. Ansatze zu einer Logik des Wissens bei Walter
Burleigh. Archiv fur Geschichte der Philosophie, vol. 61, pp. 32-45.
Weiss, Paul. 1952. The Prediction Paradox. Mind, vol. 61 (April), pp. 265-69.
Wolgast, E.H. 1977. Paradoxes of Knowledge. Ithaca: Cornell University Press.
Wonnell, C.P. 1958. On the Paradoxes of Self-Reference. Mind, vol. 67, pp.
267-271.
Wyclif, John. 1986. Johannis WyclifSumma Insolubilium. P.v. Spade and G.A.
Wuson, eds. Binghamton: Medieval and Renaissance Texts and Studies, vol.
41.
Index
Brams, Steven J., 255, 260n5,
264n6,282
Brandom, Robert, 10n13
Brecht, George, 139n2, 284
Buckle, Henry, 245
Burali-Forti, Cesare, 184, 188,282
Burgess, Theodore C., 282
Buridan, Jean, 273-74
Buridan, John 21,93, 196, 198,
281-82
Burley, Walter, 233-38
Burley's principle, 235-38
Burnet, John, 97n7
Burns, L.C., 282
absolutism, semantic, 51-53

acceptability,20-22
Albert of Saxony, 4402,19502,
209n13
Aldrich, Henry, 21On15
Allais, Maurice, 267-69
ambiguity, paradoxes of, 91-114
Ameriks, Karl, 281
antinomies, Kantian, 147-151
antinomy, 7
aporetics, 3-21
Aristotle, 3, 50, 59, 60, 61, 70, 77,
83,91-92,96,99,102,104,
117nl,118, 119,142, 181,
199n6,199n7,281
arithmetical incompleteness, 205
Arnold, Karen, 139n3
Ashworth, E.J., 167n6, 194, 195n2,
199n5,210nI5,281
Axelrod, Robert M., 281
Caesar, Julius, 103-04

Campbell, Richmond, 264n6
Cantor, Georg, 161, 162, 165, 185,
188,189
Cargile, James, 282
Carnap, Rudolf, 224n4, 282
Carrol, Lewis, 106n16
Bandmann, H., 281

Bar-Hillel, Maya, 112nI9, 285
Bar-Hillel, Yehoshua, 282
Barker, Stephen, 228n7
Barwise, Jon, 200n1O, 282
Behmann, Henrich, 282
belief, conflicts of, 233-252
Bernardete, Jose A., 282
Bernoulli, Daniel, 269
Bernoulli, Nicholas, 269n9
Berry, G.G., 176, 182, 185, 188
Biard, Joel, 281, 282
Bizet, Georges, 68, 69
Boh, Ivan, 281
Bolzano, Bernard, 282
Borel, Emile, 282
Borges, Jorge Luis, 5, 5n4
Bottin, Francesco, 281, 282
box puzzles, 209-213
cassatio, 81
Casti, John L., 152n11

Cervantes, Miguel de, Saavedra, 5,
5n4,62
Chammah, Albert M., 286
Champlin, T.S., 143n8, 221n2,
282
Chesterton, G.K, 4-5
Chihara, Charles, 282
Chisholm, Roderick M., 234n3
Chrysippus, 81-82, 104n4
Church, Alonzo, 184n15,282
Cicero, 6, 82n13, 199n6, 199n7
circle principle, 172-73
circle, hermeneutic, 87-88
classification, of paradoxes, 70-74
Cohen, L.r, 112n19
coincident;a oppositorum, 91
289
290
Index
Colie, Rosalie L., 5, 5n4, 30n4,

281-82
completeness, mathematical, 186
compound paradoxes, 38-39
conflicts of belief, 233-252
consistency, mathematical, 186
Curry, Haskell B.,173-76, 185, 188
Davis, Martin, 282
De Laguna, Theodore, 283
De Morgan, Augustus, 4nl, 164,283
Dennett, Daniel C., 284
Descartes, Rene, 19, 19n19, 124
Dilthey, Wilhelm, 87n17
Diodorus, Cronus, 107-08
Diogenes Laertius, 12n16, 62n2,
77n2,78n3,81n9,102nI2,
139n3,140n4,200n8
discrepant information, 11
disjunctive paradox resolutions,
35-36
dissolution of paradoxes, 57--61
distinctions, 126-133
divine foreknowledge, 122-23
Dresher, Melvin, 260
Dumitriu, Anton, 283
Earman,John,121n4
endoxa,50
Epimenides,199-200
equivocation, paradoxes of, 91-114
eristic,77
Escher, Maurits Cornelis, 138,212,
283
Etchemendy, John, 200nl0, 282
Eubulides ofMiletus, 12,77-78,
102, 104, 111, 140n4, 199-200
Ferber, Rafael, 283
Fermat, Pierre de, 250n8
Finsler, Paul, 283
Fitch, F.B., 283
Flood, Merril M., 260
foreknowledge, divine, 122-23
Franck, Sebastian, 5, 283
Frege, GottIob, 110
Frontera, G., 283

fuzzy logic, 190
Gaifman, H., 258n3, 283
Gale, Richard M., 30n5, 120n2,
122n5,131,247n7,283
Garciadiego, A.R., 283
Gardner, Martin, 200n9, 264n6, 283
Gettier, Edmund, 67
Gilbert, W.S., 138, 142
Gooel, Kurt, 153n12, 205
Goldwin, Sam, 4
Good, I.J., 283
Goodman, Nelson, 227-230, 283
Gorgias, 106n15
Grabmann, Martin, 281, 283
Grelling, Kurt, 173-76, 185, 188,
283-84
Hamblin, C.L., 165n2, 284
Hartmann, Nicolai, 10n14
Heath, T.L., 246n6
Heijenoort, Jean van, 172n8, 176,
181n13, 182n14, 185n17,
186n18, 284, 287
Heiss, Robert, 284
Helmer, Olaf, 284
Helmont, J.B. van, 284
Hempel, Carl Gustav, 224-26,
230nl0
Herbert, George, 5
hermeneutic circle, 87-88
heterological predicates, 174-75
Heytesbury, William, 281, 284
Hinske, Norbert, 148n9, 284
Hintikka, Jaakko, 284
Hippasus of Metapontum, 246
Hofstadter, Douglas R., 284
Holmes, Oliver Wendell, Jr., 141
Hoy, D.C., 87n17
Hughes, George Edward, 195n2,
281,284
Hughes, Patrick, 139n2, 284
Husserl, Edmund 43nl
Hyde, M.J., 284
hypothetical reasoning, 233-252
291
Index
ideas, Plato's theory of, 117-19
impredicable predicates, 173-74
improper totalities, 137-155
inappropriate presupposition,
137-155
incompleteness of arithmetic, 205
inconsistency, 25-27
inductive paradoxes, 219-230
information, discrepancy, II
Intisar-Ul-Haque, 284
mathematical paradoxes, 159-190

maximal consistent subsets, 25
McGee, Vann, 285
~elhuish, George, 285
~ll, John Stuart, 28, 29
Milton, John, 123
minimal inconsistent subsets, 25
minimax potential-loss strategy,
19-20
Mirimanoff, D., 285
Moore, G.E., 44-45
Jourdain, P.E.B., 284

Kainz, Howard P., 284
Kant, Immanuel, 3, 147-151
Kantian antinomies, 147-151
Kierkegaard, SlIlren, 119, 120
Kirk, Geoffry S., 98n8, 100n9,
101nll
Kneale,~~a,209nI4,284
Nelson, Leonard, 174n9

Newcomb, William A., 264,266,
272
Nicholas ofCusa, 37n8, 38n8, 91,

150,281,285
Nordau, ~ax 4, 4n2, 285
Northrop, Eugene Purdy, 285
Nozick, Robert, 264n6, 285
Kneale,~illtiam,209nI4,284
Koons, Robert C., 259n3, 285

Koyre, Alexandre, 177, 178n12, 285
Kretzmann, Norman, 44n2, 195n2,
209nI3,235n4,281,285
Kutschera, Franz von, 285-86
Kyburg, Henry E., Jr., 222n3, 228n7
Lando, Ortensio, 285
Langford, C.H., 285
Langford, ~., 285
Lazerowitz, Morris, 44n2
Leibniz, G.W., 83, 124
levels of language, 186
levels ofparadoxicality, 57-70
Levi, Isaac, 264n6
Lewis, David, 243
Luce, R. Duncan, 19n21, 267n8
~ackie,
J.L., 285
D.C., 214n17
~andeviUe, Bernard de, 7
~argalit, Avishai, 112nI9, 285
~artin, Robert L., 200nlO, 285
mathematical completeness, 186
mathematical consistency, 186
~akinson,
O'Carrol, M.l., 286.

O'Connor, D.J., 112nI9, 286
Owen, G.E.L., 94n5
Pap, Arthur, 286
paradox dissolution, 57-61
paradox resolution, 35-56, 61-70
paradox, compound, 38-39
paradoxes in philosophy, 117-133
paradoxes of ambiguity, 91-114
paradoxes of equivocation, 91-114
paradoxes of reference, 11 O-ll
paradoxes of vagueness, 77-88
paradoxes, classification of, 70-74
paradoxes, inductive, 219-230
paradoxes, mathematical, 159-190
paradoxicality, levels of, 57-70
Parmenides, 95, 117nl
Pascal, Blaise, 9, 9nl0, 120
Paul of Venice, 198n3, 199-200
Peano, Giuseppe, 182, 185
Perelman, C.H., 286
per-impossibile reasoning, 249-252
Perreiah, Alan, 281,286
Phietas of Cos, 200n8
292
philosophical paradoxes, 117-133
Plato, 3, 9n11, 12n12, 92n3, 93, 95,
106,107,117-18,128-29,
208-09
Plato's theory of ideas, 117-19
plausibility, 15-20,43-53
Plutarch, 86
Poincare, Henri, 172, 173, 186
Popper,~IFL, 153n12,154n12
Prantl, Carl, 12n17, 62n2, 64n3,
65n4, 77n2, 78n5, 79n6, 81n7,
81n8, 82nI3,102nI2, 104n13,
117nl, 165n3, 167n5, 195n2,
198n3, 199n4, 199n5,199n6,
199n7,276,281
predicates, heterological, 174-75
predicates, impredicable, 173-74
prediction, problems of, 151-54
presupposition, inappropriate,
137-155
principle of noncircularity, 123
principle of sufficient reason, 123
principles of comprehensiveness,
123
Prior, A.N., 286
prioritization, propositional, 27-31
Probst, P, 287
propositional plausibility, 43-53
propositional prioritization, 27-31
Protagoras, 9, 63, 64, 65
Pruss, Alexander, 211n16
pseudo-objects,167-170
Quine, W.v.O., 68, 69, 112nI9,
185n17,286
Rabelais, Fran.;:ois, 5, 5n4
Raiffa, Howard, 19n2, 267n8
Ramsey, Frank P., 185
Rapaport, Anatol, 286
Raven, J.E., 98n8, 100n9, 101n11
Read, Stephan, 281, 286
reductio-ad-absurdum, 245-49
referential paradoxes, 110-11
referential singularities, 163-64
Resnick, Michael D., 264n6
Index
resolution, of paradox, 61-70
Richards, T.]., 181-85, 188,286
Ricoeur, Paul, 87n17
riddles, 102-07
Rijk, Lambertus Marie de, 281,
286
Riverso, Emmanuele, 286.
Rivetti, Barbo, 286
Rozeboom, W.W., 286
Russell, Bertrand, 67, 143n8, 166n4,
167,170-76,185-86,188-89,
204,286--87
Riistow, Alexander, 82n12, 200,
281,287
Ryle, Gilbert, 287
Sainsbury, Richard Mark, 79n6,
220nl,287
Salmon, Wesley C., 281, 287
Saperstein, Milton FL, 287
Scheffer, Israel, 19n20
Schilder, Klaas, 287
Schofield, 98n8, 100n9, 101n11
Schopenhauer, Arthur, 4
SchrOer, Helmut, 120n2, 287
Scriven, Michael, 112nI9, 287
self~eception, 221-22
self-falsification, 193-206
self-involvement, totalitarian,
143-47
self-negation, 203
self-reference, 186,203-07
semantic absolutism, 51-53
semantical paradoxes, 193-206
Seneca, 10n12
Sextus Empiricus, 82n13
Shakespeare, William, 4
Shaw, G.B., 4
singularity, referential, 163-64
Skyrms, Brian, 287
Slaatte, Howard A., 120n3, 287
Smullyan, Raymond M., 287
Socrates, 32-33, 78,130, 133, 195,
208-09
sophisms, 3, 167
Sowden, Lanning, 264n6
Index
Spencer, Herbert, 245
Spinoza, Benedictus de, 45n3, 124
Sr. Paul, 5, 120, 200n8
Stalnaker, Robert, 222n3, 227n5,
230n11
Steinberg, Saul, 138nl
Stenius, Erik, 287
Sterne, Laurence, 5, 159
Strawson, P.F., 287
Stroll, Avrum, 287
Stump, Eleonore, 44n2, 195n2,
209nI3,235n4,285
successful identification principle,
181,204,207-08
successful introduction principle,
164-67,176,188
Sullivan, Arthur, 138, 142
Tarski, Alfred, 287
Tertullian, 5, 120
theory of type levels, 186-87
Thines, Georges, 287
Thompson, James, 247-49
Thompson, M.H., Jr., 287
Thompson's lamp, 247-49
Todhunter, Isaac, 269
Toms, E., 288
totalitarian self-involvement,
143-47
totalities, improper, 137-155
truth claims, 193-206
Urbach, B., 288
Urbach, Peter, 153n12
Urquhart, W.S., 288
293
vagueness, paradoxes of, 77-88

Vailati, G., 288
van Fraassen, B.C., 288
Venning, Ralph, 288
Verdi, Guiseppe, 68
viable introduction requirement,
163-64
vicious circle principle, 186
Visser, Albert, 288
Vlastos, Gregory, 94n5
von Wright, G.H., 288
Vredenduin, P.G.J., 288
Walton, Douglas N., 288
Weidemann, Hermann, 281, 288
Weiss, Paul, 112nI9, 288
Weyl, Hermann, 174n9
Whately, Archbishop Roberts, 18
Whitehead, A.N., 172n8, 186,
287
Wilde, Oscar, 4
William of Heytesbury, 44n2
William of Occam, 166, 167n6
Wolgast, Elizabeth H., 288
Wormell, C.P., 288
WyclifJohn,281,288
Zeller, Eduard, 9nll, 59nl, 77n2,
78n3,78n4,81n7, 81n9, 81nl0,
81nl1,82nI3, 101nl0, 107n17,
281
Zeno of Citium, 81
Zeno ofElea, 3, 59-60, 78, 81,
81n9,94-102,107-08,128-29,
281
PHILOSOPHY
'AIl generalizations are false.' The smallest integer that cannot be identified in less than
fourteen words.' 'If your dog is a father then he must be your father.'
Paradoxes. which start from the seemingly acceptahle and arrive at the seemingly
unacceptable, have puzzled and inspired thinkers since the dawn of history. Paradoxes
range from the trivial and easily solvable to the genuinely mystifying and profound.
Attempts to solve them have sometimes changed the course of philosophy, logic, and
mathematics.
Nicholas Rescher's new hook not only presents a readable account of more than 130
parddoxes, including all the best-knmvn ones, but also develops a new general theory of
parddoxes. The author's theOlY encompasses several innovations, such as: the exploration
of the distinction bet\veen truth and plausihility; cycles of inconsistency; the distinction
hetween simple and complex paradoxes; epistemic prioritization in paradox resolution;
and acceptance profiles for the comparative assessment of alternative resolutions of a paradox.
Professor Reschcr demonstrates in detail that paradox analysis is more straightforward
than generally believed, and that it can be standardized by the system he calls "general
aporetics." offering a practi<:al method of paradox management.
"Nicholas Rescher~" Paradoxes is a comprehensive SIIl7'e}' q(pbilosophical, pragmatic,
semantic, and logical paradoxes. It is a mastel.tit! illtrodliCtiOIl to aporetics. an excellent
handbook. alld al/ illl'aillable companioll to cOllrses ill logic alld critical thinkillg epistem%g).', a lid other areas of philosophy. "
-Risto Hilpinen
University of Miami
"Nicholas Rescher COl'ers el'e!)' impol1allf paradox ill tbe literatllre, frum the logicosemantical alld lI1athemalicaltu the predictive alld practical. His general project is to
shou' hOll' paradox-resolution is (I special kind (if illcor1sistellC.y-resollltioll. 1be su'eepillg
thea!)' he p"ts font 'a I'd ll'ill be of illterest to epistemologists, belief-revisio/l theorists. decision-theorists, alld logicialls. Ille book will be recogllized as a landmark of COlliprehellsiveness and sopbistictltioll ami till inl'alliable resolfrce./br stlldellt alld researcber alike . ..
-Neil Tennant
Ohio State University
Nicholas Rescher is Lniversity Professor of Philosophy at the

University of Pittsburgh, where he served for many years as Director
of the Center for Philosophy of Science. He has heen a visiting lecturer at Oxford. Constance, Salamanca. Munich, and Marburg, and has
received five honor;lIY degrees from universities on three continents.
He has puhlished over 80 philosophical books and over 300 philosophical alticles, T\yelvc books and four dissertations have been written about Professor Rescher's contributions to philosophy.
Open Court
Chicago and La Salle. Illinois

/)isl rih/lled I~)' j'/lblisben, (/1'(111/' II' "sl
em .... r design by Handy :'Ibrtinaitis
ISBN 0-8126-9437-6
0-8126-9436-8 (Clolh)
I II
ISBN 0-8126-9437-6
9 7
94376
52495
II
PHILOSOPHY
'Al l generalizations are false. ' 'The smallest integer that cannot be identified in less than
fourteen words.' 'If your dog is a father then he must be your father. '
Paradoxes, which start from the seemingly acceptable and arrive at the seemingly
unacceptable, have puzzled and inspired thinkers since the dawn of history. Pamdoxes
range from the trivial and easily solvable to the genuinely mystifying and profound.
Attempts to solve them have sometimes changed the course of philosophy, logic, and
mathematics.
Nicholas Rc-<;cher's new book not only presents a readable account of more than 130
paradoxes, including aU the best-known ones. but also develops a new general theory of
paradoxes. The author's theOlY encompasses several innovations, such as: the exploration
of the distinction between truth and plausibility; cycles of inconsistency; the distinction
between simple and complex paradoxes; epistemic prioritization in paradox resolution:
and acceptance profties for the comparative assessment of alternative resolutions of a paradox.
Professor Resch!;:r demonstrates in detail that pamdox analysis is more straightforward
than generally believed, and that it can be standardized by the system he calls "general
aporetics.' offering a pmctical method of paradox management.
'Nicholas Reschers Paradoxes is a comprehensiue suroey ~r philosophical, pragmatic,

semantic, and logical paradoxes. It is a masteljitf introduction to aporetics. an excellent
handbook, and em invaluable companion to courses in logic and c1itical thinking, epistemology, and other areas ofphilosophy. "
-Risto Hilpinen
University of Miami
''lVicholas Rescher covers every impOltant paradox in the literature. from the logicoseman tical and mathematical to the predictive and practical. His general project is to
sbow how paradox-resolulion is a special kind of inconsistency-resolution. The sweeping
thealY he plllsforward ll'ill be of interest to epistemologists. belief-revision theorists, dec isian-theOlists, and logicians. The book will be recognized as a landmark of comprehensiveness and sophistication and an invaluable resource/or student and researcher alike. ,.
-Neil Tennant
Ohio State University
Nicholas Rescher is Lniversity Professor of Philosophy at the
University of Pittsburgh, where he served for many years as Director
of the Center for Philosophy of Science. He has been a visiting lecturer at Oxford, Constance, Salamanca. Munich, and Marburg, and has
received five hOO<Jr<lIY degrees from universities on three continents.
He has published over 80 philosophical books and over 300 philosophical articles. Twelve books and four dissertations have been written about Professor Rescher"s contributions to philosophy.
ilL
Open Court
Chicago and La Salle. Illinois

Oistrihllted by Pllblisbers GrollP West
Cover design by Randy Martinaitis
ISBN 0-8126-9437-6 (Paper)
0-8126-9436-8
I II
ISBN 0-8126-9437-6
9 7
94376
52495
II

Rescher 2001 Paradoxes

Uploaded by

Copyright:

Available Formats

Rescher 2001 Paradoxes

Uploaded by

Document Information

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

Rescher 2001 Paradoxes

Uploaded by

Copyright:

Available Formats

Paradoxes

Their Roots, Range, and Resolution

To order books from Open Court, call toll free 1-800-815-2280

Printed and bound in the United States of America.

Library of Congress Cataloging-in-Publication Data

For Rudolf Haller

Paradoxes Discussed in Chapter 2

Paradox Solution via Retention Prioritization

Addressing Paradox Resolution Via Priority

Paradoxes Discussed in Chapter 3

Plausibility as a Guide to Prioritization

Theses of Lowest Priority: The Meaningless

Paradoxes Discussed in Chapter 4

Paradoxes Discussed in Chapter 5

The Paradox of the Heap

Paradoxes Discussed in Chapter 6

Paradoxes of Ambiguity and Equivocation

Paradoxes Discussed in Chapter 7

Paradoxes Discussed in Chapter 8

Paradoxes of Inappropriate Presupposition

Paradoxes Discussed in Chapter 9

Cantor's Paradox and the Viable Introduction

Paradoxes Discussed in Chapter 10

10 Semantical Paradoxes that Involve

Paradoxes Discussed in Chapter 11

11 Inductive Paradoxes (Conflicts of

Paradoxes Discussed in Chapter 12

12 Paradoxes of Hypothetical Reasoning

Hypotheses and Burley's Principle

Paradoxes Discussed in Chapter 13

13 Paradoxes of Choice and Decision

A Sample Paradox of Decision

Index of Paradoxes Discussed

Index of Paradoxes Discussed

Dualized Liar Paradox

Index of Paradoxes Discussed

Mind -Body Paradox

Index of Paradoxes Discussed

Summary of Symbols, Technical Terms, and Principles

inconsistent n-et: a group of n collectively inconsistent propositions

Summary of Symbols, Technical Terms, and Principles

must be independently pre-established. Otherwise this supposed item remains undefined.

Paradoxes Considered in Chapter 1

Apories and Paradoxes

Every era in the history of philosophy has seen a concern with

curious ideas, outlandish occurrences and such-like anomalies in

phenomenon that calls for a uniform treatment. It sets out to

stance: a dissonance of endorsements. A variant way of looking at

Sensory illusion paradoxes, for example, occur in situations in

1.2 Paradoxes Root in Overcommitment

Yet once consistency is lost, how is it to be regained? Any and

The situation is aporetic. We have three contentions:

Greek philosophers of a whimsical tendency were drawn to

(2) R is not b's next-door neighbor

(3) J is not b's next-door neighbor

(4) Sis b's next-door neighbor

(2), (3), (f)

(5) S doesn't live in Chicago

(6) R doesn't live in Chicago

(7) J does live in Chicago