Anda di halaman 1dari 364




For a complete list of volumes in this series see final page of the volume.
Reflections on Reflexivity

edited by
Steven J. Bartlett
(Former professor of philosophy at Saint Louis University)

Peter Suber
(Assistant professor of philosophy, Earlham College)



jar the United States and Canada: Kluwer Academic Publishers, p.o. Box 358,
Accord Station, Hingham, MA 02018-0358, USA
jar the UK and Ireland: Kluwer Academic Publishers, MTP Press Limited, Falcon
House, Queen Square, Lancaster LAI lRN, UK
jar all other countries: Kluwer Academic Publishers Group, Distribution Center,
P.O. Box 322, 3300 AH Dordrecht, The Netherlands

Library of Congress Cataloging in Publication Data

Sel f -r ef erence : reflec tions on re flex ivi t y .

(M ar t inus Nijho ff ph ilosophy l ibr ar y ; 21)

Bi bli ography: p .
1 . Se lf-knowled ge, Theory of. I . Ba rtlett,
St ev en J . II. Sub er, Pe ter. III. Se r ie s: Hartin us
Nij hoff ph ilos oph y li bra r y ; v. 21 )
80450 . S393 4 1987 121 ' . 4 86 - 33182
ISB N 1 3.97 8-94-010-8088-0 e -ISBN -13 . 97 8 -94-009-35 51-8
DOl: 10.1007/978-94-00 9 35 51-8


1987 by Martinus Nijhoff Publishers, ''Dordrecf'it.

Softcover reprint of the hardcover 1st edition 1987

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, mechanical,
photocopying, recording, or otherwise, without the prior written permission of
the publishers,
Martinus Nijhoff Publishers, p.o. Box 163, 3300 AD Dordrecht,
The Netherlands.
Self-Reference: Reflections on Reflexivity
Edited by Steven J. Bartlett and Peter Suber

Steven J. Bartlett, 5
Varieties of Self-Reference
Part I: Informal Reflections
DA.Whewell, 31
Self-Reference and Meaning in a Natural Language
Peter Suber, 41
Logical Rudeness
Myron MiOer, 68
The Pragmatic Paradox
Henry W. Johnstone, Jr., 88
Argumentum ad Hominem With and Without Self-Reference
Douglas Odegard, 92
The Irreflexivity of Knowledge
Part n: Formal Reflections
Frederic B. Fitch, 113
Formalized Self-Reference
Raymond Smullyan, 123
Quotation and Self-Reference
Graham Priest, 145
Unstable Solutions to the Liar Paradox
Part ill: Specific Reflections
W D. Hart, 179
Causation and Self-Reference
JosephM. Boyle, Jr., 193
Is Determinism Self-Refuting?
OlafTollefsen, 209
The Equivocation Defense of Cognitive Relativism
MartinX.Moleski,SJ., 218
The Role of Retortion in the Cognitional Analyses
of Lonergan and Polanyi
James E. Swearingen, 239
Reflexivity and the Decentered Self
Part IV: Bibliography
Peter Suber, 259
A Bibliography of Works on Reflexivity
About the Authors 365

Self-reference, although a topic studied by some philosophers

and known to a number of other disciplines, has received
comparatively little explicit attention. For the most part the focus of
studies of self-reference has been on its logical and linguistic aspects,
with perhaps disproportionate emphasis placed on the reflexive
paradoxes. The eight-volume Macmillan Encyclopedia of Philosophy,
for example, does not contain a single entry in its index under
"self-reference", and in connection with "reflexivity" mentions only
"relations", "classes", and "sets".
Yet, in this volume, the introductory essay identifies some 75
varieties and occurrences of self-reference in a wide range of
disciplines, and the bibliography contains more than 1,200 citations to
English language works about reflexivity. The contributed papers
investigate a number of forms and applications of self-reference, and
examine some of the challenges posed by its difficult temperament.
The editors hope that readers of this volume will gain a richer
sense of the sti11largely unexplored frontiers of reflexivity, and of the
indispensability of reflexive concepts and methods to foundational
inquiries in philosophy, logic, language, and into the freedom,
personality and intelligence of persons.

This is a work that owes its very substance to other men and
women. It is a work that has been in the making for several years, one
which has developed out of previous researches in the field of
self-reference, and which has relied, from start to finish, upon the
friendly cooperation of a special group of researchers who have
contributed their time and ideas to make this volume possible.
Though it is not possible to single out the many individuals to
whom the editors are indebted for the growth of their own interests and
dedication to matters self-referential, this book is the result of the
inspiration of these men and women.
It is a special pleasure for Steven Bartlett to express his gratitude
to his mentor and friend, Paul Ricoeur, to whom he owes much.
Peter Suber is happy to be able to thank Douglas Hofstadter, for
supporting his work in self-reference and for introducing him to
We gratefully acknowledge a grant from Earlham College's
Professional Development Fund that supported much of the final
editorial and layout work. Without the extensive, careful typing of Bill
Lamb, Beth Powers, and John Woida, and the expertise of Ray Ontko in
homogenizing 13 different disk formats, word processors, and
operating systems in his supervision of the printing, we would not have
met our deadline or have had such a handsome volume.
Additional typing was done with good cheer at the last minute by
Scott Hinkley, Mary Kay Kidwell, and Kelvin Holland.
Warm thanks go to Mr. A. W. Schimmelpenninck, Publisher,
Martinus Nijhoff, B.V., whose interest and management have smoothed
the book's path to publication.
We would like to pay special tribute here to Frederic Brenton
Fitch, whose contributions to the subject of self-reference over more
than four decades have brought rare philosophical sensitivity to bear on
formal studies of self-reference. His. work, perhaps more than that of
any other logician, has inspired interest in the relationship between
philosophical and formal aspects of self-reference.
Finally, each editor takes this occasion to thank the other for his
assistance and encouragement.

Steven J. Bartlett

Varieties of Self-Reference
Theory of Reference

The motivation to do philosophy seems to come from a need to

grasp the most basic, the most general, features of the intelligible
world. During the last half century, an increasingly encompassing
perspective has developed due to this motivation. Theory of reference
began as a comparatively narrow and specialized examination of
elementary linguistic forms, such as definite descriptions and proper
names. Much work in theory of reference still reflects this focus. At
the same time, there is a growing realization in technical literature in
the field that referring is not wholly reducible to linguistic
mechanisms. Without a wide range of abilities to refer, we would be
bereft of thoughts, memories, and sensations: The world as we
perceive it, remember it, and conceptualize it would, in the absence of
appropriate referring capacities, collapse into impossibility. All that
we are, have been, and would be receives its form and sense in terms
of a multitude of ways of referring which together make it possible
for our individual worlds to possess an order, for us to contact others,
interpret events, and identify a structure of common experience.
A general theory of reference represents one way among others
of realizing, in the contemporary idiom which every age requires, the
perennial desire for comprehensive philosophical understanding.
General theory of reference seeks to study, with universal
theoretic inclusiveness, an essential constitutive ingredient of human
reality: the phenomenon of referring which, in different forms, is
involved in all study, all reflection, all discourse. It appears to be an
inescapably fundamental basis of all that can be thought and
expressed. General theory of reference itself must make use of the
language, ideas, and realities of referring in order to investigate
them. This fact exhibits reflexivity, or self-reference, the common
theme of the papers in this volume.
Historically, studies of self-reference had a comparatively
limited focus as did general theory of reference, restricted to
problems arising in formal systems when self-reference was
permitted, and to problems in linguistic analysis. It is my intention
here and a main purpose of this collection to begin to show that
reflexivity is an important and pervasive phenomenon, beyond logic
Bartlett Varieties of Self-Reference

and philosophy of language, and beyond the discipline of philosophy

Studies of reflexivity have gone on for some time in a variety of
disciplines, but often in total ignorance of one another, an undesirable
state of affairs in any branch of learning.
The present collection of papers on reflexivity is the first
publication of its kind: It is the first scholarly anthology to represent,
as much as it has been possible to do so in a single volume, some of the
diversity of studies of self-reference.

Self Reference

When we employ thought to understand the nature of thinking,

when we seek to know the presuppositions involved in knowing, we
define a task that essentially involves the subjects we would study.
Reflexivities of this kind are widespread: Sociology, anthropology,
biology, and many other disciplines, as we shall see, exhibit varieties
of self-reference.
Attempting to understand reflexivity gives one the sense of
trying to lift oneself by the bootstraps. Reflexivity is often clearest to
us in the case of categories which must themselves be used if we are to
explicate them, as in thought about thought, or knowledge about
knowledge. Logician Paul Lorenzen has identified a similar form of
reflexivity which comes to light in connection with "elementary
sentences", which reveal certain undeniable conditions of discourse.
He notes that the "decision to accept elementary ways of speaking is
not a matter of argument. It does not make sense to ask for an
'explanation', or to ask for a 'reason'.... If you ask such questions,
... you have already accepted at least the use of elementary
sentences." 1
As we shall see, reflexivity takes many forms. In philosophy,
self-implicating categories and preconditions of discourse have
inspired a good deal of interest. In psychotherapy, a paradigm of
reflexivity is found in the psychological attempt to bring about
changes in one's own psychological makeup. And it appears that it is
just this reflexive capacity to initiate self-change which characterizes
much of creative thought: It may underlie basic creative
problem-solving abilities, and it identifies a special property of the
mental adaptiveness that has gained one organism undisputed control
ofa planet.
I do not have an exhaustive enumeration of varieties of
reflexivity to offer here. I will make no attempt to specify defining
properties of self-reference which are invariant over all of their

Bartlett Varieties of Self-Reference

manifestations. The ideas and terms which we use to describe

reflexive realities will be defined here only by the contextually
determined meanings they possess in the variety of uses to which the
reader himself is accustomed. Awareness of reflexivity ought
properly--i.e., self-referentially--to begin at home, in individual
mental space: It is something best understood informally by its
experience, rather than by stipulated or hypothetical definition.
In the following section, I have brought together forms of
self-reference which are encountered in diverse disciplines; many
comprise separate varieties of self-reference, and not simply
formally equivalent examples in different garb. I have, however, not
been concerned here to establish th~ir independence. What follows is
an annotated inventory of varieties of reflexivity. I have tried to keep
the annotations to a minimum. For some readers, I will doubtless say
too much, and for others, too little. My purpose is to identify main
forms of reflexivity which may interest some readers sufficiently to
explore further.
This broad but self-avowedly incomplete enumeration of
varieties of self-reference offers perhaps the clearest and strongest
evidence--sometimes only through the simple act of pointing, through
ostensive reference--of the extent to which reflexivity permeates our
intellectual and practical worlds.
We begin with a review of several varieties of self-reference
which have become most familiar in the literature:

Better-Known Varieties of Self-Reference

Reflexivity o/indexical signs, egocentric particulars, and

token-reflexive words:

Indexical signs (C.S. Peirce) refer in a manner that is relative to

the speaker: The referents of'!', 'here', 'now', and 'you' are
relative to the individual who uses them, to the place, to the
time, and to whom he speaks.

Egocentric particulars consist of pronouns, demonstratives, and

tenses. Russell called them 'egocentric particulars' because their
reference, like that of indexical signs, is determined relative to
the speaker who uses them. But Russell sought, unlike Peirce, to
reduce all such expressions to the logically proper name, the
egocentric particular, 'this': For example, 'I' refers to the set of
memories, enduring physical features, and abilities which make
up this. 'Now' refers to events simultaneous with this.

Bartlett Varieties of Self-Reference

Token-reflexive words (H. Reichenbach) refer to physical

tokens or instances of an individual act of expression, whether in
speech or in writing. Each token of a specific token-reflexive
word refers to a different physical token--that is to say, each
refers to itself: 'I' refers to the person who utters this token;
'here' refers to the place where this token is uttered. The
identifier 'this token' is token-reflexive: every token of it is a
different token, a physically distinct sound or ink pattern.

Semantical reflexivity:
Natural as well as a few formal languages possess ways of
referring to their own semantical concepts. These concepts
normally link a language to the class of objects to which the
language can refer, as do the semantical concepts truth and
falsity. When a language is allowed to become semantically
self-referential, inconsistencies mayor may not be produced.
The most famous of the semantical paradoxes is that of the Liar,
also called Epimenides' paradox, attributed to Eubulides: "This
man says he is lying. Is what he says true or not?"

Tautological reflexivity:
Tautological propositions have been considered to be reflexive
in an extended sense: Each comprises a closure over a
truth-functional domain expressed by the proposition.
Phenomenology of immediate experience suggests a form of
tautological relation, between the description of such an
experience and the experience itself. The so-called
incorrigibility of certain claims about immediate experience
may be understood in this way.

Set-theoretical reflexivity:
When set-membership is used reflexively, paradoxes may again
be generated. Of these, Russell's paradox, formulated by him in
1901, is probably best known, resulting from the specification
of the set of all and only those sets that do not contain themselves
as elements. A set so defined will contain itself as a member if
and only if it does not.

During the last few years of the 19th century, and the first
decade of the 20th, semantical and set-theoretical paradoxes

Bartlett Varieties of Self-Reference

magnified greatly the difficulties attributable to self-reference.

This was a period of heightened paradox-sensitivity. The

landmarks were:

the flrst published modem paradox, the Burali-Forti

paradox (1897), concerning the greatest ordinal number;
Cantor's paradox (1899), concerning the greatest cardinal
Russel/'s paradox (1901);
the Richard paradox (1905), identified by Jules Richard at a
lycee in Dijon, concerning the nondenumerability of real
the Zermelo-Konig paradox (1905), relating to the finite
definability of real numbers;
Berry's paradox (described by Russell in 1908), relating
to "the least integer not nameable in fewer than
nineteen syllables"; and
Grelling's (or the Grelling-Nelson) paradox (1908),
produced by the self-predication of the predicate

"heterological": a predicate is heterological if a sentence

ascribing the predicate to itself is false.

As formalized semantics fell into step with set theory, an earlier

division made among these paradoxes (by logician Paul Ramsey,
into "syntactic" and "semantic" categories) gave way to their
being grouped under the single category of set-theoretical
paradoxes .

Pragmatical, or performative, self-reference:

When a statement is made, there are two dimensions of its
assertion which mayor may not conflict. One dimension has to
do with what the statement asserts; the other, with the way in
which the statement is made, or how the speaker intends the
statement to be understood. A statement made in such a way that
these two dimensions come together and refer the one to the
other is pragmatically, or performatively, self-referential. The
implicit claim to truth, "There are no truths", is
self-referentially inconsistent in a pragmatical or performative

Bartlett Varieties of Self-Reference

sense. Uttering the word 'cake' in Ramsey's claim, "I can't say
'cake''', similarly is pragmatically self-defeating or
self-refuting. On the other hand, a person who, with face
beet-red, yells, "I'm really mad", utters a statement that is
pragmatically reflexive but not self-referentially inconsistent.
A substantial literature has been devoted to the study of
pragmatical self-reference: One might begin with works by
John Passmore, Henry W. Johnstone, Jr., and J.L. Mackie .

Metalogical, or transcendental, reflexivity:

There is a special kind of relation between a truth-functional
referring proposition and the set of conditions which are
necessary in order for the proposition to be capable of referring
at all. The relation is metalogically, or transcendentally,
reflexive, and it forms a distinct variety of self-reference first
studied by Steven 1. Bartlett.

This relation, interpreted in diverse ways, has been the often not
clearly defined object of attention of various philosophers and
logicians, over a long period of time. Kant, for example, was
especially fond of a similar relation, and made it the basis for his
transcendental deduction: He attempted to demonstrate the
existence of what may be regarded as expressing such a
metalogical (in Kant's terms, transcendental) relationship
between his Categories and the possibility of objective
knowledge. The relationship is reflexive: To assert objective
knowledge while denying one or more of the Categories would,
provided Kant was right, result in a metalogically
self-referentially inconsistent proposition.

A general metalogic of reference makes it possible to undertake

a logically compelling, theoretically neutral, and reflexive
evaluation of many philosophical and other positions. 2

We now move on to a more inclusive enumeration of varieties

of self-reference:

Linguistic Reflexivity

The discipline of linguistics has studied general reflexive aspects

Bartlett Varieties of Self-Reference

of languages, including

self-referential capabilities of natural and formalized languages,

and the paradoxical consequences of their unbridled use;
reflexive properties of generative grammars; and
linguistic and conceptual limitative hypotheses concerning the
structure of natural languages, first formulated by Benjamin
Lee Whorf in his linguistic relativity hypothesis, and later
qualified by other theorists.
Philosophical Reflexivity

In philosophical argument, pragmatical self-reference has been

used to make evident what one in fact is committed to in making a
given assertion; metalogical self-reference makes clear what one must
be committed to if an assertion in principle is to be meaningful.
Philosophical arguments using pragmatical self-reference are,
accordingly, normally expressed as ad hominem arguments.
Argumentation involving metalogical reflexivity has an unmistakable
transcendental orientation.

Other examples of self-reference used in philosophical

argument include petitio principii, circular reasoning, reductio ad
absurdum, and applications of semantical and set-theoretical
Reflexive approaches to philosophical argument are inclined to
focus on

the self-application of principles, predicates, and categories;

the self-justification (self-validation) or self-refutation of
theories, inferences, or individual propositions; or
the self-supporting character of certain inductive arguments.

Reflexivity has been significant to philosophy, beyond its

application in argumentation, in the descriptive context of
phenomenology. Hussert's theory of phenomenology is essentially
reflexive: For him, phenomenology is a science of science, a theory
of theories, which contains itself within its own proper subject
matter, as it attempts to reach a radical degree of self-understanding.
Reflexivity also appears as a specific phenomenological topic of
research in connection with reflective experience, specifically

Bartlett Varieties of Self-Reference

self-consciousness and reflexive knowledge.

Proof-Theoretical Reflexivity

Some of the most dramatic 20th century contributions to

mathematics and the theory of formal systems have resulted from
self-referential techniques of proof. The family of limitative
theorems has grown appreciably since the foundational work of
Cantor and Godel, relating to incompleteness, undecidability, and
In addition to limitative results obtained through reflexive
techniques of proof, important contributions by Frederic Brenton
Fitch and Raymond Smullyan, contributors to this volume, have
examined ways in which formal systems may be constructed so as to
permit self-reference without thereby automatically becoming
Another major area in which reflexivity plays a central role in
mathematics should also be mentioned, namely, the theory of
recursive functions, also known as computability theory, which
immediately leads into:

Artificial Intelligence: Mechanizing Reflexivity

Several varieties of self-reference currently form a part of the

subject matter of artificial intelligence. They include studies of

self-correcting systems
self-regulating systems
systems capable of self-initiated learning
self-organizing systems
self-reproducing systems.

A variety of results now exists which relates to reflexivity in the

context of computability theory. John von Neumann and C.E.
Shannon, for example, have studied self-correcting procedures for
general computations and information transmission, respectively.
(See Reflexivity in Information Theory and General Systems Theory,
below.) Thoralf Skolem, A.M. Turing, Kurt Godel, Alonzo Church,
Emil Post, Andrzej Mostowski, and others have made contributions
to recursive function theory that are basic to current research on
reflexive systems in the field of artificial intelligence.

Bartlett Varieties of Self-Reference

Reflexivity in Physics

Some of the most interesting conceptual puzzles in physics as

well as in philosophy of science stem from theories in physics which
apparently reveal forms of reflexivity in certain classes of physical
phenomena. Reflexivity seems to be involved in both contemporary
quantum mechanics and general relativity.
Quantum mechanics continues to encounter conceptually
baffling phenomena: In the last several years, for example, physicist
Alain Aspect has designed a group of experiments which study the
apparent fact that quantum results are often determined by states of
comparatively distant components of the measuring apparatus.
Aspect's attempts to isolate quantum phenomena from the physically
reflexive influence of the measuring system appear to confirm the
role of a variety of physically reflexive determination. What seems
to be in question is not an instance of physically propagated causal
influence, but rather a situation in which the measuring apparatus, the
observer, and the quantum phenomena to be measured functionally
constitute a system which itself reflexively defines properties of the
phenomena which may be measured. 3
In a similar vein, quantum indeterminism and uncertainty
appear to manifest relations of reflexivity that are involved in the
system formed by the theoretical framework, the physical apparatus
and observer, and the phenomena under study.4
General relativity offers two instances of reflexivity, the first of
which exhibits the reflexivity of functionally interdependent
descriptions, and the second, a variety of topological recurvature:

the geometric-topological model which expresses density of

matter and gravitation as functions of the metric of space
curvature, and vice-versa; and
closed universe models which, although unbounded, are finite.

Reflexivity of Space and Time

In topology, there are configurations formed by lines, surfaces,

and volumes which exhibit spatial reflexivity: Some of these are
frequently used as models of, or as spatial metaphors for, specific
forms of self-reference. 5 They include

Bartlett Varieties of Self-Reference

the closed loop, as in a circle, a line curving back on itself in a

the Moebius strip, a two-dimensional band curving back on
itself in three-dimensional space;
the Klein bottle, a three-dimensional container whose "inner"
space recurves so as to be continuous with its "outer" space (a
geometrical object not representable in three dimensions); and
the Riemannian model of the closed universe, comprising one
space-time continuum, unbounded yet finite in volume.

Also worth mentioning is a special area of topology devoted to the

study of Banach spaces, in which properties of general reflexive
spaces may be investigated.
As yet, we have not witnessed the development of a discipline
whose special subject matter is time, as is space from the standpoint of
geometry and topology. "Chronology" does not yet exist as an
independent field of study. History and futures-studies do exist, but
neither explicitly studies reflexive temporal structures such as closed
temporal loops, which may be applicable in the context of particle
physics, and cyclic and spiral periodicity.

Biological Reflexivity

Biology has encountered reflexivity in connection with

self-replicating structures, investigated in studies of genetic

replication, in particular in connection with viral reproduction;
self-organizing biological systems, for which functional analysis
and a systems-approach to the study of living organisms
(organismic biology) are essential: "[t] he whole acts as a causal
. ,
umt. .. on Its own parts. "6

Reflexivity in Political Science

Systems of political administration may involve reflexivity in

several ways:
First, there is reflexivity relating to the set of beliefs espoused
by the administration, which is basic to a country's sense of
self-identity as symbolized by flag and nationalistic creed. Here lies
the reflexivity of ideology: It forms a self-validating belief-system

Bartlett Varieties of Self-Reference

which is self-reinforcing, and self-isolating. Its self-isolating

character precludes effective communication with alien
belief-systems. Philosophical positions frequently are ideologically
bound in this way,7 as are religious belief-systems.
Second, there is self-reference which concerns the
self-limitation or self-augmentation of political power. (See
Reflexivity in Law, below.)
Third, the internal dynamics of political systems may become
dysfunctional and self-destructive. The study of reflexively
destructive political systems is the focus of political theories of

Reflexivity in Law

Self-reference takes several forms in the context of

jurisprudence, which will only be listed here. 8 They include

legislative approaches to self-limitation

self-amendment, and paradoxes which self-amending laws may
problems and puzzles engendered by self-referring laws
circularity of liens
mutuality in contract law.

Sociological Reflexivity

In recent years, sociologists have studied the question whether

publicizing a prediction about public behavior will influence,
positively or negatively, the events predicted. This is the so-called
problem of reflexive prediction in behavioral science. It has direct
application to whether public disclosure of results of a public opinion
survey prejudices the survey's results by acting as a self-fulfilling
prophesy. The problem of reflexive prediction is immediately
relevant to the policy of releasing vote tallies from the east coast of
the United States before voters on the west coast have gone to the
An analogous reflexive problem in philosophy concerns the
problem of self-prediction: Here, arguments have been advanced, as
well as countered, to show that prior to the occurrence of a decision it
is, or is not, impossible to know what that decision, or the behavior
consequent to it, will be. These arguments form a contemporary

Bartlett Varieties of Self-Reference

approach to the problem of the freedom of the will, and they have, in
the context of philosophy of science, addressed the question whether
objective knowledge in human behavioral science is actually possible.

Reflexivity in Economics

In economics, reflexivity has taken several forms; a partial list


reflexive monetary adjustment theory

theory of business cycles
self-correcting investment management strategies
the dynamics of self-fueling inflationary and deflationary
analysis of exponential growth, usually relating to compounding
of reinvestments.

Game-Theory, Decision-Theory, and Reflexivity

Game-theory and decision-theory encounter varieties of

reflexivity in connection with:

rules permitting self-modification,

self-undermining or self-guaranteeing strategies, and
decision methods which concern the ordering of individual

In relation to the latter, Kenneth Arrow is known for his

impossibility theorem, implied by Arrow's paradox, which
demonstrates, for example, that the order of voting for or against
bills, and on amendments to them, can seriously affect the outcome:
hence there does not exist a rational and equitable general social
decision method. 9

In Anthropology

Probably the most famous variety of self-reference in

anthropology occurs in Benjamin Lee Whorfs linguistic relativity
hypothesis, briefly mentioned under the heading of Linguistic
Reflexivity. According to Whorf, thinking

Bartlett Varieties of Self-Reference

follows a network of tracks laid down in a given language, an

organization which may concentrate systematically upon certain
phases of reality, certain aspects of intelligence, and systematically
discard others featured by other languages. The individual is
completely unaware of -their organization and is constrained
completely within its unbreakable bounds. to

This claim about the determination of thought by language,

upon which thought relies for its expression, is itself reflexive, for
the linguistic relativity hypothesis is precisely a set of thoughts
expressed by language.
Another more general variety of reflexive determination of
special interest to anthropologists is the framework-relativity of
culturally-based values: Culturally relative values appear to have the
self-reinforcing and self-isolating character of political ideologies.
The unquestioned acceptance of such values by members of a society,
because these values constitute social experience from their point of
view, is at the basis of our "culture-shock" when we find ourselves in
a dramatically foreign society.

Reflexivity in Mythology and Theology

In mythology, reflexivity is found in myths having to do with

the self-embodiment of a deity in the universe created by him,

general cosmic periodicity, and
the myth of eternal return (or eternal recurrence): the notion
that there exist cosmic cycles which are such that each event in
the universe will recur in exact detail an infinite number of
times in the future as it already has in the past.

In theology, we encounter a certain reflexivity on the part of the

predicate of perfection in ontological arguments. Also, in some
religious rituals, for example, communion, there is an evident
expression of reflexivity: The faith of worshippers at once is a means
for, and a symbolization of, their own reflexive participation in the
self-embodiment of god, i.e., his incarnation in his own creation,
itself a second manifestation of reflexivity, as just noted.

In Literature

Literary imagination and ingenuity are especially evident in

Bartlett Varieties of Self-Reference

certain works of poetry and fiction which have self-referential

characteristics. Members of this class of literary works are known as
self-begetting, self-describing, or self-reflexive. ll
I cannot refrain from including a specific reference here, one
which was communicated to me by Martin Gardner. This is the
delightful and extremely clever reflexive story by Max Beerbohm,
"Enoch Soames" .12
Some of the best science fiction has made reflexivity a theme of
central importance, as in works which involve closed loops in time,
paradoxes of self-identity, shifts in and out of higher dimensions, etc.

In Music

Because music exists sequentially in time--even though it has

both the "diachronic" and "synchronic" dimensions of melody and
harmony--music must rely on the auditory memory of listeners to
attain the musical equivalent of reflexivity. Cyclical structures,
recurrent thematic material, and forms such as the fugue and canon,
offer opportunities for the expression of the varieties of reflexivity so
far realized in music.
Close neighbors of music, physical acoustics and the physiology
of hearing, have identified phenomena with reflexive characteristics:
In acoustical resonance, for example, sound waves cause standing
waves to be established in a resonating material, which contributes in
a self-reinforcing loop to the propagation of sound waves of greater
amplitude. In the physiology of hearing, it has been discovered that
there is a kind of circularity in judgments which seek to determine
musical pitch.

Reflexivity in Art

A self-depicting, recurrent subject in painting, for example,

may consist of a picture of a scene containing an easel and canvas, on
which is painted the same scene with easel and canvas, on which the
picture recurs to the limits of resolution. Salvador Dali and M.C.
Escher were skilled in their successful renditions of self-referential
subjects. The picture-in-a-picture variety of self-reference in art has
been .examined by D. Carrier. B. Ernst has studied reflexivity in
Escher, while Douglas Hofstadter has explored self-reference in the
art of Escher, the music of Bach, and the limitative results of Godel.


Bartlett Varieties of Self-Reference

In humor, reflexivity is essential in certain puns and

double-entendres: Here is a play on words, which potentially
involves a cyclical oscillation between homonyms:

Three brothers move to California to start a cattle ranch.

When they have bought the land, they phone their
mother, asking that she name their ranch. The name she
suggests is: "Where the sun's rays meet."

Reframing is frequently involved in humor, as when there is a

sudden retroactive change of meaning, normally the purpose of the
Humor, the capacity to apprehend different levels of meaning
quickly, reframing, creativity, and play are interwoven abilities
which can involve self-reference. These varieties of self-reference
appear to make up one side of a coin, the other side of which
represents conditions which come about when these "healthy" forms
of reflexivity short-circuit, producing such dysfunctions as autism
and schizophrenia.

Psychiatry and Psychotherapy: Reflexivity Awry

Reflexivity in psychiatry and psychotherapy is of more critical

importance than perhaps in any of the other applied sciences. The
assistance they attempt to offer is heavily dependent upon the
reflexive capacity of patients themselves:
[O]ne's destiny is shaped from within....This is ... a process of
change that originates in one's heart and expands outward, always
within the purview and direction of a knowing consciousness,
begins with a vision of freedom, with an "I want to become...",
with a sense of the potentiality to become what one is not. One
gropes toward this vision in the dark, with no guide, no map, no
guarantee. Here one acts as subject, author, creator. I3

Psychiatry and psychotherapy treat certain conditions which

either appear to be produced by reflexive incapacities of patients, or
by reflexivity which has become distorted or excessive.
Furthermore, as we shall see, some techniques used in
non-pharmacological psychotherapy may themselves have a reflexive
Conditions which appear to implicate reflexivity gone awry

Bartlett Varieties of Self-Reference

autism, the psychiatric equivalent of solipsism;

narcissism, a form of adult-level autism which is self-isolating,

built upon layers of pretense, and marked by extreme

willfullness', 14
schizophrenia, characterized by a breakdown of contact with
reality, distorted and disjointed thinking, and behavioral
a desire for self-destruction which, unchecked, may become
suicidal; and
self-undermining patterns of communication in troubled

Different hypotheses have been offered to explain how these

conditions come about. Of particular relevance to a study of
reflexivity are three hypotheses concerning, respectively, narcissism,
schizophrenia, and families at war with themselves:

It has been hypothesized that narcissism is brought about by a

traumatic episode that precipitates a progressive flight inward:
--a trauma of this kind, for example, might be the suicide of a
boy's father after a quarrel between them.
It is thought that schizophrenia may result either from a
generalized breakdown of the capacity to distinguish levels of
meaning in communications (Gregory Bateson's double-bind
theory), or from a child's chronic experience of and
involvement in conflict-based family impasses and blocks to
open, undistorted expression (Ronald D. Laing's "knot" theory) .
Interactions between members of a family are thought to form a
system which can short-circuit in pain and frustration, or--also
through a feedback loop--can become integrated so as to offer a
context for personal growth.

A reflexive psychotherapeutic technique which sometimes is

dramatically effective is captured by the title of Allen Pay's book,
Making Things Better By Making Them Worse. 1S The technique is
known as "symptom prescription" or "paradoxical injunction". It
bears some similarity to the Eastern technique of using koans to force
an individual, through the use of his established patterns of thinking,
to break free from the constraints of his own conceptual system. In

Bartlett Varieties of Self-Reference

symptom-prescription, for example, a former draftsman, whose

hands now shake uncontrollably but because of no physical illness,
may be asked to stand in front of a mirror at regular intervals, and
attempt to shake. The theory, and the actual effect, is that, often, an
uncontrollable pattern is thereby brought under voluntary control.
Symptom-prescription has now been used extensively in the context
of family therapy and individual behavior modification.
Symptom-prescription may be understood as the use of controlled
positive feedback (see below) to regain control of a system--here, the
personal world of an individual--that has gone out of control.

Reflexivity in Information Theory and General Systems


These two disciplines share responsibility for the concept of

feedback, which is of sufficient importance to the phenomenon of
reflexivity that w.e should look at it somewhat closely.
Self-reference, translated into the terms of dynamic systems
analysis, and viewed as a liability rather than an asset, may be likened

On the other hand, in a more constructive sense, here is an

example of a positive feedback loop, representing the exponential
increase of money in a savings account:

In a positive feedback loop, a chain of cause-and-effect

relationships closes on itself, so that increasing anyone element in the
loop will start a sequence of changes that will result in the originally

Bartlett Varieties of Self-Reference

changed element being increased even more.

In a negative feedback loop, a change in one element is
propagated around a chain of events until it comes back to change that
element in a direction opposite to the initial change. Consider, for

() ~
Births/year (+) .-,P-o-pu-la-ti-on--', (-) Deaths/year

%~fpopu1ation ~
dymg each year
~ giving
birth each

It follows that positive feedback loops tend to lead to runaway

growth, while the addition of negative feedback loops tends to
regulate growth and hold a system in a steady state.
Many of the varieties of reflexivity we have identified have this
character: Some paradoxes may be understood as runaway logical
feedback loops. Self-reference itself appears essentially to require a
positive feedback loop. In formal logic, a restrictive theory of types
acts as negative feedback, eliminating the runaway cycling of an
original paradox. Further applications of feedback loops to other
varieties of self-reference will likely occur to the reader.
Information theory has contributed to the study of reflexivity in
two ways which will be mentioned here, and has collaborated with
general systems theory in connection with two others:
C.E. Shannon and R.W. Hamming have contributed to the
development of error-detecting codes, now used in telemetry, which
help to insure accuracy in the transmission and reception of messages.
An error-detecting code is reflexive, since the message transmitted is
encoded so as to reveal errors not only in the message content but in
the code check bits themselves. 16
Gregory Bateson's double-bind theory of schizophrenia,
already discussed briefly, strongly relies on concepts drawn from
information theory.
General systems theory, in comparison, has sought to offer a
method of analysis which can show how the components of a system
possess the identity they do as a function of the system as a whole.
General systems theory endeavors to construct abstract models which
can make explicit the isomorphism of concepts, laws, and theories,

Bartlett Varieties of Self-Reference

often as they are expressed in a plurality of more specialized

General systems theory is presently working to develop
reflexive models, for example, of

the processes of development, specialization of organs, and

growth of organisms; and
the maintenance of homeostasis in biological systems (in
medicine, called "health")

General systems theory and information theory have joined

forces outside of their own individual frameworks, in connection

approaches in the field of family therapy which emphasize

effective communication and a systems-based understanding of
family dynamics; and
the formulation of a theoretical basis for the psychotherapeutic
technique of symptom-prescription.

Hermeneutics, Paradigms, and the Theory of Research


Although Imre Lakatos' theory of research programs has

become an area of independent interest in the philosophy of science
along with Kuhn's studies of paradigm-shifts and the nature of theory
change, both may be considered to belong to hermeneutics: both are
theories of interpretative models. But, no matter how one wishes to
make the classification, studies of paradigms, research programs, and
general hermeneutics jointly examine a single form of self-reference
which is at the root of tendencies of systems of interpretation so to
construe their subject matters that they become self-validating,
standing comparatively immune to "recalcitrant experience". One of
the topics of central interest in hermeneutics is how rival systems of
interpretation may be self-amending in the face of problematic
experimental results, and yet continue to adhere, in a self-validating
manner, to differing heuristics.
The variety of self-reference involved here is closely related to
the reflexivity of position-taking in general, as in political ideology
and religious dogma, psychological narcissism, and much

Bartlett Varieties of Self-Reference

philosophical argumentation.

Reflexivity in Neurophysiology

From a phenomenological point of view, we appear to share an

experience of forms of reflexivity in connection with self-referential
thoughts and feelings such as pain and anxiety. It is perhaps natural to
suppose that subjective experiences of reflexivity have associated with
them underlying neurophysiological processes which are similarly
reflexive in character.
As yet, studies of the neurophysiology of the human brain are
inconclusive here. However, research conducted by Karl Pribram
and others suggests that the junctional microstructure of the brain
may serve as a substrate for holographic interference patterns or
resonance circuits. Descriptions of these hypothesized processes,
whose existence is not yet certain, are indeed unmistakably reflexive
in character:
Recent developments in optical storage and information
processing have been helpful to neurophysiology by offering
theoretical models which may reveal a good deal about human
cognitive abilities. For example, three-dimensional holograms are
now possible, in which information is distributed throughout a
volume; the applications of holography to model the human brain
appear to be promising. The patterns of electrical activity involved in
memory, vision, taste, smell, and touch appear to be of holographic
At present, these studies suggest that the brain responds as a
general system to patterns of excitation which do not have localized
paths of conduction. The entire system is organized as a network,
involving a multiplicity of loops of different lengths and orders of
complexity. There are neuronal circuits and holographic return
loops which appear to be able to reexcite initial patterns of excitation.
The resulting system exhibits a dynamic which involves feedback and
feedforward. It may form the reality that underlies and sustains our
perhaps distinctive abilities to utilize the contents of memory, to
reflect upon our experience, to anticipate, and to gain a progressively
integrated sense of personal identity.


The individual papers appearing in this collection study

reflexivity from several perspectives:

Bartlett Varieties of Self-Reference

From a general standpoint: Papers by

D.A. Whewell, on the reflexive flexibility of natural


Douglas Odegard, concerning the irreflexivity of

and W.D. Hart, discussing a causal conception of

On two philosophical theories of reflexivity: An essay by

Martin X. Moleski, describing retortion III

Lonergan and Polanyi.

From a formal point of view: Papers by

Frederic B. Fitch, formalizing self-reference; and

Raymond Smullyan, on self-reference and quotation.

On semantical reflexivity: Papers by

Myron Miller, on the Liar's paradox; and

Graham Priest, discussing a recent theory of semantical

With pragmatical reflexivity in view: Essays by

Peter Suber, on the norms of debate; and

Henry W. Johnstone, Jr., on self-reference and ad
hominem argument.

On specific applications: Essays by

Joseph M. Boyle, Jr., offering a self-referential

argument for free choice;
Olaf Tollefsen, on the self-defeating nature of relativism;
and James Swearingen, characterizing reflexivity in the
novel, Tristram Shandy.

Steven J. Bartlett
Former Professor of Philosophy
St. Louis University
St. Louis, Missouri
Bartlett Varieties of Self-Reference


1. Paul Lorenzen, Normative Logic and Ethics,

Bibliographisches Institut, 1969, p. 14. See also his conception
of operative logic in Einfuhrung in die operative Logik und
Mathematik, Springer Verlag, 1969.

2. See citations under 'Bartlett' in the Bibliography. (This is a

variety of self-reference not mentioned in this introduction.)

3. To this author, these experiments appear, in an almost obvious

way, to point toward the relevance of a metalogical analysis of
preconditions of possible quantum measurement of (Le.,
reference to) the micro events in question.

4. Cf. Steven J. Bartlett, "Self-reference, Phenomenology, and

Philosophy of Science", Methodology and Science; 13, 3,
(1980) 143-167.

5. See, e.g., the essay by Graham Priest in this volume.

6. Wilfred Eade Agar, A Contribution to the Theory of the

Living Organism, Melbourne University Press, 1943. See also
Ludwig Bertalanffy, Modern Theories of Development: An
Introduction to Theoretical Biology, trans. by J.H. Woodger,
Oxford University Press, 1933; John Richard Gregg, ed., Form
and Strategy in Science, D. Reidel, 1964; Joseph Henry
Woodger, The Axiomatic Method in Biology, Cambridge
University Press, 1937, and Biological Principles: A Critical
Study, K. Paul, Trench, Trubner, 1929).

7. See Steven J. Bartlett, "Philosophy as Ideology",

Metaphilosophy, 17, 1 (1986) 1-13.

8. For further information, see the section "Reflexivity in Law" in

the Bibliography.

9. Kenneth J. Arrow, Social Choice and Individual Values, Wiley,

1951; Collected Papers, Harvard University Press, 1983; Social
Choice and Justice, Harvard University Press, 1983. See also
John von Neumann and Oskar Morgenstern, Theory of Games
and Economic Behavior, Princeton University Press, 1947; R.

Bartlett Varieties of Self-Reference

Duncan Luce and Howard Raiffa, Games and Decisions, Wiley


10. Benjamin Lee Whorf, "Langage, Mind and Reality", ETC.: A

Review of General Semantics, 9,3 (1952) 177.

11. For works describing these, see the Bibliography, "Reflexivity

in Literature".

12. In Max Beerbohm's Seven Men, William Heinemann, 1926

(first published in 1919), pp. 3-48.

13. Allen Wheelis, M.D., How People Change, Harper and Row,
1973, p. 105.

14. See Steven J. Bartlett, "Narcissism and Philosophy",

Methodology and Science, 19, 1 (1986) 16-26.

15. Allen Fay, Making Things Better by Making Them Worse,

Hawthorne Books, 1978.

16. See, e.g., Richard Wesley Hamming, Coding and Information

Theory, Prentice-Hall, 1980; Claude E. Shannon, The
Mathematical Theory of Communication, University of Illinois
Press, 1949; and Steven 1. Bartlett, "Lower Bounds of
Ambiguity and Redundancy", Poznan Studies in the Philosophy
of the Sciences, 4, 1-4 (1978) 37-48.

17. See Paul Greguis, ed., Holography in Medicine, IPC Science

and Technology Press, 1976; and Karl H. Pribram, Languages
of the Brain: Experimental Paradoxes and Principles zn
Neurophysiology, Prentice-Hall, 1971.

Part I. Informal Reflections
D.A. Whewell

Self-Reference and Meaning in a Natural

In this short article, I attempt to refute the theory that no
proposition can properly make an assertion about itself; in other words,
that all self-referring statements are, for either logical or semantic
reasons, illegitimate. I argue, firstly, that it is serously misleading in
this sort of discussion to treat all self-referring statements as if they
belonged to a single homogeneous class; that in fact, there are a number
of ways in which a statement might refer to itself. As secondly, I argue
that the logical or semantic objections to treating self-referring
statements in ordinary language as legitimately self-referring are not in
all cases insurmountable, although in some other cases the objections
are clearly insurmountable, as in the case of the totally self-referring
statement "this statement is false".
There already exists a formidable quantity of literature in this
field, most of it of a highly technical nature. The reason for this is
historical, for the view that all self-referring statements are
meaningless was originally advanced by Russell as a direct consequence
of his ramified theory of types. This theory was specifically designed
to dispose of the so-called self-referential paradoxes which cropped up
in the course of his attempts to formalize the foundations of
mathematics. These paradoxes were also encountered by other
logicians in their attempts to formalize set theory and to describe exact
language structures in which the rules for determining truth and
falsehood were precisely stated. Clearly, in constructing any sort of
formal system it is vitally important to ensure that the formation and
derivation rules that are put in the system do not generate logical
paradoxes; that is to say, that they do not lead to inconsistencies and
incoherencies within the system. This can in principle be achieved
quite simply by inserting some suitable exclusion device. But if the
exclusion device is at all arbitrary, as is quite often the case, then it
cannot legitimately be used to dispose of the paradoxes as they occur in
ordinary language. I .should perhaps point out that I am only concerned
with one specific type oflogical paradox, namely that which appears to
be generated by the self-reflexiveness of a self-referring statement.
The notorious Liar Paradox is a good example. If the statement "what I
am now saying is false" is interpreted as being about itself then if it is
true it is false and if it is false it is true. What I wish to determine is
whether in an informal and flexible language like the English language,
Whewell Self Reference and Meaning

it is possible for a statement to refer meaningfully to itself, or whether

there are always decisive objections to its so doing. The mere fact that a
particular move would be impossible within a formally rigorous
language does not seem to me to be, in itself, a decisive objection to
such a move's being made within a natural langauge. The very
untidiness of a natural language makes it more flexible and more
generally useful than a formalized language. What can be achieved in
the one may not be achievable in the other.
Even in a natural language, however, one does not wish to allow
the kind of incoherence exemplified by the Liar Paradox. On the other
hand, one does not need special reasons for rejecting logical paradoxes
of this type, other than that one is obliged to ascribe contradictory truth
values to them and therefore cannot understand anything by them. This
may perhaps be called an exclusion device, but if so there is nothing in
the least arbitrary about it. Moreover, it has the advantage of not
automatically excluding statements which do not themselves generate
paradoxes, as the exclusion devices of formal systems invariable do.
This leaves us free to decide whether there are any positive grounds for
treating those seemingly non-paradoxical statements as legitimately
self-referring. To this end, I shall be discussing a whole range of such
statements. I begin by distinguishing three basic types of
self-referential statement: the totally self-referring, the partially
self-referring, and the incidentally self-referring.
A statement is totally self-referring if it refers explicitly to itself
by means of a singular referring expression. For instance, "this is a
true statement" or "what I am now saying is false". Such a statement
may, in addition, refer to statements other than itself, but in so doing
must make use of an other-referring expression; e.g. "this statement
and the previous one are false". A statement may also refer to itself via
another statement. For instance, on one side of a blackboard is written
"the statement on the other side of this blackboard is true", and on the
other side of the blackboard is written "the statement on the other side
of this blackboard is false". As is evident from this example, indirect
self-reference may give rise to precisely the same type of logical
paradox as direct self-reference. This is important because it shows
that the totally self referring statement "this statement is false" cannot
be regarded as measningless just because itgives rise to logical paradox.
For one would then have to say that the statement "the statement on the
other side of this blackboard is true" is meaningless whenever the
statement on the other side of the blackboard is such as to give rise to a
logical paradox. Such a conclusion would be unacceptable because it is
a contingent fact that the statement on the other side of the blackboard
says what it does say. 1 On the other hand, the mere fact that the

Whewell Self Reference and Meaning

paradox arises is a sufficient reason for denying that the sentence

conveys a genuine assertion in this particular context, although one
knows of course what it would be like for it to convey a genuine
assertion, and to that extent one has no difficulty in understanding it.
A statement is partially self-referring if it is about a whole class
of statements of which it is itself a member. For instance, "all the
statements on this page are true" and the statement "every meaningful
but non-tautological statement must be in principle empirically
verifiable". Sometimes I may have to go outside the statement in order
to know whether it is self-referring or not. For instance, "everything
said by Whewell is false". This proposition is self-referring, and also
paradoxical, if said by Whewell, but not otherwise. Qua proposition, it
describes a logically possible state of affairs. It would therefore be
absurd to claim that it was meaningless just because it happened to be
asserted by Whewell. Another point that deserves to be mentioned here
is that a partially self-referring statement, unlike one that is totally
self-referring, does not need to contain a referring expression, as is the
case with the statement "all general statements are false".
A statement is incidentally self-referring if it refers to some
members of a class of which it not only happens to be a member, but
also a member of the sub-class of members referred to. An example of
this would be the assertion "some of my utterances make sense". One
peculiarity of such a statement is that it has the appearance of being
self-validating, but it is not self-validating, because in order to
determine whether or not the statement is self-referring I must first
determine whether or not it makes sense. Merely to say: "it does not
matter what truth-value is assigned to it, it still makes sense and so is
true", is of course to prejudge the issue. Nevertheless, one can, and
indeed must, know it makes sense before attempting to determine its
truth value as an other-referring statement. The point is: I only have to
know what it would be like for it to have a truth-value as an
other-referring statement to know that it has a meaning. Hence,
without first deciding the truth-value of any other statement I can see
that this statement does indeed belong to the class of those of my
utterances which make sense, and is therefore an example of the very
thing I am talking about. And in this way it does become its own
illustration, though not, in any logically vicious sense, its own proof.
In statements of this type it is natural to add the rider "like this one, for
instance". Thus, in this case we get" some of my utterances make sense,
like this one for instance".
Even a statement that is self-referring in this rather weak sense
may give rise to a logical patadox. Take the following statement as an
example: "I sometimes say things which are false". One might be

Whewell Self Reference and Meaning

tempted to argue that this statement is self-validating, for if it is true

that I sometimes say things which are false then, because this is
something I say and it is false then it is true that I sometimes say things
which are false. Hence, if true it is true, and if false it is still true.
However, this argument would only work if one could treat the
statement, when false, as legitimately, though incidentally,
self-referring, but this is impossible. For I must first know that it is
itself a member of the class of things which I say that are false, before I
can say that it is self-referring. I am not entitled arbitrarily to assume
that it is false; I must have a positive reason for saying so. But clearly I
cannot have a positive reason for saying that it is false without first
determining its truth-value as an other-referring statement, and to do
this I must go outside the statement and examine all of my other
statements. If, on examining my other statements, I find that none of
them is false then I have in effect discovered that my claim that I
sometimes assert things which are false, is false, and so incidentally
self-referring. But this of course gives rise to the liar paradox, for
when we add the rider "like this one, for instance" we get "some of my
assertions are false like this one, for instance". This shows that the
statement is illegitimately self-referring when interpreted this way.
If the above statement had been legitimately self-referring then
Russell's claim that he once succeeded in tricking the notoriously
truthful Moore into telling a lie would have been false. The incident is
reported in Russell's autobiography.2 "I have never but once succeeded
in making Moore tell a lie, and that was by subterfuge. 'Moore', I said,
'do you always speak the truth?' 'No', he replied. I believe this to be the
only lie he ever told." If the claim "I do not always speak the truth" is
allowed to be self-referring then it cannot be assigned a determinate
truth-value and so (qua self-referring statement) cannot be a lie.
However, as an other-referring statement, it can be a lie, and so
Russell's trick can work. It is important to realize, however, that the
statement does not thereby become its own illustration, much less its
own proof.
Before moving on to a consideration of the class of totally
self-referring statements I wish to explain my criteria (though perhaps
'guidelines' would be a better term to use as they are not formally
rigorous) for distinguishing legitimate from illegitimate self-reference
withing the framework of a natural language. The criteria, or
guidelines, to be acceptablee, should not only appeal to common sense,
or logical intuition, but should allow for a greater degree of logical
flexibility in the use of language than would be possible within a
formalized language structure. Since the criteria are not formally
rigorous, their application to particular cases will be, to some extent, ad

Whewell Self Reference and Meaning

hoc, but I do not see this as a decisive objection, since they are not
supposed to function like the formation and derivation rules of a formal
system that must, of necessity, be both rigid and highly specific. Thus I
say that a statement is to count as legitimately self-referring if it
satisfies the following three conditions: (1) it must be in principle
possible to ascribe a determinate truth-value to it. This is to be
construed in such a way as to exclude those, and I hope, only those
individual statements that result in logical paradox (e.g. this statement
is false). (2) It must be possible to give a logically plausible reason for
assigning one truth-value to the statement and not another. In other
words, the assignment of a truth-value should not be arbitrary. This is
intended to exclude those self-referring statements which no amount of
human ingenuity could represent as informative; e.g. "this statement is
true". This statement is vacuous when interpreted self-referringly
because, although a determinate truth-value can be assigned to it (unlike
the previous example), there is no logically plausible reason why one
should say that it was true rather than false, or false rather than true.
(3) It must make a substantial difference to the statement whether it be
interpreted self-referringly or not. This is designed to exclude those
cases where self-reference is possible but necessarily makes no
difference to the truth or falsity of the statement, e.g. "all the statements
on this page are true".
One is not, of course, logically obliged to treat any
self-referring statement which satisfies these criteria (even if suitably
tightened up) as legitimate. Some of the statements, perhaps most of the
statements, that I characterize as legitimate would have to be excluded
as illegitimate in a formalized language system. For instance, on the
basis of my purely minimal criteria, the statement "this statement is not
a statement about Socrates" is construable as legitimately
self-referring, even though such a statement would be impossible
within a self-consistent formalized language of the type envisaged by
Tarski. 3 I would justify my ascription of a specific truth-value to the
statement in the following way: In order to be about itself and
therefore totally self-referring, the statement "this is not a statement
about Socrates" must function as a meta-statement, and a
meta-statement is a statement about a statement and therefore not a
statement about a physical object. In other words, "this is not a
statement about Socrates" is a statement about a statement and therefore
not a statement about Socrates. It is therefore a true statement and,
indeed, a self-validating one. The same would be true of the statement
"this is a self-referring statement". Such a statement could provide an
informative answer to the question "what is a self-referring statement?"
Conversely, the self-referring statement "this is a statement about

Whewell Self Reference and Meaning

Socrates" is self-refuting and so false, as is the statement "this is not a

self-referring statement". Tarski, on the other hand, insists that a
statement must not be allowed to operate on two different levels at
once, which is what these statements are doing. Thus, from Tarski's
point of view, the statement "this is not a statement about Socrates"
naturally has the form of a second order statement, but in order to have
any content must be logically equivalent to the third-order statement
"the statement that this is not a statement about Socrates is not a
statement about Socrates", but no statement can be allowed to function
simultaneously on different logical levels in this way. However, even if
Tarski's rule were to prove invaluable as an exclusion device for
ridding formal systems of possible antinomies, it may prove to be
unduly restrictive when applied to a natural language. For, if we wish
to rid our own language of actual antimonies we Cflll do so on an ad hoc
and yet non-arbitrary basis as they arise. Why should we wish to
legislate in advance against the mere possibility of their arising by
formulating a strict general rule which would inevitably exclude more
than the antimonies themselves, including perhaps some statements that
we would not wish to exclude? If we are prepared to sacrifice
consistency throughout the language for flexibility within it, we are
free to do so.
A possible objection to this claim is that the sort of flexibility
which allows us to say such things as "this is not a statement about
Socrates" is, to say the least, a dubious asset. But this just happens to be
a rather trivial example. Other less trivial examples could be cited.
Take for instance, the partially self-referring statement "God knows
everything there is to be known". The statement is partially
self-referring because it refers to a whole class of propositions, namely
all possible true propositions, and says of each individual member of
that class that it is known by God. But of course, the statement itself is a
member of that class. In other words, to be omniscient, God must not
only know everything there is to be known, he must also know that he
knows it. If God does not know that he is omniscient, then there is
something he does not know, and so he is not omniscient. In this case at
least, the meta-statement "all true propositions are known by God"
must be interpreted self-referringly even if it cannot be so in a Tarski
A possible objection to this sort of approach to the problem is
that it overemphasizes the distinction between a natural language and a
formalized language. Thus it might be argued that even a natural
language must have rules which are analogous to the formation and
derivation rules of a formal system, and that these rules must be applied
consistently throughout the language. Accordingly, if one wishes to

Whewell Self Reference and Meaning

distinguish between legitimately and illegitimately self-referring

statements one must discover the appropriate rules for so doing.
General guide-lines, such as those I have suggested, are applicable only
on an ad hoc basis and so cannot do the job.
My answer to this is simply that there are no rules which we
apply consistently throughout the language. Any putative rule may be
ignored from time to time or replaced by one that is apparently
inconsistent with it. Why should this matter? You may say: because it
would lead to incoherencies and inconsistencies in our use of language.
But again, why should this matter? After all, a language is not a theory,
which must, of necessity, be internally consistent in order to function as
a theory, but a vehicle of communication, and as long as the rules we
use do not change too drastically from one area to the next, this end
may be achieved. When the incoherence is such that it makes
communication impossible, then it may be disposed of on an ad hoc
basis. Furthermore, although it is true that, in treating some
self-referring statements as legitimate and others as illegitimate, I may
have to make use of a rule that I must sometimes ignore, the way in
which I do so is not entirely arbitrary. For it is governed by certain
minimal criteria or guide-lines which are intended to legislate against
only those self-referring statements to which there are insurmountable
objections; statements in which the incoherence or lack of content is
absolute, i.e. that cannot be disposed of whatever rules we adopt, e.g.
"this is a true statement", but not such statements as "this is a
self-referring statement" or "God knows everything" .
A further argument for disposing of illegitimately
self-referring statements on an ad hoc basis is that they do not, as is
sometimes assumed, form a single homogeneous class. This fact alone
strongly suggests that a general solution of the type favored by formal
logicians to the problem of harmful self-reference is likely to prove
unattainable. According to the most radical of these theories all
self-referring statements are meaningless. This theory was originally
advanced by Bertrand Russell as a direct consequence of his theory of
types.4 Since then, it has been strongly advocated on other grounds by
such philosophers as JlISrgen JlISrgensens and AlfRoss6. This theory has
been strongly attacked by, among others, H.L.A. Hart,7 Karl Popper,s
and Niels Christensen. 9 Thus both Popper and Hart give several
examples of statements which are in a sense self-referring, but at the
same time perfectly intelligible. One of Popper's examples is the
statement "I am now whispering so softly that dear old Socrates cannot
make out what I am saying" when spoken in the presence of Socrates.
One of Hart's examples is the statement "this sentence is written in
English." Since the truth or falsity of both these statements is

Whewell Self Reference and Meaning

determined in a straightforwardly empirical manner, they are clearly

However, as Alf Ross has pointed out in his article on the
subject, if one accepts a slightly modified version of the theory, one is
not bound to regard these last two statements as devoid of meaning.
For, as he claims, it seems a perfectly reasonable hypothesis that "the
vice of self-reference" occurs when one tries to express in a sentence a
meaning that refers to the meaning of the same sentence (e.g. this
statement is false), and that no such defect occurs "in a sentence
expressing a meaning that refers either to the sentence itself as a
grammatical construction or to the speech-act as a sound-sequence".
According to this account, the first example may be regarded as
intelligible because it refers to itself as a sound-sequence, and the
second because it refers to the sentence which bears its meaning and not
to the sentence itself. I am not myself entirely convinced of the validity
of this distinction, partly because I am not sure that I fully understand
it. Where precisely does one draw the line? For instance, on what side
of the line does the statement "this sentence is not being used to assert a
proposition" fall?
Nevertheless, the distinction does seem to have some
plausibility, so I shall let it stand. Those statements on the
harmless-looking side of the line I shall call self-referring expressions,
those on the dangerous-looking side of the line I shall call self-referring
propositions. This distinction, as can be seen from the chart at the end
of this article, occurs in each of the three main types of self-referring
statement. I do not propose to consider them further.
A much more radical modification has been suggested by Hart.
In his opinion, Russell's theory is true only of totally self-referring
statements, and not of statements which are only partially
self-referring. Thus he regards a totally self-referring statement like
"this statement is true" as unintelligible, but a partially self-referring
statement like "all the statements on this page are true" as intelligible.
On an intuitive level Hart's theory may at first seem quite
plausible, because nothing at all seems to be involved in saying that the
first statement is true, whereas one is tempted to say that one knows
perfectly well what it would be like for the second statement to be true.
For if all the other statements on the page were true, the meta-statement
would also be true whether it were interpreted self-referringly or not.
In other words, if we let a, b, c, etc., represent all the other statements,
and if we let S1represent the meta-statement that they are all true, then
S1 must be true, and if Slis true then it follows that all the statements
on the page, including the meta-statement, are true, and this we may
call S2;S2 being the meta-statement interpreted self-referringly. But I

Whewell Self Reference and Meaning

would argue that this partially self-rerring statement is illegitimate.

For the mere fact that SI and S2 are true under precisely the same
conditions means that they cannot be properly distinguished from one
another, and therefore the move from SI to S2 is empty. Hence my
third condition of legitimacy in self-referring statements is not met.
The general point here is that if the truth conditions of a statement are
in no way altered by interpreting it self-referringly, then the statement
is vacuous in so far as it is interpreted self-referringly. Here, on the
other hand, is an example of a non-vacuous partially self-referring
statement, "all the general statements on this page are about finite
classes of physical objects". This is not subject to the previous
objection, for it certainly does make a difference whether the statement
is interpreted as a self-referring one or not. For in order to check its
truth as an other-referring statement we at least have to look at the
other general statements on the page, but as a self-referring statement,
we do not have to go outside it to see that it is false. We can see just
from looking at it that it is not itself a general statement which is about a
class of finite physical objects.
I also disagree with Hart's claim that all totally self-referring
propositions are illegitimate. I have already cited several legitimate
examples, but here is another: "this statement has no meaning". We
know that this statement must be false simply because we understand
what it says. If the statement were an other-referring statement we
would first have to understand what it said and then see if it were true
or not, but in this case to understand what it says just is to see that it is
false. Conversely, the statement "this sentence has a meaning" is true.
It does indeed have a meaning, not because it says it has a meaning, but
because we understand it when it says it has a meaning. Thus although
it is a true totally self-referring statement it is not an illegitimate
self-validating one.
Once the theory that all self-referring propositions are
illegitimate, or that some easily specifiable sub-class of such
propositions is illegitimate, is abandoned, as it must be, then some
means has to be found for distinguishing harmless from harmful types
of self':'reference. My contention is that this task can be performed
perfectly adequately by reference to the three simple criteria I have laid

D. A. Whewell
University of Durham

Whewell Self Reference and Meaning


1. This point has been made many times before, e.g., by J. L.

Mackie in his Truth, Probability, and Paradox, Oxford
University Press, 1973, p. 243.

2. Russell, B., The Autobiography of Bertrand Russell, Geroge

Allen and Unwin, 1967, vol. 1, p. 64.

3. Tarski, Alfred, "The Semantic Conception of Truth and the

Foundations of Semantics", in Herbert Feigl and Wilfrid Sellars,
eds., Readings in Philosophical Analysis, Appleton - Century -
Crofts, 1949, reprinted from Philosophy and Phenomenological
Research, IV (1944) 52;-84.

4. Russell, B. and A. N. Whitehead, Principia Mathematica,

Cambridge University Press, 2nd ed. 1962, vol. 1, esp. pp. 37,

5. J0rgensen, J0rgen, "Some Reflections on Reflexivity", Mind, 63

(1953) 289-300.

6. Ross, Alf, "On Self-Reference and a Puzzle in Constitutional

Law", Mind, 78 (1969) 1-24.

7. Hart, H. L. A., "Self-Referring Laws", in Festskrift Tillagnad

Karl Olivecrona, Stockholm: Kunglia Boktryckeriet, 1964, pp.

8. Popper, Karl, "Self-Reference and Meaning in Ordinary

Language", Mind, 63 (1954) 162-169.

9. Christensen, N. E., Juristen, 1960, pp. 23lff.

Peter Suber
Logical Rudeness
Section 1

Consider the following exchanges:

1. Gerda: So you believe that all belief is the product of custom

and circumstance (or: childhood buffets, class struggle ... ). Isn't that
position self-limiting? Mustn't you see yourself as reflecting only a
single complex of circumstances?
Grobian: Your objection is inapplicable, for it is merely the
product of blind forces. Moreover, your childhood buffets were
pernicious and regrettable, for they have set you against this truth.

2. Gerda: So you believe that all knowledge comes from God in

proportion to our virtue or worth, and that all ignorance, error, and
uncertainty come from the Devil in proportion to our vices. May I ask
what evidence you have for this remarkable thesis?
Grobian: I pity you infinitely for your sins.

3. Gerda: Doctor Grobian, I am not crazy! I stole the bread

because my children were hungry. Why do you assume that every
crime is caused by illness?
Grobian: Why do you deny it?
Gerda: I am not playing a game. I really want an answer to my
Grobian: Obviously your ego cannot cope with the truth and
you display this inadequacy in hostility to your doctor. I will not
recommend your release.

4. Gerda: So you believe x, y, and z. But you are mistaken.

Consider evidence a, b,and c. What do you say?
Grobian: It's a mystery. If I could understand it, I wouldn't
believe it. I can't help it if it's the truth. One day perhaps you'll see the
light too.

In each of these cases something has gone wrong with the

process of debate. In his self-insulating replies Grobian has raised the
ire of more open and more dogged inquirers. We are put off, perhaps
indignant or angry. What's more, we feel justified in taking offense.
Suber Logical Rudeness

We may concede for the sake of argument that Grobian's positions are
strong candidates for truth on their merits, and that he has only devout
good faith to motivate his use and defense of them. Yet we feel that
strength on the merits and good faith do not justify his responses. We
wish he would, like us, concede the strength and good faith of his
opponents, if only for the sake of argument. But must he do this to be
called rational, or merely to be called polite?
Does our sense of justified indignation derive from principles
that we are willing to defend in the open? Or are we merely offended
by seeing 'our side' lose an exchange? Has Grobian committed any sort
of fallacy that might be generalized and generally proscribed? Or does
his offense lie simply in hurting our feelings? Or in his maneuvering to
escape criticism or disagreement? Can we complain if a theory can
evade refutation? Is that a sign of truth, or merely a source of friction?
May we say that a theory that authorizes its proponents to use such
arguments in self-defense is therefore false? Inadequately defended?
Undebatable? If Grobian has violated norms of debate, might it be
because debate is one game and he has chosen to play another?
I will call Grobian's offense "logical rudeness" .. Specifying its
nature will not be as difficult as explaining why it is objectionable and
discovering whether it is unavoidable. I deliberately use the alogical
term "rudeness" to avoid prejudicing the question of its logical status.
Logical rudeness may not be fallacious. But at least it is offensive.
"Rudeness" captures this sense of impropriety. The word derives from
the same root as "erudite", which literally means "not rude" in the
original sense, not rudimentary or rough-hewn. The question of this
essay is whether erudition can always be achieved, or rudeness avoided,
by honest, logical, good faith inquirers for truth. The informality of
the term should not hide the fact that the topic is the ethics of argument.
In the final section I ask what our disdain for rudeness reveals about the
activities we cherish under the names of reasoned inquiry and debate.

Section 2: Preliminary Description. of Rudeness

Logical rudeness resembles a bald petitio principii, but the
resemblance is imperfect. Rude replies presuppose the truth of the
theory being rudely defended, like a petitio. But rudeness is usually a
defensive weapon only. It is a form of self-defense that turns away all
objections, or at least all objections of a certain kind. Unlike a petitio, it
does not purport to justify a conclusion or belie!; it purports to justify
believers in disregarding criticism of their beliefs as if such criticism
were inapplicable, irrelevant, or symptomatic of error. This is not
self-justification in the manner of a petitio, in which assumed premises
Suber Logical Rudeness

can validly imply the disputed conclusion. It is self-justification for the

human proponent of the conclusion, who finds a license, authority, or
justification in his theory itself for refusing to answer objections. Its
success in insulating the believer and the belief of which it is a part
seems independent of the merits or truth-value of the theory. That is
one of the rudest jolts. It strikes us that theories that are false or
implausible could use a rude defense as well as true or plausible
theories. For this reason we suspect that the license to brush off
objections is not a sign of truth or even a supporting argument. It is a
gimmick, a piece of insolence that 'civilized' and 'reasonable' people
will not stoop to use.
A related reflexivity is the self-licensing of debating behavior
by the theory being debated. Rudeness highlights the sense in which
beliefs authorize believers to act in certain ways, solely by virtue of the
the content of the beliefs and the mechanics of good faith and loyalty. If
I believe that fast-talkers are usually liars, then that belief will guide my
responses to a fast-talking critic. But this is merely a psychological or
descriptive observation. Normatively, we tend to want it this way. We
want people to have freedom of inquiry and belief; and when people
come to conclusions we want them to be free (within limits) to act
accordingly. Such a free society is a society of self-licensed actors. If
we respect freedom of conscience in our laws and in our own minds,
then these self-licensed actors are genuinely licensed; what good faith
belief authorizes, we believe, is authorized-at least until it conflicts
with a higher rule. In cases of logical rudeness, belief in certain
theories authorizes believers to be incredibly smug. Is this a price, or
an abuse, of freedom?
If the consequences of a 'bad' belief are intolerable to public
order, we may deal with it through the criminal law, as when we
prohibit polygamous marriages while permitting, indeed protecting,
the freedom of Mormons to advocate the religious obligation to marry
polygamously. But if the consequences of a rude belief are inimical
only to conversation or reasoned persuasion with the believer, then
how shall we deal with it? We cannot revoke or refute the believer's
license to be rude, say, by converting him from his iniquitous faith, for
a barrier of rudeness prevents our arguments from having any effect.
As inquirers we may deal with the rude believer's belief without
dealing with the rude believer; but we admit that this is to abandon a
valuable practice that is valued for its contribution to inquiry -debate.
The most common form of rude theory is that which contains an
explanation of error that fits certain kinds -perhaps all kinds- of
critics and dissenters. The theory is especially rude, but also especially
implausible, if it directly equates error and disagreement (more on this
Suber Logical Rudeness

in Section 4). But it may more plausibly equate error with certain
states of mind or symptoms of belief, when it (not accidentally) happens
that these states characterize the doubters and disbelievers. In the
second example in Section 1 above, which may be called the demon
theory of error, Grobian easily applies his theory of error to Gerda. In
that case it seems that he could as easily have refrained, and offered any
evidence he possessed. But suppose he did offer evidence and it failed
to persuade Gerda (which is the likely result). Then is it as apparent
that he could refrain from his rude explanation of Gerda's failure to
agree? A faithful believer of the demon theory of error must apply it
to Gerda sooner or later, silently or aloud.
A recurring reflexive feature of logical rudeness is the
application of a theory to the context of its own defense. This is
unobjectionable if the theory's subject matter includes truth and
falsehood, validity and invalidity, meaning and nonsense, or other
parameters of debate or demonstration. In this way, rudeness hangs in
the air most around theories about theorizing or meta-theories about
meta-theorizing. But when the application of the theory to the context
of its own defense justifies the theory's proponent in ignoring critics,
then something objectionable has entered the picture. For example, a
certain sort of disciple of Wittgenstein might put forth the theory that
there is no such thing as mind as traditionally conceived, although there
is a word "mind" that is used in certain ways. The theorist might also
claim, more radically, that all questions of existence are meaningless or
reducible to questions of word usage. A critic might begin by asserting
that both of them have minds, and offer reasons or evidence. The
proponent might deflect such criticism by saying, yes, the word "mind"
is properly used as the critic has used it. All further criticism could be
deflected in a similar way. The theorist clearly is applying her theory
to its own proper subject matter, and is striving to preserve her
theory's consistency and her own good faith as a believer in its truth.
Yet these virtues add up to the vice of treating the critic rudely and
disserving inquiry by leaving the critic unanswered.
If a philosopher had a nervous tick that was triggered every time
inquiry threatened to interfere with belief, and if he (not
coincidentally) held the theory that inquiry creates nervous anxiety,
then we could not engage that philosopher on the merits of the anxiety
theory of inquiry without causing him anxiety. This whimsical case is
an easy way to raise a serious question: in the name of cooperative

truth-seeking, can we expect believers to put aside their beliefs or

compromise their loyalty?
Some theories do not obviously apply to the context of their
Suber Logical Rudeness

own debate. Grobian may believe p and add that all error is caused by
the confusion brought about by pain. Gerda may object that pain-free
inquirers may commit errors, and that pained inquirers may speak the
truth. If Grobian is satisfied that Gerda is not suffering physical pain as
she speaks, he will be obliged (by logical courtesy or erudition) to
answer the objection as best he can. Logical rudeness is closed to him
unless he can believe the objection is raised under the duress of pain;
but in that case he is licensed by his beliefs to explain the objection away
rather than answer it. When the theory on the defensive mayor may
not apply to the context of its own debate, further inquiry or bald
presumptions are required before the proponent can defend it rudely.
The point of the examples so far is that rudeness follows from
unobjectionable, even praiseworthy, features of believers and their
beliefs. True as well as false theories, if believed true with good faith,
will be applied to all relevant contexts and will not be compromised to
salve the feelings of dissenters or to serve their ideas of inquiry. Even
if the tenacious good ~aith that leads to this result is not praiseworthy
(explored in Section 5), it might be found in a believer of a true theory.
Because even true theories might be believed in this way, and perhaps
ought to be, we cannot automatically infer falsehood from rudeness.
But if rudeness does not imply falsehood, how do we evaluate
theories that are rudely defended? It seems that they cannot be debated,
at least with their proponents. If we abandon debate and examine such
theories in silence or apart from their proponents, we feel that we have
abandoned a valuable practice, perhaps a practice indispensable to
reliable inquiry. Moreover, we may feel that a negative judgment not
tested in debate with the 'defendants' will be rude in its own way.
Finally, even in the isolated inquiry at our desks we may fail to get
around the theory's rudeness if our method requires us to imagine and
answer the likely responses of the good faith believer. Then we
replicate in drama what we were spared in history.
Rudeness will be possible, as noted, for any theory that
properly applies to virtually any aspect of debate or demonstration,
such as the truth or knowability of theories, the validity of arguments,
the meaning of statements, the sincerity of believers, or the methods of
inquiry. This is disturbing because it shows that most philosophical
theories will be capable of rudeness in this way. And note that this
rudeness is 'legitimate' in the sense that it is permitted by the content of
the theory being defended and the good faith of the believer. It is not
like telling critics to shut up, even though this too is always possible.
More generally as well as more precisely, a theory may be rude
if only it treats any sub-activity of theorizing or debating and identifies
any sort of flaw, fallacy, foible, or fault that could justify a theorist in
Suber Logical Rudeness

dismissing an objection as false, flawed, fallacious, irrelevant, or

inapplicable. Call any such theory a theory of "justified dismissal".
Examples are theories of error, illogic, or nonsense. To explain and
evaluate rudeness we need not reach the question when dismissal is
really justified. If a theory permits dismissal of competing theories
when they are consistent with the writings of Karl Marx, or might lead
to disrespect for law if generally affirmed, or is unintelligible to five
year old children, then that theory can be rude whenever a critic's
contending theory fits the fatal mold. Any attempt to judge the theory
of justified dismissal will be deflected as just another attempt to pierce
the shield of rudeness. Judging the theory of justified dismissal may be
done, of course, but not in debate. If a believer dismisses theories that
are consistent with Marxism, then an objection to that theory will
probably be dismissed as consistent with Marxism. This kind of
self-applicability arises not from praiseworthy good faith and
consistency alone, but also from belief in a theory of justified dismissal.
But holding a theory of justified dismissal also seems harmless.
In fact, in philosophy it is almost obligatory.
Our problem as 'civilized' inquirers is that we want philosophies
complete enough to explain error, illogic, nonsense, and other grounds
of justified dismissal; we expect believers to apply their beliefs with
consistency and good faith to all the applicable contexts of life; and yet
we do not want them to apply their grounds of justified dismissal to the
critics and dissenters in the realm of debate who help us decide the
theory's truth. Are we asking too much? Are we demanding
inconsistent tasks of our opponents? Is debate a privileged process in
which beliefs can be examined without the distortions introduced by
believing, or (from the believer's standpoint) is it a damnable realm in
which one is expected to give up one's faith to defend it?
(Note that I use "belief' and'tfaith" in a weak sense. Any claims
to truth will be called "beliefs" or "faith", even if the proponent also
considers them to constitute "knowledge".)
A theory may explain away the criticism or disagreement of
critics descriptively or normatively. The first example in Section 1
above is descriptive, the second normative. If the critic's disagreement
is put down to an unfortunate series of childhood buffets or to any other
source independent of the merits or truth-value of the theory he
criticized, then he is rudely treated. He is not answered, but reduced to
ineffectual squealing from the standpoint of the proponent. Once
stigmatized as suffering from the defect ascribed to him, a defect
well..explained by the theory, the critic is put out of court. The well of
discourse is poisoned. Nothing he says afterward can affect the theory,
at least in the judgment of the proponent. If the critic's disagreement is
Suber Logical Rudeness

put down to vice, sin, or a normative weakness, then he is equally not

answered and relegated to limbo -a limbo either of well-explained
incompetency or of well-explained ineligibility for our attention and
answers. Descriptive rudeness imputes a foible, prescriptive rudeness a
fault, to critics or dissenters.
The authority to be rude consists in heeding the terms of the
theory that describe the foible or fault and who deserves to be branded
with it. The terms of the theory may be false or implausible, but it is
futile to hope to persuade the rude proponent that that is so when our
attempts only feed self-righteousness.
Rudeness of this type makes debate much like an unnamed
childhood game I recall with pleasure and frustration. One player asks
yes-or-no questions, and the other answers "yes" or "no" according to a
secret algorithm. The object of the game is to guess the algorithm. It
might be, "answer 'yes' whenever the question begins with a vowel or
ends with a two-syllable word; otherwise answer 'no'." (One must
always answer "yes" and capitulate when the correct algorithm is
proposed.) In such a game the words "yes" and "no" are not used with
their ordinary meanings. Hence the questioner will be seriously misled
if she asks, "does the algorithm concern syllabification?" and takes the
"yes" or "no" answer in its ordinary sense. In the game, which I will
call "Noyes" for convenience (for the pun on "no-yes" and the
homonym of "noise"), "yes" and "no" are tokens of exchange, not signs
of affirmation and negation. The questioner cannot begin to play
meta-Noyes by asking, "seriously, is syllabification involved?" The
questioner cannot get traditional "yes" and "no" answers as long as the
'oracle' maintains his role and plays the game. The analogy to logical
rudeness is that the critic cannot get the believer to give up his good .
faith for the purposes of debate, and perhaps should not want to. It is
equivalent to asking the Noyes oracle to give up his algorithm for the
sake of play. Because the believer is ruled by his beliefs in selecting
responses in debate, as the Noyes oracle is ruled by his algorithm, the
questioner is apt to find her questions and objections translated from
the genre of criticism to the genre of noise, and dealt with as input to an
unknown algorithm. The difference of course is that Noyes is plainly a
game, and the refusal of the oracle to play meta-Noyes is part of his
role in playing Noyes. Is debate equally a game, and are some believers
equally bound to refuse to play meta-debate?
Noyes makes play out of what can be a serious problem.
Consider the case of a rapist who believes that "no" means "yes" and
that struggle indicates pleasure. Recent law in England has allowed
rape defendants to argue good faith (that is, sincere) belief in the no-yes
equation, and a few 'rapists' have won acquittal with that defense. The
Suber Logical Rudeness

effect is to equate a woman's consent with a man's belief in a woman's

consent. The result is nothing short of evil in practice, though it rests
on the slender theoretical reed that people are ruled, not by what is real,
but by their belief about what is real. This is one case in which the
'authority' one receives from good faith belief leads to intolerable
consequences and should be barred by the criminal law. The
Anglo-American criminal law occasionally (but rarely) excuses
conduct or mitigates punishment for crimes performed in good faith
error of the facts. But to prevent 'good faith rape' and similar abuses,
usually an objectivity requirement is added that the belief be
'reasonable'. The peculiarity of the English law is that good faith
belief, no matter how unreasonable under the circumstances, suffices to
acquit. (This astonishing doctrine was first asserted in Director of
Public Prosecutions v.Morgan et aI., 61 Crim. App. Reports 136
(1975).) .
Some political regimes may be Noyes games writ large. Suppose
one is in a despotic state where the officials act according to rules that
ordinary citizens are not allowed to know or to criticize. These
meta-rules about criticism are sometimes enforced against critics with
imprisonment and other forms of violence, but for most people most of
the time they are enforced by social pressure. If one engages one's
neighbor in conversation on the wisdom of such policies, one will be
surprised that one's very desire to examine the wisdom of the policies is
considered suspect and criminal. If the topic of conversation shifts (it is
not much of a shift) to the desirability of open discussion of every
question, one will be more surprised to hear one diagnosed as
'bourgeois' or 'reactionary' or (from the other end of the ideological
spectrum) as 'anarchical'.
One may be aware of theories of government according to
which free discussion is inimical to good order, revolutionary
initiatives, or reeducation; but one would at least like to debate the
merits of such theories of government. The loyal proponents of such
positions, however, like most loyal proponents, apply their beliefs to
the context of their debate, as they apply their beliefs to all the contexts
of history. From their own point of view this is only good faith and
consistency. One cannot get such proponents to 'jump out of the
system' for the time and labor of a joint inquiry into the merits of their
beliefs; and one should not expect to be able to. Much like the
questioner in a game of Noyes or the victim of a rapist who believes
that "no" means "yes", one's criticism of a rude state policy will be
interpreted in that state as something other than a criticism to be
answered as criticism. In this case it will be interpreted as a violation,
and one's attempt to reach a meta-level at which one could discuss the
Suber Logical Rudeness

propriety of such an interpretation will be interpreted as another

violation. Like the critic of the demon theory of error, or the hapless
victim of the tarbaby, one's struggles to escape the verdict of one's
opponent only confirm his confidence in one's miserable fate.
The rude regime raises important issues of political theory,
particularly the question whether commitments to principles or results
should supersede commitments to method or process. This and related
issues of 'procedural' democracy will be explored to some extent in
Section 5.
The Noyes regime and rapist suggest a closely related species of
rudeness: the tactic of the proponent in disregarding the logical or
illative dimension of the critic's words and treating them solely as
behavior to be explained by his theory. The same effect is achieved
when criticism is interpreted as a symptom of historical, economic, or
psychological forces, or as ideology. In many ways this is merely a
different perspective on the same species of rudeness considered above.
If the proponent's theory contains an explanation of behavior (which
we also expect a good philosophy or social science to have), then the
critic may find herself unable to escape the object-language of the
theory she is attacking and reach its meta-language. All criticism and
disagreement may be seen as behavior, and to that extent fall into the
arena of the subject-matter of the theory. Like birdsong or ritualistic
dancing, they are colorful bits of the explanandum, logically
subordinate to the explanation and incapable of refuting it except as
counter-examples or anomalies.
The difference between disagreement as behavior to be
explained and as criticism to be answered is at least partly a matter of
perspective within the discretion of the proponent. Again we encounter
the question whether his choice is ever fixed by the content of the
beliefs he is defending and his general commitments to consistency and
good faith. And again, we are as reluctant to close off any option by
normative force. Just as explanations of error are desirable, so are
explanations of behavior. Even behavior with a logical or illative
dimension is worth studying merely as behavior to such disciplines as
anthropology, the sociology of knowledge, psycho-history, and the
descriptive parts of comparative jurisprudence. But we want to
discourage the sort of rudeness that studies critics as specimens to the
exclusion of (rather than in addition to) hearing their criticism.
Religious belief has been studied as a psychological condition
and social phenomenon. Some schools of linguistics study 'verbal
behavior'. There is no epistemological or scientific reason why a social
science could not study 'argumentative or critical behavior'. The
theories of such a social science would be fraught with great potential,
Suber Logical Rudeness

from birth, to license their proponents to treat their critics rudely.

Such a science might use the term "refutationary behavior" to refer to
arguments, refutations, criticisms, and polemics intended to
demonstrate falsehood. Refutationary behavior is fascinating. People
thrust and parry, advance and retreat, concede small points and lay
traps on large ones, take disagreement personally, get angry, resort to
ad hominem attacks, decoy the opponent with false camaraderie or
uncertainty, sting in the heel with irony, trip up with sophisms and
paradoxes, fall back on definitions, and refuse to fall back on
definitions. In our large universe, any theory of refutationary
behavior, like theories of other kinds, will encounter disagreement. If
a sociologist of polemics proposes that refutationary behavior is
motivated by class interests, then a critic may be as erudite as can be,
but the proponent can study the proffered criticism as another example
of refutationary behavior, perhaps as one that confirms the theory.
Rudeness that views arguments only as a special class of
behavior for empirical study highlights a feature of all rudeness, which
is that the rude believer is not summoned or elicited to be rude until
criticism is expounded or uttered or made into behavior. A theory may
be refuted in abstracto, in silence, in thought, in ideality, or in private
at one's desk, but this kind of refutation does not put the rude proponent
on the defensive or calion him to use his rude defenses.
The necessity of expounded criticism to trigger logical rudeness
in tum highlights another feature of all rudeness, which is that the
theory may 'really' be refuted while the proponent is 'justifiably'
unconverted. Rudeness insulates believers, not beliefs. Rudeness
suggests the presence of logical perspective: even sound refutations,
those that might work at one's desk or in the journals, might fail to
convert the proponent, and the proponent may have a 'sufficient'
warrant from his theory for his intractability. If good faith belief in a
theory suffices to warrant the believer to act under its terms (a
political, not a logical, principle), then the believer is 'really' justified
in disregarding the sound refutation. RudenesS drives a wedge in
between logical argument and rhetorical persuasion, preventing the
power of the former from aiding the power of the latter. The rude,
insulated believer need not be illogical to be protected by the mantle of
rudeness; he must believe a theory of a certain kind, with the sort of
good faith devotion that seeks to preserve the theory's consistency and
to apply it to all explananda within its domain. This is also disturbing,
for it suggests that generally praisworthy traits of inquirers may make
argumentation, on its logical side (as opposed to its personal or political
side) nugatory.
We might be tempted to say that it is always rude to interpret
Suber Logical Rudeness

criticism as unwitting confirmation of one's theory. A good example is

the theory that the subtlest, and therefore most likely, action of the
devil would be to deny his own existence and cause others to deny it.
Opponents who doubt the existence of devils are hopelessly trapped; no
objection can fail to confirm the believer in his belief. When this tactic
is rude, it is like the empirical study of refutationary behavior in
refusing to see a meta-level in the critic's criticism.
We should be careful here, however. Some criticism does
confirm the theory being criticized, in which case a response by
retortion is appropriate. Critics may resent this sort of intellectual
judo, but we may not call it logically rude unless the critic is deprived
of a response on the merits, or cannot have his criticism taken as
criticism, although perhaps it is also taken as symptom, behavior, or
confirming instance.
Suppose a disciple of David Hume adapted Grobian's buffet
theory of belief (in Section 1), and claimed that all belief was based on
local custom and habit. This theory might have met comparatively
warm approval in late eighteenth century Britain. But
contemporaneous Germans would have denied it in unison. The
Humean could interpret the German choir as simple corroboration:
their consensus and their Teutonism would explain one another. Like
the student of refutationary behavior, such a Humean would be guilty
of little more than applying her theory to its subject matter, which
happens to include the context of its own debate. And that, by itself, is
not blameworthy. But in each case we feel that such application is
hasty. Before the critic is used against himself, he should be told why
he is wrong. But while the student of refutationary behavior is clearly
failing to explain the errors of his critics, the Humean is not. The
former merely says, "That's about what I'd expect from a middle-class
white male," while the Humean has found a putative cause of the
opponents' error in Germanic national character.
Rudeness that twists objections into confirmations highlights a
feature of all rudeness, which is that the proponent of a theory must
struggle to avoid perceiving criticism as applicable to him or his
theories, qua criticism. The proponent must see criticism as false,
non-cognitive, meaningless, irrelevant, unwitting confirmation,
undebatable, unknowable, self-contradictory, or generally
inapplicable, ripe for justified dismissal.
Both the proponent of the class theory of refutationary
behavior and the proponent of the custom theory of belief have traced
the beliefs of their opponents to their supposed sources. The difference
is that the proponent of the class theory of refutationary behavior does
not (necessarily) believe that such a genealogy is equivalent to a
Suber Logical Rudeness

refutation, while the Humean does. The former is constantly, even

professionally, tracing refutationary behavior to its source. One may
pursue such a course and still believe that the truth-value of ideas is not
affected by their origin. No empirical study is per se guilty of the
genetic fallacy. But the Humean relativizes any belief that she succeeds
in tracing to its source; if the belief is not already self-consciously
relativistic (as eighteenth century German philosophy typically was
not), then it is subjected to a supposed refutation. A rude slap has been
added to the initial reductionism.
But is not the Humean's own claim about custom relativized by
itself? The Humean may evade this consequence by making the custom
theory of belief an exception to its own tenets; the exception may be
hard to justify, but at least to claim it avoids paradox. lnitiallyshe
would resemble Arcesilas, Carneades, and the other skeptics of the New
Academy who claimed that all was uncertain. They were urged by
Antipater to make an exception for their very claim that all (else) was
uncertain; but in fear of implausibility or in pursuit of mischief they
This paradox and its avoidance raise an important point. Some
kinds of rudeness are fallacious, and the inference of falsehood or
inconsistency is justified. For example, the verificationist theory of
meaning is meaningless by its own criterion. However, any objection
along these lines is also meaningless by that criterion. Hence, the
proponent of the theory may seem able to sit smugly on his criterion
and refuse to allow any objection to enter his realm of debate. But that
would commit a fallacy. The weapon raised by the verification theorist
to slay his opponent slays himself. This is not always so with rude
defenses, but it is so here and for the Humean proponent of the custom
theory of belief, as well as for Grobian's buffet theory of belief in
Section 1. The verificationist apparently has two choices in the face of
the charge of self-referential inconsistency: he may make his theory an
exception to its own tenets, which would be odd and implausible but
consistent, or he may try to fend off the objection by classifying it
meaningless ab initio, which his theory apparently entitles him to do.
But the latter choice is not really open, or it does not really preserve the
theory's consistency in the face of the objection. If the theory is not
excepted from its own standards, then it must suffer the very fate
contemplated for the opponent.
We may generalize. Normally one may not infer falsehood
from rudeness. But one may do so with rude theories whose grounds of
justified dismissal properly apply to the theories themselves. One may
at least infer the presence of a fallacious defense, beyond a merely rude
one, and the presence of self-referential inconsistency.
Suber Logical Rudeness

The proponent of the custom theory of belief is rude; if she does

not make her theory an exception to itself, she will be fallaciously rude.
Her condition should be distinguished from that of another kind of
debater who likes to trace criticism to its source. If a religious
fundamentalist objects to the theory of evolution, a biologist may say,
"Ah, that is because he believes in the account in Genesis, and takes it
literally." This would be rude only if the imputation of the cause of the
objection is considered an elliptical refutation, shorthand for the claim
"that is false because it derives from a system of superstition long
disproved." But it need not be rude in this sense; it may be-shorthand a
more complex evasion. The biologist. may believe that the origin of
ideas is irrelevant to their truth-value; she is not rude if her statement is
merely an elliptical way of postponing or deferring an answer on the
Discovering that an objection to one's theories originated in a
religious belief, or from any source other than the objectionable
character of one's theories, is not a refutation; if it is not used as a
refutation, then it is not rude to point out the discovery. For example,
objections to certain theories of astronomy from astrology are often
tossed aside because of their origin. This mayor may not be rude. It is
not rude if the astronomer is saying, "Astrology has been answered
before; if I don't take this astrologer seriously it is only because the
reasons are shared by all the members of my profession, and even if
those reasons are inadequate, obsolete, or subject to the criticism before
me now, they can go without saying."
To subsume an objection under the larger faith that gave rise to
it, however accurately, does not help a bit in answering or disarming
the objection. It is pure postponement. It serves communication, not
refutation. In context it usually informs all interested parties of one's
position, and even the source of one's counter-evidence and
counter-arguments. But it does not actually answer the criticism or
refute the body of beliefs that gave rise to it. Even when it is shorthand
for a definitive refutation, it does not recapitulate the reasons against
the position, but only alludes to them, and only indirectly, by alluding
to the faith that is presumed to be long refuted. Logical courtesy
(erudition) demands that the objection be answered on its merits,
although no logics themselves demand it. To allude to a supposed
definitive refutation without restating it is on the face of it nothing
more than a weak display of disagreement. But to subsume a belief
under a larger system as if that constituted refutation begs the question,
and worse. It is like any other reductio ad absurdum in which; the
absurdum is not a contradiction but simply unacceptable or unheard of.
One is not acting with the courage of conviction, that believes truth is
Suber Logical Rudeness

demonstrable, but only with the complacency of conviction, that

believes dissenters are pitifully benighted.
This discussion brings us back to the beginning. For a theory of
justified dismissal may focus on a fault or foible of the believer, or on
the body of beliefs that gave rise to the objectionable theory. Both can
be rude; but the second can also be mere postponement. Both involve
the explanation of the objection. If we explain the criticism of critics in
a way that justifies us in dismissing it, according to our theory of
justified dismissal, then we have treated the critic rudely. But if we
explain the objection as originating in a possible flaw in our own
theories, then we are as polite as can be. We are then granting 'for the
sake of argument' that our beliefs might be objectionable or false.
Another type of rudeness arises when a proponent feels
authorized in holding a theory independent of the authority that comes
from correctness. Many government officials are guilty of this kind of
rudeness, and seem to believe that their ideas are sufficiently authorized
by the election results and thereafter need not be defended or debated.
When critics or reporters ask why a course of action was not taken
(requesting a reason), many officials will answer, "We decided it would
not be appropriate at this time." This could be translated as, "I don't
have to explain or defend myself as long as the people let me stay in
office." Grobian's fourth response in Section 1 is of this type: he felt
authorized in his faith, not by sharable evidence and reasons, but by a
private inner light.
There are certainly many other kinds of logical rudeness. I do
not mean to give an exhaustive taxonomy. One final type, similar to the
government official's, may be mentioned. Suppose someone believes
that (1) ESP exists, (2) only some people possess it, (3) it may be
acquired but that doubt is an obstacle to its acquisition, and (4) it cannot
be displayed in the presence of hostile or unbelieving witnesses. This
theory is rude in two novel ways. First, it is unfalsifiable. All negative
results from experiments may be answered with the all-purpose
subterfuge, "The researchers must have doubted." Any unfalsifiable
theory may be called rude in a weak or attenuated sense. Critics are
teased, because they may disagree all they want, but no applicable or
decisive refutation may be found. For ordinary empirical theories,
amassing contrary evidence is never a conclusive refutation, but at least
the strength of a negative inference mounts; amassing contrary
evidence to such an ESP theory would not even strengthen a negative
inference in the judgment of the proponent.
A stronger sense of rudeness derives from the first. A critic who
denies that ESP exists can be told, "I guess you just don't have it" This
reply makes the ESP theory a case of a more general type. Max
Suber Logical Rudeness

Scheler's theory of value and value-blindness is another case. Probably

the most infuriating case may be called the blessing theory of truth
-the theory that knowledge is a gift from a god, that only some receive
it, and that those receiving it know it when they see it by unmistakable
internal signs. I suppose it is optional for a proponent of a blessing
theory of truth to claim that the blessing theory itself is knowable only
as part of such a gift.
The general feature shared by rude theories of this type is the
belief that some valued capacity, relevant to truth-seeking or knowing,
is either present or absent in one, and that possessors know they are
possessors and nonpossessors do not (or sometimes cannot) know that
the race divides into possessors and nonpossessors. This general type of
theory takes two equally rude forms: (1) the "born loser" theories,
according to which nonpossessors of the gift are doomed to remain
nonpossessors, and therefore ignorant, and (2) the "one path" or "trust
me" theories, according to which nonpossessors may become
possessors only by following a regimen set for them by self-proclaimed
possessors. The regimen may include a code of conduct as well as of
faith, all of which must be taken on faith or without evidence in the
beginning. Proof comes only to those who take the path to the end. A
cross between the born-loser and the one-path theories may hold that
the gift falls on possessors gratuitously.
The general type may be called "boon theories". We are all
familiar with boon theories of knowledge, wisdom, virtue, and
salvation. The first ESP example was a one-path boon theory. Max
Scheler's view that some people 'see' values rightly and others are
value-blind is a one-path boon theory. A social Darwinist theory that
held that males and whites deserve their privileged positions simply
because they have acquired them is a born-loser boon theory. Note that
in boon theories in which the boon is not gratuitous, nonpossession is a
stigma. Hence the critic is not only excluded from grace and ignorant,
but is blameworthy. The smugness of rude proponents and the rude
immunity to conversion are thereby justified all the more.

Section 3: What Sort of Delict is Logical Rudeness?

Let me summarize the species of rudeness sketched in Section 2.

The primary type is probably the application of a theory of justified
dismissal, such as a theory of error or insanity, to critics and dissenters.
Another major type is the interpretation of criticism as behavior to be
explained rather than answered. This is closely connected to the type
that refuses to see a meta-level in the critic's criticism, and will not
allow critics to escape the object-language of the theory. A rude theory
Suber Logical Rudeness

may reinterpret criticism as a special kind of noise, or as unwitting

corroboration. A theory may evade criticism without rudeness by
postponing an answer or referring the critic to the answer of another.
The abuse of postponement may be rude, however, as when the motions
of postponement are made shorthand for dismissal, or when the
subsumption of an objection under a targer system of belief is made
shorthand for refutation. A rude theory may be held for reasons other
than its correctness, such as the support for the believer shown by
voters or grant-giving agencies. A weak sort of rudeness lies in any
unfalsifiable theory, and a strong sort lies in boon theories that identify
critics as nonpossessors of a special boon. The theories of justified
dismissal and the boon theories tell critics that they are disqualified
from knowing truth or even deserving answers because of some
well-explained foible or fault in themslves. All the types have in
common an evasion of a responsibility to answer criticism on the
merits, when that evasion is authorized by the theory criticized. All
types are triggered only by expounded criticism, and only insulate the
proponent from conversion or capitulation, not the theory from
Only one type was found fallacious, the dismissal of an objection
on grounds that would suffice to dismiss the theory itself. Such
dismissal is self-referentially inconsistent unless the theory were made
an exception to its own tenets, a move that usually cures inconsistency at
the price of implausibility. The kinds of rudeness seen here may
apparently be used with true beliefs as well as false, unless one is
already a partisan of theories that would make any rude theory false. If
we admit the adaptibility of rudeness to true and false theories, then we
must find another avenue of complaint. What is wrong with it?
The only obvious delict of non-fallacious rude defenses is that
they separate the believer from the belief in such a way that the belief
may be criticized or refuted and the believer left smug and unswayed.
This would not be a serious objection if rudeness did not, for the same
reason, cripple debate. A rude defense terminates all debate with the
rude theorist. Critics see that they can make no progress against rude
believers, and tum to fellow travelers and the journals. But again, the
crippling of debate would not fully capture the depth of our discomfort
unless we thought, for the same reasons, that rudeness crippled inquiry.
Does rudeness cripple inquiry? Does the crippling of debate
cripple inquiry? Is rudeness an epistemic sin or just plain impolite?
With these questions in the background I would like to start off
on an apparent digression with the aim of returning to them shortly.
Rudeness insulates the believer from expounded criticism. The rude
believer need not answer criticism, but may deflect or explain it away.
Suber Logical Rudeness

In legal tenns, the rude believer's refusal to answer his opponent is a

refusal to recognize a burden of going forward created by the critic's
Anglo-American law distinguishes the burden ofproof from the
burden of going forward. The burden of proof is a tie-breaker rule;
when the evidence and arguments on each side seem balanced, then the
party with the burden of proof loses. The burden of going forward is
the obligation to respond after the opponent has made a preliminary
case. When a philosophical inquirer puts forth a theory, and when
critics publish their disagreement along with erudite arsenals of
evidence and arguments, then can we say that the 'burden of going
forward' has 'shifted' to the theorist? Do those who publish theories, in
print or orally, have a duty to respond to critics who make a minimally
plausible case that they are wrong? What we have called rude defenses
seem reducible to different ways of shirking a supposed burden of
going forward. Is there such a burden in philosophy?
We should remember that the use of burdens in law furthers
certain policies. When one party in court has made a case for herself,
the judge turns to the other, in effect, and says, "Your tum! I have to
decide this case and cannot wait forever. I want to be fair. Speak now
or forever hold your peace." This boils down to, "Your tum or you
lose!" Parties that fail to meet their burden, either of proof or of going
forward, will nonnally lose the case, either by judgment or by default.
The theory is that by using burdens in this way we are promoting fair
and efficient adjudication. First, judges must decide the cases before
them. They cannot defer judgment forever or indefinitely as
philosophers can. Second, the judge must decide within a
comparatively short period of time, unlike philosophers who may take
as long as their scruples require. Third, the judge may (and usually
does) have to decide on imperfect infonnation, when some facts are
missing or contested or both. Fourth, the judge wants her judgment to
be infonned by the merits of each side as they are perceived by each
side. All these policies are served by compelling one party to speak or
suffer default when the other has spoken.
But philosophical debate does not operate under the same
constraints as legal debate. Nobody has to decide philosophical
questions at all, let alone soon or on imperfect infonnation. At least the
sense in which people 'must' answer philosophical questions (such as,
when pregnant, the morality of abortion, or when tenninally ill, the
morality of suicide) does not give rise to prudential, procedural rules
for allocating burdens of proof and going forward in the same way as
in law. Moreover, there is no adversarial process in the same sense.
Hence, there appears to be no comparable reason why philosophers
Suber Logical Rudeness

must speak up after their opponents have made a preliminary or even a

formidable case against them.
Is this equivalent to saying that there is no logical reason why we
must answer our critics? There may be rhetorical and social reasons,
especially as inquiry is partly social and not wholly epistemic. We do
not exclusively strive for true knowledge in inquiry, but also for social
integration, the cooperation of different inquirers, the communication
and application of results, the preservation of a milieu in which inquiry
is free and fruitful, and the satisfaction of the human purposes in having
knowledge or ideas at all. Logical rudeness is certainly not prohibited
by logic; it is prohibited, I maintain, only by social norms. It is
objectionable, but not in the manner of illogic or hypocrisy; it is
objectionable more in the manner of refusing to speak to one's spouse,
putting urgent callers 'on hold', or meeting student questions with
sardonic laughter.
Philosophers have no equivalent of default except the
presumption that the silent or rude theorist has no answer on the merits
to offer, and (qua individual proponent) may be presumed ignorant or
incorrect and dismissed. This presumption, however, is very legalistic,
and in many cases will be false. The limits of the applicability of legal
procedures to philosophical argument may lead us to rethink this
presumption. At the moment, however, the presumption looks like a
theory of justified dismissal: theorists that resort to rude defenses may
be dismissed; their theories may be true, but we must await another
proponent to find out how that position responds to certain questions
and objections before we can judge it fairly on the merits.
Courteous or erudite philosophers tend to use the concept of
burden. Indeed, the concept of a burden of going forward is an element
of the positive system of logical etiquette that defines rudeness. It is not
a part of logic itself, but part of the practical implementation of logical
courtesy and social norms in debate. It furthers social policies and
inquiry, but its absence would also serve inquiry, though to a different
degree. The truth-value of a rude theory is not affected by the silence
or rudeness of its proponents in the face of disagreement. In short,
philosophical inquiry may be crippled by logical rudeness, but the
legalistic remedy of a burden of going forward would cripple
philosophical inquiry even more. Rudeness cripples inquiry by
obstructing cooperation, not by silencing contenders for truth or by
deceiving inquirers. Rudeness, like a boulder in a stream, makes
inquiry pass around it. If inquiry proceeds without debate, something
is 'lost. But because falsehood cannot be inferred from rudeness, much
more would be lost if we dismissed rude proponents, as if in error, for
violating some imported rules of procedure. Legal inquiry is
Suber Logical Rudeness

successful when it is both fair and probative. Philosophical inquiry

may be successful if it is only probative, that is, if it only brings us
closer to truth. Respect for the parties is secondary; to put it higher is
to put persons on a par with truth, which may be proper for every
purpose except inquiry for truth.

Section 4: Must Some Theories Be Rude?

It may seem that the imputation of a foible or fault to a critic
simply qua critic is always optional, never necessary to preserve the
consistency of the theory or the good faith of the proponent. But this is
not true. First, there is the case of the brazen theory that includes as a
tenet the forthright equation of disagreement and error. This tenet is
not as rare, nor probably as naive, as one might at first suspect. It may
be called (using legal jargon) the "exclusivity clause" of the theory.
Any theory may have an exclusivity clause, and most theories may
have them without contradicting their own content. The 'clause'
merely states that the set of tenets comprising the theory is the truth and
the only truth on its precise subject. It does not imply completeness; but
it does imply that propositions inconsistent with the theory are false. It
may be tacit and understood, and indeed it does seem to follow from the
mere claim of truth according to the principle of excluded middle (tacit
in many theories) and most classical notions of truth. If a theory
contains an exclusivity clause, even a tacit one, it impels the good faith
proponent to equate disagreement and error. Critics may courteously
be indulged in the realm of debate, and cajoled into seeing the light, if
possible, but that would be supererogatory under the canons of logic
and good faith. One premise of 'civilized' debate -that any contender
might be speaking the truth and debate is one way to tell whom- is not
shared by all the contenders. For this reason it is disturbing to note that
almost any claim to truth may bear a tacit exclusivity clause.
Even more disturbing is the case of philosophical systems. The
paradigm of good philosophy for several western traditions -the
complete, consistent system- is impelled to be rude. This is not news
to Kierkegaard, who felt rudely subsumed by Hege1's system, and was
told by contemporary Hegelians that he was logically incapable of
attaining a perspective outside the system sufficient to attack it.
If the system is supposed to be complete as well as true, then the
good faith proponent must believe the critic in error, and therefore
must apply the system's explanation of error to her. Note that mere
belief in the completeness and truth of the system suffices here to justify
the conclusion that disagreement is error. The good faith proponent
need not immediately act on this belief in the critic's error, but neither
Suber Logical Rudeness

can he escape concluding it, any more than he could willingly suspend
judgment on the truth of his beliefs. Proponents of what are supposed
to be true, complete, consistent systems must choose between apostasy
and rudeness. They must defend their beliefs either by appeal to
premises and principles from outside the system, which they believe are
false, or by appeal to premises and principles from within the system,
which is question-begging and liable to be very rude. This may be
called the dilemma of systematic self-defense. To ask such a believer to
be logically polite 'just for the sake of argument' is equivalent to asking
him to give up some tenets of the faith he wishes to defend just to enter a
realm of debate to defend it. This is why systems with pretentions to
completeness have traditionally seemed rude, have traditionally
authorized rude defenses in their proponents, or have gone undefended
at fundamental levels.
It is this feature in political systems that allows the equation of
dissent and mental illness, dissent and crime, and dissent and error, and
the feature that led modern philosophers like Kierkegaard and
Nietzsche to abjure the pursuit of philosophical systems per se.
There may be more than rudeness to tum one from systems, but
one should note that rudeness should not suffice, for falsehood cannot
be inferred from mere rudeness. On the other hand, if systems are still
attractive, this analysis indicates at least that the logic of defending
systems is peculiar, and that if we still cherish both the pursuit of
systems and the classical forms of debate, then we will have to forgive
some question-begging and rudeness. Moreover, if this is so, we should
expect a true system to take these peculiarities into account and present
a logic in which some circular arguments and rude defenses are
permissible. Hegel's system fulfills this expectation more than others,
and perhaps the reason is that it is more self-conscious of the logic of
systematicity than others.

Section 5: What is Debate?

Logical rudeness may be considered a complex form of ad
hominem argument. It tells critics and dissenters that they are defective
human beings whose ignorance or error is well explained as frailty,
fault, foible, or the absence of a boon. Moreover, this form of ad
hominem is justified by the theory under attack. When our questions
are answered by ad hominem assaults, we are angered. Our anger
cannot be reduced to hurt feelings because we were not merely
wounded in our dignity; we were put off in our inquiries for truth by a
refusal to cooperate. A rude response can therefore trigger three levels
of indignation: personal affront, thwarted cooperation, and crippled
Suber Logical Rudeness

inquiry. The first is personal, the second social and political, and the
third epistemic.
Rudeness thwarts cooperation, which in tum thwarts inquiry, at
least under some concepts of inquiry. Rudeness prevents inquiry from
being optimally fruitful. But logic does not tell us to make inquiry
optimally fruitful; human interests do. Rudeness therefore is not so
much a fallacy as a violation of human community. The rub is that we
want to permit all possible truths to be propounded and debated: some
of the candidate-truths will deny any role to cooperation in inquiry and
others will license rude defenses. Opening the realm of debate this
much will therefore permit logical rudeness to enter, which in tum will
make inquiry sub-optimal, at least under some concepts of inquiry.
The tensions within the concepts of debate and inquiry between
openness and fruitfulness can be seen from a wider perspective. The
epistemic principle violated by rudeness is not merely that inquiry must
go on. If we are told, in effect, that we do not deserve to be answered
on the merits, or are disqualified from knowing truth, on account of a
foible or fault in ourselves, then we are being excluded from the
universe of possible knowers in which we thought we had enlisted by
inquiring and debating. If the truth is not (yet) known, but is subject to
inquiry and debate, then we cannot (yet) exclude any person from the
universe of possible knowers. That is, we cannot do so a priori,
although once we know truth we may be able to do so a posteriori
-when we learn, for example, about color-blindnessand the diversity
of mental illness.
Logical rudeness violates what might be called the principle of
epistemic democracy: the principle that all persons have an equal
entitlement to know the truth. This might well be reclassified as a norm
of logical etiquette, and denied the name of an epistemic principle, for
it is a mere presumption. If we stated it more completely, it would say:
all persons should be presumed to have an equal entitlement to know the
truth, until and unless we discover some truth to the contrary. As long
as we are confessedly ignorant, it is a methodological assumption that
results in fair and courteous treatment to our co-workers.
The problem is that the rude proponent believes he does possess
some true knowledge that justifies the cancellation of the presumption.
His rudeness from this angle derives equally from (1) the content of his
belief, that it disqualifies some people from knowledge, people who
tum out to be his critics, and (2) his unshakeable faith that he is right to
believe it. The latter dimension will be explored more fully toward the
end of this section. First I would like to examine the former dimension.
The principle of epistemic democracy is normative, not
descriptive. As long as we are confessedly ignorant, we just do not
Suber Logical Rudeness

know whether all of us have equal right to the truth. We think we ought
to act as if our entitlements were equal, because that is a demand of
fairness or courtesy. The rude proponent who denies this principle by
his ad hominem methods, therefore, seems to us to deny an important
normative rule; he is not just rude, then, but also unfair.
The principle of epistemic democracy conflicts with another
principle that we hold dear: it might be called the 'no holds barred'
principle of debate. It states that philosophers can (should be permitted
to) ask any question, propose any answer as true, challenge any theory
as false or unproved, make any argument, and generally debate any
theory on the merits. The conflict between' the no-holds-barred
principle and the principle of epistemic democracy is simply that,
under the former, the latter (like any other principle) may be
challenged and denied. The no-holds-barred principle conflicts with
itself in the same way that it conflicts with the principle of epistemic
democracy: under its terms, it may itself be challenged and denied.
In this the no-holds-barred principle is like the First
Amendment to the federal constitution. The principle of freedom or
toleration embodied in the First Amendment may be challenged in
public speech. The Amendment has been interpreted to protect even
those who oppose its values. But what is our rationale for this
super-toleration? It could be that only in this way can we preserve the
First Amendment (or no-holds-barred principle), since to prohibit
opposition to it in any degree would compromise the principle itself. In
this it would resemble the Humean custom theorist or the Academic
skeptic: the principle could be made an exception to itself to save itself.
But we might well feel that that would destroy the value we cherish in
the principle itself. The alternative is to allow challenges and denials
(and advocacy of repeal) and to accept the outcomes of fair procedures,
even if the sky should fall. That is, we might use the First Amendment
to protect a movement to repeal the First Amendment, and trust the
amendment process and public intelligence to do the best thing. We
might use the no-holds-barred principle to protect a philosophical
school that denied its value or truth, and trust to the realm of debate (or
the 'marketplace of ideas') to deal with the proposal justly. Note that
both the latter scenarios presuppose independent norms of just
procedure. These would have to be something like norms of logical
courtesy. In this sense, the principles of logical etiquette cannot be
debated properly or fairly except in a realm of debate already
constituted by them or their cognates.
Both the principle of epistemic democracy and the
no-holds-barred principle seem to be principles of logical courtesy. In
fact, violating them creates paradigmatic types of rudeness. Violating
Suber Logical Rudeness

the principle of epistemic democracy allows the proponent to believe

her critics are disqualified from knowing truth or deserving answers,
and violating the no-holds-barred principle allows the proponent to
deny that the critic's criticism is a permissible move in the game she is
playing. Their conflict, therefore, suggests that perfect courtesy, or
simultaneous compliance with all ruling principles of etiquette, is
We may consider the conflict between the two principles a
reflection of a broader conflict between equality and freedom. The
conflict may be avoided by ranking the principles so that one always
takes priority in cases of conflict. But no such strategem can eliminate
the conflict of the freedom principle with itself. Moreover, ranking
either above the other would allow just those infringements of the
'inferior' principle that the 'superior' principle authorized. These
would be rude infringements. For example, to rank the equality
principle higher would justify limiting the freedom of inquirers to
challenge the equality principle. To rank the freedom principle higher
would justify an a priori dismissal of theorists who proceeded in denial
of the freedom principle.
Some form of rudeness seems inevitable. Either the equality
principle will be violated by the rude theory that critics are unequally
entitled to know the truth, or the freedom principle will be violated by
the rude theory that critics are making impermissible moves in a game.
These two fundamental types of rudeness can be barred only by one
another. To secure some courtesies, then, we must impose other rude
principles. There is something Godelian about this result. No system
of logical etiquette can be both complete and consistent. For every such
system there will be a permissible but rude theory.
There are other ways in which rudeness may be inevitable, as
seen in Section 4. Some theories must be defended rudely to preserve
their own consistency and their proponent's good faith. Some are
caught in the dilemma of systematic self-defense. Under the
no-holds-barred principle we want proponents to be free to propound
and defend these and all other theories. This is another way of seeing
our general conclusion that rudeness per se does not imply falsehood.
We want to allow inquirers to propose the demon theory of error and
the buffet theory of belief. The alternative is rudely to impose a code
of debate on debaters, compromising the no-holds-barred principle,
and presumptuously presupposing an exclusive vision of truth prior to
debate. We may keep the hope alive that this may be done later, when
we know more, i.e., that toleration is just a makeshift until truth is
known to be known. But like the task of set theorists selecting axioms
that eliminate paradox and preserve 'good' mathematics, this cannot be
Suber Logical Rudeness

done without controversy. The no-holds-barred principle says we are

better off hearing this controversy. Toleration should not disappear
with the advent of knowledge unless inquiry is also to disappear.
The automatic inference of falsehood from rudeness or
undebatability may be called the fallacy of petulance -in which we
peevishly allow our hurt feelings to supersede our better judgment.
The fallacy of petulance is to use the criteria of courtesy as criteria (or
as a subset of the criteria) of truth. Sociability in debate may be
important for many reasons, even for the fundamental epistemic reason
of keeping debate a fruitful avenue of inquiry and for basic ethical
duties to other inquirers; but its norms do not thereby become criteria
of truth.
We may now consider the second element of a rude defense, the
firmness of the proponent's faith that the first element, the content of
the belief, authorizes a rude defense.
Can there be any theories that are inconsistent with the polite
concession of their corrigibility or possible falsehood? If some
theories have 'exclusivity clauses' and if no theory with such a clause is
consistent with the concession of its corrigibility, then the demands of
consistency would subvert the demands of courtesy. Then the system of
logical etiquette would be as reactionary as foot-kissing for demanding
courtesy at the expense of consistency. This is especially embarrassing
if most or all theories contain tacit exclusivity clauses, or if
corrigibility per se contradicts the claim of truth.
Rather than introduce the modal complexities of possible
falsehood, I will ask the question from a slightly different angle: not
whether a theory can be consistent with its possible falsehood, but
whether a theorist can retain her good faith while sincerely conceding
the corrigibility of her theory and herself.
Shifting the question this way is legitimate because, for the
purposes of logical etiquette, good faith is equivalent to truth. To the
proponent of a theory, the truth of the theory alone justifies rude
treatment of critics; but all inquirers outside the warmth of the
proponent's faith can see that it is his good faith that the theory is true,
and not its truth, that grounds this justification. The obligation to be
rude is not conditional upon the truth of the theory; it arises as much
from faith, and could not arise even in a true theory without good faith.
As we have seen, rudeness insulates believers, not beliefs, or
theorists, not theories. In Section 2 we saw that a kind of tenacious
good faith can require that a theorist apply her theory to all the
explananda within its scope, which frequently includes the context of its
own debate. I will call the kind of tenacious good faith that cannot bend
to concede the corrigibility of its object "fixed belief', after Charles
Suber Logical Rudeness

Peirce. A less tenacious kind of good faith -one in which sincerity

coexists with the concession of corrigibility- may be called "critical
belief' . Clearly it is attainable. What is not clear is whether it is
attainable for all our beliefs, or ought to be.
Insofar as fixed belief justifies rudeness to the believer, a canon
of logical courtesy prefers critical belief to fixed belief. This is
consonant with the 'civilized' demand that no inquirer be a fanatic, or
that all should hold their beliefs with detachment, and be prepared to
defend them with evidence and reason and to give them up in the face of
superior evidence and reason. The epistemology implicit in this
'civilized' demand is not merely that some faith is blind, but that fixed
belief blinds. Once critical detachment is lost in fixation, ignorance is
invincible. Those who will not concede the corrigibility of their beliefs
must directly equate disagreement and error, and fit their explanation
of error on the heads of all critics and dissenters. Fixed belief per se
authorizes rudeness to its possessors. This rude dimension of
immovable complacency or confidence explains the perjorative
overtones of the (originally neutral) term" dogmatism".
While this is the demand of courtesy we recognize from the
western tradition, particularly from the Englightenment, it by no
means follows that it conforms to the ethics or epistemology of the late
twentieth century. The traditional etiquette includes an aging concept
of debate that may be truncated roughly as follows. Debate serves
inquiry; its values are epistemic; it is neutral in that the truth (whatever
it may tum out to be) may be approached by debate; debate is joint
inquiry; debate is the marketplace of ideas in which the epistemic worth
of ideas is tested and evaluated and reevaluated; success in debate may
occasionally go to the unworthy, and true ideas may lay unpersuasive
for generations, but in the long run debate will reward all good ideas
and punish all bad ones; it is a self-correcting method; it is a method
without presupposition or principle; it works best when proponents of
theories state their position publicly for all to examine, offer all
evidence and reasoning for public examination, answer all questions,
reply to all criticisms on the merits, and interact with those with
differing opinions by propounding their own questions and criticisms;
it works best when the participants and spectators allow their assent to
follow the evidence and reasons exchanged in debate, and do not enter
with prejudice or simply for sport.
It is according to such a concept of debate that the examples at
the top of Section 1 were said to betray 'something wrong'. Note that
the activity outlined by these principles is ineliminably that of a
cooperative enterprise.
Do these norms of logical etiquette reflect a pattern of social
Suber Logical Rudeness

interaction, or even of reason and inquiry, that died in the

Enlightenment, and that is impossible, and reactionary, to wish back to
life? Doubts of this order have forced me to put 'civilized' and
'well-mannered' in quotation marks throughout the essay. Our distaste
for rudeness is certainly not the same as the aristocratic distaste for
commerce and trade. Nor is our distaste for rudeness reducible to bad
sportsmanship. But is it similar to the wistful sighs of aristocracy in
that, its epistemological merit notwithstanding, it is inseparable from a
certain nostalgic longing for the days when the logic of self-insulation
was not freely practiced by every ignoramus and heretic, the days when
the elegant tools of logic were not made to serve the boorishness of
every comer? Have we romanticized the 'classica1' forms of debate,
idealizing the tradition from the Athens of Socrates to the London of
Joseph Addison? In our code of logical etiquette have we legislated a
form of argumentative geniality that never existed? Or one that ought
to exist no longer? Or one that distorts our enterprise to pretend that
we practice?
The danger of legislating a style of thinking in order to secure a
form of cooperation is real. So I take these questions seriously,
whether I am in a mood to favor good epistemology and hope that good
ethics will follow, or vice versa. But answering these questions is
beyond the present topic. Here it is enough to point out that debate has
norms other than the norms and rules of any shared logic, and that these
norms may be leftovers of bygone social structures. If they have merit,
it is not that of logics, but of manners.
My authority in saying just what logical courtesy demands is
simply that of a native of the realm of whose customs and ideals are
being described. It is that of mere acquaintance, and may be corrected
by others of wider acquaintance or more acute perception. It is not like
saying what a formal logic demands. Hence, we should be careful that
we do not allow descriptive inquiries into the normative domain of
logical etiquette to be swayed by normative disagreements among
debaters as to correct style, cooperative harmony, and civilized
behavior. We should not legislate in the name of description. My
purposes here have not been wholly descriptive, of course. In our
descriptive inquiries we should try to resist the temptation to describe
as rude (and therefore to stigmatize) practices whose only vice is their
endorsement by the beliefs and theories of our opponents. That would
be rude. But in dealing with the challenges of the descriptive inquiry,
we should not overlook the normative. For the canons of logical
etiquette we use without reflection, those we urge falsely in the name of
logic itself, and those that we tolerate in our comrades and resent in our
critics, create the ethics of argument that govern discussion.
Suber Logical Rudeness

Peter Suber
Department of Philosophy
Earlham College
Richmond, Indiana

Myron Miller

The Pragmatic Paradox

I The Problem

Some of the English verbs and verb phrases associated with

illocutionary acts are: state, assert, describe, warn, remark,
comment, approve, welcome, promise, express approval, and
express regret. 1
In terms ... of the triadic relation of semiosis, ... [o]ne may study
the relations ... of signs to interpreters. This relation will be called
the pragmaticai dimension of semiosis,symbolized by 'Dp " and the
study of this dimension will be named pragmatics. 2

That these two positions each describe the same linguistic

feature is further supported by the choice of examples which Rudolph
Camap chooses to illustrate those relations which he identifies as
pragmatic. These examples are: " .. .they intend to assert ... ," "When
using this name ... ," " ... serves to express the assertion ... ," " ... is
acknowledged ... ," etc. 3 It is well known of course that Camap was
skeptical about being able to develop a systematic pragmatics in the way
that it is done for syntax and semantics. Morris, however, identifies a
distinction which he attributes to C. S. Peirce which should have given
pause to Camap's skepticism, namely Peirce's distinction between
"interpreter" and "interpretant." Morris says,
The interpreter of a sign is an organism; the interpretant is the habit
of the organism to respond, because of the sign vehicle, to absent
objects which are relevant to a present problematic situation as if
they were present.4

While it is an empirical, and not a logical, matter as to what

some interpreter does with a sign on some specific occasion and under
some specific set of circumstances, the description of the habit that the
interpreter exhibits in using signs is a logical entity, not an empirical
one. These habits are captured in just those verbs the likes of which are
listed by Searle above. A logic for illocutionary acts, then, if one were
constructed, would be a pragmatics for some specific language.
As straight-forward as this may seem, there still seems to be
some confusion about what belongs to Morris' "dimension," or what
Miller The Pragmatic Paradox

might better be called "domain," of pragmatics. Moreover, the

confusion is of the sort as to raise doubts about whether we have yet
gotten a clear account of the distinction between semantics and
pragmatics. Interestingly, this lack of clarity arises in a concept that has
long been identified as semantic, one on which P. F. Strawson wrote an
important and influential paper.s Strawson thought that the relation of
"referring" better accounted for what logicians like Russell had
identified as "denotation." The latter, of course, is a relation which has
been central in defining a systematic semantics.
The problem becomes acute when we tum to the class of
sentences which have been identified as semantically self-referring.
Tarski gives rise to this kind of language in his famous study of the
concept of truth, using the liar's paradox as a test case for showing how
his definition of truth solves this ancient conundrum. Tarski
emphasizes this in the followiIlg way:
Semantics is a discipline which, speaking loosely, deals with
certain relations between expressions of a language and the objects
(or "states of affairs") "referred to" by those expressions,

and just below this,

While the words "designates," "satisfies," and "defines" express

relations (between certain expressions and the objects "referred to"
by these expressions) the word "true" is of a different logical
nature: it expresses a prope7 (or denotes a class) of certain
expressions, viz., of sentences.

Thus, a sentence which denotes or designates itself in some way

might" loosely" be said to be referring to itself. When we try to tighten
up what is meant by "referring," however, it becomes problematic
whether reference is a semantic or a pragmatic concept.

"Tarski's Doctrine" is that the reason why sentences like:

1. Sentence 1 in the paper "The Pragmatic Paradox" is false,

are logically peculiar is that they conflate two levels of discourse.' One
level contains the sentences of the language and the other contains the
names of the sentences of this very language. This forms a closed
language and closed languages are inconsistent. Both the arguments
supporting this doctrine and the doctrine itself have been extremely

Miller The Pragmatic Paradox

influential in twentieth century philosophy. In spite of this influence,

however, Tarski's analysis of the liar's paradox rests on a mistake, a
mistake which when clarified will help us to deal successfully with the
liar's paradox and make sharper the distinction between semantics and

IT The Mistake

Tarski's reasoning in support of the closed language doctrine is

that once we allow substitution by variables to obtain a generalization
for all such self-referring sentences as sentence 1 above, we will
immediately detect the flaw. Robert L. Martin has refined and
generalized this reasoning in an interesting article on the Liar's
Paradox, but his reconstruction does change the stance of the argument
which Tarski used. Tarski was insistent on remaining in the area of the
logic of these sentences, while R. L. Martin's modification introduces
the concept of people using these sentences. This shift is very
important, as will be shown below but not for the reason (alone) which
Martin alleges.
Tarski emphasizes that once we have made out our case that
".. .is true" and " ... is false" are predicates of sentences, then we can see
at once what has gone wrong with the sentences which say of themselves
that they are false. Moreover, as Tarski stressed, we now have a ready
way of defining 'true,' at least from a logical point of view. Given the
mathematical concept of 'satisfaction,'8 we can now say that a sentence
is true if it is satisfied in all its recursive forms. Actually, this very
happy outcome, viz., the development of a discrete and exact way of
handling 'true,' tends to overshadow the argument which was intended
to eliminate the puzzle of sentences which refer to themselves as being
false. Yet it is just that reasoning which contains an important mistake.
Tarski's suggestion is that if x = sentence 1, then we can
substitute x in every sentence where sentence 1 is named. This is very
straight-forward, much like the observation that if x = Venus, and
Venus is the morning star, then x is the morning star. This merely
applies Leibniz's law of identity to sentence 1. After substituting x for
the name 'sentence 1 in "The Pragmatic Pardox" is false,' which in
single quotes is the name of the sentence which is sentence 1 above, the
result is the following:

1a. x is true if and only if sentence 1 in "The Pragmatic

Paradox" is false.

Miller The Pragmatic Paradox

The next step would be, on the analogy of 'Snow is white' is true if and
only if snow is white, to substitute a description of the state of affairs
that is named by the sentence. However, following Tarski again, since
what falls to the right of the biconditional is also sentence 1, the
paradoxical result seems to be

1b. x is true if and only if x is false.

Once we see how directly contradictory 1b is when we apply

Leibniz's law, we can see at the same time that Leibniz's law is either
wrong or we do not have a proper instance where Leibniz's law applies.
Tarksi, for good reasons, does not explore the first alternative, which
leaves only the second. The result is the assumption that what makes it
incorrect to apply Leibniz's law is that we have different levels of
language. In turn there are different types of entities yet treated first as
the subject and then as the predicate of the same sentence in sentences
like sentence one.
This leads then to the solution stated in part one of this paper.
Moreover, this solution accords with what we know must be
maintained, viz., the distinction between use and mention. The
predicates in sentences like those of sentence 1 are both mentioning the
sentence, and being used in the sentence, of which they are a part. No
wonder, then, that such sentences are puzzling--they are not, in fact,
genuine sentences after all. They are but two fragments of a sentence, a
subject without a (proper) predicate and a predicate without a (proper)
subject, "glued" together with the copula "is." The result is something
like trying to nail a poster to a waterfall. The predicate is simply not
the kind of thing that can be said of the subject.
However, this same analysis should also invalidate the following

2. Sentence 2 in the paper "The Pragmatic Paradox" is not


Yet far from being inconsistent, or contradictory, it is simply

false. Moreover, we do not escape this objection by simply noting that
'English' is not, while 'false' is, a semantic concept. Both are semantic
concepts in the sense that they describe a feature of language; 'is false'
describes a feature of a set of sentences in any natural language, while
'is not English' describes a feature missing from many sentences in

Miller The Pragmatic Paradox

natural languages. Once we have put the matter this way we begin to
have doubts about the original argument. First, since only sentences
can be true or false, it would not be proper to say 'x is true' when we
substitute x for sentence 1. Sentence 1 is not a sentence on Tarski's
analysis and, therefore, can be neither true nor false.
We were lead to think that sentence one allowed the substitution
on the analogy with the paradigm "Snow is White." In the paradigm
case, to say that "'Snow is white' is true" allows for the substitution of x
for 'Snow is White' is to admit that what is named by 'Snow is white' is
a genuine sentence. This is not the case for" Sentence 1 in the paper
'The Pragmatic Paradox' is false." In this case all we have for which x
could be a substitute is 'Sentence 1 in the paper "The Pragmatic
Even if we grant for the sake of argument, however, that we at
least take sentence 1 to be a sentence (thus producing the psychological,
though not a logical, puzzle of a self-refuting sentence) there is an even
more important problem. This is that when we make the second
substitution resulting in 1b from la, we have not, strictly speaking,
made the same substitution that was made in transforming 1 to 1a. If
we say that x names the sentence 1, it names the whole sentence,
including the predicate "is false." But in the second substitution while
we do have 'x is true if x ...," we also explicitly retain" ... is false" in
the form lb.
Either 'is false' is already referred to by x or it is not. If it is,
then sentence la already contains a contradiction an.d the second
substitution is at least redundant. If it is not, then x, in its second
substitution, does not name sentence 1. Rather, it names only the
subject of sentence 1. In this case the first substitution transforming 1
to la is simply wrong. Subjects can be neither true nor false. The same
is true of predicates alone. The redundancy of the first alternative of
course is generated by accepting already the assumption that 'is false' is
a predicate of sentences. Thus, this part of the proof begs the question.
Tarski's doctrine, then, however right it seems, is not
established by the line of argument that he uses. Still, the assumption
that 'is true' and 'is false' properly apply only to sentences, when used
as semantic concepts, is undoubtedly correct. Thus, there has to be
another way to view the difficulty that is raised by the Liar's Paradox.
R. L. Martin's assessment of the texture of the language which the
paradox reveals introduces a fruitful suggestion, though not, perhaps,
among those which he had intended.

Miller The Pragmatic Paradox

ITI The Confusion

Robert L. Martin's interesting essay on the Liar's Paradox

recognizes the necessity of giving an account of self-reference that
allows some sentences which refer to themselves to be either true or
false. Such would be the sentence, for example, of sentence 2 above.
Yet his solution to the Liar's Paradox seems to confuse semantics and
pragmatics. Note his suggestion for determining semantical
correctness in self-referring sentences:
The second step of our procedure is to determine whether or not
the sentence is self-referential. Let us say that a sentence type is
self-referential if and only if every token of that type is
self-referential and that a sentence token is self-referential if and
only if what one mentions with the subject expression of the
sentence token is the sentence itself (type or token) which is being
used. 9

The mention and use of sentences are acts performed by those who
mention and use the sentences. Semantic correctness on this approach
will rest on pragmatic correctness, which means that the semantic
paradox will not be solved until we introduce pragmatic considerations.
This would be puzzling since, in the way that Tarski frames the Liar's
Paradox, there is no need to introduce the concept of use. In fact,
Tarski explicitly says,

.. .it would be quite wrong and dangerous from the standpoint of

scientific progress to deprecate the importance of this and other
antinomies, ... the antinomy of the liar and other semantic
antinomies give rise to the construction of theoretical semantics. 10

Moreover, it is easy to see why the antinomy, or paradox, is for

Tarski a semantic issue. Sentences like sentence 1 are not merely
psychologically puzzling. Any rational being or rule-driven machine
would abort such sentences. The reason, however, is not that a
predicate which names one class of objects is attached to a subject which

names another class of objects. This happens in sentences generally,

e.g.,'color' in

A. "The ball is red"

is a different class of "objects" than the geometrical name which forms

Miller The Pragmatic Paradox

the subject of the sentence. The class of colors is a different class from
the class of geometrical shapes.
Nor is the problem of the Liar's Paradox a violation of the sort
where the predicate combines names of incompatible classes. In the

B. "The lasagna tastes five feet long"

the predicate in B, viz., "tastes five feet long," contains incompatible

class characteristics. The feature named by "tastes" is simply a
different class of characteristcs than the feature of measurement. Some
may think that such sentences are meaningless because they are neither
true nor false. In the sentence B, it is neither true that lasagna tastes
five feet long, nor it is false to say this. In fact, however, the subject
"lasagna," is compatible with what is named in the predicate, i.e., "five
feet long." Garfield the cat, among other lasagna devotees, could make
such a statement in the metaphorical sense meaning that its flavor might
be said to "go on forever." Even here, however, the metaphor is one to
communicate the longevity of remembering the taste, not the physical
dimensions of the taste. What is incompatible is the literal use of the
copula with the predicate, viz., "is a thing which tastes five feet long."
This use violates the rules in a natural language governing what can be
named by "taste." Thus, it is simply false, not meaningless, to say that
lasagna has such a taste. Because nothing could have this taste, lasagna
certainly does not have it, and to say it does is to say what is not the case,
i.e., what is false.
Yet the Liar's Paradox does not exhibit an incompatibility
within its copula/predicate name structure. "Is a thing which is false" is
properly used of very many sentences. Nor would anyone be tempted
to say that the problem is within the predicate alone. The problem
arises in adding another predicate to sentences like sentence 1, viz., "is
This produces an explicit logical contradiction if what is
predicated "true" has already been predicated "false." It is this, of
course, that appears to have happened with sentences like sentence 1.
Yet as argued in Part II above, it is not clear that these two predicates
("is true" and "is false") do, in fact, name the same thing. The subject
"Sentence 1 in the paper 'The Pragmatic Paradox'" is a name which
apparently identifies an object, viz., that sentence numbered 1 in the
paper "The Pragmatic Paradox." But that item so named is a sentence
form which contains a predicate which apparently names a quality of

Miller The Pragmatic Paradox

sentence 1 in the same paper. So long as it remains a sentence form,

i.e., a string with a subject, a copula, and a predicate, no rules of
language are violated. The violation enters when we take this string to
be the concatenated name of the sentence and add another predicate to
create another sentence, i.e., a sentence which is not sentence 1 in the
paper "The Pragmatic Paradox."
This new sentence, let us call it sentence 2, is

2. 'Sentence 2 in the paper "The Pragmatic Paradox" is false' is


But since this is now sentence 2, not sentence 1, and by it we have

introduced the "new level," or additional language, needed to keep the
predicates separate, again there is no contradiction. A contradiction
results when, and only when, these two levels of language are taken to
be one level. But, again, the concept of "taking" a sentence to be
something or other is a pragmatic one. It is what language users do with
the pieces of their language.
The Liar's Paradox, then, is in fact a pragmatic, not a semantic,
paradox even though semantic concepts are at work (just as syntactic
ones are at work, as well, but of course we do not call it a syntactic
paradox for all of that). This takes nothing away from the important
semantic achievements which Tarski's work accomplished. It does,
however, .stress the importance of finishing the task which the Liar's
Paradox generated for Tarski. This undoubtedly is what Richard M.
Martin is aiming at when he says,

Semantics should have no truck with self-reference, which is a

liberty the speaker allows himself on certain occasions. It may lead
him to self-contradiction or it may not. This is the chance one
takes. 11

R. M. Martin, then, takes reference to be a pragmatic relation, in

contrast to Strawson's interpretation of reference as a semantic
relation. One advantage to taking reference to be pragmatic is that it
allows a distinction to be made between" contradiction," which is either
a syntactical or semantical concept, depending on whether the conflict
is generated from the sentence form or sentence content, and
"self-contradiction," which would then be a pragmatic concept, the
conflict arising from the sentence use.
With this distinction, sentences like sentence 1, while not

Miller The Pragmatic Paradox

contradictory since there is no incompatibility between the subject

(,Sentence 1 in the paper "The Pragmatic Paradox"') and the predicate
('is false'), could be self-contradictory. This would happen, however,
only if a pragmatic relation were understood to be at work on sentences
like sentence 1. This would transform sentences like sentence 2 into
something like:

3. I (you, we, etc.) believe (assert, take, state, etc.) that

'Sentence 1 in the paper "The Pragmatic Paradox" is false' is true.

The addition of the string placement variable for a person and a

variable for a relation between the person and the linguistic item being
used by that person makes explicit what kind of logic is needed. It
would be a set of rules that govern each of these relations. Such rules
would require that the sentence named by the single quotes be the very
same sentence that is named by the supressed quote "that." This
amounts to taking a sentence to be both true and false at the same time.
To put the matter this way makes what is going wrong in the Liar's
Paradox much more plain. The clarity was achieved, however, by
going explicitly to the level of pragmatics. While this is not surprising,
it is important if we are to get a logical picture of the language needed
to treat the paradox.

IV APromise

This shift in analysis of the language structure that is revealed by

the Liar's Paradox achieves some distinct advantages that are in need of
exploration. Both Daniel Vanderveken12 and 1,13 in different places
and in different ways, have pointed out that there is a set of interesting
paradoxical sentences. The informal analysis of the two approaches,
however, is quite compatible. Actually Vanderveken develops three
general categories of such sentences, each with sub-set distinctions. 14
Whatever the proper structuring of the data involved, there seems to
me to be only one general characteristic of these sentences. Fine tuning
the kinds of paradoxes involved will depend on the number of uses, or
kinds of uses, to which a rational being might put elements of a
language. The set of such uses, however, is open with any listing of the
kinds of paradoxes--a procedure that is in principle incomplete. In
general, however, problems can arise either with first person (singular
or plural) descriptions or with second and third person (singular and

Miller The Pragmatic Paradox

plural) commands.
Thus, a person might utter either of the following which result
in self-contradiction, in the sense that we have above restricted the
category of self-contradiction.

4. I have forgotten everything I ever knew.

5. You are to forget everything.

The relation of "forgetting" is one which interfaces a ,person with,

among other things, language. Memory is a necessary condition of
having the use of a language. This does not mean that sentences like 4
and 5 are not used in everyday discourse. They, in fact, are used from
time to time. When they are used in ordinary language, the strict rules
for these relations are suspended so that the speaker or writer is
understood to exclude that which is being uttered or written, in this case
the very sentences 4 and 5, from that to which 4 and 5 refer. It is very
much like taking the position which Sj1jren Kierkegaard identified as
sub specie aeternitatus. The person uttering the statement is viewing
him or herself or the person being commanded, as a special exclusion to
the class of beings to whom the specific utterance being made normally
If we analyze sentences like 4 and 5 in terms of there being an
illocutionary force involved in uttering or writing them, we would
distinguish between the statement being made, the meaning of the
sentence, and the force of saying what is meant, i.e., in 4 a describing
and in 5 a commanding. Let us suppose that it is possible to construct a
logic of the act of utterances (more properly, a logic of the descriptions
of the acts of utterances) which would constitute a logic of illocutionary
force. ls Perhaps it would look like a systematic pragmatics in which
the rules for use contain logical primitives defined in empirical ways.
Commanding, questioning, pleading, etc., could all be given a logical
analysis fitting them into a common set of logical rules and provided
with a content which is explicated in empirical generalizations. But if
this is able to be done, we will need to know first what it is to use a
segment of language, whatever its specific form.
Thus the theme stressed by John R. Searle in his own
contribution to Speech, Act, Theory, and Pragmatics, viz., that

... in general the meaning of a sentence only has application (it

only,for example, determines a set of truth conditions) against a

Miller The Pragmatic Paradox

background of assumptions and practices that are not representable

as a part of the meaning, 16

is itself true only if there are truth conditions apart from a background
of assumptions and practices to which the assumptions and practices
must adhere, or by which they are judged as meaningful. Without this
we would have to accept Camap's original skepticism about being able
to develop a logic of the relations between humans and their language.
All that would be left would be a set of empirical generalizations about
the ways that humans characteristically have made utterances on some
occasions in the past and in some specific places~ This would not be a
science of such utterances, however, since we would not have a theory,
or set of axiomatic statements about the way we are to interpret the
This point becomes important as we tum to the most general of
all such relations, viz., the relation of 'use.' Wherever there is a
paradox, there is a breakdown of rules--no paradox would be possible
if there were no linguistic rules. Thus the sentence,

6. I (descriptively; you, he, she, etc., prescriptively) take all

sentences to be unuseable,

taken in its strict sense would be self-contradictory. Such a person

would be using a sentence, an item which is taken by the same person to
be unuseable. But this reference to "strict sense" is the promise of a set
of rules which determine the pragmatic meaning that can be modified
for specific occasions in specific circumstances, in the light of any
restrictive assumptions and practices. If so, then all sentences must be
useable. How they are used will depend on the restricting conditions of
use. What is needed, then, in the spirit of Tarski's definition (T) is a
definition of the relational predicate 'use'; a Tarskian definition (U).
On this all of pragmatics will rest in much the same way that the
development of a systematic semantics rests on a clear notion of truth.
All sentences after all are either used or unused though the parallel of
unused with 'false' is not exact. Particular pragmatic relations specify
the way they are used.
Such a definition would force us to make explicit that we have
three elements in our sentences, viz., each sentence involves a
relationship between a variable for persons (or a constant if a proper
name is used)17 and a segment of that language which is able to be
properly (by grammatical rule) attached to such a relation. In addition

Miller The Pragmatic Paradox

to the advantage of making the object of pragmatic analysis explicit,

this also, while including performatives, includes a much wider set of
locutions. Whatever we need to capture as the illocutionary force,
then, will be able to be included in the definition of th particular 'use'
relations (plead, assert, question, etc.) themselves.
This latter point is important, for in such apparently
non-epistemic locutions as

7. I promise you that I will be on time,

the lexeme "promise" functions as a three place predicate relation.

Whatever 'promise' means (something that must be defined), the
"propositional" content, or statement, viz., "I will be on time," the
individual named by 'you' in the context of use, and the person named
by 'I' in the context of use are all related by the use of 'promise.'
Imbedded in the relation, then, is the statement which is something that
is asserted as a part of the promise. Another part is the description of
what it means to pledge something, so that to include in the sentential
object of the relation a content that negates the relation will naturally
result in a contradition. Note that in

8. I promise that no promises are kept,

there are no mysteries about what kind of act is involved in the

sentence,since it is not the promising act which produces the
contradiction, any more than

9. Plato is the father of no children,

involves the act of fathering while not fathering. It is not the act of
uttering but that which is uttered which produces the logical problem.
Ordinarily we would simply take sentence 9 to mean that Plato is not a
father. Similarly we would ordinarily take 8 to mean that there are not
promises and that uttering 8 is not able to be taken as a promise--it is an
assertion namely the claim that there are no such things as promises.
This tum in the analysis does not mean that there are no such
things as illocutions. On the contrary, it requires that we look more
closely at the very language used in describing what has been called the
illocutionary act in order to determine what its linguistic structure is.
The analogy in sentence 9, with its appeal to our linguistic intuitions,
should force us to ask what is the confusion in 6, especially, though not

Miller The Pragmatic Paradox

exclusively, in contexts where the use of the language seems to form the
meaning conveyed. Using Tarski's semantic analysis; then, as our
guide to this necessary expansion into the pragmatic elements we have
found in the Liar's Paradox, we have a more generalized escape. It is
an escape, in fact, which provides foundations for pragmatics in just the
way that the semantic "escape" provided foundations for semantics. It
would be well to state the problem in the Liar's Paradox again, but this
time without any specific prejudice on the kind of use, e.g., the
epistemic use noted above, and yet make explicit the reference to the
user. Let us, then, restate sentence 1 of this paper in the following way:

1'. I am not using sentence l' in the paper "The Pragmatic


It is immediately clear that l' contains no difficulty so long as it is not

uttered or written by anyone. However it looks like that is just exactly
what has been done, i.e., l' is being used both by the writer of this
paper and the person who is reading it, so long as "reading" is taken to
be a use of sentences, viz., as a means of gaining the meaning of that
sentence. The first person position has the delightful ability to provide
dramatic objectivity, a feature that allows us psychologically to escape
the self-contradiction in 1'. It is the sort of attitude that we take when
reading a novel; the character in the story says "I will now stand on my
head," and we do not think for a moment that we are uttering that
claim. 'I' refers to the one uttering or writing the sentence when it is
being used, not to the observer of the one using the sentence.

V A Definition

It is even more important given the empirical context of our

pragmatic language, that we follow Tarski's emphasis that we must
relate the question of 'use' (as with 'truth') to natural languages. Since
"use" in l' involves a case of linguistic use, we must specify the
language in which this sentence is being used. A sentence in a language
is being used if it is uttered, written, read, accepted, questioned,
shouted, believed, etc. The result in general is that a sentence may be
used in one language, but not in another, e.g.Jch habe keine Zeit is not
read, written, nor uttered in English, though it has an English meaning
equivalent. While there are many sentences of a language which do not
manifest in their structure the fact that they are being used, we have
seen that if they are being used,that use requires a structure that makes

Miller The Pragmatic Paradox

explicit the reference to the rational being using the sentence. This
requirement forms anew, and pragmatic, sentence.
The subject of pragmatic a sentence is the name, noun, or
descriptive reference to the user. The object of a pragmatic sentence is
the sentential expression or statement. The feature that identifies the
sentence as pragmatic is the relation specifying the kind of use to which
the subject is putting the object. Pragmatic sentences do not always
wear the fact that they are pragmatic on their surface. This is the great
beauty of the work that has been produced by those who have identified
and explicated the notion of illocutionary force. 1s To say "Stop
following orders" does not reveal itself as "I order that you are not to
obey any order." But until we put it into the latter form, we will not be
sure of the source of our perplexity, even though we will undoubtedly
immediately sense the "Catch-22" into which we would be forced
should we try to follow this direction.
Let us see, then, how this analysis removes the perplexity in I',
which is simply one form of the general pragmatic paradox. If we
adopt the sign 'U', which will stand for the disjunction of all pragmatic
relations, and a sign 'p', which will be a person variable, then for any
sentence, such as "Snow is white," there will be a logical picture for the
form of the use of that sentence. For instance,

"Snow is white" is used if and only if (at least) some person

writes, or utters, or believes, or questions, or etc., "Snow is

To picture this pragmatic relation using 'X' as a name place marker for
sentences, we would get the pragmatic parallel with Tarski's (T), let us
call it (U), i.e.,

(U) X is used if and only if (Ep)pUX,

where '(Ep)pUX' is the "state of affairs" in which there is some

person who is using the sentence named by X in some specific way.
Note that the definition presupposes, then, that (T) is imbedded in it.
Moreover, (U) requires that the variable 'p' range over any rational
being (even when being irrational). This might require, of course, that
it not be the case that non-rational beings (like computers) use
sentences. The difference between (T) and (U) is that the name of the
sentence appears on both sides of the biconditional. This can be called a
definition of use if we say that 'pUX' is to be replaced by any sentence

Miller The Pragmatic Paradox

of the language to which the word "use" refers, or which "use"

designates, and X is replaced by the name of that sentence. The
language to which 'use' refers, then, would have to contain appropriate
predicates to relate the name of the sentence and to the person using the
sentence in that language. For example, if a person did not know how
to speak, read, or write, etc., German, it is not clear how the names for
German sentences could be used by such a person. This does not mean
that such a person could not use the name of a German sentence for
some other purpose, one that did not require a knowledge of German.
If this were to happen it would be quite clear that what is named is in a
different language than that sentence in which the (non-understood)
German sentence was placed. Thus, the definition (U) does not allow
that 'X' and 'pUX' be of the same language. On the contrary, it
requires that 'X' be the name of a sentence to which the use of that
sentence is referred by the language in which the sentence that 'pUX'
names occurs. Like the definition of truth, then, the definition of use
becomes precise only when the languages involved are clearly
specified, their structure spelled out in terms of the inferences allowed,
Another way of showing what has happend to I' by applying our
definition of use, i.e., (U), is by picturing the way in which I' is
defective given the requirement of (U). The disjunction which (U)
designates is very large and open ended. Languages have indefinitely
many uses to which they can be put. There is a restriction, however,
and that is that these uses must be linguistic ones. Carving a word out of
a block of stone and beating someone with it would not constitute the
use of a language. The use of language requires, then, that we engage
the rules that allow some fragment of language to work as a conveyer
of meaning. To utter a German sentence in an English context might
have a use other than to speak German, but then the use of that fragment
would have to obey the rules of English, not German.
Given this restriction on (U), another. parallel with (T)
emerges. The linguistic object of a pragmatic relation must be either an
implict or explicit statement, i.e., a complete sentence. It is true that
this is not always on the surface of pragmatic sentences. The statement
"I remember nothing" contains no clear sentential object for
"remembering." In context, however, we often use sentence fragments
to express what is in fact a complete sentence. In this modification of
sentence 4 we might be saying something like "I remember that nothing
like what you are saying occurred." This removes the puzzle that
would arise out of generalizing "nothing" and makes clear a context of

Miller The Pragmatic Paradox

its use.
But now sentences like l' fail to meet the requirements of the
restriction we have just noted. If we take sentence 4 to include sentence
4, we would seem to be insisting that 4 be forgotten too. Would it be
possible to assert such a claim? People do apparently assert many
strange and wonderful things, why could someone not assert 4? If
someone did do this, however, we could only take them to be joking,
trying to mislead us on the collapse of the categories of the language
involved. As Ryle long ago emphasized in another context, only
philosophers would have trouble with such sentences, confusing the
grammatical form of the sentence with the logical form. 19 More
perspicuously, from sentence 4, on a strict literalist reading, we would

4'. I forgot sentence 4'.

Applying the substitutions which (U) allows we would first get the
following from 4' though this time let us use '... x .. .' as our sentence

4". '... sentence 4' .. .' is used if! forgot sentence 4'.

Keeping in mind that by definition (U) is a disjunctive definition, one

that will include 'to remember' as one of the ways that sentences can be
used, sentence 4" then negates all the other disjunctions in (U).
Finally, we substitute once more to get

4'" '... sentence 4' .. .' is remembered if I forgot '... sentence

4'.. .'
It now becomes clear, in a way that the semantical analysis did not make
clear, that we are dealing with one sentence, a complete sentence in each
expression of it, that is contradicted by the relations at the two
pragmatic levels, one indicated by "is remembered" and the other
indicated by "I forgot."
Thus, the pragmatic analysis of the Liar's Paradox includes by
generalization all the kinds of contradictions that can be expressed in a
language. At the same time it promises to give a whole structure to a
systematic pragmatics, one that rests on a clearly definable concept of
use. The implications of this reconstruction of paradox should both aid
in the development of a more thoroughly systematic semiotic, but also
unite what might otherwise have been thought to be warring camps in

Miller The PrngmwticPrumdox

philosophical analysis, viz., those investigating the illocutionary force

which defines much of language use, and those investigating

Myron Miller
Division of Social and Behavioral Studies
St. Petersburg Junior College
St. Petersburg, Florida

Miller The Pragmatic Paradox

1. John R. Searle, "What is a Speech Act?" The Philosophy of
Language, ed. J. R. Searle Oxford University Press, 1971, p. 39.

2. Charles Morris, "Foundations of the Theory of Signs,"

Foundations of the Unity of Science, eds., O. Neurath, R. L.
Camap, and C. Morris. Vol. 1, The University of Chicago
Press, 1938, p. 84.

3. Rudolph Camap, "Foundations of Logic and Mathematics,"

Ibid., pp. 147-148, italics mine.

4. Morris, p. 109.

5. P. F. Stawson, "On Referring." Mind (1950), reprinted in

several places.

6. Alfred Tarski, "The Semantic Conception of Truth."

Philosophy and Phenomenological Reasearch, IV, 3 (1944) 345.

7. The more sophisticated presentation of this doctrine, of course,

is in Part V of his "The Concept of Truth," reprinted in Logic,
Semantics, Metamathematics: Papers from 1923 to 1928. Tr. J.
R. Woodger. Oxford University Press, 1956.

8. Though, as Rartry Field has skillfully argued in "Tarski's

Theory of Truth," The Journal of Philosophy, LXIX, 13 (July
13, 1972), particularly parts I and II, Tarski has not succeeded
in reducing the concept of truth to non-semantic concepts. This
problem does not affect the argument of this paper since
whether or not the definition Tarski gives is free of semantic
concepts, such as denotation in addition to satisfaction, the claim
that I am making is that semantics is not sufficient to
disambiguate the problems raised by the Liar's Paradox. This is
particularly important when we generalize to the
"self-defeating" or "self-stultifying" paradoxes that are clearly
pragmatic in nature. If we can show that pragmatics is involved
at the level that Tarski was satisfied to leave with a semantic
analysis, then surely these insights will be important where the

Miller The Pragmatic Paradox

use of the sentences generates the contradiction.

9. Robert L. Martin, "Toward a Solution To The Liar Paradox,"

The Philosophical Review. LXXVI, 3 (July, 1967) 293.

10. Tarski, p. 348.

11. Richard M. Martin, "The Pragmatics of Self-Reference,"

Chapter 5 of Pragmatics, Truth, and Language.; Vol. 38 of the
Boston Studies in the Philosophy of Science. D. Reidel
Publishing Company, 1979, p. 68.

12. Daniel Vanderveken, "Illocutionary Logic and Self-Defeating

Speech Acts," Speech, Act, theory, and Pragmatics, eds., John
R. Searle, Ferenc Kiefer, and Manfred Bierwisch. D. Reidel
Publishing Company, 1980, pp. 247-272.

13. Myron Miller. The Logic of Self-Referring Knowledge Claims.

Unpublished dissertation. New York University, 1977.

14. In section 2, where Vanderveken actually identifies what he calls

characteristic of the nature of self-defeating illocutionary acts.
Broadly he picks out examples of sentences which are
impossible because of conflicting requirements of achievement,
examples which are impossible because of a conflict in the
"propositional" content of what is uttered, and, finally,
examples which illustrate impossibility because of a conflict
with what must be presupposed. .

15. The foundations of such a logic are promised. Vandervekan, in a

bibliographical entry at the end of the work cited above, says
that he has co-authored with John R. Searle a work entitled
Foundations of Illocutionary Logic, which is forthcoming.

16. John R. Searle, "The Background of Meaning." Speech, Act,

Theory, and Pragmatics. Op. Cit., p. 221.

17. The calculus of individuals, of course, is assumed to be

incorporated into our logical reconstruction, along with all the
other syntactical features of elementary logic.

Miller The Pragmatic Paradox

18. Rooted, of course, in the work of J.L. Austin, but pressed much
beyond what he envisioned by John L. Searle, et al ..

19. Gilbert Ryle, "Systematically Misleading Expressions,"

Proceedings of the Aristotelean Society, XXXXII (1931-32),
reprinted in The Linguistic Turn, ed. Richard Rorty. The
University of Chicago Press, 1967, pp. 86-7, particularly the
point "There are many expressions ...couched in grammatical or
syntactical forms which are in a demonstrable way improper to
the states of affairs which they record ... " I am not saying, of
course, that the distinction between grammatical form and
logical form as I have presented it in this paper is what Ryle has
in mind by the distinction he makes in this remark.

Henry W Johnstone, Jr.

Argumentum ad Hominem
with and without Self-Reference
In his splendid article "Self-Reference, Phenomenology, and
Philosophy of Science"l Professor Bartlett lucidly spells out his
understanding of the use I make of self-reference in developing the
thesis that there are valid ad hominem arguments in philosophy. That I
did at times regard all such arguments as self-referential is strongly
suggested in some of my writings. 2 I see now, however, that my
thinking on this issue has not been altogether clear or unambiguous.
For I have also published accounts of argumentum ad hominem which
it would be difficult to reconcile with the claim that all such arguments
are self-referential.
But let me begin with self-referential ad hominems. Such
arguments claim that there is a self-referential inconsistency between
the utterance and the presuppositions of a view. An example, to cite
Bartlett's citation of me, is "the statement 'life is a dream,' which is
meaningful only if it is presupposed that a meaningful distinction
between dreams and waking states is possible. But this possibility is
precisely what is denied by the statement.',3 My reason for claiming
that such argumenta ad hominem are valid was that if it can be shown
that a proposition implies a contradiction, there cannot be any question
that that proposition is false. Notice that validity here does not depend
on the use of self-reference in the argument; it depends only on the
existence of a contradiction (which mayor may not be self-referential).
For many years I have had doubts about the possibility of
conclusively establishing the existence of a contradiction in
philosophical discourse.4 But that is not the reason I have fudged about
self-reference. Indeed, it is easiest to fudge if it be assumed that
philosophical contradictions are clearly identifiable. For it is when we
are faced with what we are confident is a contradiction that it makes the
most sense to characterize what faces us as either self-referential or not.
An example of a contradiction which is not self-referential
appears in my discussion of Berkely's argument to the effect that it is
hopeless to appeal to external bodies in the attempt to explain the causes
of our ideas. "For though we give the materialists their external
bodies, they by their own confession are none the nearer to knowing
how our ideas are produced; since they own themselves unable to
Johnstone Argumentum ad Hominem with and without Self-reference

comprehend in what manner body can act upon spirit."s

Berkeley is claiming that there is a contradiction between holding that

external bodies cause our ideas and holding that it is incomprehensible
how such bodies can act upon our minds. He is offering the materialists
a choice: either give up the thesis that external bodies cause our ideas
or else explain how they do. I cannot now see how self-reference is in
any way involved in this argument, even though I concluded my
discussion by saying it was involved. 6 What is essential is simply the
inconsistency of materialism on this score.
It is important to understand that even though Berkeley's
argument is valid, it does not compel the materialist to join the idealist
camp. For in order to eliminate the contradiction, he need not give up
the claim that external bodies cause our ideas. He can instead give up the
thesis that it is incomprehensible how external bodies cause our ideas.
If he does, Berkeley's argument has forced him to revise his position;
he can no longer maintain his materialism in its original form. It seems
likely that Berkeley can now find another argumentum ad hominem to
use against this new version of materialism.
There is a similar argument in Plato's Phaedo which, however,
differs from Berkeley's argument in an important respect. When
Socrates tells Simmias to "see which of the theories you prefer: that
learning is recollection, or that the soul is attunement,"7 he assumes that
this injunction is warranted by the inconsistency of the two theories.
But in this case Simmias has no real choice. For in 75-76 Simmias has
already been forced to admit that learning is recollection. To withdraw
this thesis now would accordingly plunge him into another
This Phaedo argument is like what Gregory Vlastos calls "The
standard Socratic elenchus,"s except that the premise "Learning is
recollection" is logically secured, while in the standard elenchus the
corresponding premise is not; it is, as Vlastos says, "ad hoc." But in
either case the refutation is ad hominem. The difference is only that in
one case the interlocutor is forced to accept a specific proposition,
while in the other he is compelled to revise his position in one way or
Professor Bartlett's discussion of self-reference provides a solid
rock on which to build. I wish I could build my theory of philosophical
argumentation entirely upon that rock. Unfortunately, however, my

Johnstone Argumentum ad Hominem with and without Self-reference

theory extends beyond this base, and parts of it are doomed to rest upon
nothing more substantial than the sand of contradiction without

Henry W. Johnstone, Jr.

Department of Philosophy
The Pennsylvania State University
University Park, Pennsylvania

Johnstone Argumentum ad Hominem with and without Self-reference

1. Steven J. Bartlett, Methodology and Science, 13 (1980)

2. Primarily in "Self-Refutation and Validity," The Monist 48

(1964) 467-485, reprinted in Validity and Rhetoric in
Philosophical Argument, An Outlook in Transition, The
Dialogue Press of Man and World, pp. 29-38, and in
"Philosophy and Argumentum ad Hominem,"The Journal of
Philosophy 49 (1952) 489-498 (reprinted as pp. 5-12 of
Validity and Rhetoric in Philosophical Argument). See also
Philosophy and Argument, The Pennsylvania State
University Press, 1959, especially pp. 69, 75.

3. Bartlett, op. cit., p. 149.

4. An early expression of these doubts is to be found in

"Argumentation and Inconsistency," Revue Internationale de
Philosophie 15 (1961) 353-365, reprinted in Validity and
Rhetoric in Philosophical Argument, pp. 45-52. The doubts
over the years are summarized in the "Epilogue" to that book,
especially on pp. 134-135.

5. Philosophy and Argument, p. 67. The Berkeley quotation is

from Principles, Sec. 19.

6. Philosophy and Argument, p. 69.

7. 92C8-10.

8. See "The Socratic Elenchus," abstract of a paper presented in

an APA Symposium, December 29, 1982, The Journal of
Philosophy 79 (1982) 711-714.

Douglas Odegard

The Irreflexivity of Knowledge

1. Classifying knowledge as a relation.

Treating knowing-that as a relation can misleadingly suggest

that knowing something is like kicking something, i.e. that knowing is a
simple two-termed relation between knower and known. Also, it can
falsely imply that when we claim knowledge, or attribute knowledge to
someone else, we are simply describing how someone stands in relation
to the world. Both pitfalls can be avoided by making it clear that to
claim or attribute knowledge is, in part, to assess how what is known
stands in relation to the evidence, how the knower's evidence stands in
relation to what is known, and how the knower's evidence stands in
relation to the rest of the evidence. If we remember this, we shall not
be tempted to treat knowing as a two-termed relation and we shall avoid
treating a knowledge claim as a mere description.
Is knowing-that a transitive, symmetrical, or reflexive relation?
If these technical terms are understood in their familiar senses, the
question is absurd. For instance, knowledge obviously cannot be
literally transitive, since a relation R is literally transitive only if 'a is R
to b, and b is R to c' implies 'a is R to c'; and if a knows b, then b is a
fact, or a truth, and hence not the sort of thing to know something else
in tum.
But perhaps the meaning of the terms can be altered to allow a
sensible question to arise. Instead of understanding 'transitive' in
accordance with the formula

(x)(y)(z)Kxy . Kyz) :J Kxz)

let us understand it in accordance with

(x)(y)(z)(Kx : Kyz :J Kxz)

where 'Kx:Kyz' means 'x knows that y knows that z'. If knowledge is
transitive in this altered sense, we merely have to say that anyone who
knows that another person knows that p must also know that p.
Although such a principle may be false, at least it is not absurd.
If the meaning of 'symmetry' is similarly altered, to say that
Odegard The Irreflexivity of Knowledge

knowledge is symmetrical would be to say

If one person knows that another person knows that p,

the second person knows that the first person knows that p.

But, although this remark is coherent, it seems pretty clearly false.

Whether I know that you know something seems clearly independent of
whether you know that I know it. Thus, although we can create a sense
in which it is not absurd to ask whether knowledge is symmetrical, the
answer is pretty obvious from the start that knowledge is
The term 'reflexive' sometimes means slightly different things
in ordinary contexts. For instance, calling a relation reflexive can
imply that anything that has the relation must have the relation to itself
and only to itself, as in the case of 'being in exactly the same place as'.
In symbols,

(x)(y) (Rxy ::J (x = y

Or, calling a relation reflexive can imply simply that anything that has
the relation must have the relation to itself, leaving open the possibility
of its having the relation to other things as well, as in the case of 'being
the same colour as'. In symbols,

(x)(y) (Rxy ::J Rxx)

Analogues of both concepts are available in the case of

knowledge. We might say that knowledge is reflexive in the sense that a
person knows something only if the person's knowing it is known to
that person and only to that person. That is,

(x) (y)(w) [(Kxy ::J (3 z)(z=Kxy (Kwz::J (w = x]

Or, we might say simply that a person knows something only if the
person's knowing it is known to that person, leaving open the
possibility of others knowing of the knowledge as well. That is,

(x)(y) (Kxy ::J (3 z)(Kxz . (z = Kxy

Since it is unlikely that someone's knowledge can be known only to that

Odegard The Irreflexivity of Knowledge

person, the correct answer to the question of whether knowledge is

reflexive in the first sen&e is pretty clearly negative. Let us therefore
use 'reflexive' in the second sense.
We thus have two serious questions before us: 'Is knowledge
transitive?' and 'Is knowledge reflexive?'. The second is more
important than the first and I shall begin with it.

2. The Concept of Knowledge

I shall argue that, in the strongest sense in which we can
realistically expect to acquire knowledge, knowledge is ineflexive; i.e.
that when we know something we never know that we know it. The
issue is complicated, however, by the fact that the word 'knowledge'
has a shifting sense in philosophical contexts and that in at least one very
strong sense knowledge is reflexive, whereas in other, less demanding
senses it is not irreflexive and may also be reflexive.
The strong reflexive sense is employed by Malcolm, following
suggestions in Prichard and Wittgenstein. 1 He uses 'knowledge' so
strongly that, by definition, anyone who knows something must be
entitled to dismiss even the remotest chance of something's counting as
successful evidence against what is known. If a knower by definition
must be in such a strong position, a knower must also be in a position to
know of the knowledge, in which case there is very little to keep the
knower from having the higher-order knowledge. Granted,
knowledge in this sense also requires some sort of psychological
attitude on the knower's part; and there is some room to argue that
someone might know that p, hence have the necessary attitude toward
p, without having the corresponding attitude toward 'I know that p'.
But even if this objection is sound, Malcolm can at least insist on a
limited form of reflexivity, by insisting that anyone who both knows
something in this very strong sense and has the necessary higher-order
attitude must also have the higher-order knowledge. If 'B' stands for
the psychological attitude and 'Bx:Kxy' means 'x has the psychological
attitude toward Kxy', then in symbols,

(x)(y) Kxy Bx:Kxy) :::> Kx:Kxy)

Although knowledge in this sense is reflexive, or almost

reflexive, I do not think that we can ever expect to acquire knowledge
in such a strong sense. The reason for this will become clearer later
when I discuss the way in which we are justified in being sure of things.

Odegard The Irreflexivity of Knowledge

Very briefly, it is because we have to rely on the track record of past

evidence when we anticipate what evidence the future will bring, and
there is no kind of evidence that has had a perfect track record.
Granted, this track-record argument has the form 'Since some A's have
been F, this A may be F'; and as Malcolm points out, not all arguments
of this form are sound. But, as even Malcolm concedes, some
arguments of this form do work. And the present argument is one of
them, precisely because we have no other way of estimating what future
evidence will be like except by relying on the past record of given types
of evidence. Hence, the argument manages to create a universal doubt
as to whether our evidence is as fool-proof as it unquestionably has to
be if we are to have knowledge in such a strong sense. The
track-record argument therefore establishes that we never have such
At the other end of the spectrum, Malcolm introduces a sense of
'knowledge' in which a correct conviction supported by some reason is
sufficient for knowledge. 2 Knowledge in this sense is not irreflexive,
since we are sometimes correctly convinced that we have knowledge
and have reasons to support our conviction; accordingly, we in this
sense know of our knowledge. Indeed, perhaps we always satisfy such
weak requirements for knowing that we know, in which case
knowledge in this sense is reflexive.
Similar comments apply to many senses that can be explained in
terms of justified true belief. When we know, we are often justified in
believing that we know, indeed may have to be justified in holding the
second-order belief, in which case very little stands in the way of our
having second-order knowledge. For instance, when Keith Lehrer
defines 'knowledge' to mean 'true belief that is completely justified
independently of any false statement' and maintains that knowledge is
reflexive, he probably is not far off the mark. 3 Similarly, when
Roderick Chisholm defines knowledge in terms of truth, acceptance,
and nondefective evidentness, and then'offers the principle

If S considers the proposition that S knows that p,

and if S does know that p,
then S knows that S knows that p.

he too is probably close to the truth. 4

Although knowledge in senses taken from Malcolm, Lehrer, and
Chisholm is reflexive, or at least nearly so, it is not knowledge in the
strongest sense in which we can realistically expect to acquire

Odegard The Irreflexivity of Knowledge

knowledge. In each case, the author deliberately avoids making

knowledge require certainty and thereby makes the sense weaker than it
has to be in order to avoid a universal scepticism. We can in fact
acquire knowledge even if, by definition, what is known must be
certain and even if, by definition, we must be justified in claiming that
it is certain. Accordingly, we can expect to acquire knowledge in a
sense that is stronger than any of the senses in Malcolm, Lehrer, or
Chisholm, although it is not as strong as Malcolm's very strong sense.
And in this intermediate sense, knowledge is irreflexive. But what
exactly is this intermediate sense?

It is a sense governed by five conditions.

(i) What is known to be the case must be certain.

(ii) A knower must be sure of what is known, in the sense that a

knower must be prepared to claim of what is known that it is

(iii) A knower must be justified in being sure.

(iv) Whatever justifies a knower in being sure would have justified

the same attitude regardless of what further evidence had been
added from the evidence that the knower does not possess.

(v) Any proposition essential to a knower's being justified must be


Let us assume that the five conditions are suitably co-ordinated;

e.g. that whatever makes the knower sure in condition (ii) is what
justifies the knower's sureness in (iii). We can thereby avoid having to
allow as knowledge a case in which our being sure of something for
which we have a justification is merely fortuitious. For example, we do
not have to count as knowledge a case in which we are sure because of a
memory that, unknown to us, is unreliable and we are nonetheless
justified in being sure because of testimony that we mistakenly think is
Conditions (i) and (iii) are the important ones here, since they
serve to distinguish the present sense from weaker senses in Malcolm,
Lehrer, and Chisholm. Condition (iii) is especially critical, since,
although it can be satisfied in cases of first-order certainty claims, it

Odegard The Irreflexivity of Knowledge

cannot be satisfied in the case of higher-order claims. Hence it is the

condition that makes knowledge irreflexive. But since justification in
(iii) is a matter of justifying certainty claims, we cannot understand (iii)
unless we understand (i). Hence I shall start with (i). As for the other
conditions, they have no bearing on the question of whether knowledge
is reflexive and I shall not bother to discuss them in the present context.

3. Certainty
By definition, something is certain only if there is no sound
evidence against it at any time; i.e. only if there never has been and
never will be any sound evidence against it. The expression 'evidence
against p' will include both evidence that p is false and evidence that in
some way undermines the evidence for p. The expression 'sound
evidence' means evidence that is not completely discredited or
overridden by further evidence in the universe, past, present, or future.
The expression 'sound evidence against p' includes evidence that makes
p to some degree doubtful, as well as evidence that makes p probably
Something certain can be opposed by evidence if the latter is
overridden by further evidence. For example, a defendant in a murder
trial can certainly have committed the crime in spite of the arguments
produced by the defence, if those arguments are completely overridden
by the prosecutor's case. On the other hand, if defence arguments make
the charge against the defendant even the least bit doubtful, the
defendant is technically innocent. Indeed, in the present sense of
'certain' it is not certain that the defendant has committed the crime so
long as there is some sound evidence to the contrary, even if that
evidence is not possessed by defence counselor the court.
Certainty is therefore a matter of what the evidence throughout
the universe is like. By contrast, justification is more a matter of what
someone's evidence at a given time is like. As a result, we can be
justified in claiming certainty even if the claim is incorrect. For
example, a murder-trial jury can be justified in finding the defendant
guilty, i.e. in being sure that the defendant did it, even if there is
evidence not possessed by the court that discredits the prosecutor's case
and establishes that the defendant did not certainly do it. Conversely, a
defendant can certainly have committed the crime even though a jury
fails to be justified in claiming as much because it fails to have enough
of the evidence. Certainty and justification are therefore mutually
Although certainty requires the complete absence of sound

Odegard The Irreflexivity of Knowledge

counterevidence, it does not require the impossibility of such evidence.

Although a prosecutor ha~ to deny the existence of sound evidence
indicating that the defendant did not commit the crime, the prosecutor
does not have to insist that such evidence cannot ever tum up. The
claim is simply that none ever will tum up, not that none ever could
tum up. Otherwise the prosecutor would have to build so strong a case
that a retrial in the light of new evidence would be completely pointless.
And this is a larger burden than even prosecutors in murder trials have
to shoulder.
Since certainty is a matter of the way evidence (tenselessly) is,
certainty should not be confused with indubitability, if'x is indubitable'
means 'There is no possible reason to doubt x'. Although claiming
certainty implies that there is no good ground for doubt anywhere in
the actual universe, past, present, or future, it does not imply that there
is no such ground in any possible universe. As a result, defining
knowledge in terms of certainty in the present sense makes a knowledge
claim less demanding than a claim to have knowledge in Malcolm's
strong sense. We have a Malcolmian knowledge of something only if
we can rule out the slightest possibility of there being a good ground
for doubting what we claim to know. For example, if we claim
Malcolmian knowledge that there is an ink-bottle in front of us, we
have to be entitled to say, "There is nothing whatever that could happen
in the next moment or the next year that would by me be called
evidence that there is not an ink-bottle here now"S - where 'could
happen' is deliberately used instead of 'will happen'. By claiming
knowledge in the present sense, on the other hand, we need simply be
justified in predicting that nothing of the sort will happen, while
leaving open the possibility that the prediction is mistaken.
Yet knowledge in the present sense is still stronger than
knowledge in familiar weaker senses of the word. To claim knowledge
in a weak sense is not to predict that there will never be any sound
evidence against the proposition known. As a result, a weak claim
avoids implying that the proposition known must be true. It settles for
some kind of probability. Indeed, some weak claims imply that the
proposition known must always be possibly false (even in the case of
propositions that are necessarily true in a logician's sense of
'necessarily'). Hence some weak claims imply that there is always some
good reason to be doubtful of a proposition for which we have
supporting evidence. In short, some weak claims not only refrain from
implying certainty. They actually imply that nothing is certain.
Claiming knowledge in the present sense therefore avoids claiming a
Malcolmian indubitability while going further than claims that either

Odegard The Irreflexivity of Knowledge

refrain from, or challenge, certainty claims.

4. Justifying Certainty Claims

But are we ever justified in claiming certainty, in the sense in

which I have defined it? By definition, we know something only if we
are justified in being sure of it; and if justification is always beyond our
reach, we never have knowledge. But if we never have knowledge in
the present sense, I still have not found the strongest sense in which we
can realistically expect to -acquire knowledge and the question 'Is
knowledge in the present sense reflexive?' becomes irrelevant.
Justification is roughly a matter of two things, one concerning
what our relevant evidence at a given time is like, the other concerning
what our evidence at the time indicates about the rest of the evidence.
For example, whether I am currently justified in being sure that the UK
issued its first postage stamp in 1840 depends partly on what my
current evidence from reading stamp catalogues and journals is like and
partly on what my current evidence indicates about the remaining
relevant evidence. I am justified only if I have no awkward evidence
that 1840 was not the year of first issue and only if, given my current
evidence, there probably is no such evidence in the remaining evidence.
Accordingly, let us say that we are justified (at time t) in being
sure that p only if (at t).

(a) any evidence against p that we possess is overridden by the rest

of our evidence, and

(b) given our evidence, probably any evidence against p that we do

not possess is overridden by further evidence.

Granted, for some purposes, conditions (a) and (b) may have to be
refined. For example, if justification is possible at all, I can be justified
in being sure that people cannot survive being sawn in half even though
I watch a magician apparently saw a live person in half and am unable
to explain the trick. Possibly (a) needs to be reformulated to
accommodate this case, with corresponding changes written into (b).
Such refinements have no bearing on the present issue, however, since
they leave untouched the question of whether we are ever actually
justifed in being sure of anything.
We could, of course, choose to use 'justification' in a way that is
not governed by conditions (a) and (b). For instance, we could decide
to allow someone to be justified in being sure of each of two logically

Odegard The Jrreflexivity of Knowledge

inconsistent propositions, in the way that we sometimes commend a

philosopher like John Locke for adopting inconsistent views. But this
would not be a wise decision given our present goal. The aim here is to
identify the most demanding sense in which we can have knowledge,
and then to see whether we can have higher-order knowledge in that
sense. We cannot achieve this goal if we allow someone to know
something while also being justified in being sure of the contrary.
Consequently, conditions (a) and (b) should be preserved.
Can our evidence ever satisfy conditions (a) and (b)? In
particular, can it ever make it probable that the evidence we do not
possess contains no sound evidence against a given proposition? I think
that it can, in the following type of situation.
Consider the proposition 'I have two hands'. What I see and feel
affords evidence for this proposition, and I have no evidence against it
that is not overridden or discredited by my remaining evidence. Hence
I satisfy condition (a). Moreover, my evidence in this case is of a kind,
K, such that I remember that when I have had K-type evidence for a
proposition, I have usually not subsequently acquired sound evidence
against that proposition. For example, my evidence that I have two
hands involves clear and mutually corroborative sense impressions; it is
conjoined with a clear and firm memory that I have always had two
hands and that nothing has happened to them; it is not just the product of
a desire to think of myself as having two hands; and it is conjoined with
the memory that I have satisfactorily answered philosophical objections
to asserting the existence of objects like hands. Although I cannot
specify features that are individually necessary and jointly sufficient
for evidence's being of type K-perhaps because there is no such set to
specify-I can give examples of characteristic features. And I
remember that evidence of that type has generally not been followed by
sound counterevidence, just as I remember that dogs generally chase
after cats even though I cannot list a set of individually necessary and
jointly sufficient features of something's being a dog.
Let us call evidence of type K 'solid evidence'. I hesitate to use
this expression because in ordinary contexts to say that someone's
evidence is solid can imply that such evidence will not be augmented by
sound counterevidence. And the question here is, in part, whether we
are ever entitled to claim that our evidence has such a standing.
Consequently, the expression must be stripped of such an implication.
It must be understood in a way that leaves room to ask whether a given
piece of solid evidence will be augmented by sound counterevidence. If
we remember this, there should be no conceptual difficulties.
Let us also say that evidence that is not usually followed by

Odegard The Irreflexivity of Knowledge

sound counterevidence is reliable evidence. I am in a situation, then, in

which I can justifiably say that my evidence that I have two hands is
solid and that I remember that solid evidence has been reliable in my
case. Moreover, in my present situation, I have no evidence against
thinking that solid evidence has been reliable in my case. Thus, I am in
a situation where I have evidence, afforded by my memory, that solid
evidence has been reliable in my case and I have no contrary evidence.
Consequently, in accordance with the principle

If e contains evidence that p and no evidence to the contrary,

then, relative to e, it is probable that p.

given my evidence, it is probable that solid evidence has been reliable in

my case.
Now, the probability that solid evidence has been reliable in my
case is evidence that solid evidence is reliable in every case. In other
words, if it is probable that my solid evidence has not usually been
followed by sound counterevidence, there is evidence that solid
evidence in general is not usually followed by sound counterevidence.
We can take this step in accordance with the principle

If it is probable that all given Fls are GIS,

there is evidence that all Fls are GIS.

provided we remember that evidence here is quite defeasible. (For

example, although the fact that a given set of swans are all white is
evidence that all swans are white, this evidence is easily defeated by
observations of black swans.)
I therefore have evidence that solid evidence is always reliable.
Moreover, I have no evidence to the contrary. Consequently, in
accordance with the above principles, relative to my evidence,
probably solid evidence is reliable.
Since my evidence that I have two hands is solid and since,
relative to my evidence, probably solid evidence is reliable, then,
relative to my evidence, probably my evidence that I have two hands is
of a reliable type. In other words, probably my evidence that I have
two hands is of a type that is not usually followed by sound
counterevidence. In that case, relative to my evidence, it is probable
that my evidence that I have two hands probably will not be followed by
sound counterevidence. And I can see no reason why the two
probability operators cannot be collapsed in this case, yielding the
conclusion that, relative to my evidence, probably my evidence that I

Odegard The Irreflexivity of Knowledge

have two hands will not be followed by sound counterevidence.

The probability that my evidence that 1 have two hands will not
be followed by sound counterevidence is evidence that there is no sound
counterevidence in the evidence that 1 do not possess, whether future,
present, or past. Moreover, 1 have no evidence that there is sound
counterevidence in any of the evidence that 1 do not possess.
Consequently, relative to my evidence, probably there is no sound
evidence against thinking that 1 have two hands in any of the evidence
that 1 do not possess. In that case, my evidence that 1 have two hands

(a) contains no contrary evidence that is not overridden by

remaining evidence, and

(b) makes it probable that there is no contrary evidence in any of the

evidence that 1 do not possess.

1 am therefore justified in being sure that 1 have two hands.

5. The Irreflexivity of Knowledge

If we are to claim knowledge in the current strong sense, and in

particular if we are to maintain that certainty claims are justified by our
evidence, we have to be in a position to go through steps of the
preceding kind for each claim. And this cannot be done in the case of
higher-order knowledge.
The steps needed to justify claiming knowledge that 1 have two
hands merely establish that, relative to my evidence, probably there is
no sound evidence against 'I have two hands'. They do not establish the
certainty of there being no sound counterevidence. Yet in order to
know that 1 know that 1 have two hands, 1 have to be justified in
claiming that it is certain that 'I have two hands' is certain; hence 1 have
to be justified in claiming that there certainly is no sound evidence
against 'I have two hands'. And 1 have no basis for this claim.
Indeed, when 1 claim that there probably is no sound evidence
against 'I have two hands', 1 do so because my evidence that 1 have two
hands is of a type that is usually not followed by sound counterevidence,
not because it is of a type that is never followed by sound
counterevidence. 1 am not familiar with any type of evidence that has a
flawless track record. Consequently, my evidence contains frequency
evidence that supports conceding the possibility of sound evidence
against 'I have two hands', in which case it contains evidence for saying
that 'It is certain that I have two hands' is not certain. Therefore 1 do

Odegard The Irreflexivity of Knowledge

not have solid evidence for the certainty claim and am not justified in
claiming that it is certain. Rather, I am justified in claiming that it is
uncertain, since my evidence establishes the possibility of the claim's
being mistaken.
The existence of seductive sceptical arguments against certainty
supports the same conclusion. Although arguments against the
existence of objects like hands currently hold no attraction, arguments
that disclaim certainty offer an option that is very much alive, even
though such arguments are outweighed by considerations in favour of
being sure. Thus, in this respect as well, my evidence contains evidence
that indicates that 'It is certain that I have two hands' is uncertain and I
am not justified in being sure of the certainty claim. Since the same
point holds for any other certainty claim, knowledge is irreflexive.
Now, the conclusion that knowledge is irreflexive implies
higher-order scepticism, since it implies that we never know that we
know, that we never know that we know that we know, and so on. And
some may think that higher-order scepticism eventually leads back to
first-order scepticism. If they are right, my current project is a failure.
I am trying to identify a strong sense of 'knowledge' in which, on the
one hand, knowledge is a realistic possibility and, on the other hand, we
never know that we have knowledge. I think that I have defined a
strong sense in which we never have higher-order knowledge. But if
we never have first-order knowledge in this sense either, I have failed
to pick out a realistic possibility.
I doubt that attempts to extract first-order scepticism from
higher-order scepticism will succeed in this case. Granted, I am
conceding that there is always a possibility that sound evidence against a
first-order proposition exists, no matter what that proposition is. But
this is not to concede that there always ia sound evidence against such a
proposition. Consequently, I can make the concession and consistently
insist, for example, that there is no sound evidence against 'I have two
hands'. I can concede that the first-order proposition may not be
certain while still insisting that it is certain. Whereas the track record
of my evidence that I have two hands leaves open the possibility that
new awkward evidence will tum up, it does not leave open the
possibility that I do not have two hands. Using that track record as a
reason for wondering whether I have two hands would be like using the
fact that some philosophers have been female as a reason for wondering
whether Leibniz was female.
Granted too, the way I have justified claiming certainty makes a
certainty claim probabilistic. It therefore puts certainty claims within
the scope of the so-called "lottery paradox". For a sufficiently large

Odegard The Irreflexivity of Knowledge

number of conjuncts, a conjunction with individually probable

conjuncts is itself improbable, since the conjunction's value is the
product of multiplying the individual values. And the number of
first-order things of which I am justifiably sure is large enough that
they probably are not all certain. As a result, although I am justified in
being sure of each of them, I am also justified in claiming that at least
one of them is uncertain. Hence my position is "inconsistent" in the
sense that it endorses claims not all of which can be correct. But this
type of inconsistency does not bother me. It does not involve the
outright contradiction of both being justified in rejecting a conjunction
and being justified in accepting it, since being justified in accepting each
conjunct of a conjunction does not entail being justified in accepting the
conjunction. I am simply justified in making each of several claims that,
taken jointly, form a conjunction the contrary of which I am justified in
claiming. There is nothing impossible, or reprehensible, about that.6
Nor am I settling for a probabilistic claim when I could get
more. I am not like someone who wants to find out which horse won
the Kentucky Derby in 1980 and examines the pre-race odds instead of
reading reports of the result. Since claiming certainty is unavoidably a
matter of predicting what future evidence will be like, I have to rely on
the odds concerning evidence in this case. Nothing else is available to a
finite mind. Consequently, living with the attendant inconsistency is
not irrational, given that the goal is to maximize correct certainty
claims while minimizing incorrect ones.
I think that I can say, then, that in the strongest sense in which
knowledge is a realistic possibility, knowledge is irreflexive.

6. Klein's Objections
Most philosophers who deny the irreflexivity of knowledge use
'knowledge' in either a stronger or a weaker sense than I am using it
here. Since I concede that knowledge is not irreflexive in such senses, I
am not disagreeing with as many positions as one might think. I am,
however, at genuine odds with the position recently adoped by Peter
Klein.7 He thinks that we sometimes know that we have knowledge,
and he uses 'knowledge' in much the same sense as it is used here, since
knowledge in his sense requires certainty and certainty is a matter of
the absence of genuine defeating evidence. Thus, in claiming to know
that he knows, he implicitly claims that it is certain that there is no
genuine defeating evidence, which is to make a claim that I think is

Odegard The Irreflexivity of Knowledge

Klein does not profess to be able to produce an argument to

show that there is no genuine defeating evidence in a given case. But he
does think that he can still show that there is no such evidence by doing
three things: first, explaining carefully what the expression 'genuine
defeating evidence' means; secondly, assembling the considerations
that, in the given case, are most plausibly thought to be genuine
defeaters; and thirdly, establishing that such candidates are either not
defeaters or not genuine. 8 Now, I agree that the procedure he
recommends can help to justify claiming the absence of genuine
defeaters. But I do not think that it can justify the claim that there
certainly are no genuine defeaters. Establishing that the currently most
promising defeaters are all spurious merely helps to establish the
probability that any others are spurious as well. The possibility of a
surprise always remains, since there have been occasions when the type
of procedure he recommends has been followed and yet genuine
defeaters have subsequently turned up (e.g. consider the rigorous
procedures adopted by the engineers who built the Titanic). As long as
the procedure has occasionally ended in defeat, the possibility of future
defeat cannot be eliminated.
Klein might be inclined to reject this line of reasoning as an
aberration. He says "that a person would either be feigning disbelief or
entranced by scepticism who, understanding what I mean when I
asserted that there are good reasons for believing that some contingent
propositions are absolutely certain, claimed that he/she did not have
those good reasons in situations in which he/she had evidence [that we
normally take to be optimal] and no evidence for the existence of a
genuine initiating defeater.,,9 But I think that his appraisal of the
situation is unfair to scepticism. Some sceptics who reject first-order
certainty claims understand the meaning of the expression 'good
reasons for claiming certainty' and yet offer serious arguments for
thinking that the expression is never applicable. They are not all simply
"entranced" by such arguments. Granted, none of the arguments is
completely successful and we are in fact sometimes justified in claiming
certainty; hence we do have good reasons for believing that some
contingent propositions are certain. But the arguments are still
sufficiently worthwhile that they may tum out to be vindicated by
future inquiry. They still offer a live option in that sense.
Consequently, although their existence does not make any first-order
propositions doubtful, it does cast some degree of doubt on certainty
claims. The very existence of serious philosophical arguments against
claiming certainty constitutes evidence that inakes certainty claims

Odegard The Irreflexivity of Knowledge

themselves less than certain.

Klein overlooks this point when he constructs his case for
claiming higher-order knowledge. He seems to concede that we know
that we have knowledge only if anything that, if genuine, would cast
some degree of doubt on our knowledge claim is in fact misleading.
And he defends the view that we sometimes satisfy this condition by
arguing for four propositions. 10

(1) There is no good reason to believe that we never have


(2) There are good reasons to believe that we sometimes have


(3) There is a good argument for claiming that we sometimes have


(4) There are better reasons for believing that we sometimes have
knowledge than there are for believing that we never have it.

Propositions (1)-(4) are all true and hence first-order scepticism is

false. But (1) need not mean that the reasons sceptics offer for
believing that we never have knowledge are so bad that they fail to cast
the least degree of doubt on knowledge claims. Similarly, (4) need not
mean that the reasons for believing that we sometimes have knowledge
are so superior to the sceptic's reasons that the latter are completely
overridden. It is enough to say that the reasons for claiming knowledge
outweigh those against. And I think that this more moderate judgment
better reflects the current relative worth of the two positions. Hence
we do not know that we sometimes have knowledge, even though
(1)-(4) are true.
Moreover, among the reasons for believing that we sometimes
have knowledge are premises describing the track record of the type of
evidence we rely on when we claim knowledge. And those
descriptions, if accurate, have to concede that no type of evidence has a
flawless track record, that refutations sometimes occur no matter how
rigorous the investigative procedures have been. Consequently, our
very reasons for claiming knowledge imply that our evidence justifies
claiming knowledge only if it also justifies claiming the possibility that
we do not have knowledge. In short, we are justified in claiming
firsf-order knowledge only if we are justified in disclaiming
higher-order knowledge.

Odegard The Irreflexivity of Knowledge

Thus, the irreflexivity of knowledge survives Klein's criticisms.

7. The Transitivity of Knowledge

If a knower must satisfy some kind of psychological condition,

knowledge is not transitive without further qualification. Transitivity
means that anyone who knows that S knows that p also knows that p;
and logically, someone can have the necessary attitude toward'S knows
that p' without also having it toward p. Thus, we have to qualify
transitivity by saying that anyone who knows that S knows that p and
has the necessary attitude toward p also knows that p. In this sense,
knowledge is transitive.
Yet because we never have higher-order knowledge, transivity
becomes a matter of purely academic concern. Arguments that show
that we never have knowledge of our own knowledge equally show that
we never have knowledge of anyone else's knowledge either. Such
arguments establish that we are never justified in being sure of
certainly claims; and third-person knowledge claims imply certainty
claims just as much as first-person claims. Knowledge is therefore
transitive only in the hypothetical sense that, if one person were to
know that a second person knows that p, and the first person had the
necessary attitudes toward p, the first person would also have to know
that p.
The knowledge relationship is therefore best classified as
nonsymmetrical, irreflexive, and, in a rather academic mode,

8. Postscript

Although the question of transitivity is somewhat academic, the

question of irreflexivity can have very important consequences in
practice. By combining first-order knowledge claims with
higher-order ~cepticism, we put ourselves in a position to claim
first-order knowledge without having to lose our respect for our
opponents. As a result, we can avoid the dogmatism that is sometimes
associated with claiming knowledge. This has obvious practical
benefits across the whole range of human inquiry. And it is an
especially useful result in the area of morals, where the need for
preserving autonomy is often thought to be a fatal obstacle to claiming
knowledge. By acknowledging the irreflexivity of knowledge, we can

Odegard The Irreflexivity of Knowledge

assign our moral adversaries a full measure of intellectual respect,

without on that account having to withdraw the claim that we know that
their moral judgments are mistaken. Granted, we have to treat our own
knowledge claims with a degree of doubt (and of course abandon them
should they tum out to be false). But we do not have to transfer that
doubt to our moral judgments themselves. Thus, the doctrine of
irreflexivity can be an important weapon in the battle against both
moral scepticism and moral authoritarianism.

Douglas Odegard
Philosophy Department
University of Ontario
Guelph, Ontario

Odegard The Irreflexivity of Knowledge


1. See Norman Malcolm, Knowledge and Certainty, Prentice-Hall,

1963, pp. 58-72.

2. See Malcolm, Knowledge and Certainty.

3. See Keith Lehrer, Knowledge, Oxford University Press, 1974.

4. See Roderick Chisholm, Theory of Knowledge, 2/e,

Prentice-Hall, 1977, esp. pp. 110, 116.

5. Malcolm, Knowledge and Certainty, p.67.

6. Cpo Henry Kyburg, "Conjunctivitis", in Marshall Swain (ed.),

Induction, Acceptance and Rational Belief, Reidel, 1970, pp.

7. See Peter Klein, Certainty: A Refutation of Scepticism,

University of Minnesota Press, 1981.

8. See Klein, Certainty, p. 208.

9. Klein, Certainty, pp. 209-210.

10. See Klein, Certainty, pp. 212-215.

Part II. Formal Reflections
Frederic B. Fitch

Formalized Self-Reference
Elsewhere [1] I have argued that the concept of self-reference
has important applications in philosophy. These applications are a
reason for trying to handle the concept as rigorously and
formalistically as possible. In this connection, I have constructed a
system of logic cr [3] which can be shown to be consistent if the rule
for '11' is omitted, and within which at least some aspects of
self-reference can be represented. I will now sketch this logic and show
how it deals with the Russell Paradox and the Grelling Paradox.
Among the well-formed formulas (wffs) of the system are the
primitive symbols: '=' (equality), 'N' (the class of natural numbers, i.e.
0, 1,2, 3, ... ), ,~, (negation), 'v' (disjunction), 'E' (non-emptiness), '0'
(zero), 'T' (exponentiation), 'a' (successor as among natural numbers),
and 'B' (composition of functions and multiplication). (When a wff is
placed within single quotes, reference is to the wff itself.)
Starting with the primitive symbols, the class of wffs may be
defined by induction as follows:

(1) Every primitive symbol is a wff.

(2) If 'a' and 'b' are wffs, so is 'Cab)'. (Here lowercase italicized
letters are being used as variables of the metalanguage. When
one of these letters, or a combination of them, possibly together
with wffs, is placed within single quotes, reference is to the
various wffs got by replacing the italicized letters by wffs.)

We abbreviate 'Cab)' as 'ab', and 'ab)e)' as 'abc' and

'(ab)e)d)' as 'abed', and so on, so, for example, '(ab)e)de)j)'
as 'abe(def)'. Also, 'ab)e)' may be abbreviated as '[e a b]'
provided that spaces are inserted as indicated, so that, for example, '[e
= b]' would stand for '=b)e)' or, more concisely, for '=be'.
Outermost square brackets can be omitted (if spaces have been
correctly inserted), so that 'c = b' would stand for '=bc'.
For typographical reasons, three substitutions have been made in the
author's notation. (1) Instead of his inverted iota, Greek eta (1'\) has been
used. (2) In building class names, circumflex (") is placed following aletter
instead of over it. (3) A small raised circle is used instead of a square to
symbolize necessity. [Editors]
Fitch Formalized Self-Reference

Expressions of the form 'a = b', that is, '=ba' or ultimately,

'(=b)a)', are called "equations," Given any class 4 of equations,
we can form a system based on 4 (as ifthe equations 4 were axioms
for the system) and called C4, The class of provable wffs (theorems)
of system C4 is called P 4, and the class of refutable wffs (denials
of theorems) of system C4 is called R4, The rules for system C4,
defining membership in classes P 4 and R4, are as follows (omitting
the rule for '11' as being slightly defective and. not needed), The
parenthetical parts of these rules are rules for membership in R4 '
"Iff' means "if and only if,"
Rulefor'=', 'a = b', that is, '=ba', is inP4 (inR4)iff'a=b'
is one of the equations (not one of the equations) in 4,
Rule for 'N', 'Na'is in P 4(R4) iff at least one of (each of) the
equations 'A = 0', 'A = 0'0', 'A = 0'(0'0)', 'A = 0'(0'(0'0', and so
on, is in P4 (R4), The latter condition could also be expressed by
saying that at least one of (each of) the equations 'a = 0', 'a = 1', 'a =
2', 'a = 3', and so on, is in P 4 (R4),

Rule for '~', '~a' is in P4 (RA) iff 'a' is in R4 (P4),

Rule for 'v', 'a v b', that is, 'vba 'is in P~ (R~) iff at least
one of (each of) 'a' and 'b' is in P4 (R4),
Rule for 'E', 'Ea' is in P ~ (R4) iff for some 'b' (for every
'b') 'ab' is in P 4 (R4),
Rule for '0', 'Oab' is in P ~ (R4) iff 'b' is in P ~ (R4),
Rule for 'T', 'Tab' is in P4 (R4) iff 'ba' is in P4 (R4),
Rule for 'a', 'aabe' is in P 4 (R~) iff 'ab(be)' is in P 4 (R~),
Rule for 'B', 'Babe', also abbreviated as '[b B ale' or as
'[b a]e', is in P ~ (R4) iff 'a(be)' is in P ~ (R~),

Fitch Fonnalized Self-Reference

Supplementary rule. Each of the above rules can be further

extended by adding a finite sequence of italicized letters, standing for
wffs, to the wffbeing classified as in Pi,\ (Ri,\) and to the wff on which
its membership in P i,\ (Ri,\) depends, except for the rightmost
equation in the rule for '='. Thus the supplemented rules for '=', 'E',
and 'a' are as follows:
Supplemented rule for '='. '[a = b]c1C2 cn' is in Pi,\ (R1,\)

iff 'a =b' is one of the equations (not one of the equations) in 1,\.
Supplemented rule for 'E'. 'Ea C 1C2 .00 cn' is in P1,\ (R1,\) iff

for some 'b' (for every 'b'), 'ab C1C2o. co' is in P 1,\ (R1,\).

Supplemented rule for 'a'. 'aabcd1d 2o' do' is in P 1,\ (R1,\).

Similarly for the other supplemented rules. Another example:
Supplemented rule for '.... '. '....abl b2.oobn' is in P 1,\ (R1,\) iff

'ab1broo b n' is in P 1,\ (R.~).

Using the above rules, the following "principle of abstraction"
can be derived in a well-known way[2]: If '(00' a )' is any wff in [2]
which 'a' occurs one or more times, then a wff 'b' can be found
such that for every wff 'c', 'bc' is in P 1,\ (R1,\) iff' (00. c )' is in
P1,\(R1,\), where '( c )' is the result of replacing 'a' by 'c'
throughout '( a ..)'. Strictly speaking, for reasons given in 10.5 of
[2], 'a' cannot involve any of the operators 'B', 'T', 'W', '1', or 'K', but
this is always easily arranged. 'W' is definable as 'C(B(a(aOT)'
and 'C' is definable as 'BBT(BBT)(BBT)'. It can be shown that 'W'
satisfies the rule that 'Wab' is in P1,\ (RJ,\) iff 'abb' is in PJ,\ (RJ,\),
and 'C' the rule that 'Cabc' is in P1,\ (Ri,\) iff 'acb' is in Pi,\ (RJ,\).
'K' is definable as 'CO' and satisifies the rule that 'Kab' is in P i,\ (RJ,\ )
iff 'a' is in P i,\ (Ri,\). These rules may be supplemented in the usual
way. Roughly speaking, the method for finding 'b' of the principle of
abstraction is to move all the occurences of 'a' in ' ( a )' as far to
the right as possible by using appropriate rules, and then cause all those
Fitch Formalized Self-Reference

occurences to coalesce into one occurrence by using the rule for 'W'.
The complex expression then remaining to the left of that one
occurence is the required 'b'.,
Notice that the principle of abstraction in effect converts a
context into an operator that acts equivalently to the context but that is
"real" in the sense that operators are real, i.e., if we had variables in
our object language, our operators would be values for them. Thus we
have a kind of "Platonism of contexts" of which a "Platonism of
properties" would be merely a special case. Also, every context
represents a way that one entity is "in" another (or even within itself, as
in cases where we have 'a aa'). These different kinds of "inness"
throw a metaphysical light on the richness of different kinds of being,
and they go far beyond the narrow concept of mere spatio-temporal
inness. If the system or systems were further expanded by. the
introduction of empirical operators, as could be done, this richness of
different kinds of being would be vastly greater.
The wff 'b' of the principle of abstraction can be abbreviated as
'[xl ( x )' and is analogous to what Russell would write as 'x"( ...
x )' ("The class of things x such that ( x ..)"), or what Church
would write as 'AX ( x .)' ("the function that has the value ( x
) for the argument x"). Thus the principle of abstraction asserts that
'[xl (.. x .) a' is in P 4 (R4), iff' ( a ..)' is in P 4 (R4) so that
'[xl (.. x ..)' operates on 'a', equivalently to the context '(... )',
to give' ( a ... )'. As a further extension of this principle of
abstraction, it can be shown that '[xl [y] ( x y )ab' is in P4
(R4), iff' ( a ..b )' is in P 4 (R4), and similarly for '[xl [Yl
[zl (.. x
y . z )abc', and so on. These results hold regardless
of what class of equations the class 4 is chosen to be. Of course ,[x] [Yl
(... x ... y )' may be viewed as denoting the relation of x to y such
that ( x y ) is true, and '[x] ( x )' may be viewed as
denoting the class of things x such that ( x ..) is true, or as denoting
the property assigned to x by ( x ).
It is possible to choose 4 as r of [3] so that equalities have the
following properties:
(1) 'a a' is in r and in pr (for every wff 'a').
(7) If 'a b' is in r (and Pr) so is 'b a'.=
(3) If 'a = b' and 'b = c' are in r (and Pr), so is 'a = c'.
(4) If 'a b' is in r (and Pr), and if '(.. a ..)' is in pr, then '( ...

Fitch Fonnalized Self-Reference

b )' is in pr, so that equals may be substituted for equals in

every provable wff of system cr.
(5) A necessary and sufficient condition for an equation 'a b' to =
be in r, and hence in pr, is this: For all D, for all' CI" 'c2' ... ,

'c n ': 'a CIC r cn ' is in p~ (R~) iff 'b clc r ,. cn ' is in p~

(R~); and this must hold for every class ~ of equations that has
r as a subclass and that is closed for substitution of equals for
equals into equations. Details will be found in [3].

Using criterion (5) together with the principle of abstraction,

we get such equations as
,[xl (.. x .) a = ( a ...)',
'[xl [Yl (. x .. y )ab =(... a ... b ...)',
and so on. These may be called abstraction equations.

We tum now to the Russell paradox and the Grelling paradox as

illustrating kinds of self-reference that the system cr is able to handle.
Let the class of all classes that are not members of themselves be called
lithe Russell class. The Russell paradox purports to show that the

Russell class both is and is not a member of itself. (Because, if it is not a

member of itself, then it belongs to the Russell class, and so is a member
of itself, while if it is a member of itself, then it does not belong to the
Russell class, and so is not a member of itself. But either it is a member
of itself, or else it is not a member of itself. In either case, it both is a
member of itself and is not a member of itself.) This paradox,
however, can be shown to be avoided by every system p~, regardless
of which class of equations the class ~ is chosen to be. For suppose that
the Russell class is denoted by '[xl (-(xx' (lithe class of classes x such
that it is false that x is a member of x"), approximately to what Russell
might write as 'x 1\ (-(XEX'. Let' R' be an abbreviation for '[xl
(-(xx', so that'R' denotes the Russell class. Then ,[x] (-(xx R'
is in p~ (R~) iff '-eRR)' is in p~ (R~) by the principle of
abstraction, so that' RR' is in p~ (R~) iff '-(RR)' is in p~ (R~).
We even get' RR = -(RR)' as an abstraction equation in r and pro
But there is no contradiction here because 'RR' is not required to
satisfy excluded middle (that is, required to be such that either 'RR' or

Fitch Fonnalized Self-Reference

'.... (RR)' is in P4), so that both 'RR' and '~(RR)' may, and in fact
must, on pain of contradiction, fail to be in P 4 (and fail to be in R4).
Only equations, or wffs depending wholly on equations, satisfy
excluded middle. If 'RR' satisfied excluded middle in system C4,
then both'RR' and '.... (RR)' would be in P4, and system C4 would
be inconsistent.
But no matter how the class.:\. of equations is chosen, the system
C.:\. is consistent (is such that 'a' and, '....a' are not both provable for any
wff 'a'). This is easily seen as follows: First of all, the rule for '='
clearly does not by itself lead to a contradiction. Next, consider any of
the remaining rules, namely those for 'N', '.... ', 'y', 'E', '0', 'T', 'a',
and 'B', each of these expanded by use of the Supplementary Rule (and
of course omitting the rule for 11, which is not used), and suppose that a
contradiction is obtained by employing that rule. Then it can be shown
that a contradiction must previously have been obtained (as will be
indicated below), so that there cannot be afirst contradiction and so not
any at all. For example suppose that the rule for 'y' was used to obtain
the result that 'Yab' and '....(yab)' are both in P 4, so that, using the rule
for '..... ', the wff 'yab' must have already been shown to be in P 4 and
R4. But if 'yab' is shown to be in P 4, then 'a' or 'b' must have been
shown to be in P 4, and if 'yab' is shown to be in R4, then 'a' and 'b'
must have been shown to be in R.:\.. So either 'a' has been shown to be
in P4 and R4, or 'b' has been shown to be in P4 and R4. In other
words (using the rule for '..... '), 'a' and '-a' have been shown to be
provable in system C4, or 'b' and '-b' have been. In either case, a
contradiction has been previously established in system C4. A similar
result can be achieved in the case of each of the other rules. Thus C4
must be a consistent system, and we know "ahead" that no contradiction
can be established in it, either by way of the Russell class or by any
other method. Even the fact that the-equation 'R R = . . eRR)' is
provable in system cr and that equals may be substituted for equals in

Fitch Fonnalized Self-Reference

that system, does not give rise to a contradiction, since neither 'RR'
nor ....(RN.)' is provable in system cr.
Next it will be shown how the Grelling paradox may be handled
in the system pro It is possible to assign non-negative integers to wffs
("GOdel numbers" of wffs) in such a way that two wffs have the same
Godel number only by being equal. Thus if 'a' and 'b' have the same
Godel number, then 'a b' will be provable in system cr. A wff
actually has assigned to it all GOdel numbers of other wffs equal to it,
but this makes no trouble. We may think of GOdel numbers assigned to
a wff as being "names" of a property expressed by that wff. (Details of
the assignment will be given later.) In order to formulate the Grelling
paradox, we first assert that there is a property of being "heterological"
which a name has provided that the name names a property which the
name itself does not have. Thus the word "long" names a property that
the word "long" does not have, so that the word "long" is heterological.
Similarly the word "French" names a property the word "French" does
not have, so the word "French" is heterological. On the other hand, the
word "English" clearly is not heterological. Now we ask if the word
(or name) "heterological" is itself heterological. If it is heterological,
then by definition it is not heterological, while if it is not heterological,
then by definition it is heterological. This dilemma is avoided in the
system cr because the wff representing the statement that the name
"heterological" is heterological does not satisfy excluded middle in that
system, and so is neither provable (true) nor refutable (false). The
detailed definition of "heterological" using Godel numbers as "names"
of properties, will now be presented.
There are actually many different ways in which GOdel numbers
could be assigned. We choose one way which seems fairly convenient:
First, it is assumed that the nine primitive symbols '=', 'N','.... ', 'v',
'E',' 0', 'T', '0", and 'B' are successively assigned Godel numbers 1
through 9. Next, it is supposed that if wffs 'a' and 'b' have been
assigned Godel numbers n and m respectively, then the compound wff
'ab' is assigned the GOdel number 3n 5m For example, , 0'0' would be
assigned the GOdel number 38'5 6 Also, it is to be understood that if 'a
= b' is in pr, then every GOdel number assigned to 'a' is also assigned
to 'b', and vice versa, because equals may be substituted for equals in
theorems of system cr. Now we wish to define a wff 'G' such that 'a
G b' is in pr just in case that 'a' stands for a Godel number assigned to
the wff 'b', where '0' stands for 0, '0'0' stands for 1, '0'(0'0)' stands

Fitch Fonnalized Self-Reference

for 2, and so on. We first define a wff 'G', and later define 'G' as
'GG'. In defining 'G', use will be made of an operator '0' such that
'oa' may be interpreted as meaning that 'a' is necessarily true, and
where '0' is regarded as an abbreviation for '=(=00)', so that 'oa' acts
like 'a = [0 = 0)'. (In accordance with (5), it can be shown that 'a = [0
=0]' is in r, and hence in pr, iff, for every class '-' of equations that
has T' as a subclass and that is closed for substitutions of equals for
equals into equations: 'a' is in P,-,. It is easy to show that T' itself is
just such a class ''-'', so that if 'oa' is in r, so is 'a' itself.) The wff 'oa'
will be used in places where 'a' itself might be expected. The reason
for doing this is to guarantee that 'G' will be such that 'Gbctl, being
equivalent to some combination of equations, will, like equations,
satisfy excluded middle, and will do so for all 'b', 'c', and 'd'.
Otherwise, 'GGctl, like 'RR' in the case of the Russell paradox, might
not satisfy excluded middle.
'G' is now defined to be an abbreviation for '[g] [v] [u] [Nu &
[[[v = =] & [u = 1]] v [[v = N] & [u = 2]] v v [[v = B] & [u
= 9]] V 3x13x13Y13Yl(o[[XI gg YI] & [Xl gg Yl] & [v =YIY1]
& [u [3%1 5%1]]])]]', where '1', '2', ... stand for '00', '0(00)', ... ,
and where '[c' d]' stands for '[c B d],. Then, as an abstraction
equation in pr, we get 'GGba = [Na & [[[b = .... ] & [a = 1]] v [[b
= M] & [a = 2]]v v [[b = B] & [a = 9]] v V
3XI3x13YI3Yl(0[[XI GGYI & [Xl GG Yl] & [b =YIY1] & [a =
3%1'5%1]]])]]. Replacing 'GG' by its abbreviation 'G', and replacing
'Gba by the equivalent notation 'a G b', gives '[a G b] [Na & [[[b
= =] & [a = 1]] v [[b = N] & [a = 2]] v v [[b = B] & [a =
9]] v 3xI3x13YI3Yl(0[[XI G YI] & [Xl G Yl] & [b YIY1] & [a =
= [3.d 5x2 ]]])]]'. This equation, incidentally, illustrates a general
method for defining a wff that denotes a self-referential relation. The
wff 'G' has been defined in tenns of a condition (expressed by the right
side of the equation) that already involves 'G' itself, so that 'G' denotes
a self-referential relation.
The word "heterological" can now be represented by the wff 'H'
defined as an abbreviation for '[x]3y[[x G Y] & .... (yx)]', so that by

Fitch Fonnalized Self-Reference

the abstraction principle we get the result that 'Ha' is in pr (Rr) iff
'3y[[a G y] & ....(ya)]' is in pr (Rr). Suppose next that 'a' in fact
represents a GOdel number n of 'H' so that it is correct to regard 'n G
H' as in pr. Then if 'n G b' is in pr, 'b =H' must be in pr because,
as already noted, only "equal" wffs can have the same GOdel numbers in
system cr. Also, "by the abstraction principle we get the result that
'Hn' is in pr ( Rr ) iff '3y[[n G y] & .....(yn)]' is, and hence iff there
is some wff 'b' such that 'n G b' and '-(bn)' are in pr ( Rr). But
since 'b =H' is in pr, and since equals may be substituted for equals in
system cr, we have the result that 'Hn' is in pr ( Rr ) iff 'n G H' and
'-(Hn), are in pr (Rr). Furthermore, we know that 'n G H' is in
pr, so that we can conclude that 'Hn' is in pr iff '-(Hn)' is in pr; and
thus we have formalized the Grelling paradox (i.e. , the name of the
property of being heterological is itself heterological if and only if it
isn't heterological). But no contradiction arises in the system cr
because 'Hn' does not satisfy excluded middle, just as"RR' did not
satisfy excluded middle in the case of the Russell paradox. We knew
"ahead of time," as it were, that there could be no contradiction because
the consistency proof showed that there could not be any.
The Russell paradox can be formalized by use of an abstraction
equation. In the case of the GreIling paradox there does not seem to be
any such way to use an abstraction equation, but only the abstraction
principle itself. But this seems sufficient.

Frederic B. Fitch
Department of Philosophy
Yale University
New Haven, Connecticut

Fitch Formalized Self-Reference


1. "Self-Reference in Philosophy." Mind, 55 (1946) 64-73.

2. Elements of Combinatory Logic. Yale University Press, 1974.

See pp. 68-77.

3. "A Consistent Combinatory Logic with an Inverse to Equality."

Journal of Symbolic Logic, 45 (1980) 529-543.

Raymond M. Smullyan

Quotation and Self-Reference

#1. Some Preliminary Illustrations: We shall begin with some
concrete problems illustrating the types of languages that we will be

(1) For the first problem, let us consider a machine Ml that prints
out various expressions composed of the following five symbols:

.... P Nil' 1 *2
By an expression we mean any finite non-empty string of these
five symbols. An expression X is called printable if the machine can
print it. We assume the machine programmed so that any expression
the machine can print will be printed sooner or later.
By the quotation of an expression X we mean the expression
* lX* 2 [We are thus using * l' * 2 as a formal opening quote, closing
quote, respectively]. By the norm of an expression X we mean the
expression consisting of X followed by its own quotation-i.e. the
expressions X* lX* 2. By a sentence we mean any expression of one of
the following 4 forms (where X is any expression).

(1) p* lX*2
(2) PN* lX* 2
(3) .... p* 1X* 2
(4) .... PN* 1X* 2

By a positive sentence we mean a sentence of one of the forms

(1), (2) and by a negative sentence we mean a sentence of one of the
forms (3), (4). We define a sentence p* lX* 2 to be true if and only if X
is printable (by the machine); we define PN*lX*2 to be true iff the
norm of X is printable. [We crudely read p* lX* 2 as "printable X",
and PN* lX*2 as "printable the norm of X"]. This constitutes a precise
truth defmition for positive sentences. A negative sentence ....S is called
Smullyan Quotation and Self-Reference

true iff the positive sentence S is not true (thus ~PN*IX* 2 is true iff
X*IX* 2 is not printable). This now constitutes a truth definition for
all sentences.
We have here a curious loop: The machine is printing various
sentences that make assertions about what the machine can and cannot
print. We assume that the machine is completely accurate in that all
sentences printed by the machine are true. [Whether it can print
expressions that are not sentences is immaterial; the important thing is
that it never prints any false sentence.] The assumption of accuracy has
several ramifications: for one thing, given any expression X, if
P* IX* 2 ever gets printed, so does X, because if P* IX* 2 is printable,
it must be true, which means that X is printable. The converse does not
necessarily hold; if X is printable, then the sentence P*IX'" 2 is
certainly true, but we are not given that the machine is capable of
printing all true sentences, and so we have no grounds to conclude that
P* IX* 2 is also printable.
The problem now is to find a true sentence that the machine
cannot print! [The idea, ultimately deriving from GOdel, is to construct
a sentence that asserts its own non-printability. The solution will
probably occur to the majority of readers, and will be given later on.]

(2) The Machine M2 - We now consider another machine M2

based on what we will call a one-sided quotational language. This
machine was first introduced in [3]. Superficially, it will appear to be
but a trivial variant of the machine Ml' but we will later see that it has a
surprising property apparently not shared by MI'
The machine M 2 prints out various expressions composed of the
following 4 symbols:

~ P A *
By the quotation of an expression X we now mean the
expression *X. By the associate of an expression X we mean the
expression X * X. [The associate of X is X followed by its own
quotation, but the quotation of X is no longer'" IX* 2' but *X.] The
positive sentences are now expressions of one of the two forms (l)
P*X; (2) P A *X, and (1) is called true iff X is printable, and (2) is

Smullyan Quotation and Self-Reference

called true iff the associate of X is printable. The negative sentences

are again the expressions ....X, where X is a positive sentence, and again
the negative sentence .... X is called true if and only if the positive
sentence X is not true. Now the problem is to find a true sentence
which MI cannot print (assuming all sentences printed by M2 are true).

(3) The Machine M3 - The Machine M3 uses the symbol "R" in

place of "A"; the other 3 symbols are the same. By the repeat of an
expression X we mean the expression XX (X followed by itself). The
sentences are the same as those of M 2, except that "R" is used in place of
"A". A positive sentence p*x is called true iff X is printable by M 3; a
positive sentence PR *x is called true iff XX (the repeat of X) is
printable by M 3 . A negative sentence ....X is again called true iff X is
not true. Now the problem is to find a true sentence that M3 cannot
print (assuming M3 accurate).

(4) Cross-Reference: We have remarked that M2 has a property

apparently not shared by M 1: For M 2 , we can construct two sentences
X,Y such that it can be proved that at least one of them is true but not
printable by M 2 , but there is no way to tell which of them it is! [Hint:
Construct X,Y such that X is true iff Y is printable, and Y is true iff X
is not printable.] The same can be done with the machine M 3. Also,
with both machines M2 and M 3 , given any n ~ 2, one can construct
sentences Xl'""Xn such that at least one of them must be true and not
printable, but there is no way to determine which one it is.

(5) Here is a more difficult problem related to #6 of this article: We

have a machine M4 that prints out expressions composed of the
following six symbols.

By the quotation of an expression X we mean *IX* 2' We have

the following formation rules for terms, and each term designates a
unique expression according to the following rules:

Smullyan Quotation and Self-Reference

(1) For any expression X, its quotation is a term and

designates X.
(2) For any term t, the expression Rt is a term, and if t
designates X, then Rt designates the repeat of X.
(3) For any term t, the expression Qt is again a term, and if t
designates X, then Qt designates X* 2.

A positive sentence is any expression Pt, where t is a term, and

Pt is called true iff the expression designated by t is printable (by M 4).
A negative sentence is again the negation of a positive sentence, and is
true iff the positive sentence is noUrue. We again assume that M4 is

(a) Find a true sentence not printable by M 4

(b) Find a pair (X,Y) of sentences such that at least one is
true but not printable, but there is no way to tell which.
[Again one can do the same with three or more

#2. Quotational Systems: Before defining a quotational system we

consider a more general structure A. We start with some alphabet K
and we let e be the set of all (non-empty) words (expressions) in K.
The structure A is to be a collection of the following items.

(1) The set e

(2) A function which assigns to each X in e some expression
X in E called the name of X.
(3) A binary relation D(X,Y) on e, which we read "X
designates Y" such that no X designates more than one
expression, and such that for every X, the name X of X
is one of the expressions which designates X. An
expression will be called a term if it designates at least
one expression (and hence designates exactly one
(4) A set 1: of expressions called sentences.)

Smullyan Quotation and Self-Reference

(5) A set n of expressions called predicates, such that for

every predicate H and every term t, the expression Ht is
a sentence. [By Ht we mean H followed by t.]

We call two terms t l , t2 equivalent (in the structure A) if they

designate the same expression. We call a set T of sentences extensional
if for every predicate H and any two terms t l , t2 ' if ttl' t2 are
equivalent, then the sentence Htl is in T if and only if the sentence Ht2
is in T. We shall call two sentences SI' S2 equivalent if for every
extensional set T, the sentence SI is in T if and only if the sentence S2 is
in T. It is immediate that if tl' t2 are equivalent terms then for every
predicate H, the sentences Htl and Ht2 are equivalent sentences. In
particular, if t designates X, then Ht is equivalent to H X (a fact that
will be used repeatedly throughout this paper).
We might remark that extensional sets T arise in the following
manner: By an interpretation I of the set of predicates, we shall mean a
function which assigns to each predicate H a subset I(H) of E. We say
that an expression X satisfies H with respect to I if X is a member of
I(H). Call a sentence of the form Ht true under I if the expression
designated by t satisfies H. Then the set of all sentences Ht which are
true under I (we might call this set the truth set of I) is easily seen to be
extensional. Also, any extensional set T of sentences of the form Ht is
the truth set of some (in fact exactly one) interpretation-namely the
interpretation I which assigns to each predicate H the set of all X which
satisfies H under I. We call a sentence S a Glidel sentence for a
predicate H if S is equivalent to the sentence HS. [This of course
implies that under any interpretation I, S is true if and only if S satisfies
H.J We call A Glidelian if for every predicate H there is G6del
sentence for H.
We shall call an ordered pair <SI' S2> of sentences a Glidel
cross pair--or more briefly a Glidel pair-for an ordered pair
<HI ,H2> of predicates if Sl is equivalent to H IS 2 and S2 is equivalent
to HIS I . [This implies that under any interpretation I, SI is true if and
only if S2 satisfies HI and S2 is true if and only if SI satisfies H 2]. We

Smullyan Quotation and Self-Reference

call A doubly Godelian if for every pair <H I,H2> of predicates there
is a Godel pair <SI'S2> for <HI'H2>. We are primarily interested in
establishing sufficient conditions for A to be Godelian and sufficient
conditions for L to be double Godelian. Our results are extensions and
variants of some of the results of [1], but a knowledge of [1] is not
We shall call A a two-sided quotational system if there are
expressions l' 2 such that for every expression X, its name X is
IX 2' We call A a one-sided quotational system if there is an
expression such that for every expression X, its name X is *X.
Systems of the two sorts just described will be collectively called
quotational systems. [Strictly speaking, a one-sided quotational system
is a system A together with a distinguished expression satisfying the
above condition, and a two-sided quotational system is a system A
together with an ordered pair <. l' 2> of distinguished expressions
satisfying the above condition].
For any expression X, by its repeat we mean the expression XX.
For a one-sided quotational system, by the associate of an expression X
we mean the expression x*x (which is XX). For a two-sided
quotational system, by the norm of an expression X, we mean the
expression XIX 2 (which again is XX). We shall first tum our
attention to the following three types of quotational systems.
By an N-system we mean a two-sided quotational system such
that there is an expression N such that for every term t and every
expression X, if t designates X, then Nt designates the norm of X. [In
particular, N* IX 2 designates X IX 2]' By an A-system we mean a
one-sided quotational system such that there is an expression A such
that for every term t and every expression X, if t designates X, then At
designates the associate of X. [In particular, AX designates XX.]
By an P-system we mean any system A, not necessarily quotational,
such that there is an expression R such that for any term t and any
expression X, if t designates X, then Rt designates the repeat of X. [In
particular, RX designates XX.] By an PI-system we shall mean a

Smullyan Quotation and Self-Reference

one-sided quotational system which is also an P-system.

We will prove the following facts about N-systems, A-systems,
and PI-systems.

Theorem A:
N-systems, A-systems and PI-systems are all GOdelian.
Theorem B:
A-systems and PI-systems are also doubly Godelian.

That N-systems and A-systems are Godelian is no surprise (we proved

essentially this in [1] for N-systems, and the proof for A-systems is not
significantly different). That A-systems are doubly Godelian came as a
surprise to the author, who chanced upon the construction in the course
of writing [2]. Our results for PI-systems seem even more curious, in
that the simpler operation of repetition can be made to accomplish the
same end as the more complex operation of association. However,
one-sided quotational systems are curious things.
We doubt that N-systems are necessarily doubly Godelian
(though we have not found a proof to the contrary). We will later
consider some two-sided quotational systems which are both GOdelian,
and also some curious two-sided quotational systems which are doubly
Godelian, but do not appear to be necessarily GOdelian.

#3. Designational Properties: We let Ao be the substructure of A

consisting of just the first three items (the expressions, the naming
function and the designation relation). We might refer to Ao as the
designational substructure of A. Many of our results about A will be
obtained as corollaries of certain theorems about Ao. The notions of
one-sided quotational system, two-sided quotational system,
N-system, A-system, P-system, PI-system make no reference to
the sentences or predicates, and will accordingly be applied to Ao as

Smullyan Quotation and Self-Reference

We shall say that Ao (and also A) has property Dl if for every

expression E there is a term t which designates Et. We shall say that
Ao (and also A) has property D2 if for any expressions Xl' X 2 there
are terms tl' t2 such that tl designates X2t2 and t2 designates Xltl .
The following theorem is extremely general, and applies to systems A
not necessarily quotational.

Theorem 1:
(a) If A has property Dl then A is GOdelian.
(b) If A has property D2, then A is doubly GOdelian.
(a) Suppose A has property Dr Take any predicate H.
By property Dl there is some term t which designates
Ht. Therefore the sentence Ht is equivalent to the
sentence HHt. Thus S is equivalent to HS, where S =
Ht, and so S is a GOdel sentence for H.
(b) Suppose A has property D2. Let HI' H2 be any two
predicates. Then there are terms tl' t2 such that tl
designates H2t2 and t2 designates Hlt l . Since tl
designates H 2t 2 , then tl is equivalent to H 2t 2 , and so
Hltl is equivalent to HI H 2t 2. Similarly, since t2
designates HItI' then H2t2 is equivalent to H2Hltr So
Sl is equivalent to HI S2 and S2 is equivalent to H2~'
where Sl is the sentence Hltl and S2 is the sentence H 2t 2.

#3.1 Normality: We continue to consider a system A not necessarily

We call an expression V a normalizer of Ao (also of A) if for
any expression X, the expression VX is a term that designates
XX. We call Ao (also A) normal if there is a normalizer V for

Smullyan Quotation and Self-Reference

A 0 (A). Obviously N-systems are normal (N is a normalizer) and

A-systems are normal (A is a normalizer).

Theorem 2:
If Ao is normal, then it has property D I (and
hence A is GOdelian).

Let V be a normalizer of A o' Take any
expression E. Then VEV designates EVEV. We let t
be the term VE V, and we see that t designates Et. Thus
Ao has property Dr Then also A is GOdelian by (a) of
Theorem 1. [More specifically, for any predicate H,
since the term VHV designates HVHV, then HVHV is
a GOdel sentence for H.J
We might still more explicitly note that for an
N -system, for any expression E, a term t which
designates Et is N* IEN* 2' and hence if E is a predicate,
then EN* l' EN* 2 is a Godel sentence for H. For
A-systems, a term t which designates E is A *EA, and if
E is a predicate, then EA*EA is a Godel sentence for E.

The solution to Problems 1,2 of #1 should now be
apparent: For problem 1, the sentence ~PN* I~PN* 2
works; it is true iff the norm of ~PN is not printable, but
the norm of ~PN is the very sentence ~PN*I~PN*2'
And so this sentence is true iff it is not printable by Mr
Since MI is assumed accurate, the sentence must be true
but not printable. For the solution of Problem 2, we take
the sentence ~PA*~PA.

#3.2 Weak Normality: Theorem 2 does not appear to help us show

Smullyan Quotation and Self-Reference

that PI-systems have property Dl (or even that they are GOdelian); for
this we need a stronger theorem.
Call an expression W a weak-normalizer for Ao (and for A) if
for every expression X there is an expression Y such that WY
designates XY. And we call Ao ( A ) weakly normal if there is a weak
normalizer W. It is trivial that any normalizer V is also a weak
normalizer, because for any X, since VX designates X, then there is a
Y such that VY designates XY -namely Y =X. Thus every normal
system is also weakly normal. Therefore the following theorem is
stronger than Theorem 2.

Theorem 2*:
If Ao is weakly normal, then Ao has property Dl
(and again A is GOdelian).

Let W be a weak normalizer for Ao' Take any
expression E. Let X be the expression EW. Then there
is some Y such that WY designates XY -thus WY
designates EWY . We let t be the term WY , and we see
that t designates Et.

Now, PI-systems are not in general normal, but

they are weakly normal-indeed, R is a weak normalizer
(because for any expression X, since X* designates X*,
then R*X* designates the repeat of X*, which is X*X*,
and so RX * designates XX *, and thus R Y designates
XY, where Y X*).
It now follows from Theorem 2* that PI-systems
have property D 1, and are Godelian. Indeed, given an
expression E, a term t which designates Et is R *ER *,
and if E is a predicate, then ER *ER* is a GOdel sentence
for E.

Smullyan Quotation and Self-Reference

We have now proved all parts of Theorem A.

The solution of Problem 3 of #1 is the sentence

#4. The Properties G2+,D2+: We will soon see that anyone-sided

quotational system having property Dl must also have property Dz'
and hence (by Theorem 2*) any weakly normal one-sided quotation
system has property Dz. In particular, it will follow that A-systems
and PI -systems have property Dz, which by Theorem 1-(b) implies
that they are doubly GOdelian, which will prove Theorem B. However,
we will show something stronger.
Let G z be the property of being double GOdelian. We shall say
that A has property G z+ if for any predicates HI' Hz there are
sentences Sl' Sz such that Sl is equivalent to H;Sz and Sz is not only
equivalent to Hz 8 1, but actually is the sentence HzS-;. In other words
Gz+ is the property that for any predicates HI' Hz, there is a sentence S
which is equivalent to the sentence HI Hz"'S'.
We shall also say that Ao has property D z+ if for any
expressions EI' E z there are terms tl' t z such that tl designates E z t z
and t z not only designates Eltl' but is the name El tl of E 1t 1 Thus
Dz+ is the property: for any EI' E z there is a term t which designates
E z E1t 1"

Theorem 3
Property D z+ implies property G z+.

Assume property D z+. Take any predicates HI'
Hz. Then there is a term t which designates Hz Hit, and

Smullyan Quotation and Self-Reference

therefore Hit is equivalent to HiH2tr;i'. And so S is

equivalent to H1H;g, where S is the sentence HIt.

Since property G 2+ obviously implies that A is
doubly Godelian then by Theorem 3, the property D2+
implies that A is doubly GOdelian. For applications, let
us explicitly note that if t is a term that designates
H 2H 1t, then <Hit, H2 Hit> is a Godel cross-pair for
<HI' H2 >.

Theorem 4:
Anyone-sided quotational system having property DI
also has property D2+.

Suppose A is a one-sided quotational system
having property Dr Take any expressions E 1, E 2. Let
X be the expression E2 *Er By property Dl' there is a
term t which designates Xt-thus t designates E2 *E1t,
which is E2 EI t.

Note 1:
We have noted that if t is a term that designates
H2HI t, then <Hit, H2 Hit> is a GOdel pair for
<Hl'H2>. For an A-system, a term t that designates
H2 Hit is A *H 2 *H1A (it designates the associate of
H 2*H 1A, which is H 2*H 1A*H 2*H 1A, which is
H 2*H 1t, which is H 2H 1t) and so <H 1A*H 2*H 1A,
H2*H1A*H2*H1A> is a Godel pair for <HI' H 2>. For

Smullyan Quotation and Self-Reference

an P t-system, a term t that designates H zH t t is

R*Hz*HtR* (it designates the repeat of Hz*HtR*,
which is Hz *HtR*Hz*HtR"', which is Hz *H t t, which is
H zH t t ), and so a Godel pair for <HI' Hz> is
<HtR *H z*HtR *, Hz *HtR*H z*HtR *>.

Note 2:
Now we have solutions to Problem 4 of #1: For
machine M z' take Y to be ""PA*P*""PA and take X to
be P*""PA *P*""PA (which is P*Y). X obviously
asserts (i.e. is true if and only if) Y is printable, and Y
asserts that the associate of p*...,pA-which is X-is not
printable. If X were false, Y would be printable, hence
true, hence X would be printable, contrary to the
assumption that M z is accurate. Hence X is true and Y is
not printable. The truth of Y cannot be determined, but
if it is true, Y is true but not printable; if Y is false, X is
true but not printable.
An alternative solution is to take P A... ...,p ... p A for
X and "'P*PA*""P"'PA (which is ...,P*X) for Y. As for
the machine M3, the solutions can be obtained from those
of M z by simply putting'" at the end of both sentences.

#5. The Property G t ': Theorem B can also be obtained as a

consequence of the next two theorems, which strike us as being of
interest in their own right.
Let us say that A has the property Gt' if for any predicate H and
any expression E, there is a sentence S which is equivalent to HES. [If
we think of H as being the name of some property P, a Godel sentence
for H can be thought of as asserting that it has the property P, whereas a
sentence S which is equivalent to HES can be thought of as saying "the
expression consisting of E followed by me has property P".]

Theorem 5:

Smullyan Quotation and Self-Reference

Property D t implies property G t l.

Assume property Dr Then for any predicate H
and any expression E, let X be the expression EH. By
Dt there is a term t which designates Xt, so t designates
EHt. Then Ht is equivalent to HEHt. Thus S is
equivalent to HES, where S is the sentence Ht.

Theorem 6:
For a one-sided quotational system, property Gtl implies
property G 2+.

Let A be a one-sided quotational system having
property Gtl. Take any predicates H t , H 2. Let E be the
expression H 2. By G t l there is a sentence S which is
equivalent to HtES. Thus S i~ equivalent to H t H 2.S,
which is H t H 2"'S'.

#6. Some Two Sided Quotational Systems: We shall now tum our
attention to two-sided quotational systems. Except when specified to
the contrary, A will now be assumed to be a two-sided quotational
We shall say that A has property Q if there is an expression Q
such that for any term t and any expression X, if t designates X, then
Qt designates X* 2. By an NQ-system we shall mean an N-system
having property Q. Now, we already know that N-systems are
Godelian, but we have no reason to believe that they are necessarily
doubly Godelian. We will see, however, that NQ-systems are
necessarily doubly Godelian (in fact they must have the stronger
properties D 2+,G2+).
So far we have said nothing about P2-systems-i.e. two-sided

Smullyan Quotation and Self-Reference

quotational systems which are also P-systems; such systems do not

appear to be necessarily either GOdelian or doubly GOdelian. We shall
see, however, that any such system which also has property Q must be
both Godelian and doubly GOdelian (such systems might be called
The above facts are consequences of the following more general
facts: Define A to be an N' -system if there is an element N' such that
for all t, X, if t designates X, then N't designates X* IX 2*2 (which is
XX 2)' Of course every NQ system is also an N'-system-just take
N' to be QN. Define A to be an P'2-system if for any t, X, if t
designates X, then R't designates XX*2' and define A to be an
P 2" -system if for any t, X, if t designates X, then R" t designates
X X * 2* 2' Any P 2Q -system is also both an P 2'-system and an
P 2"-system (take R' to be QR and R" to be QQR). We shall prove the
following two theorems.

Theorem C:
P 2' -systems are GOdelian

N'-systems and P 2"-systems are doubly Godelian
(in fact have properties D2+ and G 2+).

We do not know whether P'2-systems are necessarily Godelian,

nor whether N' -systems or P 2" -systems are necessarily Godelian.
However, NQ-systems are both R'2-systems and R 2"-systems, and so
Theorems C, D have as immediate consequence:

Theorem E:
NQ-systems and PQ-systems are both Godelian and
doubly GOdelian.

Smullyan Quotation and Self-Reference

Now we tum to the proofs of Theorems C and D.

Proof of Theorem C :
It suffices to show that P 2'-systems are weakly
normal (result then follows by Theorem 2* and (a) of
Theorem 1).
Well, for any expression X, the expression
R'* lX* 1*2 designates the result of appending *2 to the
repeat of X* 1; this result is X* lX * 1*2. So if we take
X* 1 for Y, we see that R'Y designates XY , and so R'
is a weak normalizeF.
Without appeal to earlier theorem'", we can see
directly that for any predicate H, the sentence
HR'* IHR'* 1*2 is a GOdel sentence for H.

A solution to (a) of Problem 5 of #1 is the sentence
-PQR*I- PQR *I*2

As for the proof of Theorem D, we shall break it into several


Theorem 7
For any two-sided quotational system A, each of the
following conditions (except the last) implies the next.
(a) A is an N'-system or an P 2"-system.
(b) There is an expression 1t such that for every expression
X, there is an expression Xo such that 1t Xo designates

(c) For every expression E there is a term t that designates
(d) A has property D 2+.
(e) A is doubly GOdelian (in fact has property G 2+).

Smullyan Quotation and Self-Reference

(1) Suppose (a) holds. If A is an N'-system, take N'
for 1t and for any X, take Xo to be X. Then N'Xo =
N' X and N' X designates X X * 2' and so N' X 0
designates XXo *2.
If A is an P 2" -system, take R" for 1t, and for any

X, let Xo = X*l. Then R"Xo designates X OX O 2 2'

W h lC x* 1 x* 1 * 2* 2' W hlC hIS Xx*.
1 2' W hlCh IS

Thus (b) holds.
(2) Assume (b). Then there is some 1t such that for
all X there is some X 0 such that 1t Xo designates
XXo *2. Now take any expression E; let X =E1t. Then
there is some Xo such that 1tXo designates E1t Xo * 2'
and so t designates Et*2' where t is the term 1t XO. Thus
(c) holds.

Note: If A is an N'-system, we take N' for 1t and for any X we

take X itself for XO' and hence for any expression E, a
term t which designates E t 2 is N' EN', which is
N' '1\ EN' 2" [The reader can directly verify that
N'lEN'*2 designates EN'*lEN'2*2.] It then
follows that for an NQ-system, a term t which designates
Et*2 is QN 1EQN*2. Now, if A is an P"2-system, we
take R" for 1t, and for any X, we take x* 1 for X O' and
so for any expression E, a term t that designates Et2 is
R"ER"l' which is R"*lER"12. [Again the reader
can verify directly that

Smullyan Quotation and Self-Reference

R" ER" designates E R"* ER"* * .]

,1 1 'l \ 1 1 'l 2
t t
Also, if A is an P2Q-system, then a term t that designates
Et*2 is QQR*lEQQR*1*2'
(3) Assume property (c). Take any expressions EI'
E 2. Let E E2 *lEl' By (c) there is a term t which
designates Et*2; thus t designates E2*IElt*2' which
E 2E 1t. And so A has property D2+.

Note: By the above proof and the note of (2), we see that for an
N '-system, a term t which designates E 2E I t is
N ' * 1E 2* 1E 1N ' * 2 (it designates
E2*IEIN'*lE2*IEIN'*2*2' which is E 2E 1t). For an
R" -system, a term t which designates E2E It is
R " * IE 2 * 1 E I R " * 1* 2 (it designates
E2*IEIR"*IE2*IEIR"*1*2*2' which is E 2E 1t).
(4) We already know (Theorem 3) that property D2+
implies G 2+' which of course implies that A is doubly

Theorem 7 of course implies Theorem D. Given a pair <HI'

H2> of predicates, to actually find a Godel cross pair for <HI' H 2> we
proceed as follows: We know that <Hit, H2Hl t > is such a pair,
where t is any term that designates H 2, Hit. For N'-systems we know
that N'*lH2*lHIN'* 2 is such a term t (cfNote to (c, and therefore a
Godel cross pair for <HI' H 2> is <HIN'*IH2*IHIN'*2'
H2*lHIN'*lHIN'*2*2>' For P 2"-systems, a term t that designates
H2Hlt is R"*IH2*IHIR"*1*2' and so a GOdel pair for <HIH2> is

Smullyan Quotation and Self-Reference

<X,Y>, where

- H1R"
12 H1
Y - H 2 1H 1R" 1H 2 1HR" 2 2 2

Of course, for an P 2Q-system, we take the above <X, V>,

replacing R" by QQR. In particular, for the machine M4 of Problem
5 of #1, a solution is

(a) Suppose that A is a system (not necessarily
quotational) having the property that there is an
expression 1t such that for any expressions A, B there is
an expression Y such that 1t Y designates ABV. Show
that A has property D2+.
(b) Show that for a two-sided quotational system A,
if A has property (c) of Theorem 7, then it not only has
property D2+, but satisfies the hypothesis of (a).

#7. Concluding Remarks: We would like to briefly indicate several

directions in which many of the results of this paper can be extended.
First of all, we have assumed that the empty expression 0 is not
in e; let e' be the set of elements of e together with 0. Then Theorem
l(a) and Theorem 5 can be combined into the single statement: For any
predicate H and any element E of e', there is a sentence S which is
equivalent to HES. Now, let us say that a function f(X) from e into
e is definable in Ao (also in A ) if there is an expression F (which
would be said to define f) such that for any term t and any X in e, if t
designates X, then Ft designates f(X). Let G 1 be the property that
for any predicate H and any function f which is either definable in A or

Smullyan Quotation and Self-Reference

is the identity function, and for any E in e', there is a sentence S which
is equivalent to Hf(ES). It is not difficult to modify the proofs of
Theorem l(a) and Theorem 5 to show that property DI implies
property G I *. Therefore N-systems, A-systems and PI-systems all
have property G I ....
Secondly, for any D ~ 2, call An-fold Godelian if for any
predicates HI' , H D , there are sentences SI' , So such that for each
i < D, Si is equivalent to HiSi+~' and So is equivalent to HoSt" Let
Go be the stronger property that for any predicates HI' , H o' and
any elements EI' ,E o of E', and any functions fl' , fo' each of
which is either definable in A or is the identify function, there are
sentences S l' , So such that for each i < D, S i is the sentence
Hli(Ei+1 Si+l) and So is equivalent to Hofo(EI--S;). Let Do + be
the property that for any expressions EI' . , Eo of E there are terms
tl' , to such that for each i < D, ti Ei+1 ti+l' and to designates
Elt l . It is not difficult to show that for each D ~ 2, Do+ implies Go'
and it can also be shown that anyone-sided quotational system having
property DI has property Do+ for every D ~ 2. Therefore A-systems
and PI-systems have property Go for each D ~ 1. Also NQ-systems
and PzQ~systems can be shown to have properties Do+, Go for all D ~
There are several points about the designational substructures
Ao of A not treated here that can be found in Chapters 9-12 of [2]. For
example, for A-systems and PI-systems, given any function f(x)
definable in the system, there is an expression X that designates f(X)
(this result is closely related to Kleene's second recursion theorem).
Also, for any functions f(x), g(x) definable in the system, there are
expressions X, Y such that X designates feY) and Y designates g(X) (a
result related to the author's double recursion theorem). These results
can also be shown to hold for NQ-systems and PzQ-systems. A unified

Smullyan Quotation and Self-Reference

treatment of all this, together with some generalized recursion

theorems, fixed point theorems for combinatorial logic and various
diagonalization techniques is currently in preparation [3].

Raymond M. Smullyan
Department of Philosophy
Indiana University
Bloomington, Indiana

Smullyan Quotation and Self-Reference


1. Smullyan, Raymond M., "Languages In Which Self-Reference Is

Possible," Journal of Symbolic Logic, 22, 1 (March 1957)

2. The Lady Or The Tiger? - and Other Logical Puzzles. Alfred

A. Knopf, 1982.

3. Diagonalization and Self Reference-currently in preparation.

Graham Priest

Unstable Solutions to the Liar Paradox

1. Introduction

For about 80 years logicians have participated in a research

programme called "solve the paradox". The immediate cause of the
existence of the research programme was the proliferation of logical
paradoxes around the tum of the century. The central assumption or
"hard core" of the programme is the assumption that no contradiction is
true, and hence that the reasonings which result in the contradictions
must be fallacious. The aim has been to locate the fallacies and to
articulate a theory which explains the data: the fallacious, yet highly
plausible reasoning. Some time later, which might conventionally be
dated at the publication of Ramsey's essay "The Foundations of
Mathematics" [1926], the programme bifurcated into one for solving
the set theoretic paradoxes and one for solving the semantic paradoxes.
The strategy of divide and conquer is a familiar enough one, and often
successful. Yet in this case it was a retrograde, or at least defeatist,
step: the original aim of the founding fathers, such as Russell, to find a
unified solution to the problem had to be given up. If one now
considers the semantic branch of the program, it is difficult to avoid the
conclusion that it has been somewhat less than successful. (I think that
the same is true of the set theoretic branch. However in virtue of the
general acceptance-at least by mathematicians-of the cumulative
hierarchy, the case is much more difficult to make and 1 shall not
address it here.)
There is certainly no generally accepted solution. The strategies
or "heuristics" for solving the problem are but few. There is the
"meaningless" strategy, the "neither true nor false strategy" and so on.
Yet within this framework, purported solutions have multiplied in a
way that makes the breeding habits of rabbits look like family planning.
Show that one distinction does not work and a dozen appear in its place;
show that a theory runs into trouble with a well-supported
philosophical theory and a dozen patched-up versions appear to replace
it. This is not the place to chart the historical details of this process,
which are, in any case, widely known. However, one can see in this
process what Lakatos has called a "degenerating research programme" .
Characteristic of this is that no essential progress is made towards
solving the central problem. Rather, enormous time is spent trying to
solve problems of equal or greater acuity created by the programme
Priest Unstable Solutions to the Liar Paradox

itself.1 An extreme form of this is where the proposed problem

solution does not really solve the problem at all but merely one of its
manifestations. The original problem is then transferred, and appears
in a different place. It may sometimes appear in a slightly different
guise, whence to a cursory glance the proposed solution may appear
more successful than it actually is.
The most recent instalment in this program is an approach to the
semantic paradoxes, or rather family of approaches, provided by Anil
Gupta [1982] and Hans Herzberger [1982].2 Their construction is
elegant and of clear structural interest, and it might appear that it is a
"creative shift in the heuristic of"the programme"; but I think that in
fact it is merely another phase of the degenerating programme. The
major part of this paper is an attempt to show this. Despite this fact, it
does seem to me that, although it is incorrect, the idea points the way to
a more adequate understanding of the semantic paradoxes. I will return
to this in the final section of the paper.

2. The Construction

In order that the paper may be reasonably self contained I will

start by outlining the Gupta/Herzberger (hereafter 'GH') construction
and pointing out some of its salient features.
We take a first order language, L, with a predicate 'T', which is
thought of as the truth predicate. Let M 0 be any first order
interpretation for L. The domain of M o' D, contains a subset, S, which
is just the sentences (closed formulas) of L. The extension of T in M o'
U, is arbitrary. However it is simple and natural to let S ~ U.3 We
now define a transfinite class of structures {Ma 1 ae On} by recursion

i) Given M o' Ma+l is exactly the same as Ma except that the extension
of Tin Ma+l is exactly {cpe S 1 Ma~cp}.

ii) For limit ordinals, the Gupta and Herzberger variants differ

Let X.,+(U) ={ cpe S 13{3<1 \fa({3<a and a<1 =* Ma~cp)}

Priest Unstable Solutions to the Liar Paradox

The sentences in Xy+(U) are locally stably true at y. Those in

Xy+(U) are locally stably false at y. Now let A be a limit ordinal.
Then M,. is the same as Ma for a < A except that the extension of T in

a) Xy+(U) (Herzberger)
or b) x.t(U) u (U X,.-(U (Gupta).4

The differences in construction matter not for our purposes.

Let Xoo +(U)={cpe S I 3(3\fa~(3 Ma"cp}

and Xoo-(U) = {cpe S I 3(3\fa~(3 Ma".cp}.

We will call the members of Xoo+(U) globally stably true and those of
Xoo-(U) globally stably false. Let Xoo (U) =Xoo+(U) u Xoo-(U). The
members of Xoo(U) are globally stable (relative to U). If cp is globally
stably true with respect to all U we will call it absolutely stably true; if
globally stably false with respect to all U, it is absolutely stably false. If
it is either of these it is absolutely stable. S For fixed U, if cp is globally
stable, we will call the least ordinal at which cp or.cp enters the
extension of T never to depart, its stabilisation point. Let

l:(U) ={(313cpeXoo(U)' (3 is the stabilisation point of cpl.

And let a(U) Ul:(U). We will call Ma stabilised iff a~a(U).
It is now easy enough to establish the following facts:

0) If M a is stabilised then if cpe Xoo +(U), M a"CP and if

cpe Xoo-(U) then Ma"'CP.

1) If "Cp then cp is absolutely stably true.

Priest Unstable Solutions to the Liar Paradox

2) For any U, Xoo+(U) is closed under logical consequence, as then

is the set of absolutely stably true sentences.

3) If fPE Xoo(U) then Tm=fP E Xoo+(U) where m is a name for <p,

which, without loss of generality we may suppose L to contain.
(Thus if <p is absolutely stable, its T -sentence is absolutely stably

4) If fPe Xoo(U) it does not follow that T,m.=<p eXoo +(U).

However there is no guarantee that it is in Xoo+(U).

5) If fP does not contain 'T' then for all a, p, MalofP iff MjilofP
Hence fPE Xoo+(U) or fPE Xoo-(U).

3. Meaning and "Rules of Revision"

So much for the technical construction. Let us now tum to the

question of whether it provides a philosophically satisfactory solution
to the semantic paradoxes.
Given that a paradox is an argument with a contradictory
conclusion, a necessary condition for a solution is that it locate a step in
the argument which is fallacious. Take now a paradox such as the liar,
which is Gupta and Herzberger's most favoured case. The argument
goes essentially:

Let'l'be' 'l'is false' (1)

The T -scheme gives: 'I' is true iff 'I' is false. Hence'll is both true and
The point at which the GH construction faults this reasoning is
precisely in the application of the T -scheme to '1'. As we have seen, an
instance of the T -scheme is not bound to hold in any particular model,
let alone all stabilised models. In fact, given that we can find an Mo in
which (I) is true, the negation of the T-sentence for'll turns out to be
absolutely stably false.
Does this solve the paradox? Of course not. It is easy enough to

Priest Unstable Solutions to the Liar Paradox

flatly deny a step in the reasoning. It is also straightforward to produce

a technically constructed model in which it fails: Tarski's original
construction, which both Gupta and Herzberger reject as inadequate
does that. 6 The problem is to find an adequate philosophical
justification for the construction. This might be approached in a
variety of ways; but what, in the end, they must all come down to is that
the construction provides an adequate analysis of our conception of
truth. (A construction that is openly offered in a revisionist
fashion-which is the way the Tarski hierarchy has often been
viewed-does not solve the problem. For the aim was to explain what
was wrong with the original argument, not some revision thereof.)
Or, to put it in a form more in line with modern philosophy, the
construction must provide an analysis of the meaning of 'true'.
Gupta, at least, sees the matter in these terms too. He says:

...1 am suggesting that underlying our use of 'true' there is not an

application procedure but a revision procedure instead. When we learn the
meaning of 'true' what we learn is a rule that enables us to improve on a
proposed candidate for the extension of truth. It is the existence of such a
rule, I wish to argue, that explains the characteristic features of the concept
of truth. [1982], p.37

And again later:

In intuitive terms the conception I have tried to defend is this. When we

learn the meaning of 'true' we learn a rule that enables us to determine the
extension of truth provided that we know the denotations and extensions of
all the names, predicates, and function symbols in the language. [1982],

Thus the meaning of 'true' is the rule which takes us from the extension
of'T' in Ma to its extension in Ma+l"
Let us then ask whether the construction does give an adequate
account of the meaning of 'true'.' This question is faced with the
difficulty that at present the whole question of how meanings are to be
explained is itself a moot one. We have several different accounts of
what a theory of meaning for a language should be like: Davidson's
account of meaning; Montague semantics; Dummett's verificationism
and so on. Unfortunately Gupta's proposal fits into none of them.
None of them makes provision for "rules of revision" whatever,
exactly, they are. Of course the idea that meanings are rules was
commonly aired between the 1930's and 1950's (though the rules in
question were normally rules of application). But the theory of
meaning has long since passed beyond those tentative and piecemeal

Priest Unstable Solutions to the Liar Paradox

days. It is therefore somewhat disconcerting to find the suggestion that

meanings are rules (albeit of a new kind) thrown off this casually. If
we were to put it uncharitably we could say that since the GH account of
the meaning of 'true' is at odds with all the best developed accounts of
what a theory of meaning should be like, it is in trouble. More
charitably (and I think more accurately) the point is that the GH
construction puts a large spanner in the works of theories of meaning.
We have here, therefore, an excellent example of a proposed problem
solution posing a deep and acute problem purely of its own making.
However, let us not leave the problem there; for the acuity of the
problem needs to be emphasized. The mere occupation of a field by a
theory, or collection of theories, does not mean that these are right,
uncriticizable, or to be taken for granted. Could the GH construction
be incorporated in a fully fledged theory of meaning? Obviously it is
foolish to deny the possibility of this. (Equally obviously, the onus is
on Gupta and Herzberger to show that this is at least plausible, or claims
to have solved the paradoxes are somewhat premature.) What would
such a theory be like? I do not want to put words into other people's
mouths. However, let us try to see what it could be like.
It would seem that a theory of meaning for a language must be
an axiomatic theory which for every meaningful sentence of the
language, S, has a theorem which spells out what the meaning of Sis. 8
Exactly how it does this is a point of some substance. However there
appears to be little alternative to Frege's observation that to give the
meaning of a sentence is in some sense to spell out its truth conditions.9
The simplest suggestion (i.e. Davidson's [1967]) is that, at least for
languages which contain no indexical sentences, the meaning of S is
spelt out by the instance of the T -scheme for S in a Tarski-type truth
theory. This approach is certainly not open to Gupta and Herzberger.
Their semantics are of a model theoretic truth-in-a-structure kind,
rather than an absolute truth-definition kind. Nor is there any hope (as
can be done in possible-world semantics) of nominating a particular
structure (some Ma ) and identifying truth (simpliciter) with truth in
that structure. For (on Gupta's account) meaning is essentially
relational, concerning the generation of one structure from another,
rather than being a property of a single structure. Even worse, in
unfortunate situations, there will be instances of the T -scheme which
fail in all structures, as we have seen. Thus there is no hope of using the
T-scheme to state (truly) meanings. lO
How then to proceed? Somehow we must incorporate the idea
of there being a plurality of structures at issue, and important
relationships between them. A way to do this is suggested by

Priest Unstable Solutions to the Liar Paradox

possible-world semantics. In these, what is thought of as spelling out

the meaning of S is the statement of the truth-in-a-possible world

For any world (situation) w, S is true in w iff <p(w).

In effect, every sentence is treated as indexical with respect to a

world-context. (See, e.g. Montague [1970].)
Perhaps a theory of meaning using the GH construction could be
based on the idea that the meaning of a sentence S is spelt out by a
theorem of the form:

For any stage a, S is true at a iff <pea).

We would have to get a great deal clearer about what the stages were
stages of (refining our conception of truth?, approximating the
absolute?). However I think we can leave this murky problem aside.
For it is quite unlikely that such a theory could be an adequate theory of
The crucial question to ask is what we would be imputing to
speakers of a language with such a theory of meaning. For a theory of
meaning spells out what it is that speakers know when they understand a
language. They do not, perhaps, have to know the whole theory.
Knowledge of the content of sentences which spell out the meanings of
sentences might suffice. Neither must we suppose that the speakers can
explicitly formulate these claims. Indeed perhaps they may not even
have a language in which this can be done. None the less, the
meaning-giving sentences of the theory must express what, in some
sense, is grasped by speakers of the language, or it is hardly an account
of the meaning of their language. It follows that concepts used in
stating the meaning-giving sentences must be attributed, at least
implicitly, to speakers of the language which the theory of meaning is
Possible-world semantics are sometimes criticized on just these
grounds: that they impute to speakers concepts, such as that of possible
worlds (or at least, such set theoretic machinery as is necessary to
construct their surrogates), which they do not necessarily have. The
objection might be parried by arguing that someone who knows the
meaning of a sentence knows not only how to use it in the actual
situation but also how they would use it in different situations. In some
sense, then, a language speaker must have a conception of possibilities
different from the actual. Hence, possible-world semantics are not

Priest Unstable Solutions to the Liar Paradox

methodologically vicious. I do not wish to discuss the adequacy of this

reply, for the important point here is that there is no similar reply open
to a parallel objection addressed to the suggested theory of meaning
based on the GH construction. This theory of meaning imputes to
speakers not just the notion of possible worlds, but those of an arbitrary
ordinal, of ordinal operations and transfinite induction (in the
specification of cp). 11 It seems that there could be nothing in the
behaviour of most language speakers which would justify this
attribution, in which case the semantics cannot be an adequate account
of meaning.
It may be that Gupta and Herzberger would wish to formulate
their accounts of meaning in some other way. But however it is
formulated I do not see how it could sidestep a similar objection. For
the transfinite construction is the core of their proposal, and it would
seem impossible for meaning-giving sentences in a theory of meaning
based on the construction to avoid referring to it.
Before we leave the area of the theory of meaning, there is one
further observation worth making. Both the absolute truth definition
of the Tarski approach and the model-theoretic possible-world
approach, provide us with a notion of truth, simpliciter. Now there is
an important connection between truth and assertion. Basically it is that
truth is the aim of assertion. 12 The truth is, generically, what we aim to
assert when we assert. Thus the fact that there is no notion of truth,
simpliciter, in the GH construction poses something of a problem. Is
the class of sentences we aim at asserting time dependent; so that we can
legitimately assert and deny the liar sentence alternately every n
minutes? Surely not. 13 There must be a fixed class of sentences at
which we aim. What is it? The set of sentences true at some Ma is far
too arbitrary to be satisfactory. Similarly, those sentences globally
stably true with respect to a particular U have an air of arbitrariness.
The only satisfactory class is the class of absolutely stably true
sentences. (The class of sentences which are not absolutely stably false
will not do since this is inconsistent.) This, I suspect, would be Gupta
and Herzberger's line; it seems the only reasonable one.

4. Strengthened Paradoxes
(i) The Significance o/These

Considerations concerning the theory of meaning are not, I

Priest Unstable Solutions to the Liar Paradox

think, the major objections to the proposal that the GH construction

solves the semantic paradoxes. The major one is its failure to resolve
paradoxes of the "strengthened" variety.14 I will explain this in
subsequent sections. However I want first to say a few words to put the
situation in its correct perspective. It might be thought harsh to
criticize a novel proposal for failing to solve more contrived
paradoxes. After all, the liar paradox is the paradigm problem, and if
we sort that out properly, we can hope to get the more peripheral
problems sorted out later.
This perspective of the significance of strengthened paradoxes is
the exact opposite of the correct one. To see this, look at the liar
paradox as follows. We start with a set of sentences; we can call them
bona fide truths. These are the sentences that are genuinely assertible.
On most conceptions these will coincide with the true sentences (though
in a many-valued logic they might coincide with the ones of designated
value, and so on). Those that are left over, we will call "the Rest". The
essence of the liar paradox is a particular twisted construction which
forces a certain sentence, if it is in the bona fide truths, to be in the Rest
(too); and conversely, if it is the Rest, it is in the bonafide truths. Since
it can't play for both teams at once, the problem is posed.
Now the pristine liar "This sentence is false" is only a
manifestation of the problem arrived at by taking the Rest to be the
False (and the bonafide truths to be the True). In this particular case,
then, we can of course get out of the problem by insisting that the False
is only a proper part of the Rest. This opens up a gap in which the liar
sentence can conveniently lie. But this solves the problem only by
showing that it was inadequately posed. For if the False is only
properly contained in the Rest, then the pristine liar is not the correct
formulation of the problem. What strengthened liar paradoxes, such as
"This sentence is false or neither true nor false", do, is to remind us of
this fact: If we ever try to get out of the problem by taking a category
which is not the Rest, we can pose the original problem by describing
the Rest in some other way.
To summarise: the basic liar problem is that posed by a
construction which destroys the division between assertible sentences
and the Rest. As such it is the strengthened liar paradox which reflects
the central problem; the ordinary liar paradox is but a special case.
Thus a proposed solution to the liar paradox does not solve the essential
problem if it leaves the strengthened liar paradox wide open. Such
"solutions" are excellent examples of solutions which appear to solve
the problem, but actually merely succeed in transferring it to another
place or guise. (See sect.!.) Since most proposed solutions to the liar
paradox fall foul of some version of the strengthened liar, we could say

Priest Unstable Solutions to the Liar Paradox

that such "solutions" are only meta-stable and that the strengthening
construction is sufficient to destablise them. This is certainly true of
the GH construction as we will now see.

(ii) The Strengthened Liar: an informal account

In the particular case at hand, the bona fide truths, the assertible
ones are, as we have seen, the absolutely stably true ones (or just 'stably
true' if no confusion can result). The strengthened liar therefore takes
the form:

This sentence is not stably true.

If this sentence is stably true, it is true and hence not stably true. Thus,
the sentence is not stably true and we have established this as a logical
truth. Hence not only is it true, but it is stably true. (Since all logical
truths are stably true sect.2 Fact 1.) Contradiction. I will make this
argument more precise in the next part. But first let us examine what
Gupta has to say about it. (Herzberger does not mention it.) He says:

In response to this [i.e. the strengthened liar] I observe that the

notion of "stable truth" may be viewed in three ways. First, as belonging to
the metalanguage. This is the way we have used it above. We have used it
in the metalanguage to give an account of the concept of truth in the object
language L. This, it seems to me, does not in any way vitiate our account
of the concept of truth. Further, when the notion is viewed this way the
paradox does not arise. Second, the notion may be viewed as belonging to
L itself but under the condition that L has sufficiently weak syntax as far as
the predicate that expresses the notion of stable truth is concerned....
Under such conditions we can envisage formulating the entire
theory of truth given above in the language L itself. (Of course we will
need other notions as well, and the technical details will be messy, if not
overwhelming.) The paradox still does not arise. Third, the notion may be
viewed as belonging to L when L does not meet the condition of
sufficiently weak syntax. Now the paradox is present for the concept
"stably true in L". But we must ask how is the concept "stably true in L"
added to L? It must be added, it would appear, via a rule of revision. But
then can we not give an account of the new paradox parallel to that we gave
of the old? [1982], pp.55-6.

Gupta offers three possible responses. Let us take these in tum:

a) 'Stable truth' is part of the metalanguage but not the object

language. This merely resurrects the levels of language notion of
Tarski. I find it rather sad that Gupta, after so much ingenuity, falls
back on this tired old distinction. If this is necessary to get us out of

Priest Unstable Solutions to the Liar Paradox

trouble, we might just as well have stuck with the original Tarski
construction. Of course this construction will no do. Its inadequacies
have been pointed out by many, including Gupta himself,15 and it is a
simple matter to transfer these arguments to the new situation.

b) 'Stable truth' belongs to L itself, but the self referen~ial

machinery necessary for paradox is absent from L. This line is no
more promising. Gupta shows that a theory can consistently contain all
instances of the T -scheme provided certain syntactic machinery is
missing. 16 We can avoid the extended liar paradox in a similar way.
However, again, if this is an adequate way out of paradox, we might as
well have taken it in the first place. But it is not, as I am sure Gupta
realises. No one doubts that virtually anything can be avoided if one
weakens the expressive power of the language sufficiently. This is
beside the point. For the problem arises with our ordinary
concepts/language and our ordinary means of expression. It was the
correct semantic analysis of these that is in question and not that of
some castrated language.

c) 'Stable truth' is part of L, but its semantics is given by a "rule of

revision". What exactly Gupta has in mind here is not clear. However,
it would seem that the suggestion is somewhat disingenuous. What is in
question here is precisely a language which can express its own
semantic concepts and in which we can formulate Gupta's "entire
theory of truth.. .in the language L itself'. But then the term "stable
truth" does not have to be added to the language. It is a notion defined
on the basis of the given vocabulary (and defined moreover without the
use of'T') in just the way that Gupta and Herzberger show. Whatever
else can be "added" to the language by a "rule of revision", the
semantics is already inconsistent.

The fact that these rather inadequate remarks are thrown in,
almost as an afterthought to the paper, suggest that Gupta shares the
inverted perspective of the significance of strengthened paradoxes that
I criticized in the previous part.

(iii) The Strengthened Liar: a formal account

In virtue of some of the slippery points involved in the issue of

strengthened paradoxes, it is desirable to make the criticism of the
previous section more rigorous and precise. To this end a small
theorem is useful.

Priest Unstable Solutions to the Liar Paradox

Let L be a first order language which contains the language of

first order arithmetic and, in addition, the truth predicate 'T'. Let
{Ma I ae On} be a GH interpretation for L such that Mo (and thus all
the Ma ) are extensions of the standard model of arithmetic (or at least
are models of Peano Arithmetic). Let M~ be any stabilsed model, and
suppose that the set of absolutely stably true formulas of L with respect
to Mo is defined in M~ by a formula of L with one free variable
ST(x), i.e. cp is absolutely stably true iff M~~ST (sQ.)
Now let M be any model which is the same as Mo except perhaps
for the extension of 'T'. Since M extends the standard model of
arithmetic, we can code up formulas in the usual way and apply the
diagonallemma17 to fmd a formula cp (independent of M) such that
M ~ 'I' Ei! ...., STUll) (1)
'I' is, of course, just the strengthened liar sentence.

'I' is not absolutely stable.
Suppose 'I' is absolutely stable.Then
M~ ~ TUIl)='I' (Section 2, Fact 3).
i.e. M~ ~ T(JIl} =....,ST(JIl} by (1).

But M~ ~ ST(JIl}=> 'I' since ST defines the set of stably true

sentences and M~ is stabilised.
Hence M~ ~ ST(JIl}::J....,ST(JIl}
i.e. Mp~ ....,ST(JIl}
i.e. Mp ~ 'I' by (1).
So since 'I' is absolutely stable and true in some stablised model,
it is absolutely stably true,
i.e. Mp ~ ST(JIl}. Contradiction.

Let us, for the sake of interest, record a few corollaries.

Priest Unstable Solutions to the Liar Paradox

Corollary 1
The set of absolutely stably true formulas of L with respect to
Mo is not arithmetic.
If it were, there would be a purely arithmetic formula of one
free variable ST(x) which would define the set of stably true
formulas in every extension of the standard model of arithmetic.
Since its extension does not vary, ST Clil) is absolutely stable, as,
therefore is ..,ST{)Jl) which contradicts the theorem.

Corollary 2
The set of globally stably true formulas of L with respect to Mo
is not arithmetic.
Simply rework the whole proof of corollary 1 with 'globally'
replacing 'absolutely'.

Corollary 3
The set of absolutely (or globally) stable formulas of L with
respect to MO is not arithmetic.
The proof is essentially as for absolutely (globally) stable truths.
The major difference is that we suppose Stab(x) to define the
set of stable formulas and then let:
M 10 \jI=..,(Stab ()Jl)" T(:y!
The other modifications are relatively minor.

Let us now return to the philosophical import of the theorem.

For it is sufficient to sink the GH construction as a solution to the
paradoxes. It forces a dilemma: The theorem holds for all
(interpreted) languages which contain the language of arithmetic (with
its correct interpretation). Now take for L the language used by Gupta
and Herzberger themselves. This contains the language of set theory
and we may certainly therefore take it to contain the language of
arithmetic. Gupta and Herzberger do not tell us what the interpretation
of their language is supposed to be, yet clearly the arithmetic language
must be given the correct interpretation and, on pain of self-refutation,
the interpretation must be of the form {Ma I (Xe On} where this is a

Priest Unstable Solutions to the Liar Paradox

GH hierarchy. Now consider the predicate of this language "absolutely

stably true". Either this does not refer sensibly to the absolutely stably
true sentences, or if it does we are faced with an equal absurdity. Take
the first hom of the dilemma. In this case the predicate defines the set
of stably true sentences in no stablised model. Surely, that it should do
this is a minimum condition necessary for us to use the phrase sensibly.
We may not care what its extension is at the vicissitudes of lower
models. But we want the phrase to mean the right thing at some
sensible model. In other words, on this hom of the dilemma all the
facts amassed by Gupta and Herzberger concerning absolutely stable
sentences don't mean what they take them to mean. This is surely
The other hom of the dilemma is that the predicate does mean
what it says, at least in some stablised model Mp- But in that case, the
theorem shows that'll is not stable and a fortiori not stably true. But in
the interpreted language we are using, this is equivalent to 'II (by 1).
Thus we have proved, and are therefore committed to asserting
something unstable-which is "as true as false". This contradicts the
conclusions about assertion at which we arrived at the end of sect.3.
Either hom of the dilemma is unpleasant, but we can tum the
screws even harder. For after all, we know how "stably true" was
defined in the use-language:

x is absolutely stably true iff for all M' which differ from Mo in at
most the extension of 'T', 3aEOn'v'l3~a M'p~x.

Call the whole definiens (2). "Stably true" is thus defined in L without
using 'T'. Hence its extension is the same in any structure which differs
from MO in at most the extension of 'T'. Thus for any cp, ' cp is (not)
stably true' is stable. Hence'll is stable (by (1. But the theorem
proves that 'II is not stable. Hence the whole construction is
inconsistent. .
It seems to me that the only possible way out of these
problems-a fairly desperate one-is to deny that our use-language can
be identified with any language of the form of L. One might argue, for
example, as follows. Consider the formula (2) of L. It does not define
in M 0 (or any other M a -it doesn't matter since its extension is
constant) the set of stably true sentences. What (2) defines in Mo is the

Priest Unstable Solutions to the Liar Paradox

set of sentences stably true with respect to the ordinals of Mo,

On Mo (assuming that 'Io' receives its correct interpretation in MO ).
Now since Mo is a set, On Mo cannot contain all the ordinals. In fact, it
can contain only those ordinals a such that a<A for some A. But then
the extension of (2) in Mo is just the set of formulas that are locally
(absolutely) stably true at A. This will, in general, differ from the set
of globally (absolutely) stably true formulas. It follows that when we
refer to the stably true formulas of L we cannot be using L itself. We
must be using a different, in fact stronger language (one whose
quantifiers range over all, or at least more, ordinals). Thus L can at
best be part of our language. Our language itself must be conceived of
as a hierarchy of languages each of which is of the form of L and each
of which is adequate to express the semantics of a lower one. Hence we
are off up the Tarski hierarchy of metalanguages again, not, this time,
with respect to Tarski semantics, but with respect to GH semantics.
Nothing has been gained: we can simply tum Gupta's own arguments
(and all the others against the hierarchies of language approach) against

(iv) The Inevitability of Semantic Ascent

Moreover, semantic ascent of this kind is no accident. The aim

of the problem is to solve the semantic paradoxes. These appear to
occur when a language can express its own semantic notions. Thus the
aim requires us to give a consistent semantical theory that can handle
the semantics of the theory itself. ls But such is a classical chimera. For
to give an adequate account of the semantics of a theory we require at
least the following. First we need to spell out an interpretation of the
language in question. This may be an absolute truth definition, a GH
construction, or whatever. We then demand that the theory be proved
sound with respect to this interpretation. (We may also demand a proof
of completeness with this notion suitably formulated. However, a
proof of soundness is a minimum necessary condition for claiming to
have given a suitable semantics for the theory.) But classically,
soundness implies consistency. Hence giving a semantics for the theory
entails proving consistency. Now if all this can be done in the theory
itself, it follows that the theory can establish its own consistency. But
provided that the theory is "sufficiently strong" and based on cla~sical
logic, we know that no consistent theory can prove the canonical
assertion of its own consistency. This is Goedel's second

Priest Unstable Solutions to the Liar Paradox

incompleteness theorem, and the failure of the GH construction but a

corollary of this.
What the second Goedel incompleteness theorem shows is that,
classically, consistency can be maintained only by giving the semantics
of a theory in a different theory. Thus any (consistent) theory must fail
to be capable of giving its own semantics either by the requisite notions
failing to be expressible in the language of the theory, or by requisite
principles about them failing to be provable in the theory. The theory
must therefore be either expressively incomplete or proof-theoretically
incomplete. 19
To summarize: incompleteness is the price paid for consistency.
All the "solutions" to the semantic paradoxes ring the changes on this
theme, one way or another. In particular, when our proof procedure is
naive, so that there are no prior axiomatic constraints on provability,
we must have expressive failure, i.e. we must, to be consistent, consider
ourselves to be using a meta-language. This is why this idea comes up
again (Tarski) and again (Kripke) and again (Gupta and Herzberger).
Each of these solutions sought to improve on the former, but each falls
to what is merely a new way of expressing the same point. The
solutions are therefore inherently unstable and a fundamental cause of
the degeneration of the paradox-solution research programme is

5. In Praise of Inconsistency
We have seen that the programme of solving the paradoxes is
doomed to failure. Yet we still need a coherent approach to the
paradoxes, and we still need to understand how a language such as
English can handle its own semantic notions. How is this to be done?
We will find a way suggested if we look at a third criticism of the GH
construction. This one deals with the heuristic of the construction.
In his paper [1982a] Herzberger spells out some of the heuristic
ideas behind his technical construction. In particular, Herzberger
invites us to consider how, by simple steps of reasoning we obtain the
flip-flop pattern associated with the liar paradox. Let us spell this out
in slightly more detail. Suppose that'a is a name of the sentence 'a is not
true'. Then a is either true or not. Without loss of generality suppose it
is the former. So

a is true

i.e. 'a is not true' is true

Priest Unstable Solutions to the Liar Paradox

i.e. a is not true.

Call this progression a. This chain of reasoning is supposed to be

modeled in some sense by the construction which takes us from Ma to
M a+l' and specifically the change in the extension of T. Thus
Herzberger says:
I believe that this kind of construction does so far incorporate the "ordinary
rules" [i.e. of inference] that Wittgenstein remarked upon; ...There is indeed
an inconsistency between the valuations at one stage and those at another.. '!
offer it as a reconstruction of what some people have felt to be the
inconsistency of natural language.... [1982a], p.487-8.

But is it? The process of someone making inferences in

accordance with a is a process in time, and though Herzberger does not
say that the construction through the ordinals is supposed to model a
temporal progression, the language he uses makes it difficult not to
think of it in those terms (e.g. "On this picture, our language has an
inner dynamics of a highly regular sort, based on a process of
progressive semantic evaluation" [l982a], p.492. My italics.) And if
we think of the progression as temporal, the idea of the extension of the
truth predicate changing over time becomes quite a tempting one.
After all, novel arguments do force us to revise what we take to be true.
However, the illusion of temporality is a spurious one. Though
someone who reasons through a may make a temporal progression, the
progression a itself is not a temporal one, but a progression of logical
support, which is timeless. Once one grasps this then the thought that
when we arrive at the end of a we must change our interpretation to
bring it into line with our conclusion is not at all enticing. Indeed, such
a change, changing as it does the truth of the premise on which the
conclusion was based, undercuts the very rationale for making that
change. What the progression a shows us is that, far from anything
changing, the truth of 'a is true' commits us to the view that it is not
true too at the same time.
In fact, although Herzberger claims that in "Naive semantics
[i.e. his semantics], the paradoxes will not be made to disappear"
([l982a], pA80), this is precisely what they do. And they are made to
disappear by enforcing a misplaced temporal metaphor,2o to try to
make the talk of change natural. Thus the natural circular image of the

Priest Unstable Solutions to the Liar Paradox

a is true ais not true

is changed to a spiral one:

Time (The oniin(lls)

ais not true

a is true
ais not true

(See Herzberger [1982a], pp.483-4.) But the temporality is out of

place. The picture of the circle is much more appropriate than the
spiral. Indeed, if a picture is desired, the best one is that of a Moebius

C aisnottrue ~

To summarise: natural reasoning itself forces us to the
conclusion that both the liar sentence and its negation are true; that it is
both true and not true. It is not, therefore, the extension of the truth
predicate that needs revising (whatever, in the end, this is supposed to
mean) but our incorrect rejection of this fact. This outcome is perhaps
a strange one. But there is little that is truly novel in philosophy, and
we can find travellers that have been here before us. One such is Hegel.
Hegel claimed that any two contradictory categories, A, AI, have an
"interpenetrating" boundary. That is, that the concepts (or perhaps

Priest Unstable Solutions to the Liar Paradox

better the analytic principles characterising them) force us to recognise

the existence of something per impossibile in both categories; indeed,
such that the thing's being in A is exactly its being in AI. Hence we are
forced to recognise a novel category, the supersession (or synthesis) of
A and A I, which transcends the distinction. (It is the "being of each in
the other".) Whatever we are to make of this claim for all categories, it
is certainly true for the pair true/false. For the liar paradox forces us
to the conclusion that something is both. Indeed, the liar sentence's
being false is exactly its being true. Thus we must accept the category
of dialetheia -something both true and false. 21
Moreover, Hegel very astutely realised what would happen if
one tried to consistentise the situation: the result would be an instability
that would ultimately be futile (the only resolution of the situation
being to recognise the transcendent category). Bearing in mind that
Gupa and Herzberger have projected the oscillation into a realm that
Hegel could never have guessed at (the transfinite) let us allow Hegel to
speak for himself:

If we let somewhat and another, the elements of determinate Being, fall

asunder, the result is that some becomes other, and this other is itself a
somewhat, which then as such changes likewise, and so on ad infinitwn.
This result seems to superficial reflection something very grand, the
grandest possible... [However the] progression to infinity never gets further
than a statement of the contradiction involved in the finite, viz. that it is
somewhat as well as somewhat else. It sets up with endless iteration the
alternation between these two terms, each of which calls upon the
other... [So] such a progression to infinity is not the real infinite. That
consists in being at home with itself in its other, or, if enunciated as a
process, in coming to itself in its other. Much depends on rightly
apprehending the notion of infinity, and not stopping short at the wrong
infinity of endless progression.
Hegel [1830] sect.94. (Quotation rearranged.)

As Hegel insists, we must recognise dialetheias. I do not deny

that this raises important philosophical issues; but it is a necessary first
step in an adequate approach to the semantic paradoxes and to
semantically closed languages. I shall not try to explain this in detail
here. Much of it is already in the literature. I will, however, briefly
explain its salient features and its relation to some of the points I have
First, the idea that a sentence may be both true and false (or that
both a sentence and its negation may be true) must be built into a
coherent semantics (e.g. as in Priest [1979], [1980]). We can then take
the paradoxical arguments to be what they appear, prima facie, to be,
viz. sound arguments with contradictory conclusions. In particular, it

Priest Unstable Solutions to the Liar Paradox

is possible to produce inconsistent but non-trivial theories, such as set

theory, based on these semantics (e.g. Routley [1977], Brady [1986]).
Moreover it is possible to produce a semantically closed theory22 e.g.
an axiomatic theory which can prove all instances of its T -scheme. (See
Priest and Crosthwaite [198+].)
A few further points are worth noting in the present context.
First the semantics can be fitted into any orthodox theory of meaning
(for example a Davidsonian one; see Priest and Crosthwaite [198+]),
which can therefore be taken as providing an analysis of the meaning of
'is true'. Hence we have a satisfactory account of our naive conception
of truth. Moreover, the T -scheme is, what it appears to be, an analytic
principle governing the concept of truth, and not a "hasty
generalization" .23 Second, the standard connection between (absolute)
truth and assertibility is preserved. Third, any version of strengthened
paradoxes one cares to formulate (if indeed there is any way one can
sensibly distinguish them from the ordinary variety) can be handled in
the same way as the ordinary variety, viz. left to stand. Fourth, the
flip-flop behaviour of the sentence 'This sentence is not true'. (Suppose
it is true; then it is not true; then it is true ... ), which is cited by Gupta
and Herzberger in favour of their account, is explained in the best
possible way: it is a valid sorites. Finally, this approach provides once
again the possibility of a uniform treatment of the logical paradoxes.24
This approach to semantic closure clearly exhibits highly
desirable features. Of course, it goes without saying that the logic
generated by the dialetheic semantics is not classical. For oth~rwise the
contradictions involved would trivialise the issue. In this context it is
worth saying a final word about Herzberger's attitude to classical logic.
Herzberger defends his use of classic81logic partly on the ground that it
is classical logic which in fact produces the paradoxes:

[My construction] uses only ordinary models and classical two-valued

valuations. This seems appropriate, inasmuch as it is reasoning in
accordance with classical logic which in the fIrst instance gives rise to the
semantic paradoxes. [1982], p.61.

This is incorrect. It is naive reasoning (concerning the concept of

truth) which generates the paradoxes. The liar paradox was known
2,000 years before "classical" model theory; and intuitionistic (and
relevant) logic equally produce the contradiction. Of course it could be
that classical logic is an adequate formalization of our naive reasoning
procedures. But this certainly cannot be assumed without further ado.
It is slightly disconcerting to find that many logicians, including
apparently Herzberger, have forgotten that classical logic is just a

Priest Unstable Solutions to the Liar Paradox

theory of what naive reasoning is, with both strengths and weaknesses.
(It is clear that the logical paradoxes are an Achilles' heel of classical
logic, one that will, in the end, I think, take it in the same direction as
Achilles.) For example Herzberger says:

All standard [semantic] schemes are weaker than the classical valuation
scheme, and consequently no one of them seems to be altogether free from
intuitive wrinkles. [1982], p.92.

The remark is a casual one made in the context of discussing valuation

schemes within Kripke's theory of truth. Nonetheless, its implication is
clearly that deviation from classical logic is ipso facto a defect
("wrinkle"), and, correlatively, that classical logic has no "wrinkles".
This is, of course, not the case. For one thing, it is the very strength of
classical logic which has been under attack from various directions
throughout this century (Brouwer, C.I. Lewis, Anderson and Belnap).
In a time of normal science (to use the language of T.S. Kuhn)
the dominant theory is so much taken for granted that its problematic
nature is suppressed into the "collective subconscious" of the scientific
community. It is time that the problematic nature of classical logic is
firmly brought back into the "conscious". The logical paradoxes are
just the thing to do this.25

6. Appendix

The preceding sections of this essay were written in 1982. Since

they were written some other papers using constructions related to the
GH construction have appeared. In this appendix I want to comment
briefly on one of these, Yablo [1985].26 (Subsequent page references
are to this.) There are many of the philosophical comments in Yablo's
paper that I quite agree with. There are also a number of things that I
disagree with. 27 However, I will, in this appendix, discuss only those
aspects of the paper which relat~ directly to points made earlier in this
An important difference between Yablo's construction and the
GH construction, is that whilst the latter works with classical
two-valued evaluations the former works with four-valued evaluations,
according to which a sentence may be assigned true (t), false (1), both,
or neither. In virtue of the fact that dialetheias are explicitly
countenanced, it might be thought that I should have no quarrel with the
construction. However, Yablo argues that supposing there to be
dialetheias does not solve the problem of strengthened paradoxes (p
302), and that a construction like his is called for. I will take issue with

Priest Unstable Solutions to the Liar Paradox

both these claims.

Let us take the first point first. The argument is premised on
the claim that truth is 'strong', in the sense that if cp is true (and possibly
false as well) then T,m. is true and true only, and if cp is not true (and
possibly not false either) T.t is false and false only. In other words, the
truth predicate is always classically valued. Now consider the extended

('V) 'V is not true

'V can have neither of the classical truth values for the usual reasons.
Moreover it can have neither of the other values since the truth
predicate is always classically valued. Thus 'V can have no consistent
Now, first, I would take issue with the claim that truth is strong.
If cp is true, then I certainly agree that T.t is true, by the T -scheme; and
if cp is not true, I agree that T.m. is false. But I see no reason to suppose
that T,m. cannot be false (as well as true) if <p is. I (now) think that it
mayor may not be false, depending on cpo (The reasons are explained
in ch. 4 of Priest [198+].) Yablo argues for his position; but his
argument seems to me not to be cogent. It goes thus (p 301):

'Pa' is true iff a has the property P.

'Pa' is false iff a does not have the property P.
Hence, taking T for P and cp for a:
, T.m.' is true iff cp is true.
, T.t' is false iff cp is not true.
It follows that truth is strong.

But how is it supposed to follow that T,gt cannot be both true and
false? This would follow only if it were impossible for <p to be both
true and not true, in other words, if the truth predicate behaves
consistently. But this is exactly what we are supposed to be showing.
To bring out the question-begging nature of the argument, note that if
it were .right it would prove not only that T is a classical predicate, but
that every predicate is classical. It would therefore under-cut the whole
rationale of four-valued semantics.
But even supposing that truth is strong, does the argument

Priest Unstable Solutions to the Liar Paradox

against the dialetheic account of extended paradoxes work? No. It

shows only that'll can be given no consistent evaluation. But the point
of dialetheism is to allow precisely for this. I claim that 'II is both true
and false. Assuming truth to be strong, it follows that'll is both true
and not false. This is, indeed, a contradiction. But equally, this is
exactly what we should expect. As I argued in section 4, the liar
paradox is a construction which violates all semantic boundaries-those
of four-valued semantics included. Dialetheism is designed precisely
to cope with this situation.
Even though the objection against a dialetheic account of the
paradoxes is incorrect, it brings home an important point. One cannot
adopt a dialetheic position on paradoxes without being inconsistent
oneself. Yablo (and some other writers)28 are prepared to countenance
a half-hearted dialetheic view according to which a sentence may be
both true and false, but they balk at describing situations inconsistently
themselves. But this cannot even seem reasonable unless we enforce a
rigid distinction between the (inconsistent) object theory and the
(consistent) meta-theory. And if this is a reasonable move then we
might as well invoke the object/metalanguage distinction to get rid of an
inconsistent object theory in the first place. But this is not reasonable,
as I discussed in section 4 above. A red-blooded dialetheism is the only
viable option.
Let us tum now to Yablo's own construction. How, exactly, this
is supposed to solve the paradox is not, surprisingly enough, spelled
out.29 Still, the construction is supposed to provide an analysis of the
notion of truth, and to take account of the paradoxes, in some sense.
The construction is a good deal more complex than the GH
construction, but it is similar in the following ways. A semantics is
defined for a language with its own truth predicate. This is done by
defining a hierarchy of interpretations by transfinite induction on the
ordinals. The interpretation of the truth predicate (which is the only
thing that varies in the hierarchy) at level a + 1 is produced by a
uniform operation on its extension at level a. During the process of
ascent through the ordinals, a certain stability emerges, and this can be
used to define a set of absolute categories for formulas of the language.
The exact details of the construction need not concern us here.
All we need note is that the construction provides a pair of "equally
good" evaluations, 0 and 0 (p 331) representing the limit situation,
in terms of which we may define (j) to be:

Priest Unstable Solutions to the Liar Paradox

true if both nand n assign t to cp;

untrue if neither Q nor n assigns t to cp;
false if both n and n assign f to cp;
unfalse if neither n nor n assigns f to cpo

These categories are neither mutually exclusive (except for the first
pair and the second) nor exhaustive. We note also that (p 332):

T~ is true iff cp is true;

T~ is false iff cp is untrue.

With these details under our belts we can see that exactly the
same objections apply to this approach as apply to the GH approach. In
particular, the same features of the degenerating research programme
are present. First, the account of truth, depending as it does on
transfinite ordinals and induction, is susceptible to the argument based
on the theory of meaning that I used in section 3. Yablo distinguishes
between the psychological problem of determining how people actually
operate with the notion of truth and the descriptive problem of
characterising truth (p 229-300). He might therefore object to this
argument on the ground that to criticise the characterisation in this way
is untoward psychologism. However, it is agreed that the descriptive
problem is essentially one of giving an account of meaning (fn. 5); and
though m~aning cannot be defined in psychological terms, there are
certainly psycholo~ical constraints on what cart count as an adequate
theory of meaning. 0
More crucially, exactly the same situation concerning the
strengthened liar paradox again arises. The Rest, in this construction,
is just the set of sentences that are not true. Yablo calls these non-true
to distinguish them from the untruths (p 331). Now, consider the

('V> 'V is non-true.

By the usual argument 'V is true iff it is non-true. Hence it is both true
and not true. Note that the conclusion is not that'll' is true and
false, which would be alright. Note also that Yablo's metalanguage is
quite classical. Thus, 'V is either true or it is not, and not both.
As in section 4, we can work this argument into a proof that the

Priest Unstable Solutions to the Liar Paradox

class of non-truths cannot be represented in the theory, in the following

sense: assuming the the base interpretation extends the standard model
of arithmetic, there is no formula of one free variable, N(x), which is
true of just the non-true sentences. 31 For if there were we could, by
diagonalisation, produce a formula, '1', which has the same value in any
evaluation in the hierarchy as NOlO. It follows that 'I' is true iff N(lIO
is true. But NOlO is true iff 'I' is non-true. Contradiction.
Since Yablo talks about the non-true sentences, it follows that if
he is to be consistent and mean what he says, he must be talking in a
language other than the one he is discussing, a metalanguge; the strategy
is thus forced into semantic ascent, as we noted in section 4 that it must
be. We noted also the self-defeating nature of this move if the aim is to
produce a semantically closed theory. Of course, since the underlying
semantics of the language does allow for things to be both true and
false, it would be possible for Yablo to refuse the semantic ascent by
accepting the contradiction, but only by himself becoming a dialetheist,
and, as we noted above, he is not prepared to be this red-blooded.
Moreover, if one is, then the motivation for the construction seems to
be under-cut. 32 A semantically closed theory, with truth behaving as
we think it does, is much more simply obtainable, as I indicated in
section 5.
Thus we see that Yablo's construction merely adds another
epicycle to the "solve the paradoxes" research programme, as any
attempt to face the semantic paradoxes and remain consistent, must.

Graham Priest
Department of Philosophy
University of Western Australia
Nedlands, Australia

Priest Unstable Solutions to the Liar Paradox


1. "A successful research programme bustles with activity. There

are always dozens of puzzles to be solved and technical questions
to be answered; even if some of these--inevitably-are the
programme's own creation. But this self-propelling force of the
programme may carry away the research workers and cause
them to forget about the problem background. They tend not to
ask any more to what degree they have solved the original
problem, to what degree they gave up basic positions in order to
cope with the internal technical difficulties. Although they may
travel away from the original problem with enormous speed,
they do not notice it. Problemshifts of this kind may invest
research programmes with a remarkable tenacity in digesting
and surviving almost any criticism.
Now problem shifts are regular bedfellows of problem
solving and especially of research programmes. One frequently
solves very different problems from those which one has set out
to solve. One may solve a more interesting problem than the
original one. In such cases we may talk about a 'progressive
problemshift'. But one may solve some problems less
interesting than the original one; indeed, in extreme cases, one
may end up with solving (or trying to solve) no other problems
but those which one has oneself created while trying to solve the
original problem. In such cases we may talk about a
'degenerating problemshift '." Lakatos [1968], pp.128-9.

2. A third variant is given by Belnap [1982].

3. This is an inessential modification of the GH construction.

4. Belnap gives yet a third possibility, viz. XA, +(U) u

(ZA,-XA,"(U where ZA, is an arbitrary (sub) set (of S).

5. Some care needs to be taken over the terminology. What I call

'globally stable', Herzberger calls 'stable' and Gupta calls
'relatively stable'. What I call 'absolutely stable', Herzberger
calls 'naively stable' and Gupta calls 'stable'. Gupta's
definitions of the various notions is also slightly different but

Priest Unstable Solutions to the Liar Paradox

6. Specifically, if cp is a sentence in language of order n+1 in the

Tarski hierarchy and SR is its name in the language of order n+1
(we can allow names for the sentences of all the languages to
occur in each language), then T nSR=CP may fail, where Tn is the
truth predicate in the language of order n+1.

7. The question of the sense in which an application of the rule in

general "improves" the extension of T is, fortunately, an issue
we can avoid.

8. See, e.g. Davidson [1967].

9. There are totally psychologistic accounts of meaning, such as

that of Grice. But such theories are incapable of dealing with
the compositionality of meaning.

10. It might be suggested that only the absolutely stable sentences

are meaningful, and hence that we require of a theory of
meaning only that it deliver the T -sentences for absolutely stable
sentences. However, this will not work since, as we shall see,
for sufficiently rich languages, the set of absolutely stable
sentences is not arithmetic. Thus, assuming that we can
effectively tell a meaning-giving sentence when we see one, the
truth-theory could not be axiomatic.

11. The construction must be iterated beyond the finite since, in

general, stabilisation will not occur at finite levels.

12. The point is made in Dummett [1973]. See esp.p.320.

13. I will return to the question of temporality in the final part of

the paper.

14. For the terminology, see Haack [1978], ch.8.

15. See his [1982], e.g. pp.27-30. My own shot is in Priest [1984].

16. [1982] sect.II. A simpler proof of this fact can be found in

Priest [1984].

17. See, e.g. Boolos and Jeffrey [1974], p.176.

Priest Unstable Solutions to the Liar Paradox

18. This was, of course, emphasized by Tarski. But the current

literature trying to .show how a theory can handle its own truth
predicate is a further narrowing of the programme of solving
the paradoxes,predicated on the assumption that all semantic
relations can be defined in terms of truth (or at least
satisfaction). But all the proposed theories work with other
semantical notions-such as stability-which cannot be defined in
terms of truth (or satisfaction). The very theories therefore
show the further narrowing of the programme to be untenable.

19. Thus, ZF cannot prove its own consistency because it cannot

quantify over proper classes, whereas NBG cannot prove its
own consistency since one cannot prove certain "impredicative"
classes to exist.

20. The strategy itself of avoiding contradictions by postulating a

temporal dimension is hardly a new one; it can be found e.g. in
Kant See von Wright [1968] sect.11.

21. The term was coined in Priest, Routley and Norman [1986].

22. Pace Herzberger [1982a], p.481.

23. Pace Gupta [1982], p.51.

24. There is perhaps one final observation worth making. Gupta

isolates a phenomenon he calls failure of local determination
[1982], p.2lff. What this amounts to in effect is a failure of
compositionality. The semantic value of a sentence is
determined by things other than the relevant semantic values of
its components. When we look at his proof that local
determination may fail, we meet an old friend: Curry's
paradox. In a recursively based account of (absolute) truth,
compositionality must hold. However, in a semantically closed
theory, the principle AH(A~B}/B must fail if Curry
paradoxes are to be avoided.

25. For a further discussion of these matters, see Priest [1986].

26. Another is Woodruff [1984], which I have discussed in Priest

[1984a]. A number of the comments made there also apply to

Priest Unstable Solutions to the Liar Paradox

Yablo's construction.

27. The comments on the genuineness, inevitability etc. of paradox

I quite endorse. But because of this, I do not think that it is
necessary to produce yet another construction which tries to
avoid them. I also disagree with Yablo's "cosmological
argument" for groundedness, pp 316-7.

28. E.g., Rescher and Brandom [1979]. See esp. ch 26.

29. A rather swift comment is made on the matter in fn.1. This is


30. For a discussion of these see Davies [1981] ch.1. Note also that
Yablo is not against using psychologistic considerations to try to
make his account plausible. (See, e.g., p 330.)

31. Note that, in particular, -,T(x) will not do. -,T(x) is true iff
T(x) is false iff x is untrue.

32. See Priest [1984a] section 4.

Priest Unstable Solutions to the Liar Paradox


Belnap, N., "Gupta's Theory of Truth," Journal of Philosophical

Logic, 1 (1982) 103-16.
Boolos, G. and Jeffrey, R. Computability and Logic. Cambridge
University Press, 1974.
Brady, R., "The Non-Triviality of Dialectical Set Theory," in Priest,
Routley and Norman [1986].
Davidson, D., "Truth and Meaning,"Synthese, XVII (1967) 304-23.
Davies, M. Meaning, Quantification, Necessity. Routledge and Kegan
Dummett, M. Frege. Duckworth, 1973.
Gupta, A., "Truth and Paradox,"Journal of Philosophical Logic, 11
(1982) 1-60.
Haack, S. Philosophy of Logics. Cambridge University Press, 1978.
Herzberger, H., "Notes on Naive Semantics," Journal of Philosophical
Logic, 11 (1982) 61-102.
Herzberger, H., "Naive Semantics and the Liar Paradox," Journal of
Philosophy, LXXIX (1982) 479-497. [1982a]
Hegel, G.W.F. Logic: Part One of the Encyclopedia of the
Philosophical Sciences: English translation by W. Wallace.
Oxford University Press, 1975 (originally published, 1830).
Lakatos, I., "Changes in the Problem of Inductive Logic," Ch.8 of
Vo1.2 of Lakatos' Philosophical Papers. Eds. J. Worrall and G.
Currie. Cambridge University Press, 1978 (originally
published 1968).
Montague, R., "English as a Formal Language," Ch.6 of Montague's
Formal Philosophy. Ed. R.H. Thomason. Yale University
Press, 1974.
Priest, G., "Logic of Paradox," Journal of Philosophical Logic, 8
(1979) 219-41.
Priest, G., "Sense, Entailment and Modus Ponens," Journal of
Philosophical Logic, 9 (1980) 415-35.
Priest, G., "Semantic Closure," Studia Logica, 43 (1984) 117-129.
Priest, G. "Logic of Paradox Revisited," Journal of Philosophical
Logic, 13 (1984) 153-179. [1984a]
Priest, G., "Classical Logic Aufgehoben," in Priest, Routley and
Norman [1986].
Priest, G. In Contradiction, to appear. [198+]
Priest, G. and Crosthwaite, J., "Relevance, Truth and Meaning," in
Directions of Relevant Logic. Eds. R. Routley and J. Norman.
Martinus Nijhoff (forthcoming). [198+]

Priest Unstable Solutions to the Liar Paradox

Priest, G., Routley, R. and Norman, J. Paraconsistent Logics

Philosophia Verlag, 1986.
Ramsey, F., "The Foundations of Mathematics," in The Foundations of
Mathematics and other Logical Essays. Ed. R.B. Braithwaite.
Routledge and Kegan Paul, 1931 (originally published 1926).
Rescher, N. and Brandom, R. The Logic of Inconsistency. Blackwell,
Routley, R., "Ultralogic as Universal," Appendix to Exploring
Meinong's Jungle and Beyond. Australian National University,
Research School of Social Sciences, 1980 (originally published
Woodruff, P., "Paradox, Truth and Logic. Part I: Paradox and Truth,"
Journal of Philosophical Logic, 13 (1984) 213-232.
von Wright, G. Time, Change and Contradiction. Cambridge
University Press, 1968.
Yablo, S., "Truth and Reflection," Journal of Philosophical Logic, 14
(1985) 297-349.

Part III. Specific Relations
w. D. Hart
Causation and Self-Reference
Here, in broad strokes, is a complex of ideas: It seems patent
that there is self-reference in natural languages like English: a sentence
like, "This sentence contains five words," seems undeniably true and,
therefore, perfectly in order both syntactically and semantically. As is
well known, causal conceptions of reference have recently been bruited
about the literature. But if reference is causal, then it would seem to
follow that self-reference must involve self-causation. On the other
hand, even if it is not exactly clear why, it does seem that the causal
relation cannot be reflexive. It seems to me premature to conclude
forthwith that these ideas are a reductio ad absurdum of causal
conceptions of reference: consequently, I should like to explore them
There has been considerable philosophical suspicion of
self-reference in this century. Much of this is best understood as a
reaction to semantic paradoxes like the liar. As a solution to the liar,
ruling out self-reference seems to me like using cannon to hunt
sparrows. It probably will not work, since gOdelization can imitate the
net effect of self-reference. It is hard to do, since there is no guarantee
that what Kripke calls accidental self-reference can be recognized as
such. 1 (Suppose I say, "My last sentence will be self-referential," fall
silent for a while, and then am killed unexpectedly; when did my
remark become illegitimate? How much should grammar or semantics,
or even pragmatics, depend on the vicissitudes of my career?) But most
of all, it is excessive. Even if one could be sure that the paradox of the
liar is the fault of self-reference in liar sentences, that is not enough to
show that all self-reference is illicit. Some self-referential sentences,
like "This sentence is the second sentence mentioned but not used in this
paragraph," look plainly true, and their truth should create an
overwhelming presumption of their syntactic, semantic, and pragmatic
legitimacy.2 I think that it is a desirable material adequacy condition on
solutions to the semantic paradoxes that some self-reference be
But this is not to say that we understand self-reference. First,
there are perplexing analogous cases. For example, could there be a
representational picture whose subject is wholly and entirely itself?
And if not, why not? Following Kaplan, 3 we might suggest that the
subject of a photograph must figure in its causation, and then appeal to
Hart Causation and Self-Reference

the irreflexivity of causation: perhaps, but that might seem to reinforce

the complex of ideas rehearsed at the outset. Or on the psychological
side, could there be a thought whose subject is wholly and entirely it
self? Could the sentence, "I am now having the thought expressed by
this very sentence," or "I am now having this thought," ever be true?
And if not, why not? If there is self-reference in language but not in
thought, what does that show about the relation between thought and
language? The extent of self-reference is not clear.
Second, and perhaps slightly less speculative, there may be a
problem for the theoretical characterization of self-reference. It has
become a commonplace of structural linguistics that grammar should
be generative. Generation requires an induction. That this should be
possible and non-trivial, it is at the very least necessary that there shold
be an irreflexive ordering of sentences by grammatical complexity.
Then too, it is now a received view in some philosophical quarters that
the core of semantics should be a truth theory in the style moulded by
Davidson out of Tarski. On such a view, denotations for proper names,
satisfaction conditions for simple predicates, and truth conditions for
atomic sentences are given outright; then truth conditions for more
complex sentences are given in terms of those for less complex
sentences. Such a semantics also requires an induction. That this
should be possible and non-trivial, it is at the very least necessary that
there should be an irreflexive ordering of sentences by something like
semantical complexity. David Kaplan asks whether generative
grammar and inductive truth theory mesh; he calls a language logically
perfect if there is a certain one-to-one correspondence between its
formation rules and its evaluation rules.4 This is most naturally and
interestingly the case when the (minimally) irreflexive orderings
underlying its generative grammar and its inductive semantics
coincide. Call such a language strictly logically perfect. What is not
clear to me is this: are the irreflexivities required for the generative
grammar and inductive semantics o~ stricly logically perfect language
compatible with a natural account of the confusing but obvious
reflexivities of its self-referential sentences?
To be more explicit, let me introduce a device due to my former
student, Graham Curtis. Suppose that in a semi-formal language, say a
regimented fragment of English, we have a singular term 'this S'
intended to be read as 'this sentence.' In order to make the denotation
of an occurence of it clear, and in order to preserve a small scope
reading, and thus truth, in the or-introduction inference from "This
sentence contains five words" to "Grass is orange or this sentence
contains five words," we want to attach a pair of arms to each
occurence of 'this S'. These arms should reach out and grasp the ends

Hart Causation and Self-Reference

of the sentence denoted by an occurence of 'this Sf. Thus, for example,

(this S)

would represent, "This sentence [namely, cll is 'If." and

nn 'If (this S)

would represent the self-referential, "This very sentence is 'If."

What should the formation rules for 'this S' be? A mechanical
use of past experience might suggest: If cI is a sentence and 'If(x) is a
wff with just one free variable, then

'If (this S)

is a sentence. This syntax makes self-reference ungrammatical. It

requires that the sentence cI be generated before the sentence "This
sentence [namely, cll is 'V" referring to it, and that would require
self-referential sentences to be generated before themselves, an obvious
violation of the irrefiexivity of the ordering by grammatical
We might instead think that since its arms point to the denotation
of an occurence of 'this S; and since denotation is part of semantics
rather than syntax, arms have no place in grammar. Thus, we can form
sentences as usual, treating 'this S' as a unique ordinary term and
amputating its arms. Then, after all the sentences have been generated,
arms may be atached to occurrences of our demonstrative in order to
specify for semantics their denotations. Done this way, there is no
grammatical problem about getting self-referential sentences.
But what should the truth conditions for sentences containing
'this S' be? A natural idea is to say that a sequence s satisfies

Hart Causation and Self-Reference

(this S)

if and only if the sequence which differs from s by having the sentence
cit in its ith place satisfies 'I'(xi). (The variable' Xi' should be foreign to
(this S). This idea requires sentences of the language to be members of
its domain of interpretation. I so construe reference to sentences that it
is possible to mention a sentence without asserting it: thus the truth
condition above can be met even if cit is false.) In the self-referential
case, these truth conditions become: s satisfies

nn 'I' (this S)

if and only if the sequence differing from s (only) by having

nn 'I' (this S)

in its ith place satisfies V(xi).

Whether or not there exists a well-founded satisfaction relation
of the sort required by these truth conditions, and certainly whether
one can be proved to exist aby the usual techniques, depends on the
expressive powers of our semi-formal language, on what substitutions
'I' may have.5 If 'I' may be "is false," then the liar paradox is a proof
that there are no such truth conditions. When'll is replaced by a purely
syntactical predicate, like "is a sentence" or "contains five words," then
granted that all the relevant syntax can be done prior to the necessary
semantics, the usual inductive argument for the existence of the desired
satisfaction relation should go through. But prima facie natural
languages also contain some of their own semantics. For example, a
sentence like "There occurs in this sentence a singular term which
denotes this sentence" seems plainly true and, therefore, syntactically
and semantically perfectly in order. We are thus brought up short
against the great old question: What are the maxims of its own
semantics which a language with self-reference can contain without

Hart Causation and Self-Reference

paradox? It is not obvious to me that achieving such a maximum is

always compatible with strict logical perfection. A sentence like "This
sentence is true" is not paradoxical but involves an obvious irreflexivity
that makes me dubious about inductive theories of truth.
That is as much as I wish to say here about self-reference per se.
Now let us look at the irreflexivity of causality. I do not wish to argue
that causal chains cannot close. That thesis involves, I think, physical
questions (from, most importantly, general relativity) most of which I
am not competent to discuss. Then too, since it would require
considerable chutzpah for a causal theorist of reference to invoke
general relativity to account for self-reference, heavy physics seems
out of place here. Instead, I want to ask what reasons we could give for
our usual presumption that causal chains cannot, or at least do not,
close; we want that sort of understanding which consists in seeing
This seems as much a matter of the nature of causation as
anything else. Take, for example, a regularity theory of causation.
Crudely put, the idea is that a particular event a causes a particular
event b if and only if there are event types (or, perhaps better, kinds),
X and Y, such that a is of type X, b is of type Y, and events of type X
are constantly conjoined with events of type Y. It is well-known that
constant conjunction is not enough for causation. The falling of
barometers is constantly conjoined with rain, but does not cause it;
instead they are joint effects of common causes. Moreover, bare
constant conjunction seems to require that causality be reflexive unless
we just baldly insist that constant conjunction be irreflexive. Let a be
any event and let X be any kind of type of event such that a is of kind X.
Then events of kind X are constantly conjoined with events of kind X,
namely, themselves; so a causes itself. To build irreflexivity into
constant conjunction by fiat would simply push our question back a
Of course, Hume never suggested that constant conjunction
alone is sufficient for causality. He also required spatial continuity and
temporal priority. This will not solve the problem about rain and
barometers, but if temporal priority is always irreflexive, it would
entail that causation is also irreflexive. But this seems to me at best a
shallow solution; I do not think I know any more why events cannot
precede themselves than I know why events cannot cause themselves. I
am not satisfied with regularity theories.
But there is another view. Writing about causation, Quine says

It may have had its prehistoric beginnings in man's sense of effort,

Hart Causation and Self-Reference

as in pushing. The imparting of energy still seems to be the central

idea The transfer of momentum from one billiard ball to another is
persistently cited as a paradigm case of causality. Thus we might
seek a simpleminded or root notion of causality in terms of the flow
of energy. Cause and effect are events such that all the energy in
the effect flowed from the cause. This thermodynamical image
requires us to picture energy, like matter, as traceable from point to
point through time. Thus let us picture an event simply as any
fragment of space-time, or the material and energetic content
thereof. Given an event e, then, imagine all its energy traced
backward through time. Any earlier event that intercepts all of
these energetic world lines qualifies as a cause of e. 6

I think Quine is right. I do not much like his uncharacteristic mention

here of a notion of causation, since I think a person could have this idea
but lack the idea of conserved energy, while I think it is the
conservation of energy which makes it central to causation. I would
rather say Kripke-like that we now know (if we do) a posteriori that
energy flow is the essence of causality. (Much too briefly, my thought
is that we know a priori that the conservation of some quantity or other
which is traceable along causal chains is a necessary condition for
(thorough-going) causal explanations in natural science. This I think of
as a transcendental thesis in something like Kant's sense: I also think it
yields an intrinsic explanation of the overwhelmi~gly quantitative (Le.,
mathematical) character of mature natural science, a phenomenon
which on most philosophies of natural science looks utterly accidental.
Then nature co-operates, and we learn from experience with her that
there exists in her a satisfactory basic conserved quantity traceable
along causal chains. This is energy, or mass-energy.)
Assume then with Quine that causation is energy flow. We can
then do (at least) two pleasant things. First, falling barometers precede
and are constantly conjoined with rain but do not cause it. In fact, none
of the energy involved in condensing rain drops out of water vapour
flows from the barometer to the atmosphere, (and if it did, falling
barometers would cause rain.) Instead, as heat energy flows out of the
air, its capacity to retain water vapour declines until eventually some
condenses out as rain drops; and in the oldest barometer, human hair,
this increaased relative humidity also results in some of the water
vapour being picked up by hair which then contracts, a contraction we
now register as falling. In short, as heat in the air declines, the
attractive forces of water molecules for the nuclei of raindrops can
overcome the dissociative heat energy; then this chemical energy is
transformed into the mechanical energy of the motion of barometer's
needle. So the energy flow view gets exactly right a prime
counter-example to the regularity theory of causation.

Hart Causation and Self-Reference

Second, we can account for the irreflexivity of causality if we

assume that energy is conserved. Conservation principles take two
foons, global and local. Global conservation means that you cannot get
something for nothing (or vice versa): energy is neither created nor
destroyed, so the total amount of it in the universe is fixed. Local
conservation entails that there is no action at a distance. Energy is
everywhere locally conserved if the amount of energy in any volume
varies directly with the net amount of energy entering through the
surface of the volume. (Thus, there is no real energy source or sink in
the volume. If there were action at a distance, a cause, instantaneously
and without any intervening transmission, would lose energy acquired
by an effect at some remove from the cause. Since the effect is at a
distance from the cause, there is a volume containing the effect but not
the cause. So even if the loss at the cause balances the gain at the affect,
thus preserving global conservation, in that volume the amount of
energy increases without energy passing through the surface of the
volume. Hence, energy would not be everywhere locally conserved. In
ruling out action at a distance, local conservation is a continuity
principle. (In action at a distance, there must be no transmission, not
even infinitely fast: for otherwise the cause acts through all
inteonediate points on a path from it to the effect, and thus not
genuinely at a distance.
Suppose that energy is locally and globally conserved.
Reflexive causation would seem to require that energy, say five ergs
worth, come from an event e, qua cause, to e itself, qua effect; this
energy might pass through other events or it might not. Assume that
prior to e, the universe always contained a total of ten ergs. Then
unless the universe loses five ergs and loses it where e is, the occurence
of e will violate global and local conservation. But if the universe loses
five ergs at e, the conclusion seems forced that the five ergs lost by (the
rest of) the universe are lost to e and that the five ergs of e come not
from e but from (the rest of) the universe. (Recall that energy is not a
stuff, a fluid dripping through time. If two rocks, x and y, collide with
a third, z, and give up their kinetic energy to z, and if Z then collides
with two more rocks, u and v, and gives up its kinetic energy to them,
then there is no fact of the matter about whether the kinetic energy of u
is or comes from that of x or y. Parcels of energy are thus not
sufficiently individuated for there to be a distinction between (a) five
ergs being destroyed in the universe at e and a different five ergs being
created at e, and (b) the (rest of the) universe transmitting five ergs to
e.) Symmetrical remarks apply to the universe subsequent to e; thus, e
is fitted into the causal history of the universe. A similar argument will
work if the energy of e passes through other events so long as we can

Hart Causation and Self-Reference

speak of the energic contents of the universe before and after this causal
loop. Thus my argument does not show that local and global
conservation of energy entail that the entire universe could not have a
cyclical causal history (or that spacetime could not have a cylindrical
topology), but then neither do I think that follows. 7 If the regularity
theory yields this stronger result and does so by insisting a priori that
time is linearly ordered, then so much the worse for the regularity
theory. In sum then, I think Quine's energic view of causality yields a
more accurate, more physically sensitive, and deeper perspective on the
irreflexivity of causality (if it is irreflexive) than the regularity theory
of causality.s
That is as much as I wish to say here about the irreflexivity of
causality. Now let us look at our original apparent difficulty in
applying causal theories of reference to self-reference. One obstacle to
deciding whether this difficulty is real is the nonexistence of any
genuine theories of reference stated in causal terms. For example, it is
not always crystal clear to me whether such views should be taken to be
about speaker's reference or about term's reference. But I doubt that
making this last distinction will add much to straightening out our
current difficulty. The core idea seems to be that for a singular term
used by a speaker to refer to an object (or, for a speaker to refer to an
object by using a term) it is necessary that his use of the term be the last
link in some causal chain tracing back to, at least, the initial dubbing of
the object with the term, or more interestingly, to some person's
interacting causally with the object. (Presumably Gettier problems,
and not just looseness or circularity, block claiming such a condition
sufficient as well. The more interesting form blocks a causal theory of
reference to abstract objects, but if dubbing need not be a causal
process, the other form does not.) This idea is not intended to apply to
all singular terms; in particular, for most definitive descriptions,
unique satisfaction of their constituent predicates, rather than causal
origin of their production, determines their references. The idea was
originally intended to apply especially to proper names. I, on the other
hand, have used mostly demonstrative self-reference. Intuitively it
seems to me that a causal view of demonstrating should be at least as
appealing as a causal view of naming, at least that is, if one thinks there
is pure demonstrating as opposed to demonstrating with constituent
predicates. However, my demonstratives always had constituent
predicates, so it behooves us to pose our problem with proper names
and see whether it survives.
To spell it out in detail, begin by supposing that there is a class of
expressions recognizable as potential proper names prior to their
receiving, in a context, a denotation. Suppose that "Kofi" (a proper

Hart Causation and Self-Reference

name in the Ashanti language, Twi) is such an expression, and that in

the present context it has as yet no reference. Next, we inscribe

Kofi contains four words.

This, I assume, is a token of a grammatical sentence. On the other

hand, because its subject singular term has no reference as yet, this
sentence has as yet no truth conditions, and so on some views, has as yet
no meaning (expresses no proposition). (Thus, on such views, not all
grammatical sentences are meaningful.) But then we dub the type of
the last displayed inscription with the name "Kofi"; we use the
expression, "the type of the last displayed inscription," to fix a unique
reference for the proper name "Kofi". Now the last displayed sentence,
that is, Kofi, has truth conditions (and so, on some views, meaning);
namely, Kofi is true if and only if Kofi contains four words. 9 We
inspect our inscription of Kofi, count up the words to it, see that there
occur exactly four words in it, and conclude that Kofi is true.
It follows from the core idea of the causal theory of reference
that for the name "Kofi" used by a speaker uttering the sentence Kori to
refere to the sentence Kofi (or, for a speaker uttering the sentence Kofi
to refer to the sentence Kofi by his use of the name "Kofi"), it is
necessary that his use of the name "Kofi" be the last link in some causal
chain starting with the initial dubbing of the sentence Kofi with the
name "Kofi" , or with someone's interacting causally with (an
inscription of) the sentence Kofi. For example, if we now utter a token
of the sentence Kofi, our use therein of the name "Kofi" presumably
traces back along a causal chain to our perceptions of that other, earlier
token of the sentence "Kofi" os tended when the reference of the name
"Kofi" was fixed. We thus now meet the necessary condition that our
later use of the name "Kofi" should refer to the sentence Kofi. But
what presumably cannot happen is that our use of the name "Kofi" is the
very first instance of an inscription of the sentence Kofi, before Kofi
has been dubbed "Kofi," should refer to the sentence Kofi. This is not
just because in that context no reference for the name "Kofi" has been
fixed, but more because that use of the name "Kofi" would have to trace
'back' along a causal chain to that very use of the name "Kofi" made in
inscribing the sentence Kofi, and so much reflexivity of causality here
is unacceptable. A causal theory of reference for proper names can
eschew self-causation in modelling cases in which a name denotes a
sentence in which that name is used if fixing the name's reference is
never the same event as a use of the name to refer to that sentence. Since
in such cases causation presumably cannot be reflexive, a name's being
used to refer cannot be the same event as its reference being fixed.

Hart Causation and Self-Reference

Thus the name must be somehow mentioned, rather than used to refer,
in the event in which its reference is being fixed by ostending a token of
a sentence in which the name is used (perhaps even to refer to that
sentence.) (That is, dubbing need not come after exhibiting a token of
the sentence dubbed. It may come before, as in, "I hereby dub my next
sentence 'Joe'. Joe contains four words." Here the second, used
occurence of the name "Joe" denotes the sentence in which it is used.
Dubbing may also be simultaneous with the first self-referential use of
the name. Suppose Tom utters, "Joe contains four words," and Peter in
collusion with Tom simultaneously dubs Tom's sentence "Joe." Tom
uses "Joe" to refer, but this is a different event from Peter's fixing the
reference of "Joe," even though. they are simultaneous. In sum, then,
a causal theory of reference for proper names which is careful about
the distinction between use and mention (and the distinction between
types and tokens) ought to be able to avoid reflexive causation in
modelling proper names which denote sentences in which they are used.
I close with a brief worry about demonstratives. The solution
sketched above to the problems of self-causation in self-reference in the
case of proper names would seem to work for any singular terms whose
references must be fixed (in a context) one by one, rather than being
determined uniformly by 'semantic rules' for singular terms of a
certain sort. Of course, there is no problem of self-causation for
(most) definite descriptions; their denotations are not hooked onto them
by causal chains one by one, but rather uniformly by unique satisfaction
of their constituent predicates. But where do demonstratives fit in this
classification? Demonstratives, some with and some without
constituent predicates, seem to be singular terms such that

(a) their references need not always be fixed one by one; they can be
used straightway to refer without a separate and distinct mention
of them in an act of reference fixing;
(b) causal chains from denotation to use are often crucial to the
reference of a demonstrative. Even when a demonstrative has
constituent predicates, it would often seem not to refer, and
epecially not to do so uniquely, when used by a person unless he
perceives, or is otherwise at least in part causally linked to, the
denotation of the demonstrative. One might, for example, view
pointing and other ostension in the use of demonstratives as
retracing backwards through the causal chain from denotation
to use in order to transmit reference to an audience (the
constituent predicates showing how far back along the chain to
go, assuming they must be true of whatever the demonstrative
denotes.) Or, one might view ostension as a separate act of

Hart Causation and Self-Reference

reference fixing for demonstratives, a kind of implicit mention

of them separate from their explicit use to refer to their

The problem is this: How much are demonstratives like proper names
and how much are they like definite descriptions? The problem is acute
for us since the most natural examples of self-reference, ones like "This
very sentence contains six words," use demonstratives. If such
demonstratives are like definite descriptions in not requiring their
references to be fixed one by one (and no ostension seems necessary
here), and like proper names in that when used by a person, they denote
the sentences in which they occur only if his use traces back along a
causal chain to his acquaintance with that token of that sentence, then
perhaps our original problem arises once again. But I am too far from
commanding a clear view of the semantics of demonstratives to see my
way out of that problem.

W.D. Hart
Department of Philosophy
University College London

Hart Causation and Self-Reference

1. Saul Kripke, "Outline of a Theory of Truth," The J oumal of
Philosophy, LXII (6 November 1975), 691-692.

2. Cf. my "On Self-Reference," The Philosophical Review,

LXXIX (October 1970), 523-538.

3. David Kaplan, "Quantifying In," reprinted in Reference and

Modality, Leonard Linsky, ed., OUP, 1971, p. 132.

4. David Kaplan, "What is Russell's Theory of Descriptions?" in

Betrand Russell, D.F. Pears, ed., Anchor, 1972, p. 236.

5. Cf. Saul Kripke, "Is Theory a Problem About Substitutional

Quantification?" in Truth and Meaning, Gareth Evans and John
McDowell, eds., Oxford University Press, 1970, pp. 325-419;
section 1 is especially good on such matters.

6. W.V. Quine, The Roots of Reference, Open Court, 1973, p. 5.

7. Cf. Lawrence Sklar, Space, Time and Spacetime, University of

California Press, 1977, pg. 311 for a discussion of causal loops.

8. If a regularity theory were true, then since every disease is

followed by death, all diseases would cause death, that is, be
No simple analysis of causal constructions in terms of the
subjenctive mood will do. First, reductio ad absurdum
arguments where no causal questions are at issue are often
correctly put in the subjunctive. For example, it is true that if
the circle could be squared, then 1t would be a zero of a
polynomial with rational coefficients, which Lindemann proved
impossible. There being no mathematical causality, the
subjunctive is not sufficient for causality. Second, the sentence,
"If you were to catch cold, you would die, but not because you
caught cold, but only because all men are mortal," is true of
most of us. So even if there is a Gricean conversational
implication of causality from some subjunctive conditionals, the
fact that it readily cancels shows again that they do not entail
Quine'S energic view easily accomodates an account of
the fact that lawlike, but not accidental, regularities support

Hart Causation and Self-Reference

counter-factuals (cs. Nelson Goodman, Fact, Fiction, and

Forecase, Athlone, 1954, pp. 24-31). Typically, energy flows
in patterns, and such patterns are the esssence of natural
mechanisms. So if we follow Quine in viewing counter-factuals
asserting the existence of an underlying mechanism (cf. his
"Necessary Truth," reprinted in The Ways of Paradox and
Other Essays, revised and enlarged edition, Harvard University
Press, 1976, pp.68-76, it becomes understandable that causal
laws will entailcounter-factuals, but that accidental re~ularities,
which are true without patterns of energy flow, will not. This is
a third advantage of the energic model over regularity theories.

9. One might approximate a general axiom of (linguistic)

self-reference in a truth theory thus: For every open sentence A
with exactly one free variable, there is a closed sentence S such
that S is true if and only if A is true of S. A liar paradox folows
from this axiom if there is an open sentence whose extension is
the set of closed sentences which are not true. A good solution
to the liar should preserve as much as possible of the axiom of
Consider a language each of whose singular terms has a
unique denotation. Suppose we have a sort of generalized
subject-predicate principle, viz for each singular term t and
each monadic open sentence A, there is a closed sentence S such
that S is true if and only if A is true of the denotation of t. By
the axiom of choice, it follows that for each term t there is a
function ft from open to closed sentences such that for each
monadic open sentence A, ft(A) is true if and only if A is true
of the denotation of t. Now suppose that the functions f t have
fixed points in the sense that for any monadic open sentence A
there is a term t such that ft(A) is the denotation of t. This fixed
point property is one abstract view of the essence of
self-reference, for from it (and the subject-predicate principle)
the axiom of self-reference now follows; and demonstrative
self-reference secures the fixed point property all too easily.
An extra-linguistic form of the axiom of self-reference,
stated in terms of properties and propositions, can also be
formulated. It is deducible from a parallel modification of
Tarski's ConventionT, which is in turn deducible from naIve
comprehension for properties in a theory whose language
includes demonstrative self-reference and an expression relation

Hart Causation and Self-Reference

which serves merely to codify the information stated in the form

of Convention T used); such results incline me to think Russell
was right to assimilate the set theoretic and the semantic
paradoxes, despite Ramsey's later dichotomy between them.

Joseph M. Boyle, Jr.

Is Determinism Self-Refuting?

One of the relatively minor skirmishes in the ongoing

philosophical controversy over the reality of free will is the debate
about whether the determinist position is somehow self-refuting. This
debate has seemed to be of minor importance because the argument that
determinism is self-refuting has the appearance of sophistry; it seems
like a short, fast way out of a very complicated problem.
Still, this line of argument-particularly in its more developed
forms-is intriguing. 1 For whatever is ultimately to be said about the
relevant experiences, their interpretations, the complicated arguments,
and assumptions, it is clear that determinism must be asserted in a
proper, justified way. If determinism is incompatible with some feature
of its own assertion, then we need look no farther to reject it. The
argument that determinism is self-defeating maintains that there is such
and incompatibility; thus, the determinist is held to be "hoist on his own
petard" in much the same way as the the thoroughgoing sceptic or the
all-out relativist. Just as the sceptic or relativist seems, in asserting his
thesis, to be making the sort of knowledge claim his thesis excludes, so
also the determinist is said to be doing something in asserting
determinism which this very thesis excludes.
However, scepticism and relativism are general claims about
human knowledge, and so it is not surprising that they should make
reference to their own assertions and in such a way that they might have
difficulty in allowing for them. By contrast, determinism is a claim
about human actions and volitions, and, while it is true that assertions are
human acts of a kind, it does not seem that freedom of the will must be
an ingredient in a proper or justified assertion. The determinist is
required to hold that there are sufficient causal conditions for his act of
asserting determinism, but why should this in any way compromise his
assertion as an epistemically proper act? Surely, one does not simply
choose to believe or assert determinism.
Those who hold the standard version of the thesis that
determinism is self-refuting could be more forthcoming in answering
this question; moreover, their answer is less than persuasive. But it does
have considerable intuitive plausibility. Essentially, the answer is that
an assertion is a justifiable and epistemically responsible act only if
Boyle Is Detenninism Self-Refuting?

there is an unbiased consideration of the evidence for the proposition

asserted in the light of the relevant standares of evidence. This unbaised
consideration is where the free choice must come in, for if this
consideration has sufficient causal conditions, it cannot be unbaised.
The causal conditions sufficient to detennine one's assent to the
proposition and one's assertion of it might or might not have a
connection with the relevant evidence and standards. Thus, one's
assertion--even if it were an assertion of a true proposition-would be
at best only accidentally related to what is the case. It would not be
known to be based on the relevant evidence and standards, even if one
were detennined to believe it to be so, for the evidence and standards
could not be considered with an open mind. One's mind would be
closed by the factors sufficient to detennine assent and affinnation.
This argument seems to beg the question, for it cannot be stated
without the premise that rational assent is incompatible with causal
determination. Since this premise is not itself established in the
argument, it simply states an implication of what the detenninist means
to deny.2 Perhaps it is true that reasons cannot be reduced to causes,
and perhaps this argument suggests why this is so; nevertheless, it fails
to show that the reasons sufficient to justify the assertion of a
proposition are not, either by themselves or with other conditions,
sufficient to detennine the assertion.
Moreover, there surely are rational assertions that do not
involve any choice-let alone a free one. Judgments about immediate
experience do not seem to involve choice: we see something, and
spontaneously, and perhaps nesessarily, judge it to be as it appears.
Likewise, with certain immediately evident, necessary truths: we
consider the proposition that the whole is greater than any of its proper
parts and we simply cannot think otherwise than that it is true. Surely,
the lack of anything like a choice in the assertions of these propositions
does not compromise their epistemic credentials. Quite the contrary,
we would be far more suspicious if we thought that choices were
involved in making these assertions.
Likewise with the assertion of detenninism: it is proposed as
based upon-and required by-the evidence. In general, philosophical
positions are not things we choose to accept or reject, but are
propositions which the evidence and argumentation lead us-in the
ideal case, inevitably-to accept. It seems a mistake, therefore, to hold
that free choice must be involved in the assertion of detenninism if the
assertion is to be justifiable.
In spite of the failure of the standard version of the argument
that detenninism is self-refuting, one can wonder whether there is not

Boyle Is Determinism Self-Refuting?

a better argument for the same conclusion. For asserting propositions is

an inherently normative thing; it is not simply believing a proposition,
but being convinced that one is justified in believing it, and that others
ought to accept it as true. Since there is a kind of norm which seems to
imply the capacity for free choice, if the norm involved in the assertion
of determinism is a norm of this kind, then determinism could not be
asserted without invoking a norm that implied the capacity for free
If an argument were developed along these lines, it would avoid
the obvious difficulties of the standard argument that determinism is
self-refuting: it would not make use of the question-begging assumption
that rational assertion is incompatible with causal determination; it
would proceed, instead, by showing that the assertion of determinism
involves appeal to a norm which implies the capacity for free choice.
This approach would have to involve the defense of two controversial
propositions-that the assertion of determinism necessarily involves
appeal to a norm, and that the norm somehow implies the capacity for
free choice. But it would not involve any general, question begging
assumptions about the relationship of reasons and causes. Similarly, it
would not involve any general view about the conditions necessary for
rational affirmation, but only a view about the conditions for the
rational affirmation of determinism and propositions like it. Nor would
it suppose that the determinist or his opponent actually made a free
choice in affirming or assenting to determinism; instead, it would seek
to show that a norm necessarily involved in asserting determinism
implies the capacity for free choice.
An argument of this kind exists, 3 and I believe it to be sound.
But before developing it, it is necessary to define some of the key terms
to be used. The deterministic proposition at issue in this discussion is
the claim that human beings have the capacity to make free choices. For
purposes of brevity and clarity I will name this proposition Nfc, as in
"'No Free Choice." It is a universal negative proposition; it is not-at
least on its face-a logically necessary proposition. It does, of course,
make a claim about what human beings cannot be; but this claim is
about causal or physical necessity. It is also put forth as having some
sort of epistemic necessity-as being the proper and justified
conclusion of the evidence and arguments adduced in its behalf.
The phrase "free choice" can be used to refer to a number of
things. I will use the phrase narrowly to refer to only one of the things
that determinists wish to deny and some proponents of free will wish to
assert-namely, choices in which there is no sufficient cause for the
choice a person actually makes which does not include the person's own

Boyle Is Determinism Self-Refuting?

choosing. A person makes a choice when he or she is faced with a set of

alternatives for action. Spontaneous action is blocked because one sees
that there are several possibilities for acting. The person must then
deliberate and choose. The person sees that he or she could do this or do
that, but not both, and settles the indeterminacy about what to do by
choosing to do this rather than that or to do that rather than this. A
choice is free, in the narrow sense used here, when the factors
determining which alternative will be selected are not sufficient to
determine the selection, unless the person's own choosing is included
among these factors. In other words, if all the causal factors relevant to
a choice-including such things as desires and reasons-are sufficient
to bring about the person's choosing one way rather than another, then
the choice is not free. Only the person's choosing one way rather than
the other settles the issue if the choice is free; and there are not
sufficient causes outside the person's very choosing for the person's
making the choice he or she makes. 4
One may wonder why this very restricted sense of free choice is
made the focus of the controversy over free will. For it is certainly true
that not all determinists are primarily concerned to deny free choice in
this sense; and it is also true that not all who defend free will are ready
to affirm the reality of free choice in this sense; some, for example,
wish only to affirm the irreducibility of determination by reasons to
ordinary causal determination. Moreover, some who defend the
freedom of the will regard it as a far richer and more pervasive part of
human life than this narrow definition allows.
It is true, of course, that this notion of free choice is narrow; it
does not refer to many things necessary for acorn plete theory of human
volition or human action. It surely does not follow from this, however,
that free choice is not central to the controversy about free will. Many
volitions seem to be determined-by reasons if not by causes. This
point is made by some proponents of free will, for example, by
Aquinas. 5 The issue between determinists and proponents of free will
concerns those will acts that seem to be free. It would seem, thefore,
that the dispute should center on deliberate choices, for these are the
part of human life in which people experience themselves as
self-determining. An aspect of the experience of choice is that nothing
seemed to make one choose one way rather than the other. The person
himself decided. 6 It is not accidental, therefore, that determinists have
focused on free choice in the narrow sense defined here. This is the
datum that their account must deal with.
Thus, although denying free choice in the narrow sense has not

Boyle Is Detenninism Self-Refuting?

been the central concern of all detenninists, all detenninists do deny it;
for free choice in this sense is something that does not fit into the
nonnal pattern of causal explanation; it seems to violate the principle of
sufficient reason, and so on. Nfc, therefore, is central to the detenninist
program-if not the very center of it.
There is another reason for the focus on free choice in the
narrow sense, and that is the moral significance of free choice. Many
proponents of free will regard it as essential for moral responsibility;
those who deny free will also deny-either explicitly or
implicitly-that there is moral responsibility of the kind connected
with free will. Of course, there is a kind of moral responsibility
compatible with determinism as the tradition of compatibilism or
soft-detenninism has made clear. But the kind of responsibility that
involves free will is different from that allowed by some fonns of
detenninism. 7 This responsibility seems to be the responsibility people
have for the choices they make. So free choice is again the focus of the
discussion. This last point requires some development. As Aristotle
pointed out, there are voluntary acts which do not involve choice, but
choice is the locus of moral responsibility.8 If choices are detennined
as soft-detenninists suppose, they are still voluntary and even in some
sense free. There is a sense in which persons would be morally
responsible for such choices. But this is not the sense of moral
responsibility affinned in the Bible and in the Catholic moral tradition,
in particular. For in this tradition, one's very self is detennined by the
choices one makes and one's eternal destiny may hang on how one
chooses. One's responsibility is, therefore, radical; it is up to the
individual to choose life or death, to choose to keep or not to keep the
commandments. If the choice in question were detennined by anything
other than the person's own choosing, then it would be unreasonable to
regard his or her responsibility as radical; for, while the person made
the choice and in some sense could have done otherwise, it was not
within the person's power to choose otherwise in the situation. In short,
the controversy over the reality of free choice in the narrow sense
defined above is an important one, however attenuated some may
regard the notion of free choice employed. This controversy is
certainly an important part of the free will debate, and so if it can be
shown that Nfc is self-refuting an important part of the issue will be
well on its way to being settled.
A final preliminary to the argument is a clarification of the
notion of self-refutation. Self-refutation arises in propositions that are
self-referential in a peculiar way. To clarify the kind of self-reference
at issue, it is useful to distinguish various aspects of a statement. What is

Boyle Is Determinism Self-Refuting?

stated is a proposition; it is the proposition that is true or false, and

which has sense and reference. Statements also include sentences-the
linguistic realities used to express propositions. They also include the
stating of the proposition, which is a performance of sorts. This
performance has several aspects; at the very least there are two: the
affirmation of the proposition, and the using of the sentence to express
the proposition and its affirmation. The affirmation is the act of
claiming that the proposition is true; it is at least in part an epistemic
act. The use of the sentence can be the uttering or inscribing of the
sentence, and includes the properties of these activities.
Propositions can be self-referential in a variety of ways.9
Strictly, a proposition is self-referential when it refers to itself. It is
well known that this often leads to paradox. But propositions can be
self-referential in a looser way when they refer to the
non-propositional components of the statement in which they are
affirmed. This looser form of self-reference can be a reference to the
sentence used to express the proposition, or to one or another aspect of
the performance of the statement. Thus, " This sentence is written in
English," and" This statement is typed" are self-referential, and there
is no mystery about how and what they refer to. They refer
respectively to the sentence used to express the first proposition
mentioned, and to the way in which the sentence used to express the
second is communicated.
The concern here is with performative self-reference, the kind
exibited in the second example. Moreover, it is with universal
propositions that are performatively self-referential, since Nfc is a
universal, performatively self-referential proposition. The
self-reference in such propositions is no more mysterious than in the
singular propositions examined above; the only difference is that one of
the instances of the universal proposition refers to the performance of
the statement of the universal proposition. Thus, one who states Nfc
refers to all human acts including whatever acts are involved in the
performance of that statement.
The notion of self-refutation has obvious application to
performatively self-referential propositions; it arises when these
propositions go wrong in virtue of their incompatibility with the
aspects of their own performance to which they refer. This can happen
in two ways. The first, which can be called "self-falsification," occurs
when the aspect of the performance to which the proposition refers has
some property incompatible with what the proposition ascribes to it.
Thus, if someone shouts "I always speak softly," his proposition is
self-falsifying. The second way in which performatively

Boyle Is Determinism Self-Refuting?

self-referential propositions can go wrong in virtue of their

self-reference is by being self-defeating. This occurs when the
proposition renders pointless the performance of stating it. For
example, one who states that all knowledge claims are equally
groundless renders his own statement pointless, for it is as groundless
as any other. Of course, if there are grounds for the claim, it would not
be self-defeating but self-falsifying. Most of the philosophically
interesting self-referential propositions which are found to have
self-referential difficulties are like this example in that they are either
self-defeating or self-falsifying. Therefore, I will call a
performatively self-referential proposition "self-refuting" if it is either
self-defeating or self-falsifying.
As was mentioned above, Nfc is self-refuting if two conditions
are met: (1) that its affirmation necessarily involves appeal to a norm,
and (2) that this norm implies the capacity for free choice. If these
conditions are met, the affirmation of Nfc will involve invoking a norm
that implies the capacity whose reality Nfc denies. Thus, Nfc is either
self-falsifying or self-defeating since the kind of norm it makes use of
implies the capacity for free choice, or its affirmation cannot have the
normative force it must have to be rationally affirmed, because norms
that imply the capacity for free choice are excluded by Nfc.
The fulfillment of the first condition is easily established; this is
not, it seems to me, a controversial matter. When a person affirms a
proposition, the person claims not only that he or she believes the
proposition to be true, but also that the belief is justified. Moreover,
the justification claimed in affirmingrming a proposition is taken as
imposing some requirement to accept the proposition on the part of
those to whom the statement is addressed. The proposition is put forth
as one that people ought to believe-whether or not they come to
believe it as a result of its being affirmed. Propositions can be put forth
without being affirmed, for example, they can be mentioned as
examples, reported as the opinions of others, and so on. When they are
put forth in any of these ways they are not affirmed-there is no claim
that they are true or ought to be accepted. Moreover, there are
affirmations that are are not in fact justified, and those for which the
justification claimed is very weak. But if there is no claim to
justification whatsoever, it is hard to see that there is an affirmation at
all, let alone an affirmation that is in any sense rational or purposeful.
One who is shown that there is no justification for his or her

Boyle Is Determinism Self-Refuting?

affirmation, may continue to believe the proposition originally

affirmed, but in this case it makes sense to say that the person ought not
to believe the proposition, or that the person has no right to believe the
proposition. In any case the person's continued uttering of the
proposition gives others no reason to think they should accept the
proposition, other than the fact that the person who utters it happens to
believe it.
Perhaps there are in ordinary discourse affirmations that are
irrational in the sense discussed above, and perhaps there are
affirmations that have no normative component. Perhaps some people
just talk to hear themselves talk, stating their beliefs with no pretense of
justification and no expectation that others will feel bound to take their
statements seriously. I would doubt that such statements should be
regarded as affirmations, but even if they were affirmations they could
not be the rational affirmations appropriate to serious scientific or
philosophical discourse. In a philosophical controversy the affirmatioll
of a position is surely normative in the way indicated above.
Presumably the proposition is affirmed in the context of a serious
effort to discover the truth; it is affirmed on the basis of considerations
that justify the affirmation, and impose an obligation on others to
accept the proposition. The affirmation of Nfc is no exception; it is
proposed as a reasonable affirmation based on considerations which
justify its affirmation and require assent even from those who are not
inclined to accept it. 10
The establishing of the second condition-that the kind of norm
involved in the affirmation of Nfc implies the capacity for free
choice-is the difficult part of the argument. For doing this requires
showing what kind of norm is necessarily involved in the affirmation of
Nfc, and showing that this kind of norm implies the capacity for free
choice. Those who believe Nfc will, of course, deny that any actual,
valid norm implies the capacity for free choice. Thus, they will hold
that the norms involved in the affirmation of Nfc are of a kind that do
not have this implication. Since there are norms that do not imply the
capacity for free choice, this way of handling the normative character
of the affirmation of Nfc has some plausibility. But the norms used
must have the kind of normative force for this affirmation and still be
of a kind that does not imply the capacity for free choice. These two
conditions, I believe, cannot be jointly fulfilled.
This supposes, of course, that there is a kind of norm which
implies the capacity for free choice. It does not suppose that any norm
of this kind is an actual, valid norm, for this supposition would beg the
question in this discussion. In fact, even the supposition that there is a

Boyle Is Determinism Self-Refuting?

conception of a kind of norm which implies the capacity for free choice
would be question begging if it were not established. So, what grounds
are there for thinking that there is a kind of norm that implies the
capacity for free choice?
The type of norm in question is one that prescribes
unconditionally one of a set of real possibilities for action. To show that
a norm of this type implies the capacity for free choice it is necessary to
explain the terms that define it. A norm is a standard which can be
fulfilled or can fail to be fulfilled; it does not indicate what is but what
is to be. Some norms have prescriptive force only if definite goals are
taken as given. Such norms-hypothetical imperatives or technical
norms-indicate the rational requirements for achieving these goals.
But some norms are not thus conditioned; they prescribe
unconditionally or categorically. On some conceptions of morality,
moral norms are unconditional in this way: they indicate how one
should act no matter what one may happen to desire. But the laws of
logic are also norms that are unconditional in this way, as may be other
types of norm as well.
The other element in the definition of the kind of norm that
implies the capacity for free choice-namely, that such a norm
prescribes among real possibilities for action-can be clarified by
considering the nature of what various types of norm exclude.
Technical norms exclude real possibilities for action. These
possibilities are articulable things a person could think of doing, and
might have some reason to do. They are excluded because they conflict
with the rational requirements for realizing the established goal. Of
course, the person governed by the technical norm can surrender the
goal on which the norm is contingent. The point here, however, is that
technical norms exclude real possibilities for action. The price for
violating the norm is simple practical irrationality; but this irrationality
does not mean that there is not a coherent possibility for action contrary
to the norm. Moral norms, however they are understood, also
prescribe in this way: what the moral norm excludes is a real possibility
for action-a possibility that has its attractions, however ill considered
they maybe.
These clarifications of the the notions of a norm's prescribing
unconditionally and among real possibilities for action make possible
the statement of the argument that any norm having both these features
implies the capacity for free choice. A norm implies the conditions
necessary for its fulfillment, in the sense that if these conditions do not
obtain the norm cannot really require what it claims to require. Unless
these conditions obtain, the norm is null; it has no prescriptive force. I I

Boyle Is Determinism Self-Refuting?

A norm which unconditionally prescribes one of a set of real

po~sibilities is null if the person to whom it is addressed does not have
the capacity to fulfill the norm by undertaking the prescribed action.
But this capacity for fulfillment would not obtain unless the person had
the capacity for free choice. This is not to say that having real
alternatives means having alternatives among which one must have the
capacity to choose. The point is, rather, that it would make no sense to
unconditionally prescribe one of a set of real possibilities, unless the
person to whom the norm is addressed could freely choose to conform
to the norm.
Of course, a person addressed by such a norm could be
determined by reasons or causes to conform to the norm. Such a person
would have, trivially, the ability to conform. But for those determined
not to conform, the norm would be null. At best its utterance might
contribute to determining them to conform; it would in no way require
or demand conformity. Thus, if there were no capacity to choose to
conform to the norm, the norm would altogether lack prescriptive
force: only those determined to fulfill it could fulfill it, and for all
others it would be null. So, there is a kind of norm that implies the
capacity for free choice. The question that remains is whether the
determinist necessarily makes use of a norm of this kind in affirming
Nfc. In brief, the response is that the affirmation of Nfc must invoke a
norm that prescribes unconditionally; Nfc is put forth as something any
reasonable person should accept. Moreover, the norms invoked in
rationally affirming Nfc exclude real possibilities for action-namely,
the affirmations of propositions contrary to Nfc or its premises.
The unconditional character of the affirmation of Nfc is
unavoidable. The assertion of any general philosophical thesis is
unconditional in this way: it is proposed as reasonable and as something
any reasonable person should accept. The force of the norm cannot be
escaped because one happens to have different but equally legitimate
goals than the determinist has. Being reasonable is not taken as a:
merely optional goal by those who make such assertions; if it were, the
affirmation of the thesis would have no force for anyone having other
goals the pursuit of which might reasonably require the refusal to
accept the philosophical thesis proposed. So, the .QYght involved in the
affirmation of Nfc is unconditional; it claims to regulate the
affirmations of people whatever their goals happen to be. 12
The second condition for a norm's being of kind that implies
the capacity for free choice-that it prescribe among real possibilities
for action-also seems intuitively to be realized in the affirmation of
Nfc; this affirmation excludes other possible affirmations-for

Boyle Is Determinism Self-Refuting?

example, that of the proposition that someone can make a free choice.
Propositions contrary to Nfc can be affirmed; they are certainly not
like self-contradictory propositions which cannot be coherently
affirmed. What is more, the affirmations of these propositions are
certainly things people might find it interesting or desirable to do. The
affirmation of Nfc must exclude the affirmation of these propositions
as something one ought not to do.
If the affirmation of Nfc were based on the claim that its
contraries were literally incoherent-self-contradictions-then what
its affirmation would exclude would not be a real possibility for action.
But only the fatalist argument even pretends to proceed in this way,and
this argument involves a modal fallacy. 13 Other arguments for Nfc do
claim that the concept of free choice is unintelligible, but this does not
mean that the claim that someone can make a free choice is formally
incoherent. Rather, it means that the doctrine of free choice makes
choice into an inexplicable, chance event, or that it violates the
principle of sufficient reason or causality. 14
Such claims about intelligibility assume some standard of
intelligibility by comparison to which free choices are unintelligible,
and in virtue of which one is entitled to dismiss their reality. This
standard of intelligibility is not a law of logic or a self-evident truth.
What it excludes is coherent, but is dubbed irrational in the sense that
one who refuses to affirm in accord with the norm is ignoring the
requirements for theoretical explanation and the discovery of truth.
Other arguments for Nfc make use of different standards of
rational inquiry, but in a way similar to that just discussed. Some of
these arguments appeal to principles of induction, some to a rule of
simplicity, and some to predictive success. All arguments for
determinism, other than the fatalist argument, make use of some such
standard. IS In all these arguments, the principle invoked provides
some standard for rational inquiry; the argument proceeds by showing
that the reality of free choice is incompatible with the standard, and this
is taken as a reason for excluding free choice and affirming Nfc.
These standards, therefore, are norms which condition the
affirmation of Nfc. They are an important element in its justification,
and, thus, are part of the normative aspect of the affirmation. These
rationality norms seem to function in the affirmation of propositions in
much the same way as moral norms are thought to function in directing
human actions generally. Moral norms exclude real possibilities for
action; the penalty for violation is immorality or moral guilt, and not
incoherence or the futility of undertaking the impossible. Likewise,

Boyle Is Determinism Self-Refuting?

rationality norms exclude real possibilities for acts of affirming; the

penalty is irrationality-a disregard for the requirements of theoretical
understanding-and not incoherence. Of course, the irrationality of
violating a rationality norm can be quite rational from another point of
view than that of theoretical explanation. For example, if the
affirmation of a proposition on the basis of a rationality norm should
violate a technical norm based on some goal one has, it might be quite
rational to affirm a proposition contrary to the one prescribed by the
rationality norm as the rational requirement of realizing this goal. This
is similar to the way in which an immoral act might be rational. 16
It seems, therefore, that the affirmation of Nfc does involve
invoking a norm that implies free choice. The affirmation of Nfc is
rational and justifiable only in virtue of the arguments which support it.
Like all arguments, these make use of formal logic, but more is
involved in these arguments than the formal logic involved in the
inferences. There are also standards of rational inquiry which
unconditionally exclude as irrational the affirmation of propositions
contrary to Nfc. These norms exclude real possibilities for action, and
are a necessary part of the justification of the affirmation of Nfc. If the
arguments for Nfc showed that these possibilities were not real but in
fact genuine impossibilities, then the norm could, no doubt, be in force
as an unconditional norm without supposing the ability on anyone's part
to choose to conform to it. For the natural, intellectual capacity of
human beings to recognize the impossibility of formal incoherence
would be sufficient for its fulfillment. But, since the affirmation of Nfc
excludes real possibilities, it will be null for anyone determined to
choose one of them. If Nfc is true, then a wide range of people,
including the philosophical opponents of determinism, will be persons
for whom the norm is null. Even if such persons were to be determined
to accept Nfc by its affirmative utterance or the presentation of the
arguments on its behalf, they might not be; and for such persons it
would be pointless to say that they unconditionally should accept it as if
this were a rational requirement. These individuals have different
motives and different goals the quite legitimate pursuit of which would
make it rational to refuse assent. Thus, on determinist grounds, the
affirmation of Nfc would have to surrender any claim to be
unconditionally normative. At this point, the claim to be a thesis that
people simply ought to accept is given up.
But must the affirmation of Nfc be normative in the way
supposed? Kenneth Konyndyk has argued that it need not be. He points
out that norms can function in two ways: as rules for the guidance of
actions, including affirmations, and as rules for judging or appraising

Boyle Is Detenniuism Self-Refuting?

actions. He admits that if taken in the first way, norms might be

supposed to imply free choice, but he denies that the rational
affirmation of Nfc must be nonnative in this way. It must be normative
in the second way only, and norms functioning in this way need not
involve the capacity for free choice. He argues as follows:
Assuming he is defending his denial of free choice as rational, it
seems sufficient for his defense to point out that his affinnation accords
with the appropriate norm. He need not deny that the norm is in force.
And he need have chosen his position over its contadictory because the
norm directed him to that choice. He may have no idea what the
causation was, nor does it matter what it was. If we consider a situation
in which the deterniinist is trying to persuade someone else to accept his
position, again he need not assume that the other person is free to
choose. He simply shows that the evidence and the appropriate norm
favor his position, and lets natural causation take over from there. All
he can do is demonstrate that his position is rationa1.!7
In response, it should be noted that the argument does not claim
that the determinist chose his position "because the norm directed him
to that choice." The claim is that the norm is part of the justified
affirmation of his position; how he came to believe it is not at stake.
Moreover, the claim is not that anyone actually made a free choice, let
alone the determinist himself. The point is rather that the norm is
null-is not in force-unless someone can make a free choice.
Likewise, with the persuasion of others; the issue is not whether the
utterance of Nfc will be persuasive in the sense that it induces belief, but
rather, whether one ought to believe. So Konyndyk's scenario is a
possibility not denied by the argument. The problem is that if Nfc is
only thus affirmed and believed, its affirmation lacks the kind of
normative force that it seems to have, and, indeed, must have to be
affirmed philosophically as a thesis that ought to be accepted. Such
theses are not uttered in the hope that they will induce belief, but
affirmed as what any person should accept.
Moreover, Konyndyk's distinction between the functions of
norms does not make room for the relevant norm's functioning as it
must in the affirmation of Nfc without implying the capacity for free
choice. For the appraisal of an affirmation must consider what other
people should do on coming to know the affinnation. If one who thinks
of affirming a proposition recognizes that others are in no way
rationally bound to assent to the proposition affirmed, then it seems that
he or she is not entitled to affirm it. Such a person might firmly believe
the proposition, but the belief cannot be considered justified if the
person acknowledges that others are perfectly free to deny it. Again,

Boyle Is Detenninism Self-Refuting?

this is not to deny that the presentation of the evidence and the
appropriate nonn might be sufficient to induce people to affinn Nfcon
the basis of the nonn; nor is it to deny that belief so induced is rational.
Quite the contrary, such a belief is based on a relevant norm of
rationality. But this does not mean that the affinnation is justified, for it
allows that others with purposes that might be quite legitimate could
have reasons for refusing to assent to the affinnation-reasons that are
unassailable if the nonn does not bind unconditionally.
In short, the unconditional character of the affinnation of Nfc is
unavoidable. The detenninist must say that people categorically ought
to accept his thesis, and this is just the sort of "ought" that implies the
capacity for free choice. So detenninism is self-refuting; this seemingly
minor skinnish in the debate over the reality of free will is really a
major victory for the defenders of free choice and of the traditional
notion of moral responsibility.
Joseph M. Boyle Jr.
Center for Thomistic Studies
University of St.Thomas
Houston, Texas
St. Michael's College
University of Toronto

Boyle Is Detenninism Self-Refuting?


1. James N. Jordan, "Determinism's Dilemma," The Review of

Metaphysics, 23 (1969) 48-66, is the most persuasive
formulation of this argument that I know of.

2. See Joseph M. Boyle Jr., Germain Grisez, and Olaf Tollefsen,

Free Choice: A Self-Referential Argument, University of Notre
Dame Press, 1976), pp. 43-46 for a development of this
argument. Hereafter this work will be cited as Free Choice.

3. See Free Choice, pp. 153-181.

4. See Free Choice, pp. 11-12 for a fuller explanation of this

definition. Kenneth Konyndyk, "Rational Affirmation and Free
Choice: A Study of Free Choice," The New Scholasticism, 53
(1979) 504-505, provides suggestions for the improvement of
the more formal definition offered there.

5. See Summa Theologica, First Part of the Second Part, question

10, article 1.

6. See Free Choice, pp. 18-23 for a more complete description of

the experience of free choice.

7. See Free Choice, pp. 105-110 for a fuller treatment of soft


8. See Nichomachean Ethics, Book 3, 1111 B 4-10, and 1112A 1-5.

9. See Free Choice, pp. 122-138 for a more systematic and

thorough treatment of of self-reference and self-refutation.

10. See Free Choice, pp. 139-144 for a fuller discussion of the
rational affirmation of Nfc.

11. See Free Choice, p. 151 for a further elaboration of this point
along with some examples.

12. See Free Choice, pp. 164-166.

Boyle Is Detenninism Self-Refuting?

13. See Free Choice, pp. 51-57 for a statement and critique of the
fatalist argument.

14. See Free Choice, pp. 57-90 for an exposition of these arguments
and references to the literature.

15. See Free Choice, for a discussion of all the major types of
arguments for Nfc which I know of.

16. See Free Choice, pp. 144-152 for a development of this


17. "Rational Affinnation and Free Choice," 512.

Olaf Tollefsen

The Equivocation Defense of Cognitive


What makes arguments from self-refutation particularly

attractive to many philosophers is that they seem specially decisive in
philosophical controversies. Arguments from self-refutation eliminate
positions as untenable without appeal to any claims other than those
made in the position attacked in that way; nothing other than what is
contained within the self-refutating position is used to show its
inadequacy. The danger of begging the question is thus circumvented,
and the proponent of a position successfully attacked in this way has no
further philosophical moves he can make from within his position; he
must alter or abandon it. Arguments from self-refutation, if
successful, are thus conclusive stopping points in the dialectic of
philosophic controversy. They establish limits beyond which that
dialectic cannot go, beyond which significant philosophic controversy
is impossible. 1
Nonetheless, successful arguments from self-refutation are
difficult to make. They can easily go wrong by the unrecognized
assumption of the truth of some proposition not actually asserted in the
position alleged to be self-refuting,2 and they can also go wrong by way
of equivocal interpretations of the terms used to express the allegedly
self-refuting position. In this sort of case, an argument from
self-refutation can be sound, but it will be beside the point; the position
shown to be self-refuting will not be the position actually under attack.
It is with this issue of equivocation in arguments from
self-refutation that this essay deals, but only in a limited way; at issue
here are only those cases in which the charge of equivocation is used as
a defense against arguments intended to show that cognitive relativism
is self-refuting. The relativist claims that the ultimate foundations for
any belief (or set of beliefs) cannot be cognitive, but are rather matters
of choice, convention, etc. It is often argued that this claim is
self-refuting; relativism, it is argued, implies that there can be no
grounds which warrant rational assent to relativism itself. One reply to
this objection relativists sometimes make is that the objection is
irrelevant because it rests on an equivocation (or series of
Tollefsen The Equivocation Defense of Cognitive Relativism

equivocations) about the meanings of key terms in the controversy.

Thus the relativist might claim that what his critic means by "grounds
which warrant rational assent" is something inconsistent with
relativism, and which the relativist need not accept.
With regard to this kind of defense of cognitive relativism I
shall argue two theses. The first is that the proponent of cognitive
relativism can successfully sustain his charge of equivocation against
his critics' attempts to show that cognitive relativism is self-refuting -
indeed, that because of the very nature of cognitive relativism, any
non-relativist attempt to show that it is self-refuting necessarily fails
because it equivocates. The second is that this very feature of cognitive
relativism renders it philosophically pointless. It escapes
self-refutation, but as an attempt to give a philosophical account of
human knowledge and belief, it is self-defeating; it makes it imposssible
for the relativist's critic to have any reason for considering cognitive
relativism as a rational alternative to his own view.
I begin the analysis with a recent example of the equivocation
defense of cognitive relativism.


In "The Strong Thesis of Sociology of Science" Mary Hesse

argues that all criteria of truth and knowledge are culturally relative,
and that as a consequence, a sociology of knowledge provides the most
plausible account that can be had of the nature of knowledge. 3 In the
course of laying out this thesis, Hesse finds it useful to consider, and
reject, several objections to the relativist thesis. Of those several
objections, she marks as the "most powerful" the "so-called argument
from self-refutation." Here is her statement of a version of the

Let P be the propostion 'All criteria of truth are relative to a local

culture; hence nothing can be known to be true except in senses of
"knowledge" and "truth" that are also relative to that culture.' Now
if P is asserted as true, it must itself be true only in the sense of
'true'relative to a local culture (in this case, ours). Hence there are
no grounds for alserting P (or, incidentally, for asserting its
contrary). (p. 42)

Hesse, however, rejects this "easy" self-refutation as fallacious on the

grounds that it involves an equivocation about the sense of the cognitive
terminology used in P ("truth," "knowledge," and "grounds"):

Tollefsen The Equivocation Defense of Cognitive Relativism

If a redefinition of cognitive terminology as relative to a local

culture is presupposed in asserting P, then P must also be judged
according to this redefinition. That is to say, it is fallicious [sic] to
ask for 'grounds' for P in some absolute sense: if P is asserted, it
is asserted relative to the truth criteria of a local culture, and if that
culture is one in which the strong thesis is accepted, then P is true
relative to that culture. (p. 42)

It seems, then, that the "easy" self-refutation fails because the real
premise used in the alleged self-refutation is not the same proposition,
P, asserted by the relativist. The proponent of P means by "truth,"
"knowledge," and "grounds" something quite different than his critic
does, and inevitably so: the meaning of "truth," knowledge," and
"grounds" is determined for the relativist by his holding P.
That this is so follows from the extension of P. P is a claim
about all beliefs; it is the claim that every true belief is so because it
meets the criteria of truth in a local culture. Now, for any belief that is
true in that sense, it is possible that it be false in some other sense; it is
not necessary that the set of beliefs true in the sense defined in P also all
be true in any other sense of "true." Hence, since P defines the class of
all true propositions, "truth" in P cannot mean anything other than
"meets the criteria of truth in a local culture." Thus P requires the
redefinition of cognitive terminology in itself.
It follows also that the alleged self-refutation of P is beside the
point. The relativist's critic wants to show that P is false in some sense
of "false" excluded by P. If P is to be judged on its own terms,
however, the alleged self-refutation of P is irrelevant; it denies of P a
property which a proponent of P cannot claim is a necessary condition
for rationally warranted assent to any proposition, including P. More
generally, any non-relativist attempt to show that P is self-refuting will
fail in this way, for every such attempt must make use of senses of
"true" excluded by P.
Here, however, the relativist's escape from self-refutation by
the redefinition of cognitive terminology creates a new problem: it is
no longer clear in what sense, if any, the relativist and his critic are
engaged in a philosophic controversy. The relativist's redefinition of
terms seems to have created a new domain of discourse, one so distinct
from that employed by the relativist's critic that the two fail to meet at
any point which could be construed as a philosophical disagreement.
That this is so is cl~ar from the following considerations.
The relativist claims for P the property (call it p) of meeting the
truth criteria of his local culture, and nothing more, as was established
earlier. However, the relativist's critic does not wish to deny that P has

Tollefsen The Equivocation Defense of Cognitive Relativism

p, or claim for his own position (call it Q) that it has p (although it is

possible that Q have p). The relativist's critic holds, instead, that some
property other than p (call it q) is what justifies rational assent to a
proposition. Hence whether P has or lacks p, and whether Q has or
lacks p, is irrelevant to the relativist's critic. Of course the relativist
can claim that no proposition can have q, or more weakly, that even if
some proposition has q, no one can ~ow that it has q, but his warrant
for either claim must be p, and thus either claim is irrelevant to the
relativist's critic. The relativist's critic would be interested in either
claim only if it were alleged to have q, and then he would be interested
in either claim only as an example of a self-refuting proposition.
Conversely, the relativist's critic claims that P lacks q, but the
relativist, because he holds P, limits the grounds for rationally
warranted assent to a proposition's having p. That P lacks q is as
irrelevant to the relativist as Q's lacking (or having) p is to his critic.
Hesse's own analysis of the cognitive status of P could be
construed as an attempt to resolve the question of irrelevance. Hesse
admits that there can be no conclusive argument for the relativist's
position (p. 42). Instead, she claims that what the "argument from
sociology" does is "suggest that we shift our concept of 'knowledge' so
that the alleged refutation of P becomes an equivocation" (p. 42). It
might well be objected that this redefinition of terms is circular, but
Hesse thinks that objection can be avoided. She proposes that the
relativist's thesis not be construed as a conclusion inferred from
established premises, but rather as a hypothesis "in the light of which
we decide to view knowledge, and consider whether its consequences
are consistent with the rest of what we wish to affirm about knowledge,
and whether it does in the end provide a more adequate and plausible
account than the various rationalist [non-relativist] positions we have
found questionable" (pp. 42-43). Thus P would be preferred to
whatever hypothesis the relativist's critic advances because P generates
a better account of the phenomena which stand in need of explanation
(e.g. , theory change in the history of science).
But this proposal, if taken as an attempt to avoid irrelevance,
works just so long as the relativist's redefinition of terms is ignored,
for then it is at least possible that the relativist and hi~ critic have some
common understanding of the cognitive terminology in Hesse's
proposal. In other words, if the proposal is to work, what is required is
a domain of discourse which is neither the relativist's nor his critic's,
and within which P and it non-relativist altemative(s) can be stated
without equivocation. But no such domain of discourse is possible, for
P sets the interpretation of all cognitive terminology; any attempt to
understand the cognitive significance of knowledge claims in ways

Tollefsen The Equivocation Defense of Cognitive Relativism

other than those set by P must be rejected as equivocal by the relativist.

Hesse's proposal, as I have interpreted it here, is pointless; it does not
allow any real cognitive conflict between the relativist's claims and his


The upshot of all this seems to be that relativism, taken as an

account of the nature of knowlege, is pointless - self-defeating in the
sense that it declares impossible what it originally seemed intended to
do: explain what knowledge really is. The relativist can give an
account of the nature of knowledge, but not a cognitive justification of
that account other than that it meets the conventional criteria of
knowledge he happens to use. In the end, the relativist cannot say what
knowledge really is; he has rejected the question itself.
Here, however, the relativist can object that his rejection of this
question about what knowledge really is has a rational foundation that is
as compelling for the relativist's critic as it is for the relativist. The
relativist can argue that the position of his critic is, on its own terms,
self-refuting, or, what is weaker, that the relativist's critic fails to
provide for his position the kind of cognitive justification that position
itself requires for rationally warranted belief. In short, the relativist
can claim that there is no non-relativist justification for his critic's
position; relativism wins by default.
Here is a recent example of the argument to show that the
relativist's critic's position is self-refuting:

The relativist might then draw our attention to the fact that as any
argument purporting to justify the choice of one criterion of rationally
warranted belief over any other (including that of a relativist persuasion)
depends for its effectiveness on the truth of its premises (premises
which being claimed as true must satisfy some criterion of truth),
anybody employing such an argument must already have at least one
criterion of truth, and therefore it follows that some criterion has to be
adopted without argument... This puts the ball firmly in the
objectivist's court for where the relativist is able, so it seems, to accept
that the problem 'truth' finally boils down to a matter of choice, where
the choice is made in accord with what one already believes, for such a
position is fully consistent with the claim that the truth of a proposition
is dependent upon its being affirmed, the objectivist, if he concedes the
strength of the foregoing argument is faced with explaining away an

The point of this argument is clear enough: the objectivist's position

ultimately rests on beliefs - beliefs about criteria for rationally
warranted belief - which are as unfounded (in a non-relativist sense) as

Tollefsen The Equivocation Defense of Cognitive Relativism

is relativism itself. The relativist is of course not claiming to know (in

a non-relativist sense) that the objectivist's position is self-refuting; that
would falsify relativism. The relativist is merely pointing out that
objectivism fails to meet its own criteria for rationally warranted
belief. Objectivism, in short, turns out to be just another species of
relativism. And if that is so, the relativist can respond easily to the
charge that the equivocation defense of relativism makes relativism
irrelevant to the philosophical enterprise.
The relativist can point out that his critic -the objectivist - must
take the relativist's argument from self-refutation seriously. It is true
that if that argument is interpreted in relativist terms, it has no force
for the relativist's critic, as the earlier analysis of the equivocation
defense showed. But it makes no difference for the objectivist that the
argument is made by the relativist, for the objectivist must interpret it
from an objectivist standpoint: if the objectivist holds that the premises
in the argument are true and that the conclusion follows validly from
them, then his position is self-refuting. Hence the relativist can claim
that relativism is not irrelevant to the philosophical enterprise, for if
objectivism is self-refuting, then objectivism is as relative - as
ideological - as relativism. The two positions thus compete as
ideologies, not as mutally exclusive (in a non-relativist, cognitive sense)
accounts of the nature of human knowledge.
Thus the thrust of the relativist's argument is to put the burden
of proof squarely on the objectivist's shoulders. As the analysis of the
equivocation defense showed, the relativist cannot claim to have shown
that objectivism is false in any non-relativist sense, and it is beside the
point to interpret the relativist's obejctions in that way. Instead, the
relativist's arguments should be considered as a kind of challenge to the
objectivist to show that whatever version of objectivism he holds meets
its own criteria of rationally warranted belief. And, in those cases
where the objectivist fails to do this, the relativist has no good reasons
(in a non-relativist sense) for giving up relativism, or for treating
objectivism as anything more than an alternative ideology.


There is, of course, nothing new in this line of argument; it is

the old demand that everything be demonstrated (validly inferred from
premises already known to be true), and the objectivist cannot meet this
demand, as Aristotle long ago recognized: there are no demonstrations
of first principles or foundational beliefs. But it is worth asking if the
lack of demonstration proper really does leave one without cognitive
grounds for belief.

Tollefsen The Equivocation Defense of Cognitive Relativism

Consider, for example, my belief that I am at this moment

typing a sentence in a philosophical essay. I believe this to be true
because I am aware - painfully aware - that I am doing what I believe I
am doing. To ask if my awareness of typing the sentence in question
really constitutes evidence for my belief, and thus to challenge my
criteria for rationally warranted belief (rather than their application in
this particular case) is to ask that I entertain a hypothesis that the world
and my cognitive powers are such that I am not in fact doing what I am
aware of doing - not in this case, or in any other case of the same kind.
Very well; suppose I entertain the hypothesis, and, when I find good
reasons for rejecting it, I am once more confronted with a challenge to
my criteria for rationally warranted belief (in this case, the criteria
which justified my rejecting the hypothesis). Once more I am asked to
entertain a hypothesis that the world and my cognitive powers are not
what they seem to be, and, if I reject this hypothesis, a new challenge
will be made, and so on ad infinitum.
Now, the earlier analysis of the equivocation defense shows that
at no point in this process can the relativist offer me good reasons for
adopting the hypothesis that is presupposed by his challenge to the
criteria of rationally warranted belief I am using at that point. Indeed,
if at that point I find good reasons for adopting his hypothesis, I am
affirming, not rejecting, objectivism; I am not giving up objectivism,
but shifting to an alternative version of that position, as I have good
reasons (in a non-relativist sense) for doing so. But if I find no such
reasons, what of the relativist's challenge to my original claim to know
that I was at that moment typing a sentence in a philosophical essay?
The relativist has failed - necessarily - to offer good reasons for
thinking it an unfounded belief, and I had (and still have) evidence for
thinking it true. And it is just having that evidence that leads me to
believe that my criteria for rationally warranted belief are sound.
Relativism, in short, is irrelevant to the philosophic enterprise.
The relativist's necessary redefinition of cognitive terminology cuts
him off from making arguments or offering evidence for his own
position or against the objectivist's which will have any force except
when reinterpreted in objectivist terms. And every such
reinterpretation is an affirmation of objectivism, for every such
reinterpretation requires the recognition of some objectively valid
criteria of rationally warranted belief. In sum, there are no good
reasons for adopting relativism, and, given the many things we do
know, there is ample evidence for thinking objectivism is true. The
cost of the equivocation defense to the relativist is thus very high; it
removes his position as such from the dialectic of philosophic
controversy. It is a move beyond the limits within which significant

Tollefsen The Equivocation Defense of Cognitive Relativism

philosophic debate is possible.

Olaf Tollefsen
Department of Philosophy
Saint Anselm College
Manchester, New Hampshire

Tollefsen The Equivocation Defense of Cognitive Relativism

1. See Joseph M. Boyle, Jr., Germain Grisez, and Olaf Tollefsen,
Free Choice: A Self-Referential Argument, University of Notre
Dame Press, 1976, esp. pp. 181-185 for the role of
self-referential arguments in philosophical controversy, and
Chapter 5 for an analysis of self-refuting statements.

2. See Boyle, Grisez, and Tollefsen, pp. 40-47, for an example.

3. Mary Hesse, "The Strong Thesis of Sociology of Science," in

Hesse, Revolutions and Reconstructions in the Philosophy of
Science, Indiana University Press, 1980, pp. 29-60. All other
references to this article will appear in the text.

4. Cultural relativism is of course not the only species of cognitive

relativism; relativists differ on the kind of non-cognitive factors
which settle which criteria of truth and knowledge an individual
or group actually use. However, these differences will be
ignored here, for all that is essential to cognitive relativism is
the claim that the factors which settle which criteria of truth and
knowledge are used do not provide a cognitive justification for
those criteria. Thus P can be transformed into P's: All criteria
of truth and knowledge are relative to a prerational choice an
individual makes; hence nothing can be known to be true except
in senses of "knowledge" and "truth" relative to that choice.
Other tranformations of P are possible as the relativists' beliefs
about the kind of non-cognitive factors that determine belief
vary. All such transformation of P are, like P, self-referential,
and are as open as P is to self-refutation.

5. James Marshall, Michael Peters, and Miles Shepeard,

"Self-Refutation Arguments Against Young's Epistemology,"
Educational Philosophy and Theory, 13 (1981) 49.

Martin X. Moleski, S.J.

The Role of Retortion in the Cognitional

Analyses of Lonergan and Polanyi
Nothing is more difficult to argue about than the fundamental
presuppositions of thought. The elaboration of the criteria of
knowledge necessarily involves reflexive or circular insights which
presuppose themselves in their very formation, since the principles of
thought must be active in the thoughts by which we grasp the nature of
thinking. There is no way to suspend thought in order to investigate
thought in the same way that a machine might be halted and dismantled
for analysis, or as a living body might be anesthetized for the sake of
surgery. Cognitional analysis always exhibits a self-referential
character because it uses the very structures which are under
The purpose of this essay is to show how retortion functions as a
strategy for grasping and articulating the fundamental structures of
human knowledge. Retortion is the name given by some
Transcendental Thomists 1 to the argument from self-referential
inconsistency, since the discovery and rejection of such inconsistency
turns an adversary's position against itself by means of a retort. This
kind of argument has been in use in different forms and under various
names since Aristotle used it to defend the principle of
non-contradiction against skepticism.
I have chosen to concentrate on the role of retortion in
Lonergan and Polanyi for several reasons. Rather than consider the
merits of the argument in the abstract, I prefer to consider the concrete
examples which these two thinkers provide. While each man's work
deserves attention on its own, the comparison of the two seems
especially valuable because of the relative independence of the two men
from each other's position. On account of their unique backgrounds
and approaches, the convergence of their positions is all the more
striking and the complementarity of their differences is that much
Bernard Lonergan (1904--1984) was a Canadian priest of the
Society of Jesus. At the time when his fundamental work took shape3 he
was lecturing on dogmatic theology in Rome. Michael Polanyi
(1891-1976) was a Hungarian medical doctor who also took a doctorate
Moleski Retortion in Lonergan and Polanyi

in chemistry; when his chief work was published,4 within a year of

Lonergan's Insight, Polanyi had relocated to England and was devoting
most of his energies to philosophical rather than chemical research.
Lonergan's work was primarily inspired by his studies in the nature of
grace, and stands in the tradition of Aristotle, Aquinas, and Marechal.
While he may be identified as a Transcendental Thomist, his spoke of
his position as 'critical realism.' Polanyi's point of departure was from
research on the adsorption of gases, and his tradition might broadly be
defined as phenomenological. He portrayed himself as a 'post-critical'
Neither man used the term retortion to designate the logic on
which he grounded the defense of his basic insights. Nevertheless, it
seems fair to assess their use of the argument, both to show the value of
retortion for their positions and to show the value of their positions for
a better understanding of retortion.

Polanyi on Persuasion

Polanyi's consistent purpose throughout his works is to convert

his readers from the paradigm of scientific 'objectivity' to a new model
of knowledge based on reasonable subjectivity. There are two chief
movements in his thought: one acts against the "delusion and ... false
ideal [of] "complete objectivity,"S while the other upholds lOa frame of
mind in which I may hold firmly to what I believe to be true, even
though I know that it might conceivably be false.,,6 To adopt this
position is to transform the criteria by which we judge the whole of
reality; although Polanyi delineates his position using material from
mathematics, the history of science, Gestalt psychology, and common
sense, he indicates that it is also significant for aesthetics, politics, and
Polanyi is conscious of the self-referential character of thinking
about thought. The principles which make the most difference in our
positions are the ones which are most deeply embedded within the
Whenever we are faced with the necessity of deciding on a
judgment, we cannot avoid relying on ultimate criteria. Even a
failure to judge demands that we rely on some ultimate criteria for
our refusal to judge. The point is, however, that we are often
unaware of what these criteria are until after we have relied on them
as subsidiary clues in a focal integration.'

Moleski Retortion in Lonergan and Polanyi

We do not suspend all the operations of the mind, pick out the key
criteria by which all other criteria are to be judged, and then apply
them in a formal fashion to deduce everything else. Instead, we must
continually reflect on the positions which we have actually taken, and
attempt to catch sight of the central tenets of the position as we use them
to determine meaning.
This process of raising self-consciousness is never able to
surface all of the principles at work simultaneously:

At some point, we must have "rules" of application (if we can call

them that) which we cannot specify, because we must simply dwell
in them in a subsidiary way; They are, part of our deepest
commitments. But for this reason they are not specifiable ... since
we are looking with them. They therefore must remain
indeterminate. 8

Polanyi shows that this is a phenomenon that we can and do live with as
long as the submerged principles are in harmony with reality. He calls
this 'tacit knowledge' or the tacit dimension of knowledge.
There is no strict proof of the existence or extent of tacit
knowledge. As we will see below, the whole thrust of Polanyi's
position calls the criterion of 'strict proof into question. The
acceptance of the tacit dimension of knowledge depends on the
observation of complex performances which not only do not require
conscious direction but which defy attempts to codify rules for success.
A child can ride a bicycle and an artist can create a masterpiece with
perfect assurance and yet not be able to explain in detail how they
accomplished these feats; these and many other examples indicate that
"we can know more than we can tell."g Such knowledge is 'personal'
because it depends on the integrating consciousness of individuals; it
cannot be filed in impersonal formats. Polanyi claims that the
operation of tacit knowledge can be recognized in every form of
While tacit knowledge can be possessed by itself, explicit
knowledge must rely on being tacitly understood and applied.
Hence, all knowledge is either tacit or rooted in tacit knowledge. A
wholly explicit knowledge is unthinkable. to

Polanyi, then, is trying to hold a middle ground between skepticism and

an ideal of absolute explanation.

Moleski Retortion in Lonergan and Polanyi

Polanyi uses retortion to reject the ideal of "wholly explicit

... any philosophy that sets up strictness of meaning as an ideal is
self-contradictory. For if the active participation of the philosopher
in meaning what he says is regarded by it as a defect which
precludes the achievement of objective validity, it must reject itself
by these standards ....
While impersonal meaning is self-contradictory, the justification of
personal meaning is self-justifying, if only it admits its own
personal character. It licenses certain conditions of articulation
which are bound to become apparent. .. ll

Human knowledge exists in the person who takes responsibility for

discovering and asserting it. Although we must affirm the nature of
humankind as knowers who rely on the tacit framework of their own
minds, the "conditions of articulation" set limits on how much we may
know and how much we may tell of what we know. These conditions
become apparent primarily through the breakdown of the network of
presuppositions which make thought and speech possible. The failure
to understand and communicate calls attention to previously unnoticed,
even unnoticeable conditions of success:

Spoken communication is the successful application by two

persons of the linguistic knowledge and skill acquired by such
apprenticeship, one person wishing to transmit, the other to
receive, information. Relying on what each has learnt, the speaker
confidently utters words and the listener confidently interprets
them, while they mutually rely on each other's correct use and
understanding of these words. A true communication will take
place if, and only if, these combined assumptions of authority and
trust are in fact justified.
We become aware of the precariousness of these conditions when
they are grossly unfulfilled, as in the conversation of children. 12

Experimental psychologists distort the conditions of human experience

precisely to estimate what significance each controllable condition
might have in more ordinary circumstances, in which its contribution
to the process is hidden in the tacit dimension. So, for example, the role
of visual stimulation in taste was verified by coloring whole meals
purple, the role of smell was isolated from taste by blocking the nasal
passages, and so on.
Polanyi trusts the fundamental capacity of humankind to know
and he expects that, on the whole, we will exercise and expand this

Moleski Retortion in Lonergan and Polanyi

capacity by learning from our mistakes. He believes our human nature

functions as a source of knowledge, even though this supposition cannot
be 'strictly proven' to be. true. It is a primary act of faith which is
exhibited in every use of reason by any person who makes a knowledge
claim. No one can deny this, says Polanyi, because no amount of
objectification can obscure.the personal act which is the foundation of
all knowledge: "You cannot formalize the act of commitment, for you
cannot express your commitment non-commitally. To attempt this is to
exercise the kind of lucidity which destroys its subject matter.,,13 On
account of this self-referential inconsistency, "any attempt to gain
complete control of thought is self-contradictory, systematically
misleading, and culturally destructive." 14
Polanyi also uses retortion to distinguish and defend his position
against skepticism. Based on his conviction that "truth is something that
can be thought of only by believing it,"IS Polanyi holds that general
skepticism is doubly discredited. First, it is a fiduciary .position itself,
and hence is inconsistent: "Since the skeptic does not consider it rational
to doubt what he himself believes, the advocacy of 'rational doubt' is
merely the skeptic's way of advocating his own beliefs."17 Second, if
one attempted to make the position self-referentially consistent, it
would eliminate both thought and speech:
... to avoid believing, one must stop thinking ...
Strict skepticism should deny itself the possibility of advocating its
own doctrine, since its consistent practice would preclude the use
of language, the meaning of which is subject to all the notorious
pitfalls of inductive reasoning. 17

Using retortion, Polanyi has staked out the bounds of within which
reason may operate self-consistently: between impersonal objectivism
and irrational silence, there is the domain of personal knowledge.
In this middle ground, there is no apodictic certitude either of
knowledge or of ignorance; rather, there is responsible judgment based
on frameworks of understanding that are open to being transformed by
rational persuasion. All of these personal viewpoints stand under the
judgment of the notion of truth, which is the "external pole" toward
which personal commitment strives as the ultimate frame of
reference. 18 Although none of the personal positions are neutral, it is
the commitment of the person to the notion of higher truth which
en~bles the transition from one partial viewpoint to another:

Moleski Retortion in Lonergan and Polanyi

A mere change of mind might be accomplished by means of

subliminal gimmicks, of lies, fraudulent claims, or, in general,
behavioral brainwashing. But this would not be persuasion as
practiced by scientists. Their sort of persuasion is of an altogether
moral kind, since it implies an "open-eyed" change of mind, one
made without pressure, threats, or psychological tricks--in other
words, one made in the face of the facts and representing what the
person ought to think when he has all the facts at his disposal and
when no one is trying to conceal anything from him.... Truth is the
highest oblect or intention in science--the greatest standard
motivation. 9

This is a picture of the ideal case of the conversion of a rational man to

a more rationally satisfying position. In actual instances of conflict, the
personal pole of knowledge may establish competing standpoints which
are not immediately capable of peaceful settlement, since the
terminology, fundamental facts, and acceptable methods of argument
all are rooted in the tacit, personal substructure of the competing
positions. Since all knowledge exhibits a personal character through
the internal pole of commitment, the character of the persons may be
called into question when the level of debate reaches the fundamental,
personal presuppositions of thought:

... so long as we argue within their framework, we can never

induce them to abandon it. Demonstration must be supplemented,
therefore, by forms of persuasion which can induce a conversion.
The refusal to enter on the opponent's way of arguing must be
justified by making it appear altogether unreasonable.
Such comprehensive rejection cannot fail to discredit the opponent.
He will be made to appear as thoroughly deluded, which in the heat
of battle will easily come to imply that he was a fool, crank or
fraud. '" In a clash of intellectual ~assions each side must
inevitably attack the opponent's person.

This is precisely what Polanyi does to demonstrate the domain of

personal knowledge. Whether he is attacked from the viewpoint of
skepticism or positivism, he responds with a retort, attempting to show
the underlying absurdity of using reason to discredit reason.
Polanyi does not justify irrational hostility toward those who
take opposing viewpoints. The ad hominem quality of retortion is an
appeal to the other to acknowledge the activity of the mind in the
formation of the particular judgments which define the position. In
conflict with the skeptic, Polanyi bases his retort on the assumption that
the other is a rational human being, all denials of rationality

Moleski Retortion in Lonergan and Polanyi

notwithstanding. In conflict with the positivist, he presumes the dignity

of the person who stands as the integrating center of every act of
objectification. The "attack" on the person in each case, then, is not
necessarily tinged with animosity, even though it is mean to reveal the
error of the position. Polanyi's goal is to induce an experience of
discrepancy in the consciousness of his adversaries so that from the
recognition of their errors concerning the role of their own mind and
will in knowledge, they can experience themselves already in the
possession of a modified interpretative framework:

For to modify our idiom is to modify the frame of reference within

which we shall henceforth interpret our experience; it is to modify
ourselves. In contrast to a formal procedure which we can
recapitulate at will and trace back to its premises, it entails a
conversion to new premises not accessible by any strict argument
from those previously held. It is a decision, originating from our
own personal judgment, to modify the premises of our judgment,
and thus to modify our intellectual existence, so as to become more
satisfying to ourselves.21

Again, in the ideal case, the goal of retortion is to introduce cognitive

dissonance in others, so that out of a personal awareness of the
performative inconsistency between what is said and what is done in the
act of speech, they may see for themselves that their position is
untenable. In practice, when our dialogue partners refuse to see what is
'perfectly evident' to us, tempers can give the discussion a tone out of
keeping with the ideals of rational discourse but entirely consistent with
the personal pole of human knowledge. The greater the differences in
the tacit dimension, on which everything else in the personal
framework depends, the greater the likelihood of mutual contempt.
In spite of the difficulty of bringing into consciousness all of the
criteria on which we rely in order to be in touch with reality, Polanyi
holds that to be human is to know. The immense weight of unanswered
and perhaps unanswerable questions about the structure of human
knowing does not crush his conviction that we are knowers by nature;
the difficulties proposed by skeptics only show that our knowledge is
limited and rooted in a level of reality over which we exercise little
direct control. Because our ability to know and to explain our
knowledge is limited, the desire for absolute certitude must be held in
check, lest it drive us to over- or under-rate our capacities. We remain
mysterious to ourselves, since we know we know and yet fall short of
being able to explain how this is so:

Moleski Retortion in Lonergan and Polanyi

While our acceptance of this [fiduciary] framework is the condition

for our having any knowledge, this matrix can claim no
self-evidence. Although our fundamental propensities are innate,
they are vastly modified and enlarged by our upbringing; moreover,
our innate interpretations of experience may be misleading, while
some of our truest acquired beliefs, though clearly demonstrable,
may be the most difficult to hold. Our minds live in action, and any
attempt to specify its presuppositions produces a set of axioms
which cannot tell us why we should accept them....
This then is our liberation from objectivism: to realize that we can
voice our ultimate convictions only from within our
convictions--from within the whole system of acceptances that are
logically prior to any particular assertion of our own, prior to the
holding of any particular piece of knowledge. 22

Retortion calls attention to the ingredients of this vital circle of

knowledge without claiming to be independent of these
presuppositions. In a sense, it catches the mind in the act of forming a
judgment, and provides a glimpse of the interdependence of each
principle of thought on all the rest. Whatever success we have in
articulating the nature of the vital circle comes from within the circle
itself. Isolated from this framework, an axiom is simply a tautology,
with no power to impress itself against any other tautology; immersed
in this framework, in the movement of the mind, an axiom becomes a
tool of thought which disappears from view as we use it to bring reason
to bear on something else.23 What we can do, as Polanyi suggests, is to
recognize what our knowing nature has accomplished, and then begin
to understand some of the conditions of that success.
Polanyi made sweeping claims for the validity of his insights
into the nature of the inquiring mind: "the acknowledgement of
understanding as a valid form of knowing will allow us to study all
human experience by essentially the same method."24 Retortion clearly
plays an important role in establishing this claim, since it helps to show
why we cannot accept alternatives which would distort our grasp of
humankind as limited-but-effective knowers.

Lonergan's Dialectical Analysis

In comparing Lonergan and Polanyi, I do not intend to suggest

that their positions can easily be harmonized with one another. There
are substantial differences in terminology, purpose, method, and
conclusions, particularly concerning the possibility of the knowledge of

Moleski Retortion in Lonergan and Polanyi

God and of metaphysical reality. Since Lonergan held that "an explicit
and adequate metaphysics is a corollary to explicit and adequate
self-knowledge, "25 we should pay some attention to his metaphysical
position in order to see what he meant by his cognitional analysis.
Although Lonergan's scholastic approach may appear to fall
into the self-defeating objectivism which Polanyi scorns, with a host of
technical terms and a tendency to geometric exposition that gives the
impression that Lonergan intends to 'prove everything,' in fact he
defends the same middle ground as Polanyi. Although the weight of
his work bears on what is explicit in thought, Lonergan acknowledges
the importance of the tacit dimension as the ground out of which those
insights grow that enable us to speak intelligently and decide rationally.
Like Polanyi, Lonergan aims to grasp the nature of understanding in
order to gain a viewpoint on the whole of human experience:
"Thoroughly understand what it is to understand, and not only will you
understand the broad lines of all there is to be understood, but also you
will possess a fued base, an invariant pattern, opening upon all further
developments of understanding" [xxviii]. This slogan is typical of
Lonergan's reflexive philosophy; it is indicative of how deeply he is
committed to self-referential consistency as a norm and wellspring of
Lonergan very clearly affirms the personal dimension of
knowledge. Every position taken is the act of a person:

One is not ready to confess to poor judgment because the question

for reflection can be answered not only by 'Yes' or 'No' but also
by 'I don't know'; it can be answered assertorically or modally,
with certitude or only probability; finally, the questions as
presented can be dismissed, distinctions introduced, and new
questions substituted. The variety of possible answers makes full
allowance for the misfortunes and shortcomings of the person
answering, and by the same stroke it closes the door on possible
excuses for mistakes. A judgment is the responsibility of the one
that judges. It is a personal commitment. [272]

Although all knowledge depends on this personal commitment, we are

not locked into an unchangeable position which is out of touch with the
extra-personal world. Our judgment can be led to increasingly more
adequate accounts of knowledge, reality, and objectivity. This
transformation of understanding is accomplished primarily through
"dialectical analysis," which I suggest is based primarily on the logic of
retortion,26 and which aims to convert the person from untenable

Moleski Retortion in Lonergan and Polanyi

positions (or "counter-positions") to those which affirm the structure

of the mind which holds them.
This conversion process is at the heart of Lonergan's program
for philosophy. In attempting to overcome obscurantism in all its
forms, he counts two dynaIl}ics as allies:

On the one hand, there is common sense and in its judgments ...
common sense tends to be profoundly sane. On the other hand,
there is dialectical analysis; the refusal of insight betrays itself; the
Babel of our day is the cumulative product of a series of refusals to
understand; and dialectical analysis can discover and expose both
the series of past refusals and the tactics of contemporary resistance
to enlightenment [242]

Common sense preserves a wealth of knowledge in concrete form, but

it lacks a theoretical base to explain how we know what we know.
When common sense refuses to be tutored by higher viewpoints, it
becomes "general bias." In its healthy form, common sense is a
function of tacit and collective knowledge, whereas in its opposition to
the development of positions and the reversal of counter-positions, it is
self-destructive and the source of increasingly catastrophic errors.
In his summary of the contest between positions and
counter-positions, Lonergan crystallizes his notion of dialectical

But as the detached, disinterested, unrestricted desire to know is

constant, so too are the principles that interfere with its unfolding.
However much at variance with one another positions and
counter-positions may be, a dialectical analysis, based upon a
sufficiently accurate cognitional theory, can proceed to a universal
viewpoint that embraces at once

(1) the positions in the contemporary stage of their development,

(2) the positions at each prior stage of their development, and

(3) the successive counter-positions of the past and present with their
essential incoherence with the claim that they are grasped
intelligibility and afflrmed coherently. [738]

The third point is an instance of retortion. Whoever attempts to deny

Lonergan's notions of reality as "the concrete universe of being,"
knowledge as the intelligent and reasonable judgment of a
self-affirming knower, and objectivity as "a consequence of intelligent

Moleski Retortion in Lonergan and Polanyi

inquiry and critical reflection" [387-8], implicitly affirms each of these

aspects of correct thought about thought: such denials address the
question of what is really the case in the universe, are based on the
claim to know what is knowable, and are offered to other rational
minds as a truth which may be shared by all and sundry. Recognizing
the absurdity of adopting such a referentially inconsistent position
provides Lonergan with a fixed point for reflection and "a single base
of operations from which any philosophy can be interpreted correctly"
[530-1]. As suggested above, Lonergan considers this adoption of this
base to fix the grounds of both epistemology and metaphysics:

If a spatial image and a military metaphor may be helpful, the

advance of metaphysical evidence is at once a break-through, an
envelopment, and a confinement. The break-through is effected in
one's affirmation of oneself as empirically, intelligently, and
rationally conscious. The envelopment is effected through the
protean notion of being as whatever one intelligently grasps and
reasonably affirms. The confinement is effected through the
dialectical opposition of twofold notions of the real, of knowing
and of objectivity, so that every attempt to escape is blocked by the
awareness that one would be merely substituting some
counter-positions for a known position, merely deserting the being
that can be intelligently grasped and reasonably affirmed, merely
distorting the consciousness that is not only empirical but also
intelligent and not only intelligent but also reasonable.
Once this foundation is laid and as long as it is retained effectively,
one can proceed rapidly with the erection of the integral heuristic
structure of proportionate being. [484]

Retortion, then, functions to confine thought within its proper limits,

no matter how uncomfortable those limits may be in view of our desire
for completely unconditioned knowledge. The counter-positions may
be attractive in part because they are simpler than the affirmation of
Lonergan's position, since they may absolve us from the admittedly
difficult task of making the best use of our finite intellectual resources.
We are tempted to hold that ignorance is bliss because this
licenses any whim or fancy of the narcissistic pattern of existence.
Lonergan contends that an honest appraisal of one's own consciousness
supplies all that is necessary to resist this temptation to shirk personal
responsibility for the preservation and expansion of knowledge. It is in
texts such as this that the convergence of Lonergan and Polanyi's
thought is most evident:

Am I a knower? The answer, Yes, is coherent, for if I am a

Moleski Retortion in Lonergan and Polanyi

knower, I can know that fact. But the answer, No, is incoherent,
for if I am not a knower, how could the question be raised and
answered by me? No less, the hedging answer, I do not know, is
incoherent. For if I know that I do not know, then I am a knower;
and if I do not know that I do not know, then I should not answer.
Am I a knower? If I am not, then I know nothing. My only course
is silence. My only course is not the excused and explained silence
of the skeptic, but the complete silence of the animal that offers
neither excuse nor explanation for its complacent absorption in
merely sensitive routines. For if I know nothing, then I cannot
know the explanation of my ignorance.

It is this conditional necessity of contingent fact that involves the

talking skeptic in contradiction. [329]
This fundamental capacity to know that I can know can be
expanded and refined by appreciating the patterns by which I move
from ignorance to knowledge and from misunderstanding to
successively more adequate understanding. Polanyi portrays the
crucial first step as a moment of discovery, guided by and founded on
the tacit powers of the mind. Lonergan proposes a three-fold
distinction in the development of knowledge, from data to insights
which correlate the data, and from conflicting insights to judgments
which decide the question (or decide not to decide, as the case may be).
Both hold that the experience of questioning my own ability to know
provides sufficient reason to answer the question affirmatively.
Whoever understands the basic terms will be able to recognize the
"confinement" of a circle of presuppositions that make the question
possible and then settle it decisively: "A reviser cannot appeal to data to
deny data, to his new insights to deny insights, to his new formulation to
deny formulation, to his reflective grasp to deny reflective grasp"
It is on the basis of this self-supporting structure of
consciousness, identified and vindicated by retortion, that Lonergan
repeatedly claims his position is incapable of radical revision. Such a
strong claim to invariance is absent from Polanyi's works, though one
might argue from the tone of his writing to an implicit claim. The kind
of certitude which Lonergan defends is not the same kind of certitude
which Polanyi rejects as self-destructive lucidity. Lonergan and
Polanyi have the same intention--to affirm the fact that knowledge is
real and is rooted in the person:

The orthodox Kantian would refer to our stand as mere

psychologism, as an appeal to the empirical that can yield no more

Moleski Retortion in Lonergan and Polanyi

than a provisional probability. But our retort is simple enough.

Without judgments of fact one cannot get beyond mere analytic
propositions. Further, though self-affmnation is no more than a
judgment of mere fact, still it is a privileged judgment.
Self-negation is incoherent One has only to inquire and reflect, to
find oneself caught in the spontaneities and inevitabilities that
supply the evidence for self-affmnation. One has only to make a
single judgment of fact, no matter what its content, to involve
oneself in a necessary self-affmnation. Finally, cognitional theory
differs from other theory; for other theory reaches explanation only
by venturing into the merely supposed; but cognitional theory
reaches explanation without any such venture; and since it contains
no merely hypothetical element, it is not subject to radical revision.

Polanyi's primary concern was to bring this fact to the attention of

philosophers of science who had wrongly exalted the ideal of
impersonalknowletige. While Lonergan shared that objective, he also
intended to show that there is one and only one kind of metaphysical
position which adequately accounts for the facts presented in our

... there are many particular sciences and each of them deals with a
variety of objects, but there is only one integrated view of one
universe and there is only one set of directives that lead to it...
... philosophy obtains its integrated view of a single universe, not
by determining the contents that fill heuristic structures, but by
relating the heuristic structures to one another. Because of these
differences in their objectives, scientific method stands to scientific
conclusions as a genetic universal to generated particulars, but
philosophic method stands to philosophic conclusions as the
genesis to the attainment of a single all-inclusive view. [426]

This "all-inclusive view" is not to be strictly identified with Lonergan's

articulation of it. Rather, it is a position which many philosophers have
adopted and used implicitly or explicitly as the base of their operations;
while the position is invariant in itself, it is variable in the concrete
expressions that attempt to describe it, since the accounts are influenced
by the exigencies of particular places and times.
Lonergan correlates the expressions that are peculiar to our age
with the classical categories of Aristotle in order to show how the
invariant substructure can, over time, give rise to better answers to the
same questions. Such is the case with the progress of science [394] and
refinements in the language we have available for posing and answering
questions about meaning. Lonergan claims that such progression of

Moleski Retortion in Lonergan and Polanyi

thought is grounded in those unchanging structures of the mind that

enable us to be rational:

The only manner in which this basic theorem [relating his notions
of experiencing, understanding, and judging to Aristotle's
categories of potency, form and act] could be modified would be to
modify its factual supposition that knowing consists in
experiencing, understanding and judging; and it has been argued
that that fact is not open in any concrete meaning of the term,
revision. For any human reviser would appeal to experience,
understanding, and judgment; and there is no use arguing that men
might be other than they are, because it is equally true that the
universe might be other than it is and the issue lies, not in the
possibility of a different metaphysics in a different universe, but in
the possibility of a different metaphysics in this universe. [735]

This passage typifies the realism which Lonergan insists on throughout

Insight. It clearly distinguishes him from the men of the Enlightenment
who sought a set of axioms from which all possible metaphysical
circumstances would be deducible and from which the actual history of
the universe could be derived as well. Lonergan adheres to his own
norm that what is to be sought through dialectical analysis is this
concrete universe of being, which we are capable of experiencing from
the inside out through knowing ourselves as real beings. So long as
human nature is what it is, the universal viewpoint established by
dialectics will remain essentially the same. If, for the sake of argument,
we imagine the possibility that human nature might change, both the
principles of the change and of the interpretation of the change would
have to be so different from our current experience that we now could
know absolutely nothing about it--the change or changes could only be
experienced, understood, and evaluated after the fact, not a priori.
The more difficult, and perhaps more realistic problematic is
the techno-scientific strategy which intends to change human nature.
Both Lonergan and Polanyi reject Marx's dialectical materialism and
our Western idolatry of science and technology. At this juncture, some
might take cognitional analysis as a ground for an ethical stance, since it
declares that human nature as presently constituted and as correctly
understood is good and is worthy of being protected. These concerns,
of course, go beyond the scope of this essay.

A Personal Evaluation of Retortion

I have been interested in retortion for fourteen years or so, ever

Moleski Retortion in Lonergan and Polanyi

since Peter Kreeft discussed it in a seminar at Boston College as an

antidote to unrestricted Cartesian doubt. I have come to believe that it
is a valid and important form of argument, even though it cannot settle
every philosophical issue. In conversation with my fellow Jesuits, it
seems that retortion causes more arguments than it settles, since the
sting of its ad hominem character rouses living opponents to ever more
clever defenses against the charge that their performance in presenting
a conclusion to others contradicts the contents of the conclusion. One
friend in particular remarks, "I'll never accept retortion as valid; it
makes me feel as though I've picked my own pocket." Others have
expressed a similar weariness and disenchantment with the strategy of
retortion, since it seems to them that even if the argument is sound, its
results are trivial at best. In all of the years that I've used the demand
for self-referential consistency as a key criterion of thought, I do not
think that I have seen one person I have spoken with change or abandon
his or her position for fear of continuing in the sin of performative
contradiction. What good is an argument that persuades so few? In
theory, retortion should be powerful enough to unify the whole of
philosophy--why is it so weak in practice?
W.H. Walsh offers an excellent statement of the case against
retortion, summing up the whole history of its inadequacies from
Aristotle to the present:

Aristotle in the Metaphysics had sought to demonstrate that the laws

of identity and contradiction, which he took as being truths of fact
as well as laws of thought, were such that their validity was
presupposed in any attempt to deny them. At a later stage the
Idealists were to suggest that there was no rational alternative to
their philosophy, since it followed from premises which had to be
accepted if anything was to be accepted at all: as Bosanquet put it, it
was a case of 'this or nothing.' Yet the unsatisfactoriness of this
procedure, and of the accompanying attempts to argue that rival
metaphysicians must be involved in self-contradiction, is
sufficiently shown in the fact that these diverse parties appealed to
it. [If the argument were truly satisfactory] ... one system of
metaphysics would long ago have been recognized as definitive. 27

Apart from the fact that this argument could be extended to discredit all
kinds of philosophy, including skepticism--no philosophy has won the
consent of all--the main difficulty I have with Walsh's argument is its
assumption that all philosophers are completely rational, and that they
are moved by nothing else except the unrestricted desire to know. Only
if we humans were all perfect would the definitive system of

Moleski Retortion in Lonergan and Polanyi

metaphysics (or anti-metaphysics) gain universal acceptance. The

inability of philosophers to agree is as much to the discredit of the
philosophers as it is to the philosophies which stand against one another.
Since Walsh's case is built on the facts of history, it cannot be used to
uphold general skepticism; since it is true that no one system has been
universally accredited, the cause of that chaos must be determined on
grounds other than the premise that "There is no truth on which we can
and should agree."
----what Lonergan and Polanyi both highlight is the role of belief
in the development of knowledge. No argument, however excellent,
can persuade a hostile adversary. The grounds on which retortion
depends for its validity can easily be neglected, since the first principles
of thought function in the tacit dimension. Both men agree that there is
no simple way to 'take a look' at the conditions of thought. These
conditions are evident only to those who have already committed
themselves to an interpretive framework which includes categories to
describe the action of the mind in the formation of concepts.
I personally experience the confining power of retortion, yet I
can understand why others do not. For those who stand within the
Aristotelian/Thomistic framework, retortion helps to grasp and affirm
synthetic principles as axioms. For those who stand outside this
viewpoint, retortion has the appearance of the con game in the story of
the king's new clothes. The skeptic cries out in bafflement against those
who cheer the procession of the naked king. Since metaphysical
realities by definition cannot be placed before the eyes of the skeptic for
direct verification, but can be affirmed only by a reflexive judgment
supported by an act of trust, it may be the defenders of the metaphysical
viewpoint who ultimately fall silent while the skeptic chatters on about
all that isn't visible. The metaphysical viewpoint can attract adherents,
but it cannot capture enemies.
Retortion is only one small piece of the Aristotelian tradition.
Apart from the other fundamental insights, it is not very fruitful.
Polanyi's analysis of the importance of the tacit dimension may help
explain the paradox that retortion becomes less and less effective the
more it occupies the center stage of philosophical argument. Retortion
sets certain boundaries for thought, but it is no substitute for the
dynamism of the inquiring mind that can take advantage of those
boundaries to guide itself toward the acquisition of truth. The part
must be integrated into the whole in order to play its role of
contributing to the vitality of every other part, and such integration is

Moleski Retortion in Lonergan and Polanyi

accomplished in the tacit dimension rather than in the partializing focus

of critical consciousness.
I believe Lonergan's presentation achieves the right balance
between the role of retortion and the roles of other complementary
insights into the structures of the mind. Dialectical analysis is used like
the logic of Sherlock Holmes to rule out every other position but one,
which must be accepted as true no matter how much this may shock
one's preconceptions. However, such analysis does not then provide a
starting-point from which everything else may be rigorously deduced.
Rather, it provides a ground that may bear fruit over the long run,
weeding out those presuppositions that are destructive to the progress
of thought and encouraging further investigation of the position. The
results are cumulative throughout history, forming a heritage which
anyone may claims as their own who is willing to adopt the common
foundation of thought. The philosophical position which Aristotle
discovered and gave expression to may become the common property
of anyone who wishes to possess it. Though the language of the position
has changed, the presuppositions of his framework and the method of
identifying those presuppositions has not changed. This capacity to
remain self-identical over time and through a variety of expressions in
particular places is what constitutes metaphysics as a science:

... the possibility of mistaken formulations has its ground in the

polymorphism of human consciousness; and the selection of correct
formulations can be effected in inasmuch as the incoherence of
counter-positions invites their reversal.
Now such a procedure eliminates mere disputation and bestows
upon metaphysics the status of a science. I do not meant that it
secures automatic solutions for metaphysical issues, nor that it
annihilates the obscurantist, the obtuse, or the mind fixed in
habitual routine. On the contrary, I regard the automatic solution as
a mere myth that springs from a non-rational hankering after a
non-rational security, for every solution is to be discovered by
intelligence and is to be accepted by reasonableness, and neither the
exercise of intelligence nor the exercise of reasonableness is
automatic. Again, like the poor soul of the Gospel, the
obscurantist, the obtuse, and the merely routine may be expected to
always be with us.28

Retortion, then, cannot be expected to be the messiah who will gather

together all the of the scattered flocks of philosophy and unite them in
eschatological peace. It may persuade some that critical realism is the
most authentic approach to understanding human understanding, so that

Moleski Retortion in Lonergan and Polanyi

the heritage of Aristotle may continue to have a voice in the courts of

reason and the marketplace of ideas. Those who choose this position
may well suffer the taunt that they are following an unclothed king;
they perhaps will judge that it is better to bear such contempt than to
participate in the madness of stripping reason of reason.


Our brief glance at Polanyi's position shows that retortion plays

a central role in staking out the ground of personal knowledge between
the doubts of skepticism and the hyper-objectivity of positivism.
Polanyi highlights the importance of real but unspeakable knowledge
for the whole of human life, and uses retortion as a form of persuasion
to call for the acceptance of oneself as a knower, in spite of the
irreducible mystery of the tacit dimension.
Lonergan's position coincides with Polanyi's cognitional
analysis, although employing a different terminology and designed
with a different purpose in mind. Lonergan uses retortion to defend
Aristotelian metaphysics as the only fully satisfying viewpoint from
which the variations of all lower viewpoints make sense.
This essay suggests that Lonergan's affirmation of a single,
invariant base of operations complements Polanyi's certitude that all
knowledge may be supposed to have the same structure, regardless of
its content. Polanyi's emphasis on the significance of the tacit
dimension and his insistence on the importance of the informal logic of
conversion provide an excellent balance to Lonergan's focus on the
articulable elements of cognition. Both men demonstrate that retortion
can be a vital tool of thought when it is embedded in a larger context of
insight into human nature and when it functions as the adjunct of an act
of trust in the mystery of our own finite intelligence.

Martin X. Moleski
Department of Philosophy
Catholic University of America
Washington, D.C.

Moleski Retortion in Lonergan and Polanyi

1. For a brief introduction to this school, see Joseph Donceel,
"Transcendental Thomism," The Monist, 58 (1974) 67-85. Otto
Muck provides an extended analysis in The Transcendental
Method, Herder and Herder, 1968. For a synthesis of a position
that was based almost entirely on retortion, see my article,
"Retortion: The Method and Metaphysics of Gaston Isaye,"
International Philosophical Quarterly, 17 (1977) 59-83.

2. For another comparison of the two positions, see J. Kroger,

"Polanyi and Lonergan on Scientific Method," Philosophy
Today, 21 (1977) 2-20, and "Theology and Notions of Reason
and Science: A Note on a Point of Comparison in Lonergan and
Polanyi," Journal of Religion, 56 (1976) 157-61.

3. Insight: A Study of Human Understanding, Harper and Row,

1978, second edition; originally published in 1957 by
Longmans, Green and Company, and revised in 1958.

4. Personal Knowledge: Towards a Post-Critical Philosophy,

University of Chicago Press, 1958.

5. ibid, p. 61.

6. Personal Knowledge, p. 214.

7. Michael Polanyi and Harry Prosch, Meaning, University of

Chicago Press, 1975, p. 104.

8. ibid., p. 61.

9. The Tacit Dimension, Anchor Books, 1967, p.4.

10. Knowing and Being: Essays by Michael Polanyi, ed. Marjorie

Grene, Routledge and Keegan Paul, 1969, p. 144.

11. Personal Knowledge, p. 252-3.

12. ibid., p. 206.

Moleski Retortion in Lonergan and Polanyi

13. Tacit Dimension, p. 25.

14. Knowing and Being, p. 156.

15. Personal Knowledge, p. 305.

16. ibid., p. 297.

17. ibid., p. 314-15.

18. Tacit Dimension, p. 87.

19. Meaning, p. 211.

20. Personal Knowledge, p. 151; emphasis added.

21. ibid., p. 105-6.

22. ibid., p. 267.

23. ibid., p. 169.

24. The Study of Man, University of Chicago Press, 1964 [1946], p.


25. Insight, p. 535; since I confine my analysis wholly to this work

in this section, I have simply inserted page references in the text
as necessary.

26. Ronald McKinney shows that there are three meanings of

'dialectic' in Lonergan, one of which corresponds to retortion;
see "Lonergan's Notion of Dialectic," The Thomist 46 (1982)
221-41. My re-reading of Insight in preparation for this article
convinces me that whenever Lonergan uses the phrase
'dialectical analysis,' he is envisioning a movement of thought
brought about by retortion.
27. Metaphysics (New York: Harcourt, Brace, and World, 1963), p.

28. Insight, p. 525.

Moleski Retortion in'Lonergan and Polanyi


Donceel, Joseph. "Transcendental Thomism," The Monist, 58 (1974)

Kroger, J. "Polanyi and Lonergan on Scientific Method," Philosophy
Today, 21 (1977) 2-20.
_ _-=-' "Theology and Notions of Reason and Science: A Note on a
Point of Comparison in Lonergan and Polanyi," Journal of
Religion, 56 (1976) 157-61.
Lonergan, Bernard J.F. Insight: A Study of Human Understanding,
Harper and Row, 1978, second edition [1957,1958].
,=,"",=","-:-:-_ _ Method in Theology, Herder and Herder, 1972.
Mckinney, Ronald. "Lonergan's Notion of Dialectic," The Thomist, 46
(1982) 221-41.
Moleski, Martin. "Retortion: The Method and Metaphysics of Gaston
Isaye," International Philosophical Quarterly, 17 (1977) 59-83.
Muck, Otto. The Transcendental Method, trans. William D.
Seidensticker, Herder and Herder, 1968.
Polanyi, Michael. Knowing and Being: Essays by Michael Polanyi, ed.
Marjorie Grene, Routledge and Kegan Paul, 1969.
, and Harry Prosch. Meaning, University of Chicago Press,
----:-1-::"':975 .
. Personal Knowledge: Towards a Post-Critical Philosophy,
University of Chicago Press, 1958.
_ _--:--;::-. Science, Faith, and Society, University of Chicago Press,
1964 [1946].
____. The Study of Man, University of Chicago Press, 1959.
=--=-=---==' The Tacit Dimension, Anchor Books, 1967.
Walsh, W.H. Metaphysics, Harcourt, Brace and World, 1963.

James E. Swearingen

Reflexivity and the Decentered Self

in Tristram Shandy
Ever since Copernicus man has been rolling down an incline, faster
and faster, away from the center-whither? Into the void? Into
the piercing sense of his emptiness?
The Genealogy a/Morals

There is something a little startling at first in the discovery that

Friedrich Nietzsche, the solitary and sickly philosopher, who wrung
from his tortured life a profound vision of the end of a cultural epoch,
should have adopted as his favorite novel Laurence Sterne's great
comic work, Tristram Shandy. The apparent incongruity, especially to
English readers, who tend to see Nietzsche as self-indulgent,
melodramatic, and blasphemous, is between a kind of extravagant
gravity in the one writer and a prankish good humor in the other. This
exuberant novel with its profusion of wit and wisdom, published in five
installments between the years 1760-67, may be alien to the style of
"blood and aphorisms," but it is one with Nietzsche's own Dionysian
wit, with his attacks on the spirit of gravity, and with his celebration of
Ubermut and laughter. Both are thinkers with "light feet," as
Zarathustra might say, who step easily from peak to mountain peak;
and they can commune in perfect harmony over Sterne's definition of
gravity (adapted from la Rochefoucauld) as "A mysterious carriage of
the body to cover the defects of the mind."l
There is nothing mysterious by now in the historical current of
thought from one centered, reified thing (person, idea) to another, each
conceived in tum as providing a stable reference point both outside,
and nonetheless at the center of the flux of the world and thus as
guaranteeing the intelligibility of change. The famous passage in The
Joyful Wisdom where the Madman announces the death of God entails
the even more radical prophecy of the end of man because it recognizes
the end of the whole series of grounding certitudes: "What did we do
when we loosened this earth from its sun? Whither does it now move?
Whither do we move? Away from all suns? Do we not dash on
unceasingly? Backwards, sideways, forwards, in all directions? Is
there still an above and below? Do we not stray, as through infinite
nothingness? Does not empty space breathe upon us? Has it not become
Swearingen Reflexivity and the Decentered Self in Tristram Shandy

colder?"2 No mode of inquiry offers better opportunities for

observing both the necessity and the dissolution of the ideality of the
self as a metaphysical entity than a work of fiction in which a character
tries to discover exactly what or who he is by a simultaneous, reflexive
archeology and teleology of his being as a self. Tristram Shandy is such
a novel, the focus of which is captured in a short passage of the seventh
volume: "-My good friend, quoth I-as sure as I am I-and you are
you- -And who are you? said he,- -Don't puzzle me; said I" (VII.
xxxiv. 525). The most extraordinary conceptual departure of this
astonishing book, and perhaps its most Nietzschean feature, is its
transgression of the presuppositions and implications of the novel as a
literary genre, especially its exploration and ultimate demystification
of the subject who is ostensibly writing his autobiography.
The rise of the realistic novel in eighteenth-century England is
widely understood as a triumph of the bourgeoise spirit. Among the
presuppositions of the form are the thesis of radical individualism and
the delineation of man as moved by the will to win and to achieve in a
serially organized world. 3 For the hero of classical epic, by contrast,
life is a totality the meaning of which is immanent in every scene and
every act. What happens to Achilles at Troy is not private fate but the
destiny of the community of Greeks. Georg Lukacs has aptly said that
the novel "is the epic of a world that has been abandoned by God."4
Where meaning is no longer immanent, where the hero is a solitary in
his "transcendental homelessness," the world has become problematic,
and the hero's adventures are a linear, temporal quest for meaning that
is incompletable in principle. Where "the totality of life resists any
attempt to find a transcendental centre within it, and refuses any of its
constituent cells the right to dominate it," life and plot are an endless
battle between an interiority that becomes problematic in the conflict,
and a recalcitrantly prosaic world. S
Where most great novels, especially those written before the
tum of the present century, uncritically accept the thesis of the ego, in
the age of the origin of this literary form Sterne devises a reflexive
procedure which, without explicitly rejecting any of the
presuppositions of the form, dismantles in substantial degree its
ideology, most conspicuously its central egological moorings. Here we
shall examine this profoundly and selfconsciously philosophical text in
two stages: first, by describing the novel as a reflexive strategy the aim
of which is to discover the ontological structure of Tristram's being,6
and second, by examining how the ego presumed to be the central
presence, is reintegrated in the matrix of the play of the world and
becomes a figure in the infinte play of discourse.

Sweanngen Reflexivity and the Decentered Self in Tristram Shandy

The comic obstacles which confront the reader of the novel

begin with the promise of the title page, "The Life and Opinions of
Tristram Shandy, Gentleman." Although Tristram promises his life
"for the amusement of the world" and his opinions "for its instruction,"
we learn almost nothing directly about what he does or what he thinks.
We get little but the apparently random collection of whatever" ideas"
pass across his mind. In the process, of course, we make the
acquaintance of as memorable and as odd a family of hobby-horsical
"originals" as may be found in the literature of the world. All ride
their hobby-horses through a nine volume "set of as pitiful
misadventures and crossaccidents as ever small Hero sustained." At the
head of the family is "my Father," the intellectualist Walter Shandy,
with his theories of names as determiners of character; of noses both
long and short, creative and Shandean; of the origins of civil
government in the family--one man, one woman, and one bull; of the
virtures of birth by "Caesarian section" to preserve the cerebellum
from "470 pounds averdupoise acting perpendicularly upon it"; and the
educational theory of his projected Tristrapaedia, during the
composition of which, says Tristram, "I was ... totally neglected and
abandoned to my mother...--every day a page or two became of no
consequence." There is the sentimental Uncle Toby, wounded "upon
the groin" in King William's wars, whose science of fortifications and
military games on the bowling green thinly veil a sexual repression and
singular inadequacy for satisfying the curiosity of the concupiscent
Widow Wadman. There is Dr. Slop, "Papist" and man-midwife, with
his newly invented forceps and science of obstetrics; and Corporal
Trim, Toby's servant, whose wound is to the knee and who, as "Non
Hobby-Horsical per se," communes without difficulty with Widow
Wadman's maid Briget one evening on the bowling green with some
damage to Uncle Toby's Dutch drawbridge. There is "my Mother,"
who, fully competent to control her husband's more destructive
projects, has the misfortune of not being "a woman of science," and
who, in the opinion of her husband, has" such a headpiece, that [a man]
cannot hang up a single inference within side of it, to save his soul from
destruction." And there is Yorick, the jester-parson and latter-day
Quixote, a man of wisdom who "loved a jest in his heart-and ... saw
himself in the true point of ridicule." In this gallery of comical
portraits Yorick is a normative figure in whom the spirits of judgment
and wit are so thoroughly blended that he was "over-power'd by

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

numbers" in the modem world and "died ... as was generally thought,
quite broken hearted."
As Tristram, the last of the Shandys, sits in his library writing
his story, the others are long since dead and he, bleeding from
consumptive lungs, is trying to the last to find out who he is. Although
the principle of organization in this digressive text appears to be free
association, Tristram's explicit model is Locke's Essay Concerning
Human Understanding. Of it he observes, "I will tell you in three
words what the book is.-It is a history.-A History!-{)f who? what?
where? when? Don't hurry yourself.-It is a history-book, Sir (which
may possibly recommend it to the world) of what passes in a man's own
mind ... elements of perception, instead of studying it empirically in
the manner of Locke and the others who "syllogize by their noses,"
Tristram actually begins with wholes and proceeds reflexively as
suggested by Locke's unfulfilled metaphor in the Introduction to the
Essay of the eye trying to observe its own seeing. Sterne's protagonist
in another novel, A Sentimental Journey, says, "There is not a more
perplexing affair in life to me, than to set about telling anyone who I
am-for there is scarce any body I cannot give a better account of than
myself.'" When Tristram sets out to give an account of himself, his
search constitutes in fact a tum toward transcendental subjectivity more
radical even than Kant's, for his account is not of the a priori conditions
of knowledge alone, but of the preconditions of the whole of his
concrete being, his life as well as his opinions.
The first problem Tristram faces is to find a place to begin: his
ill-starred birth is too late-too much has already gone awry. He tries
to begin "ab Ovo," from the egg, but his mother's untimely question to
his father at the very moment of conception-"Pray, my dear, have you
not forgot to wind up the clock?"-shows only too well that even his
conception marks no beginning. Before he has done, we know the
family history, symbolic of the history of England and of the spiritual
condition of European culture from "Harry the VIIIth's time" until
"half a century" hence, "the period beyond which I doubt likewise of
the existence of the Christian faith" and the human soul itself. What the
problem of beginning discloses is that there is no zero-point for either
his thinking or his being; all beginning is beginning in the middle of
things; all progress is regress, "one wheel within another," and a
continuous hermeneutic circling from part to whole to part again. The
subject is historically conditioned, a continuation of projects already
underway when he appears on the scene and unfinished still when he
One of the great puzzles for Shandy criticism has always been

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

why Tristram tells everyone's story but his own, whether indeed the
title page is not deliberately subverted by the plot. It is important to
distinguish clearly between two interpretive strategies: there is the
representation of the concrete, existential engagements of the other
characters with the world, and there is Tristram's reflexive
examination, not of the events of his own life, but of the preconditions
for having any experience whatever. Three prominent themes mark
the difference between the novelistic events at Shandy Hall and the
reflexive strategy by which the narrator seeks the ontological structure
of his own being as a fully present entity, the principle that will resolve
his problematic being by explaining him fully. First, he discovers in
the others a new and more reliable way of drawing character, namely,
from a person's hobby-horse. For example, when in the year 1718
Uncle Toby and Walter are talking in the parlor while Tristram is
being born upstairs, Walter says, "I wonder what's all that noise, and
running backwards and forwards for, above stairs." Toby gets no
further in his reply that "I think... " when Tristram interrupts: "But to
enter rightly into my uncle Toby's sentiments upon this matter, you
must be made to enter first a little into his character" (I. xxi. 68). He
then leads us through 36 pages of digressions about Toby's wound, his
interest in fortifications, his embarrassment at any mention of Aunt
Dinah's being "got with child by the coachman" some twenty years
earlier. When in the next volume Tristram finally returns to Toby's
answer, we understand its evasiveness because we know the "I," not as
an unconditioned ego, but as an "I" constituted by its emersion in the
world, a kindly, suffering, and somewhat prudish man who says, "I
think ... -it would not be amiss, brother, if we rung the bell" (II. vi.
99). The clue to a character drawn from his hobby-horse seems to lie
not in the analysis of experience constructed from elements of
perception, but in the rich intentionality which informs it.
Accordingly, what Toby sees and hears is vastly different from what
Walter experiences because their patterns of concern differ.
The existential engagements with the world that Tristram
observes in the other characters reveal the intentional structure of
consciousness as an ontological principle. What the eye could never see
in itself, it can see in the process of Toby's seeing and apply to itself in
tum. Tristram also has a hobby-horse, namely, writing his life, and
that intention would seem to be the interpretive principle that
permeates the field of his consciousness and guides the
selection-evaluation of every detail that comes to his attention. Instead
of his reflections being "organized" by free association, the ontological
appropriation of the hobby-horse reveals a coherence among the wildly

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

digressive structures of the novel. The subjectivity which seems to be

the focal point of interest is like a code by which what is dredged up
from the past and what is intended in the future rise to the surface of
that consciousness. As a comedy of events at Shandy Hall, the plot
moves well within the individualism of the novel as form: ego collides
with ego within a solitude only partially 'mitigated by the sentiment that
still binds the family together in the year '18, and all is certified by the
consciousness of a narrator who occupies the paradoxical position
beyond the plot of which he is also the center. Tristram's proximity
and distance from those events open counter-possibilities to the thesis of
the self as irreducible substance.
The second topic to which Tristram is inexplicably drawn is
time. The serial pattern of plot with its presumption that time is a
sequence of instants in which egos undertake to fulfill their aims and
discover meaning in the world, corresponds to the "metaphysical
dissertation upon the subject of duration and its simple modes," in
paraphrase of the second book of Locke's Essay, that Walter is want to
deliver to Toby. Time is not the measure of motion as in Aristotle, but
the regular succession of our ideas." Whether external or internal, the
notion of sequence nonetheless serves the implications of the novel as a
structure of deferral. Only in the reflexive step back does that analysis
become troublesome. In order to maintain the realism of his narrative,
Tristram professes to offer one hour of reading for each hour of living
only to discover in the middle of his fourth volume that he has got "no
farther than to my first day's life ... so that instead of advancing, as a
common writer ... I am just thrown so many volumes back ... At this
rate I should just live 364 times faster than I write-and consequently,
the more your worships read, the more your worships shall have to
read" (IV. xiii. 285-86).
The essence of his frustration is that serial presentation does not
fully reveal what he needs to express, namely, his most intimate
experience of temporality. In a passage that describes the problem
from the perspective of his writing, he says,

My mother, you must know,-but I have fifty things more

necessary to let you know first,-I have a hundred difficulties
which I have promised to clear up, and a thousand distresses and
domestic misadventures crowding in upon me thick and threefold,
one upon the neck of another ... -Of all the perplexities a mortal
author was ever seen in this is certainly the greatest, ... I have to
fmish ... all this in five minutes less than no time at all; -such a
head!-would to heaven! my enemies only saw the inside of it!
(III. xxxviii. 235)

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

What Tristram wants to convey by saying everything at once, is, with

respect to narrative strategy, the simultaneity of events. In fact it is
much more. To show who he is and reveal the meaning of his being he
would have to convey the simultaneity of his own temporal field.
Where the elder Shandys seek fulfillment in the achievements and
postponements of linearity, the reflexive strategy opens to Tristram the
prospect of meaning as the simultaneous grasp of a pattern,
Piist-present-future, in the temporal thickness of the "now." Thus when
in Volume VII he describes his visit to Auxerre, two journeys coalesce,
the one in his youth, the point he has reached in the chronology of his
life, and the one he has just made as writer years later. And, of course,
there is the third temporal frame in which he is writing the account of
the other two:
... In this last chapter ... I have been getting forwards in two
different journies together, and with the same dash of the pen-for
I have got entirely out of Auxerre in this journey which I am
writing now, and I am got half way out of Auxerre in that which I
shall write hereafter .... I am at this moment walking across the
market-place of Auxerre with my father and my uncle Toby, in our
way back to dinner-and I am this moment also entering Lyons
with my post-chaise broke into a thousand pieces- and I am
moreover this moment in a handsome pavillion ... upon the banks
of the Garonne ... where I now sit rhapsodising all these affairs.
-Let me collect myself, and pursue my journey. (Vn.
xxviii. 515-16)

For such a center of temporal transcendence the only past there is is the
past as presently retained; the only future, possibilty in anticipation.
For such a being, all is always now. Meaning disclosed in the linear
pattern assumed by the form of the novel is either never found or lost as
soon as found; but meaning as the transparency of the moment with its
imaginative vision of beginning and ending is a more radical
interpretation altogether. Tristram sees that the primary phenomenon
is not serial organization either of the objective world or of subjective
events. It is the simultaneous field of consciousness that can distinguish
before and after, by virtue of its own ecstatic being, the enabling
precondition of having the sequence of any history whatever. It is what
as readers we have when we complete the novel, the unitary pattern that
the name "Tristram Shandy" says, the history of the family, his
conception, his birth, his project of writing his life, and especially his
sense of ending.
Death is the third theme of life in Shandy Hall which becomes an
ontological determiner under the peculiar light of reflection. Mortality

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

is never far from mind: Yorick dies in volume I, though it only

intensifies his presence in the book; Tristram nearly dies at birth;
brother Bobby dies en route to the continent for the grand tour; the
pathetic story of Toby's brother officer Le Fever ends in death; all the
principles except Tristram are dead at the time of writing; and he faces
the imminent prospect of the end of the family when he dies without an
heir. Walter discourses learnedly on death as "nothing but the
separation of the soul from the body." Empirically regarded as a
worldly event coming down the stream of future events, death brings to
a period the very pursuit typically embodied in the genre and
constitutes the great negation. Considered onto logically and under the
scrutiny of reflection, death is quite a different matter. To treat it
reflexively Sterne needed a different strategy, of course, than what was
required by the others. It is insufficient for Tristram to muse on the
deaths of others. One's death as viewed by another means broken
community and a sudden solitude, but that does not even approximate
its ontological significance. Tristram's imaginative reconstruction of
his progenitors' attitudes toward death will not suffice, not even the
poignant scene at Yorick's death bed. The singularity of the
anticipation, its individuating effect must be presented in first person.
Thus in volume VII Tristram appears as an opaque character who
travels on the continent seeking relief from his consumption.
Notwithstanding his determination to elude death by "scampering away
to mount Vesuvius-from there to Joppa, and from Joppa to the
world's end, where, if he follows me, I pray God he may break his
neck-to (VII. i. 480), there is no evasiveness, no will to conceal from
himself his prospect of ceasing to be.
Ontologically, it is not the end that is important, for "end" is
already an "in the midst of'; it is the being-toward the end, a recognized
and present finitude that provides the bounding line that gives him his
life as his own. In realizing that his is a -being for whom it is possible
not to be, Tristram discovers the promise of his being, the possibilities
open to him. At one point in the last volume when he is tempted by yet
another digression, he resists the temptation with the lament,
I will not argue the matter: Time wastes too fast: every letter I trace
tells me with what rapidity Life follows my pen; the days and hours
of it, more precious, my dear Jenny! than the rubies about thy
neck, are flying over our heads like clouds of a windy day, never to
return more--every-thing presses on-whilst thou art twisting that
lock,-see! it grows grey; and every time I kiss thy hand to bid
adieu, and every absence which follows it, are preludes to that
eternal separation which we are shortly to make. (IX.viii.61O-11)

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

The disclosure of his mortality as the present boundary of his life

brings clarity to his situation. So long as life is regarded as an
indefmite sequence-and death as a future fact does nothing to alter that
perspective-one remains absorbed in the existential affairs of the
everyday. But whenever the definitive and individuating possibility of
ceasing is incorporated in Tristram's present grasp of his life as a
unitary pattern, a sea-change occurs which makes his existence his own.
Dying is not the disruption of a linear plot; it is the enabling
precondition of any plot whatever, the condition of natality, the
transcendental alien come home, though home is where he has never
been before. Thus at the end of volume VII when Tristram meets the
amiable Nannette who sings "Viva la Joil Fidon [fi-donc] la Tristessa,"
he wonders, "-Why could I not live and end my days thus? Just
disposer of our joys and sorrows ... why could not a man sit down in the
lap of content here-and dance, and sing, and say his prayers, and go to
heaven with this nut brown maid?" (VII.xliii.538). Instead of whiling
away the life that he has in this hedonic digression, the reflection on
mortality summons him back to his writing which, more than an
ordinary hobby-horse now, is the essential act of saying and being who
he is.
From the point at which we have now arrived it should be clear
that the hobby-horsical character of the mind, along with the themes of
time and death are among the preferred topics in the house of Shandy.
As mere existential concerns their prominence in the novel is largely
arbitrary. The principle of their selection is disclosed only within the
horizon of Tristram's reflexive gaze, where they may be understood as
the definitve continuities of the self that is presumed to lie at the center
of the novel. The strategy of reflexivity, however, begins by assuming
"one who" engages in reflection, and it ends by returning us to where it
began, to the entity who thinks, who has time, and who faces death. The
task that remains is to question the ontological thesis itself, to ask about
this assumption of an originating subject. Is there a controling subject
at the center of the novel? Is the subject an ideality that is effectively
displaced in the plot so that its discourse becomes a totally free play?
Or is the subject a provisional configuration in the play of discourse, a
being-in-between being and nonbeing, presence and absence, identity
and difference? It would surely be extraordinary, and extraordinarily
apt, if sUbjectivity-Ortega calls it "the root of modernity"-should be
dismantled in the very literary genre that takes that subjectivity as its
own condition of being. Where the hobby-horse is interpreted as the
principle of organization of the field of consciousness, one must ask
what kind of principle it is. As a figure of freedom, a metaphor of

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

desire, it implies less a defining limit, as in the comic projects of the

older Shandys, than a metonym for play, though not, as we shall see, a
totally free play. Tristram's hobby-horse is not a limiting rule or
original principle and thus not the key to himself as a present being
after all. He is more a mystery to himself than ever, and, as we shall
now see, the question "who is Tristram Shandy?" is transformed by the
problematic of language.

In response to the accusation of destroying the subject, Jacques
Derrida made the now famous reply, "I don't destroy the subject: I
situate it. That is to say, I believe that at a certain level both of
experience and of philosophical and scientific discourse one cannot get
along without the notion of subject. It is a question of knowing where it
comes from and how it functions."s It would be misleading to make
these remarks the occsaion for reintroducing man as a metaphysical
essence or an epistemological zero point, and it would be equally
misleading to conclude one's reading of Tristram Shandy with the
reaffirmation of an egological center, as though a stable identity were
disclosed. After nine volumes of unified reflections on himself and his
family,9 it must be admitted that the narrator remains an exception to
the notion of character that dominates the genre. The others are
decidedly there in their absence; but the positive features of Tristram's
physiognomy, his substantial presence, is withheld in the search for his
familial, temporal, and linguistic boundaries.
When he asks about his point of origin, the question arises
because there seems to be some thing to ask about, whether real or
hypothetical, one or many, present or absent; but when the protagonist
finds no point of beginning, the boundary between self and other begins
to fade, the ideal entity for which he was searching becomes doubtful,
and he becomes more problematic to himself than ever. Where he
seeks a "nature," be discovers a "culture." Where he looks for himself,
he finds a community of individual progenitors, but only by virtue of
an ossification of identities that come of their being dead, complete,
closed to further becoming. There is some kind of intervention in the
events of the family, of course; something refracts their light
sufficiently to encourage the reification about which Tristram asks,
although the point of refraction shows up only in what it transmits and
by a certain resistance or distortion in the transmission. As such a point
he is clearly not a solitary venture. He is that family, beset by a general
decline of creative force since its period of "long and jolly noses" in the

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

reign of "Harry the VIIIth" when it was "lifted up into some of the best
vacancies in the kingdom." Reflection (re)places the thesis of
individualism with(in) a communalism, a vision of identity as a
dynamic configuration within the changing life of a particular
community. Thus the individual is constituted historically and, until
death, provisionally in an essential interplay with others in a world. If
the presumed boundaries of the ego faded wholly into the family and its
surrounding neighborhood of influence, we might be tempted to settle
for an account of subjectivity as inherently intersubjective. The thesis
of intersubjectivity is a helpful clarification, but even it remains
insufficient because the lexicon of selves is too cumbersome and
paradoxical to describe the phenomenon in question. One may speak of
a reversibility or chiasmus of selves always and already infiltrated by
others, and yet the character of the exchange remains unquestioned.
The thesis of intersubjectivity leaves us still with explanations in
terms of ideal entities rather than helping us see what is given. It
"thinks away from" and obscures the phenomenon in question.
Tristram has discovered his being-with others in a sense so peculiar that
we cannot even say with the' Heidegger of Being and Time that "the
'who' is what maintains itself as something identical throughout
changes in its Experiences and ways of behaviour, and which relates
itself to this changing multiplicity in so doing."IO The reflections
cannot be contained in the language of identity, much less in a closed,
egological structure of thought. It is more economical and closer to
what is actually given to speak instead of the interplay of discourses
where "discourses" is understood as including all the diverse verbal and
nonberbal means in which the play of meaning occurs. As Walter
Shandy says at one point, "I maintain ... that a man of sense does not lay
down his hat in coming into a room, -or take it up in going out of it,
but something escapes, which discovers him" (VI. v. 414-15). The
ontology of the book is profoundly apt, for it is only dialogues in the
parlor reenacted years later in memory, imagination, hearsay.
Language is an important issue to the Shandys. If musing on the
difficulties of words and the activity of writing one's life is exactly
equal to playing toy soldiers on the bowling green, or to any other
"sporting little filly-folly" that one might choose to ride, it gains, like
the other themes we have discussed, a certain, if insufficient, priority
when considered reflexively. The phenomenon of language is the locus
of a crucial weakness in the family and of a determining strength in
Tristram. Characters rarely understand each other, ostensibly because
of the weaknesses of words. A conventional Lockean analysis prevails,
the view namely, that words are signs of ideas which stand in tum for

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

things. When understanding fails-when Walter makes metaphorical

use of the word "seige" one evening in a disquisition upon truth and
Toby's hobby-horse runs away with him to the battle of Namur-the
only recourse is to "give the world a good definition and stand to it."
Words, in short, are the unreliable tools of speaking subjects.
Tristram comes to realize that words cannot be separated from
speakers or from the life that speaks in and through them. What he
hears is words undergoing perpetual transformation, as when at the
court of Navarre the perfectly respectable word "Whiskers" gradually
shifts its meaning until it becomes "absolutely unfit for use." The
semantic slippage is the weakness of the word only on the presumption
of the sign theory, but if the dialectical origin of discourse is not
overlooked, the imprecision of the word reveals its power to follow
and embody the shifting context within which whiskers are brought to
"the beginnings of concupiscence." When Eugenius finds a fatal
ambiguity in "the word Crevice, in the fifty-second page of the second
volume of this book of books," Tristram says, "to define-is to
distrust." Words reveal something more than the intentions of isolated
speakers, more even than the dialogic life of a community of speakers;
discourse, always already underway, a full language in which
everything so far has been said is the locus within which everything
phenomenally given turns up. In the nonchronological beginning,
wherever one begins, there is already the word, word as creative eyent,
without any question of an initiating agency. In the order of
description (as opposed to the order of explanation with its perfectly
legitimate, but very different, questions of origin) word has priority,
and the speaker is a function of, or an event within, the discourse of
meaning rather than the reverse.
If Tristram begins by trying to account for himself, the effort is
a continuation of the discourse of his father, for he has come to
self-consciousness under the linguistic specifications of Walter's
Unhappy Tristram! child of wrath: child of decrepitude!
interruption! mistake! and discontent! What one misfortune or
disaster in the book of embryotic evils, that could unmechanize thy
frame, or entangle thy fllaments! which has not fallen upon thy
head, or ever thou carnest into the world-what evils in thy passage
into itt-What evils since!-produced into being, in the decline of
thy father's days-when the powers of his imagination and of his
body were waxing feeble ... and nothing left to found thy stamina
in, but negations ... The few animal spirits... with which memory,
fancy and quick parts should have been convey'd, -were all
dispersed, confused, confolJnded, scattered, and sent to the

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

devil.- (IV. xix. 296-97)

Repeating the discourse of the fathers is the mimesis by which Tristram

comes to self-consciousness; the dawn of his awareness occurs within
these categories of atomistic individualism and mechanistic physiology.
Small wonder that he is a problem to himself, an ego destined to resist
an evil destiny by trying to understand itself at the center of the stream
of meaningless experience. However, as we have seen, reflection
discloses no kernel of inviolable selfhood at the center of the process.
What he finds is conversation underway and a need to gain clarity by
applying words to his situation.
It might be tempting to argue that what belongs to Tristram's
identity is the difference or the alteration he makes in the ancestral
discourse, especially the ironic and comic tum he gives it. Because he
is the family, he cannot outlive them, as the son can never entirely
outlive the father; but the words he puts in their mouths as he plays
their parts are altered even when, especially when, they are repeated
verbatim. The reinactment of the now silent dialogues of Shandy Hall
consists in making present an absence; it is a retrieval and a translation
that is the act of understanding. As he understands his father, he does so
from a perspective at some distance from the Lamentations and the
conditions of its (presumed) original utterance. But even if the
difference could be determined with precision-we have only the
reinactment, of course-it would only point once more to the speaking
subject, not describe the dynamics of the difference. When we
recognize that Tristram is constituted moment by moment in the act of
speaking, when we surrender the explanatory ideality and philosophical
dream of an underlying identity, the challenge becomes to describe him
as the event of speaking which transforms the Shandys in the presence
of an absent audience. As the locus at which meanings are disclosed, he
extends through and acquires a distinct character in the practical act of
writing his life to others.
One of the oddest features of the novel is the inclusion of its
audience within itself: Your Honor and Madame reader are warned
against the "visious taste ... of reading straight forwards, more in quest
of adventures" than wisdom; we are invited to paint Widow Wadman's
picture on a blank page "as like your mistress as you can-as unlike
your wife as your conscience will let you." "Our Worships" are not
simply spoken to; we are engaged in a multiplicity of comic ways in the
novel as conversation:

Writing, when properly managed, (as you may be sure I

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

think mine is) is but a different name for conversation: As no one,

who knows what he is about in good company, would venture to
talk all;-so no author, who understands the just boundaries of
decorum and good breeding, would presume to think all: The
truest respect which you can pay to the reader's understanding, is
to halve this matter amicably, and leave him something to imagine,
in his tum, as well as yourself. (II. ii. 108-09)

Everywhere Tristram keeps his eye on the reader, watching the most
varied responses. It would be wholly inadequate to describe his writing
as a progressive embodiment of established meanings, cast like a stone
into the midst of the world. On the contrary, he is inspired and drawn
from in front by an extremely diverse audience to whom he wishes to
render an account. His request, "I Beg the reader will assist me ..." is a
rhetorical device with extensive meaning. His accounting is a
transformation of established meaning into a broader field of
discourse, not simply therefore to another, but to himself as his being
extends through that process and is transformed by it. The absent
audience quite literally helps him understand himself within that new
frame of reference, for his being is the process of that conversation.
Tristram can be distinguished from his audience no more
definitvely than he can from the Shandys. Privileging the speaking
subject under the notion of one person conveying a meaning to another
is completely inadequate for describing this phenomenon. The
important question of priority is whether we are actually presented
with a subject who establishes a relation with a reader by means of
words, or whether, within the play of language, a virtual (rather than
actual) speaker and auditor take shape. If the latter, then the relation is
primary; and the persons are configurations within an open field of
significance. In a rather odd sense subjects are literally figures of
speech. One's awareness of oneself as an identity that extends before
and beyond the reading of this novel does not offer evidence to the
contrary; it merely points to a larger context of discourse within which
one has oneself as a reflexive point. The situation is much like Socrates'
comic effort in the Cratylus to get behind words to being itself, only to
become embroiled in an etymological regress and impossible comic
search for the original word, the word of the gods which says being
truly. Behind every logos is a further logos, and everything
phenomenally present is already "enlanguaged." Both the Shandys and
the audience are present in their absence, the condition of discourse;
and as Tristram is indistinguishable in any absolute sense from either of
those communities, he too is the scene of an interplay of presence and
absence, the event of revealing-concealing, interpreting,

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

understanding. And the paradoxical language shows that our categories

are still insufficient for describing the play of discourse, that we remain
indebted still to the metaphysics of presence.
Tristram's analogy of conversation has a further significance.
A genuine conversation occurs when speakers engage in an exchange in
which each maintains his own perspective alongside an open receptivity
to what the other says. The result is that instead of a binary relation in
which each expresses views to the other that were already formulated,
both are caught up in the play of language that opens perspectives
beyond what either speaker previously occupied. This trinary
description highlights something important that the binary model omits
but that everyone knows from experience, namely, the sense of
overhearing oneself say things one had never thought before. It is
literally true to say that in the spontaneity of conversation we do not
know what we think until we hear what we say. The speaker who, like
the narrrator of this novel, releases himself to the play of whatever
meaning turns up is more medium for the play of discourse than
privileged point of origin. He understands what he says from a
perspective closer to that of an auditor than to the perspective of an
originating subject. Tristram mischievously and abruptly breaks off
his narrative at one point with the remarks: "but this is neither here nor
there-why do I mention it?-Ask my pen,-it governs me,-I govern
not it" (VI. vi. 416). And in the eighth volume he commends his way of
beginning a book as "the best" and "the most religious-for I begin
with writing the first sentence-and trusting Almighty God for the
second" (VIII. ii. 540).
His life is conversation in two senses then: as autobiography, it
is the book in which primarily he has his being, and as a mimesis of
human being it is a life which is a locus for the play of meanings. The
conversation in which the Shandys are explained to the reader is the life
and opinions of the narrator because he is the discourse of that
encounter. Whatever he is inspired by our presence to say about his
family is an interplay of self-explication and constitution, both
repetition and initiation, spoken no less to himself than to us. If we
maintain the language, indigenous to the genre, of an ego in pursuit of
meaning in a serially organized and alien world, the best account we
can give will be one that replaces the concept of identity and its desire
for a central presence with the notion of the self as ecstasis. But even
that does not articulate Tristram's discovery of himself as the
differential movement in between and beyond the between of family
and audience, arche and telos. At this point the novel achieves the
decentering of the subject, effectively dismantles its own ideology, and

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

moves beyond the conception of an agency acting on the world and

informing it with meaning. Here, then, is the all-important step from
will to word.
Instead of continuing Walter's aggressive efforts to dominate
the world by comprehending it in one or another of his hobby-horsical
theories, Tristram comes to see himself as an open, responsive
receptivity to whatever occurs in discourse. His hobby-horse, writing
his life,-mark the renovation of the term-makes no effort to
subordinate reality to his will. The relaxed intentionality of the
decentered, provisional self can surrender the assumptions that its
categories are the measure of reality and that the continuation and
enhancement of its being is the origin of all value. The isolated and
problematic subject inherent to the genre has already suffered the
Copernican revolution of which Nietzsche speaks in the epigraph to this
essay; it is already loosed from the center, wandering in the void with a
piercing sense of its emptiness. As Tristram sits alone memoralizing a
family that will end when shortly he dies without an heir, he struggles
to make some sense of "this night of our obscurity," for "We live in a
world beset on all sides with mysteries and riddles-" (IX. xxii. 625).
In the language of The Joyful Widom, he is straying "as through
infinite nothingness," feeling the cold "breath of empty spaces." But
that condition is where this novel begins, as one might say, not where it
What the Lamentation calls this child of wrath, decrepitude,
interruption, mistake, and discontent is somehow transformed by the
exultant comic spirit of the novel. The conceptual move of decentering
opens the way to a profound humility and acceptance of contingency
that renews the joyful wisdom of the outmoded jester-parson Yorick.
The comedy consists in submission to the process of being that relocates
necessity in the realm of play. The end of centrist thinking, surrender
of the arche, is the birth of a new Shandean freedom "to love that well
which thou must leave ere long." A meaningless world is a burden of
anxiety to the ego that tries to preserve itself unchanged, to insulate
itself from the flux, as though it stood under or beyond the process, but
the passage beyond will and all its onto theological centers is the
Nietzschean release from resentment against time and death in the
Dionysian affirmation of whatever comes, endures, and passes. The
character whose name, through an accidental misnaming, means
"sorrowful" maintains an attentive openness to all the crossaccidents
and misadventures that flesh is heir to. In the process of asking the .
unanswerable question who he is, the name is transformed: "True
Shandeism, think what you will against it ... makes the wheel of life run

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

long and chearfully round" (IV, xxxii. 337-38). Release from the
aggressive selfhood and its vision of the world as its ratio transforms
anxiety into wonder and the exuberance of play. Tristram's comic
writing is his life, and it is life as a joyful celebration of being with all
its dangers. Thus he plans "to go on leisurely, writing and publishing
two volumes of my life every year ... as long as I live" (I. xiv. 37).
Sterne's novel begins in the ostensible desire to elucidate a
solitary subjective center, to understand that complex structure and the
principle of its function, and to overcome its transcendental
homelessness. In Tristram's "solicitude" for the whole, his
philosophical dream of himself as the intelligible center of a turning
world, lies the incongruity between presence and desire. The strategy
of reflexivity gradually exposes the no-thingness, the provisionality of
the ego, and is itself transformed thereby from a quest for totality to a
celebration of the (non)originating word, the letting-be of life as
discourse. As readers of this raucous and bawdy love affair with the
world, we come gradually to see what Derrida means when he says that
in the Nietzschean critique of the tradition of centrist thinking "it was
necessary to begin thinking that there was no center, that the center
could not be thought in the form of a present being, that the center had
no natural site, that it was not a fixed locus but a function, a sort of
nonlocus in which an infinite number of sign-substitutions came into
play."n Here, I believe, is the important clue to why, without regard to
philosophical persuasion, all who have thought carefully and thought
long about his novel have come to appreciate the profound implications
of its happy alliance of comedy and thought and to credit Tristram with
the success of his intention: "I write a careless kind of a civil,
nonsensical, good humoured Shandean book, which will do all your
hearts good-And all yours heads too,-provided you understand it."
(VI. xvii. 436).

James E. Swearingen
Department of English
Marquette University
Milwaukee, Wisconsin

Swearingen Reflexivity and the Decentered Self in Tristram Shandy

1. Laurence Sterne, The Life and Opinions of Tristram Shandy,
Gentleman, ed. James A. Work, Odyssey, 1940, I.xi.26.
Subsequent references will be to the Work edition and will be
given in the text by volume, chapter, and page.

2. The Joyful Wisdom, trans. Thomas Common, Frederick Ungar,

Pub. Co., 1960, 168.

3. For a discussion of the relation of the novel to the rise of

middle-class individualism, see Ian Watt, The Rise of the Novel,
University of California Press, 1962.

4. Georg Luk3.cs, The Theory of the Novel, trans. Anna Bostock,

MIT Press, 1971, p.88.

5. Lukacs, p.54.

6. The following section is a synoptic view with substantial changes

in emphasis of my detailed exegesis of the novel in Reflexivity in
Tristram Shandy, Yale University Press, 1977. The revision
consists in replacing the "totalitarian" language and egological
conclusions of that study with a somewhat different emphasis on

7. A Sentimental Journey Through France and Italy By Mr.

Yorick, ed. Ian Jack, Oxford University Press, 1972, p. 85.

8. The Languages of Criticism and the Sciences of Man, ed.

Richard Macksey and Eugenio Donato, The Johns Hopkins
University Press, 1967, p. 271.

9. On the complex problem of unity see "A Work Whole and

Unfinished," in Reflexivity, pp.252-57.

10. Being and Time, trans. John Macquarrie and Edward Robinson,
Harper and Row, 1962, p.150.

11. "Structure, Sign, and Play in the Discourse of the Human

Sciences," Writing and Difference, trans. Alan Bass, University
of Chicago Press, 1978, p. 280.

Part IV. Bibliography
Peter Suber

A Bibliography of Works on Reflexivity

"I don't have to quote
anybody else to say what I
could say for myself."

Section One: Introduction

"Reflexivity" is the generic name for all kinds and species of
circularity. It includes the self-reference of signs, the self-appplication
of principles and predicates, the self-justification and self-refutation of
propositions and inferences, the self-fulfillment and self-falsification of
predictions, the self-creation and self-destruction of logical and legal
entities, the self-augmentation and self-limitation of powers, circular
reasoning, circular causation, cyclic and spiral recurrence, feedback
systems, mutuality, reciprocity, and organic form. It includes the
fallacious, the vicious, the trivial, and the question begging, but also the
sound, the benign, the useful, and the inescapable. It ranges from the
prosaic to the numinous, from the paradoxical to the self-evident, from
science to religion. It is reality and appearance, native to the processes
of the world and to our knowledge and discourse about them.
Because this diversity is unified by a shared structure, that
structure deserves a more general name than "self-reference," which
has frequently served in the past. Usage in logic and mathematics has
already favored "reflexivity" as the general term, and usage in
philosophy and the sciences is moving in that direction.
This bibliography is the first to attempt to cover the range of
reflexivity. Only by trying to approach completeness have I realized
how limited the final product is. When collecting citations over such a
broad area one must make innumerable decisions at innumerable
borderlines. The result is not only that many borderline citations are
excluded, but that the ideal of completeness is shattered. No topic as
broad and deep as reflexivity has sharp borders; instead those borders
are set for different purposes at different places. My decisions for this
bibliography have largely been made explicit. Hence, I can say with
some precision what the scope of the bibliography is, but for the same
reason I cannot be satisfied that I have captured everyone's idea of
Decisions to exclude citations have tended to be of three kinds.
(1) Sometimes the works on a certain topic shade from a hard core of
Suber A Bibliography of Works on Reflexivity

relevancy to marginal irrelevancy. For example, many works on the

theory of types or the possibility of self-deception are strongly relevant
while others are predominantly essays on irreflexive aspects of those
topics. This kind of shading off occurs with virtually every sub-topic
of reflexivity. (2) Sometimes the works of a certain type are excluded
because their great number outweighs their limited pertinency. Either
all would have to be included, diluting the strength of the bibliography,
or blurring its focus, or arbitrary decisions would have to be made,
distinguishing the less relevant of the little-relevant from the more
relevant of the little-relevant. I found this to be the case with works on
dialectical logic, organic form, systems theory, metaphilosophy, the
(psychological or ontological) "self', recursive function theory, the
physics and astronomy (but not the 'theology') of rotary motion, the
geometry of circles, spheres and related curves, mutuality in the law of
contract, most of the law on renvoi, particular studies of physical
feedback devices, and (most regrettably) much of the literature on
Godel's theorems. (3) Finally, some kinds of citations must be
excluded, even though directly germane, because of their great
abundance and because finding them in sufficient numbers would
unduly delay the final draft or multiply the expenses of the compiler.
This was not a consideration in my collection of citations, but it
necessarily became a consideration in deciding which citations would
be included in this printed version. Under this head I have
unfortunately had to exclude works not in the English language, and
English reviews of such works. I sincerely hope that I have laid a
foundation on which others may build.
This scope was maintained for works published up to 1984. No
attempt has been made to be as complete for more recent works.
Although it is incomplete, the bibliography is large. It is so
large that it would be nearly useless without some form of index. I
have adopted two methods of cross-referencing. First, I have picked 28
rubrics or sub-species of reflexivity that usefully classify most
citations. They attempt to serve both as the major subdivisions of
reflexivity, by theme, and as the broadest categories into which the
citations fell, as it were, actuarially. The rubrics are:

Computer Science
Hermeneutic Circle

Suber A Bibliography of Works on Reflexivity

Petitio Principii
Reflexivity in Art
Reflexivity in Law
Reflexivity in Literature
Reflexivity in Natural Science
Reflexivity in Religion
Reflexivity in Social Science
Russell's Paradox
Self-Fulfilling Prophecies
Theory of Types
Vicious Circle

The last quarter or so of the text of the bibliography (Section

Three) consists of the rubrics in alphabetical order, with the surnames
of authors arrayed under them, also in alphabetical order, indicating
which citations in the main list (Section Two) are pertinent to that
sub-topic. For some rubrics I added a word or phrase to the authors'
names to indicate the specific angle of their attack on the sub-topic. In
this way, one searching for works on the paradox of the liar looks at the
names under the rubric "Liar". A more particular interest in
discussions of Ushenko's "new" version of the liar paradox takes one to
the parenthetical phrases following the authors' names.
I must emphasize that citations are provided with specifying
phrases in this way, and listed under rubrics, often on account of
connections or contents evident from their titles alone. This reason is
simply that I cannot pretend to have read all the works in the
bibliography. Hence, many citations may be devoted to Ushenko's liar,
for example, and not be so indicated in the cross-reference lists.
However, in many cases I have annotated works that do not seem, from
their titles alone, to concern reflexivity at all.
For economy, works on the liar paradox, for example, are listed
only under the rubric "Liar" and not also under the rubric "Paradox"
unless there is an independent reason for the two listings. Only four
reflexive paradoxes have rubrics to themselves: the Epimenides or liar,
Grelling's or the heterological paradox, Russell's or the set paradox,
Suber A Bibliography of Works on Reflexivity

and the paradox of omnipotence. Works on other reflexive paradoxes

may be found by scanning for appropriate qualifying phrases in the
"Paradox" rubric.
The second method of cross-referencing occurs within the main
alphabetical list (Section Two) itself. Works about particular thinkers
are listed under those thinkers. For example, Gregory Vlastos's essay
on self-predication in Plato is listed under "Plato". Under "Vlastos" the
reader will find a cross-reference to Plato.
In the main list, an author's own works are listed first,
alphabetically by title, not chronologically. Any works devoted to that
author are collected at the end of his or her space in the alphabetical
sequence. When a work is on two or more thinkers, it is listed under
the author's. own name; under each of the subject thinkers there is a
cross-reference back to the full citation. Works reviewed by others in
the list are usually so noted in the citation itself. Co-authored works are
listed in full only once, under the lead author, but a cross-reference
back to the full-citation is listed under each co-author's name.
Finally, the reader should note that I have used strict
alphabetization. Spaces and recapitalization within surnames are
ignored. "McAardvark" would come after "Macaardvark", not before
as some conventions have it. "De Aardvark", "Van Aardvark", and
"Van der Aardvark" would be listed as if spelled "Deaardvark",
"V anaard vark", and "Vanderaardvark".
I wish to acknowledge my reliance upon two excellent
bibliographies already published: that on the paradox of the liar in
Robert L. Martin's THE PARADOX OF THE LIAR (q.v.), and that on
self-fulfilling prophecies and their kin in Richard L. Henshel's
"Self-Altering Predictions" (q.v.). I also wish to thank two
hard-working student assistants, Bill Stark and Kevin Lesher, for their
accurate typing of illegible 3" x 5" cards onto disk, and two others,
Marilyn Piety and Eric Larson, for their labor in completing
incomplete citations and supplying omissions.
I apologize to authors and readers for omissions that remain,
either by a 'borderline decision' or oversight. There are sure to be as
many in the second category as in the first. Bibliographies of this size
become boundless undertakings, and after one's deliberate decisions of
scope have been made with a wincing confidence, one must face the
decisions made by the running out of time and funds, and accept them.
While the bibliography still grows at a scholarly pace needed for
completeness and accuracy, this printed version is a snapshot of its
Additions and corrections to the bibliography will be received
with gratitude. I may be reached at the Department of Philosophy,
Earlham College, Richmond, Indiana, 47374.
Section Two: Alphabetical List of Citations

Chapter 3: "The Circuitous Journey: Pilgrims and Prodigals;"
Chapter 4: "The Circuitous Journey: Through Alienation to
Reintegration;" Chapter 5: "The Circuitous Journey: From
Blake to D.H. Lawrence." W.W. Norton and Co., 1971.
Achinstein, Peter, "Circularity and Induction," ANALYSIS, 23 (1963)
Achinstein, Peter, "The Circularity of a Self-Supporting Inductive
Argument," ANALYSIS, 22 (1962) 138-41.
[Achinstein], Black, Max, "Self-Support and Circularity: A Reply to
Mr. Achinstein's 'The Circularity of a Self-Supporting
Inductive Argument," ANALYSIS, 23 (1962) 43-44.
Agassi, J., "Variations on the Liar's Paradox," STUDIA LOGICA, 15
(1964) 237-38.
Alexander, Peter, "Pragmatic Paradoxes," MIND, 59 (Oct. 1950)
536-38. (Based on essays by Cohen and O'Connor.)
Allaire, Edwin B. [See work listed under Descartes]
Allison, Henry E., "Bishop Berkeley's Petitio," THE PERSONALIST,
54 (1973) 232-45.
Alston, William, "Self-Warrant: A Neglected Form of Privileged
(1976) 257-72.
Alston, William, "Some Remarks on Chisholm's Epistemology,"
NOUS, 14, 4 (Nov. 1980) 565-86. (On 'self-presenting
SELF-CONSCIOUS GENRE. University of California Press,
Altizer, Thomas. THE SELF-EMBODIMENT OF GOD. Harper and
Row, 1977.
Altizer, Thomas. [See work listed under Nietzsche]
[Anaxagoras], See: Teloh, H.
Anderson, Alan Ross. Review of Thomson's "On Some Paradoxes"
(q.v.), JOURNAL OF SYMBOLIC LOGIC, 29 (1964) 139-40.
Anderson, Alan Ross, "St. Paul's Epistle to Titus," in R. L. Martin
(ed.), THE PARADOX OF THE LIAR (q.v.).
Suber A Bibliography of Works on Reflexivity

Anderson, C. Anthony, "The Paradox of the Knower," JOURNAL OF

PHILOSOPHY, 80,6 (June 1983) 338-55.
Anon., "Circuity of Liens -A Proposed Solution," COLUMBIA
LAWREVIEW, 38 (1938) 1267-78.
Anon., "Circuity of Priority and Liens Under Section 67 (c) (1) of the
Bankruptcy Act," YALE LAW JOURNAL, 66 (1957) 784.
Anon., "Logical Paradoxes, Puzzles, and Problems, " DELAWARE
VALLEY ANNOUNCER, 35 (1962) 30-31.
Anon., [Note], VIRGINIA LAW REVIEW, 15 (1928) 90. (On circular
Anon., [Note], HARVARD LAW REVIEW, 67 (1953) 358. (On
circular liens.)
Anon., "Priorities Between Mortgages and Mechanics Liens," YALE
LAW JOURNAL, 36 (1926) 129. (On circular liens.)
Anon., "Res Judicata and Jurisdiction: The Bootstrap Doctrine,"
HARVARD LAW REVIEW, 53 (1940) 652-660.
Anon., "The Three-Cornered Priorities Puzzle," VIRGINIA LA W
REVIEW, 8 (1922) 550. (On circular priorities.)
Anscombe, G.E.M., "A Reply to Mr. C.S. Lewis's Argument That
'Naturalism' Is Self-Refuting," in her COLLECTED
THE PHILOSOPHY OF MIND, University of Minnesota Press,
[Anselm], Davis, Stephen T., "Anselm and Question-Begging: A Reply
to William Rowe's 'Comments on Professor Davis' "Does The
Ontological Argument Beg the Question?'" INTERNATIONAL
[Anselm], Davis, Stephen T., "Does The Ontological Argument Beg
PHILOSOPHY OF RELIGION, 7 (1976) 433-42. (See review
by Rowe, W.L., listed under Anselm.)
[Anselm], Devine, Phillip E., "Does St. Anselm Beg the Question?"
PHILOSOPHY, 50, 193 (July 1975) 271-81.
[Anselm], Rowe, William L., "Comments on Professor Davis' 'Does
The Ontological Argument Beg the Question?'"
RELIGION,7 (1976) 443-47.
Apolloni, David. [See work listed under Plato]
[Aquinas], Donceal, Joseph F., "Transcendental Thomism," THE
MONIST, 58 (Jan. 1974) 67-85.
Aqvist, Lennart, "How to Handle the Liar Paradox in Modal Logic
With Sentential Quantifiers and its Own Truth Predicate,"

Suber A Bibliography of Works on Reflexivity


Aqvist, Lennart, "The Protagoras Case: An Exercise in Elementary
Logic for Lawyers," (pp. 211-24) in Wlodzimierz Rabinowicz
(ed.), TANKAR OCH TANKEFEL. Uppsala, 1981.
Archibald, Peter, "Self-Fulfilling Prophecy," PSYCHOLOGICAL
BULLETIN, 81 (1973) 74-84.
[Aristotle], Greene, Murray, "Aristotle's Circular Movement As A
Logos Doctrine," REVIEW OF METAPHYSICS, 19 (1965)
[Aristotle], See: O'Brien, J.F.; Woods, J. and Walton, D.
Armstrong, D.M., "Self-Fulfilling Beliefs," (pp. 180-83), Chapter 12,
Cambridge University Press, 1973.
California Press, 1982.
[Arrabal], DeLong-Tonelli, Beverly J., "Bicycles and Balloons in
Arrabal's Dramatic Structure," MODERN DRAMA, 14 (1971)
Ashby, W.R., "Principles of the Self-Organizing Dynamic System,"
Ashby, W.R., "Principles of the Self-Organizing System," in
CONFERENCE (May 9-11,1961), pp. 255-78.
Ashby, W.R., and Riguet, J., "The Avoidance of Over-Writing in
Self-Organizing Systems," JOURNAL OF THEORETICAL
BIOLOGY, 1 (1961) 431-39.
Ashworth, EJ., "Inconsistency and Paradox in Medieval Disputations,"
paper delivered orally at a convention of the American
Philosophical Association, April 27, 1984, Cincinnati.
Ashworth, E.J. Review of R.L. Martin (ed.), THE PARADOX OF
Ashworth, EJ., "Thomas Bricot (d. 1516) and the Liar Paradox,"
1977) 267-80.
Ashworth, E.J., "Will Socrates Cross the Bridge? A Problem in
Medieval Logic," FRANCISCAN STUDIES, 36 (1976) 75-84.
Attig, Thomas. [See work listed under Descartes]


Suber A Bibliography of Works on Reflexivity


Electrical Engineering Research Laboratory, Engineering
Experiment Station, University of Illinois, Urbana, 1960.
[Bach, Johann Sebastian], See: Hofstadter, D.R.
Dissertation, University of Texas at Austin, 1979.
Bahm, Archie, "Systems Theory: Hocus Pocus or Holistic Science?"
GENERAL SYSTEMS, 14 (1969) 175-77.
University of Southern California, 1979.
North Carolina Press, 1986.
Baker, Thomas A. [See work listed under Nietzsche.]
Baldwin, Harold Westcott. THE LOGIC OF REFLEXIVE
REFUTATIONS. Ph.D. Dissertation, University of Colorado,
Ball, Terence, "Dangerous Knowledge? The Self-Subversion of Social
Deviance Theory," INQUIRY, 23, 4 (Dec. 1980) 377-95.
Ballew, Lynne, "Straight and Circular in Parmenides and the
'Timaeus'," PHRONESIS, 19 (1974) 189-209.
Barbo, Francesca Rivetti. [See work listed under GOde1.]
Bar-Hillel, Y., "Discussion --The Revival of 'The Liar',"
(1947) 245-53.
Bar-Hillel, Y., "Do Natural Languages Contain Paradoxes?"
STUDIUM GENERALE, 19 (1966) 391-97. (See Cargile, J.)
Bar-Hillel, Y., "More on Sentences, Statements, the Cogito, and the
Liar," PHILOSOPHICAL STUDIES, 19 (1968) 55-57.
Bar-Hillel, Y., "New Light on the Liar," ANALYSIS, 18 (1957) 1-6.
(See Cargile, J.)
Bar-Hillel, Y., "The Present State of the Problem of the Antinomies:
The Semantical Antinomies," TARBITS, 12 (1940-41) 275-86.
Barker, John, "The Fallacy of Begging the Question," DIALOGUE, 15
(1976) 241-55. (See reply by Sanford, D.)
[Barthelme, Donald], See: McCaffery, L.
[Barthes, Roland], See: Jay, P.
Bartlett, Steven J., "Cognitive Skills in Philosophy: A Teacher's
Guide," AITIA, 6, 3 (1978-79) 12-21.

Suber A Bibliography of Works on Reflexivity


University, Studies in Theory and Behavior, 1983.
Bartlett, Steven J., "The Idea of a Metalogic of Reference,"
METHODOLOGY AND SCIENCE, 9, 3 (1976) 85-92.
Bartlett, Steven J., "A Metatheoretical Basis for Interpretations of
Problem-Solving Behavior," METHODOLOGY AND
SCIENCE, 11, 2 (1978) 58-85.
Bartlett, Steven J., "Narcissism and Philosophy," METHODOLOGY
AND SCIENCE, 19, 1 (1986) 16-26.
Bartlett, Steven J., "Phenomenology of the Implicit," DIALECTICA,
29,2-3 (1975) 173-88.
Bartlett, Steven J., "Philosophy as Conceptual Therapy," RESOURCES
IN EDUCATION, May 1983, Identification No. ED224402.
Bartlett, Steven J., "Philosophy as Ideology," METAPHILOSOPHY,
17, 1 (Jan. 1986) 1-13.
Bartlett, Steven J., "Referential Consistency As A Criterion of
Meaning," SYNTHESE, 52 (1982) 267-82.
Bartlett, Steven J., "Reflexive Prediction in Behavioral Science,"
Learning Cycle Series, ADAPT, Behlen Laboratories,
University of Nebraska-Lincoln, 1977.
Universite de Paris, 1970.
Bartlett, Steven J. Review of Boyle, Grisez, and Tollefsen, FREE
1979) 738-40.
Bartlett, Steven J., "Self-Reference, Phenomenology, and Philosophy
of Science," METHODOLOGY AND SCIENCE, 13, 3(1980)
Bartlett, Steven J., and Peter Suber (eds.). SELF-REFERENCE:
Bartlett, Steven J., "Towards a Unified Concept of Reality," ETC.: A
REVIEW OF GENERAL SEMANTICS, 32, 1 (1975) 43-49.
Bartlett, Steven J., "The Use of Protocol Analysis in Philosophy,"
METAPHILOSOPHY, 9, 3-4 (1978) 324-336.
Bartlett, Steven J., "Varieties of Self-Reference," in Steven J. Bartlett
and Peter Suber (eds.), SELF-REFERENCE (q.v.).

Suber A Bibliography of Works on Reflexivity

Bateson, Gregory, "A Theory of Play and Fantasy," PSYCHIATRIC

RESEARCH REPORTS, 2 (1955) 39-51. (On self-referential
character of play, compared with liar paradox.) Reprinted in
J.S. Bruner, et al. (eds.), PLAY: ITS ROLE IN
Batten, Lynn M. [Co-authored work listed under Walton, D.N.]
Bausch, A. Review of Pap's "The Linguistic Hierarchy ... " (q.v.),
JOURNAL OF SYMBOLIC LOGIC, 22 (1957) 392-93.
Baylis, C. Review of Gregory's "Heterological and Homological"
(q.v.), JOURNAL OF SYMBOLIC LOGIC, 17 (1952) 220.
Beach, Edward, "The Paradox of Cognitive Relativism: A Reply to
Jack Meiland," METAPHILOSOPHY, 15, 1 (Jan. 1984) 1-15;
typesetting errors corrected in 15, 3-4 (July-Oct. 1984)
Beard, Robert W., "Semantic Theory and the Paradox of the
Non-Communicator," PHILOSOPHICAL STUDIES, 17 (1966)
OF THE "SUDARSANA-CAKRA". New York University
Press, 1973.
Behmann, H., "The Paradoxes of Logic," MIND, 46 (1937) 218-21.
(See review by Langford, C.H.)
Benacerraf, Paul. [See work listed under GOdel]
Bendiner, R., "Quickie 'Results' Could Sway the Election," LIFE, 18
(Sept. 1964) 125-33.
Bennet, J. Review of Mackie's "Self-Refutation..." (q.v.), JOURNAL
OF SYMBOLIC LOGIC, 30 (1965) 365-66.
Bennet, J. Review of Encarnacion's "On Ushenko's Version ..." (q.v.),
Ushenko's "A Note on the Liar... " (q.v.), Toms's "The Liar
Paradox," (q.v.), Donnellan's "A Note on the Liar... " (q.v.),
Ushenko's "An Addendum to the Note ..." (q.v.), Toms's "Reply
to a Note ..." (q.v.), Rozeboom's "Is Epimenides Still Lying?"
(q.v.), Huggett's "Paradox Lost," (q.v.), Whitely'S "Let
Epimenides Lie!" (q.v.), and Sibajiban's "Mr. Eric Toms on the
Liar," (q.v.), JOURNAL OF SYMBOLIC LOGIC, 32 (1962)
Ben&on, "Circuity of Lien -A Problem in Priorities," MINNESOTA
LAW REVIEW, 19 (1935) 139.
EXPERIENCE. San Fransisco: W.H. Freeman and Co., 1977.
[Berkeley, Bishop], See: Allison, H.E.
Berleant, Arnold. [See work listed under Descartes]
Bertalanffy, Ludwig von. GENERAL SYSTEM THEORY:

Suber A Bibliography of Works on Reflexivity


George Braziller, rev. ed., 1968. (First ed. 1951.)
Bertoldi, Eugene F., "Phenomenology of Phenomenology,"
Beth, E.W., "Some Remarks on Dr. Perelman's Essay on Logical
Antinomies," MIND, 45 (1936) 487-88.
Bhattacharyya, S., "Symmetry, Transitivity, and Reflexivity,"
ANALYSIS, 19 (March 1959) 93-96.
Bhattacharyya, S., "Symmetry, Transitivity, and Reflexivity,"
(1960) 77-82.
Bhattacharyya, S., "Symmetry, Transitivity, and Reflexivity:
ASSOCIATION 7 (1960) 83-84.
Biro, J.I., "Rescuing 'Begging the Question'," METAPHILOSOPHY, 8
(Oct. 1977) 257-71.
Black, Max. [See work listed under Achinstein, Peter]
[Blanshard, Brand], Garelick, Herbert M. "Blanshard and the Law of
Contradiction," IDEALISTIC STUDIES, 4 (January 1974)
. 50-63. (On circularity and non-contradiction.)
Blauberg, LV., V.N. Sadovsky, and E.G. Yudin, "Paradoxes of
Systems Thinking," Chapter 13 of their SYSTEMS THEORY:
Moscow: Progress Publishers, 1977.
THINGS GO WRONG. Three volumes. Price/Stem/Sloan,
IDEAS, 1.621-27. Ed. Philip Wiener. N.Y.: Charles Scribner's
Sons, 1973.
Bochenski, I.M., "Formalization of a Scholastic Solution of the
Paradox of the Liar," in A. Menne (ed.),
Bonifacio, Armando F., "On Capacity Limiting Statements," MIND, 74
(1965) 87-88.
Boolos, George, "Provability, Truth, and Modal Logic," JOURNAL
OF PHILOSOPHIC LOGIC, 9 (February 1980) 1-7.
Borges, Jorge Luis, "The Fearful Sphere of Pascal," in his
WRITINGS. Donald A. Yates and James E. Irby (eds.). New
Directions Publishing Co., 1964.

Suber A Bibliography of Works on Reflexivity

[Borges Jorge Luis], See: Murillo, L.A.

THE WEST. Columbia University Press, 1981. (Defends
"cyclic linearism" as a theory of history.)
Boudreau, H.L. [See work listed under Valle-Inclan]
Bowden, L., "Heterologicality," ANALYSIS, 12 (1952), 77-81.
Bowerman, William R., "The Structure, Process and Function of
Self-Referent Causal Attributions," JOURNAL FOR THE
THEORY OF SOCIAL BEHAVIOR, 8 (March 1978) 45-75.
Boyle, Joseph M., Jr. THE ARGUMENT FROM
DISCUSSION. Ph.D. Dissertation, Georgetown University,
Boyle, Joseph M. Jr., G. Grisez, and O. Tollefsen, "Determinism,
Freedom, and Self-Referential Arguments," REVIEW OF
METAPHYSICS, 26, 1, (Sept. 1972) 3-37.
Boyle, Joseph M. Jr., G. Grisez, and O. Tollefsen. FREE CHOICE: A
Dame Press, 1976.
Boyle, Joseph M. Jr., "Is Determinism Self-Refuting?" in Steven J.
Bartlett and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Boyle, Joseph M. Jr., "Self-Referential Inconsistency, Inevitable Falsity
and Metaphysical Argumentation," METAPHILOSOPHY, 3,1,
(Jan. 1972) 25-42.
[Boyle, Grisez, Tollefsen], See: Bartlett, SJ.; Konyndyk, K.
Bradley, C.K., "The Paradoxes," ASTOUNDING SCIENCE
FICTION, 53, (1954) 93-100.
[Bradley, F.H.], See: Rorty, A.D.
Brady, Ross T., "Reply to Priest on Berry's Paradox," THE
PHILOSOPHICAL QUARTERLY, 34,135 (April, 1984).
Braines, S.N., Napalkov, A.V., and Shreider, LA., "Analysis of the
Working Principles of Some Self-Adjusting Systems in
Engineering and Biology," in PROCEEDINGS OF THE
PROCESSING (ICIP), Paris. UNESCO House, 1959.
Braines, S.N., and Napalkov, A.V. CERTAIN PROBLEMS IN THE
Publications Research Service Report 2177-N, U.S. Department
of Commerce, 1960.
Brecht, George. [Co-authored work listed under Hughes, P.]

Suber A Bibliogmphy of Works on Reflexivity

PHILOSOPHY. E.J. Brill, and Humanities Press, 1977.

[Bricot, Thomas], See: Ashworth, E.J.
Briskman, Larry B., "Historicist Relativism and Bootstrap
Rationality," MIND, 60,4 (Oct. 1977) 509-39.
Brody, Michael, "On Circular Readings," in N.V. Smith (ed.)
MUTUAL KNOWLEDGE, pp. 133-46. Academic Press, 1982.
With comments by K. Brown and R. Kempson and reply by
Brody, pp. 148-79.
Brown, D.B., "Paradox Without Tiers," ANALYSIS, 17 (1957)
Brown, James M. [See work listed under Whitehead, A.N. and Russell,
Cornell University Press, 1979. (On the theme of circularity of
German Romanticism.)
Broyles, J.E. [See work listed under GOde!.]
Brush, Stephen G. [See work listed under Nietzsche.]
Bryant, John, "The Paradox of the Lawyers," MIDWEST JOURNAL
OF PHILOSOPHY, 4 (1976) 1-2. (On Protagoras and
Buck, Roger C. "Reflexive Predictions," PHILOSOPHY OF
SCIENCE, 30 (1963) 359-374. (See review by Griinbaum, A.)
Buck, Roger C., "Rejoinder to Griinbaum's 'Comments on Professor
Roger Buck's Paper "Reflexive Predictions''','' PHILOSOPHY
OF SCIENCE, 30 (Oct. 1963) 373-74.
[Buck, Roger], Griinbaum, Adolf, "Comments on Professor Roger
Buck's Paper 'Reflexive Predictions'," PHILOSOPHY OF
SCIENCE, 30 (Oct. 1963) 370-72.
Buckner, D., and P. Smith, "Quotation and the Liar Paradox,"
ANALYSIS, 46, 2 (March 1986) 65-67.
PARADOXES. Van Nostrand Reinhold Co., 1982.
Bunge, William, "Goof --A Game to Begin All Games,"
Burge, Tyler, "Demonstrative Constructions, References, and Truth,"
JOURNAL OF PHILOSOPHY, 71, 7 (April 18, 1974). (On
token-reflexive construction and tense logic).
Burge, Tyler, "Epistemic Paradox," JOURNAL OF PHILOSOPHY,
81, 1, (Jan. 1984) 5-29. (The liar paradox applied to thoughts
and beliefs.)
Burge, Tyler, "Reasoning About Reasoning," PHILO SOPHIA:

Suber A Bibliography of Works on Reflexivity

1978) 205-223.
with introduction and commentary by G.E. Hughes. Cambridge
University Press, 1982. (Chapter 8 of the SOPHISMATA, for
which see, e.g., the next citation.)
Buridan, John. SOPHISMS ON MEANING AND TRUTH. Translated
with introduction by T.K. Scott. N.Y. 1966.
[Buridan, John], See: Prior, A.N.
Burkholder, Peter M., "Petitio in the Strife of Systems," TULANE
STUDIES IN PHILOSOPHY, 16 (1967) 19-31.
Burnaby Art Gallery (ed.). MYSTIC CIRCLE. Burnaby, British
Columbia: Burnaby Art Gallery, 1973. (Catalog of an
Burnyeat, M.F., "Protagoras and Self-Refutation in Later Greek
Philosophy," PHILOSOPHICAL REVIEW, 85, 1 (Jan. 1976)
Burnyeat, M.F., "Protagoras and Self-Refutation in Plato's
'Theatetus'," PHILOSOPHICAL REVIEW, 85, 2 (April, 1976)

- C
HISTORY. Philosophical Library, 1962. Reprinted by
Greenwood Press, 1971.
METADRAMA IN HAMLET. Columbia University Press,
[Calleja, Pi], See: Levi, B.
Callois, Roger, "Circular Time, Rectilinear Time," DIOGENES, 42
(1963) 1-13.
Cameron, S. [Co-authored work listed under Yovitts, M.]
Campbell, Joseph, "The Turning Wheel of Terror-Joy," Part I of
MYTHOLOGY,pp. 405-416. Viking Press, 1968.
Campbell, Morton Carlisle, "Protection Against Indirect Attack," in
HARVARD LEGAL ESSAYS, Harvard University Press, 1934,
pp. 3-37. (On circular liens.)
Cargile, James, "On Omnipotence," NOUS, 1 (1967) 201-05.
PREDICATION. Cambridge University Press, 1979.
Cargile, J., Review of Bar-Hillel's "New Light on The Liar" (q.v.), and

Suber A Bibliography of Works on Reflexivity

Bar-Hillel's "Do Natural Languages ..." (q.v.), JOURNAL OF

SYMBOLIC LOGIC, 34 (1969) 645.
Cargile, J., Review of Geach's "Ryle on Namely Riders" (q.v.),
Fitzpatrick's "'Heterological' and Namely-Riders" (q.v.),
Geach's "Namely- Riders Again" (q.v.), JOURNAL OF
SYMBOLIC LOGIC, 32 (1967) 408-409.
[Camap, Rudolf], See: Stegmuller, W.
Carrier, D., "On the Depiction of Figurative Representational Pictures
Within Pictures," LEONARDO, 12 (1979) 197-200.
Caws, Peter, "The Paradox of Induction and the Inductive Wager,"
22 (June 1962) 512-20.
Champlin, T.S., "Double Deception," MIND, 85 (January 1976)
Champlin, T.S., "Self-Deception: A Reflexive Dilemma,"
PHILOSOPHY, 52, 20, (July, 1977) 281-99.
[Chardin, Teilhard de], See: O'Brien, J.F.
PRINCIPLE. Cornell University Press, 1973.
Chihara, Charles S., "The Semantic Paradoxes: A Diagnostic
Investigation," PHILOSOPHICAL REVIEW, 88,4 (Oct. 1979)
Chihara, Charles S., "Truth, Meaning, and Paradox," NOUS, 10,2-3
(May-Sept. 1976) 305-12. (Response to Donald Davidson.)
Chihara, Charles S., "Wittgenstein's Analysis of the Paradoxes in his
Lectures on the Foundation of Mathematics,"
PHILOSOPHICAL REVIEW, 86, 3 (July, 1977) 365-81.
Chihara, Charles S. [See work listed under GOdel]
[Chihara, Charles S.], See: Steiner, M.
Chinn, Ewing Y. [See work listed under Descartes.]
University Press, 1981.
Chisholm, Roderick. THEORY OF KNOWLEDGE. Prentice-Hall,
1966. (On 'self-presenting' states and propositions.)
[Chisholm, Roderick], See: Alston, W.
Chowdhry, K. and Theodore M. Newcomb, "The Relative Abilities of
Leaders and Nonleaders to Estimate Opinions in Their Own
PSYCHOLOGY, 47 (1952) 51-57.
[Christ, Jesus], See: Sell, J.1.
Christensen, N.E., "On an Apparent Circularity in Some Definitions of
Logical Truth," MIND, 66 (1957) 395-97.

Suber A Bibliography of Works on Reflexivity

Church, Alonzo, Review of Bradley's "The Paradoxes" (q.v.),

Church, Alonzo, Review of Geach's "Mr. Ill-named" (q.v.),
Church, Alonzo, Review of Geach's "On Insolubilia" (q.v.),
Church, Alonzo, Review of Koyre's The Liar" (q.v.), JOURNAL OF
SYMBOLIC LOGIC, 11 (1946) 131.
Church, Alonzo, Review of Lawrence's "Heterology and Hierarchy"
(q.v.), JOURNAL OF SYMBOLIC LOGIC, 15 (1950) 216-17.
Church, Alonzo, Review of Weinbergs's "A Possible Solution of the
Heterological Paradox" (q.v.), JOURNAL OF SYMBOLIC
LOGIC, 3 (1938) 46.
Church, Alonzo, "The Richard Paradox," THE AMERICAN
MATHEMATICAL MONTHLY, 41 (1934) 356-61.
Churchland, Patricia Smith, "Is Determinism Self-Refuting?," MIND,
90,357 (Jan. 1981) 99-101. (See review by Smith, 1.W.)
Clark, Malcolm. "Circularity and Criticism," Part II, Chapter VI, 1, of
THE PHILOSOPHY OF HEGEL, pp. 194-97. Martinus Nijhoff,
Clark, W.A. [Co-authored work listed under Farley, B.G.]
Clegg, Jerry S. [See works listed under Nietzsche and Plato]
NORTH AMERICAN INDIAN ART. University of Washington
Press, 1977.
Coder, David. [See work listed under GOdel.]
Cohen, David, "Hindi 'Apnaa': A Problem in Reference Assignment,"
FOUNDATIONS OF LANGUAGE, 10 (September 1973)
Cohen, L.J., "Mr. O'Connor's 'Pragmatic Paradoxes'," MIND, 59
(1950) 85-87.
Cohen, L.J., "Why Do Cretans Have To Say So Much?"
PHILOSOPHICAL STUDIES, 12 (1961) 72-78. (See review by
Kanger, S.)
Cohen, Sybil. [See work listed under Ingarden]
[Coleridge, Samuel Taylor], See: Benzinger, 1.; Fogle, R.H.
TRADITION OF PARADOX. Princeton University Press,
1966. (Esp. Ch. 12: "'I am that I am': Problem of
Self-Reference" .)
Collier, Charles S., "Judicial Bootstraps and the General Welfare

Suber A Bibliography of Works on Reflexivity


Collingwood, R.G. AN ESSAY ON METAPHYSICS, Chapter XII: "A
Psuedo-Science Refutes Itself," pp. 122-32, and Chapter XVI:
"Suicide of Positivistic Metaphysics," pp. 162-71. Chicago:
Henry Regnery Co., 1972.
Conway, Pierre H., "The Barber Paradox," LAVAL THEOLOGIQUE
ET PHILOSOPHIQUE, 18 (1962) 161-76.
[Coover, Robert], See: McCaffery, L.
Copleston, Frederick C., "The History of Philosophy: Relativism and
Recurrence," HEYTHROP JOURNAL, 14 (1973) 123-35.
[Copleston, Frederick], See: Harrison, C.
Routledge & Kegan Paul, 1983.
University of California Press, vol. I, 1976; vol. II, 1977.
Cornforth, Maurice, "The Circularity and the Test of Dialectical
Materialism," Chapter 111.1.5 of his MARXISM AND THE
LINGUISTIC PHILOSOPHY, pp. 782-84. International
Publishers, 1965.
Cosentino, Dante A., "Self-Deception Without Paradox,"
Co., 1966.
Coval, Sam, "Self-Reference for Non-Selves," PHILOSOPHIA, 4
(October 1974) 469-83.
Cowan, J.L., "The Paradox of Omnipotence," ANALYSIS,
Supplement, 25 (1965) 102-08.
Cowan, J.L., "The Paradox of Omnipotence Revisited," CANADIAN
JOURNAL OF PHILOSOPHY, 3 (1974) 435-445.
Craig, William, "Divine Omniscience and Newcomb's Paradox," paper
delivererd orally at a convention of the American Philosophical
Convention, April 27, 1984, Cincinnati.
Humanities Press, 1965.
Crosthwaite, J. [Co-authored work listed under Priest, G.]
Csanyi, Y., and G. Kampis, "Autogenesis: The Evolution of Replicative
(1985) 303-321.
Cumming, Robert Denoon. STARTING POINT: AN

Suber A Bibliography of Works on Reflexivity

University of Chicago Press, 1979.

Cummins, Robert. [See work listed under Descartes]
Curry, B., "The Paradox of Kleene and Rosser," TRANSACTIONS OF
454-516. (See review by Kalmer, L.)

Daemmrich, Horst S. [See work listed under Hoffmann]

Dahlstrom, Daniel O. [See work listed under Hegel]
Dalla Chiara, Maria Luisa, "Logical Self-Reference, Set Theoretical
Paradoxes and the Measurement Problem in Quantum
1977) 331-347.
Daniels, Charles B., "Self-Deception and Interpersonal Deception,"
PERSONALIST, 55 (Summer 1974) 244-52.
Danquah, Joseph. [See work listed under Whitehead, A.N. and Russell,
Danto, Arthur. [See work listed under Nietzsche.]
TROMPE-L'OEIL. Phaidon, 1979.
David, H.A. [Co-authored work listed under John, J.A.]
Davis, Stephen T. [See works listed under Anselm]
Davis, William H., "Is Perspectivism Just Another Perspective?"
JOURNAL OF THOUGHT, 13 (July 1978) 205-10.
Daya, "Symmetry, Transitivity, and Reflexivity," ANALYSIS, 19
(Oct. 1958) 7-11.
Daya, "Symmetry, Transitivity, and Reflexivity," JOURNAL OF THE
Daya, "Symmetry, Transitivity, and Reflexivity: Concluding Note,"
(1960) 83-84.
Daye, Douglas Dunsmore, "Circularity in the Inductive Justification of
Formal Arguments ("Tarko") in the Twelfth-Century Indian
Jaina Logic," PHILOSOPHY EAST AND WEST, 29 (1979)
Daye, Douglas Dunsmore, "Reflexivity and Metalanguage Games in
Buddhist Causality," PHILOSOPHY EAST AND WEST, 25
(1975) 95-101.
De Comulier, BenoIt, "Paradoxical Self-Reference," LINGUISTICS
AND PHILOSOPHY, 2 (1978) 435.
Dees, J. Gregory. [Co-authored work listed under Hart, 1.A.]
De George, Richard, "Reason, Truth, and Context," IDEALISTIC

Suber A Bibliography of Works on Reflexivity

STUDIES, 4 (Jan. 1974) 35-49. (On self-reference.)

De Laguna, T., "On Certain Logical Paradoxes," PHILOSOPHICAL
REVIEW, 25 (1916), 16-27.
[DeMorgan, Augustus], See: Walton, D.
Denhard, Charles H. THE AUTOBIOGRAPHY OF A BOOK. H. Wolff
Book Mfg. Co., 1936.
Dennett, Daniel. [Co-edited work listed under Hofstadter, D.R.]
Denzin, Norman K., "The Self-Fulfilling Prophecy and
Patient-Therapist Interaction." Paper presented at conference of
the American Sociological Association, New York, 1966.
De-Rijk, L.M., "Some Notes on the Medieval Tract 'De Insolubilibus',
With the Edition of a Tl1:lct Dating From the End of the Twelfth
Century," VIVARIUM, 4 (November 1966) 83-115.
Derr, Patrick G., "Reflexivity and the Methodology of Scientific
Research Programmes," THE NEW SCHOLASTICISM, 55,4
(Autumn, 1981) 500-03.
[Descartes], Allaire, Edwin B., "The Circle of Ideas and the
Circularity of the Meditations," DIALOGUE,S (1966) 131-53.
[Descartes], Attig, Thomas, "Descartes and Circularity: The
Precipitous Rush to Defense," MODERN SCHOOLMAN, 54
(1977) 368-78.
[Descartes], Berleant, Arnold, "On The Circularity of the Cogito,"
26,3 (March 1966) 431-433.
[Descartes], Chinn, Ewing Y., "A Journey Around the Cartesian
[Descartes], Cummins, Robert, "Epistemology and the Cartesian
Circle," THEORIA, 41 (1975) 112-24.
[Descartes], Doney, Willis, "The Cartesian Circle," JOURNAL OF
THE HISTORY OF IDEAS, 16 (1955) 324-38.
[Descartes], Dreisbach, Donald F., "Circularity and Consistency in
(1978) 59-78.
[Descartes], Feldman, Fred, "Epistemic Appraisal and the Cartesian
Circle," PHILOSOPHICAL STUDIES, 27 (Jan. 1975) 37-55.
[Descartes], Frankfurt, Harry G., "Memory and the Cartesian Circle,"
PHILOSOPHICAL REVIEW, 71 (Oct. 1962) 504-11. (See
review by Nelson, J.O.)
[Descartes], Frankfurt, Harry G., "A Reply to Mr. Nelson's Comments
on 'Memory and the Cartesian Circle'," DIALOGUE, 4 (1965)
[Descartes], Kelly, Matthew J., "The Cartesian Circle: Descartes'

Suber A Bibliography of Works on Reflexivity

Response to Scepticism," JOURNAL OF THOUGHT, 5,2 (April

1970) 64-71.
[Descartes], Kretzmann, Norman, "On Rose's 'Cartesian Circle',"
26 (1965) 90-92.
[Descartes], Markie, Peter. THE CARTESIAN CIRCLE. Ph.D.
Dissertation, University of Massachusettes at Amherst,
University Microfilms, 1977.
[Descartes], Nelson, John 0., "In Defence of Descartes: Squaring a
Reputed Circle," DIALOGUE, 3 (1964) 262-72. (A reply to
Frankfurt, H.G.)
[Descartes], Odegard, Douglas, "Escaping the Cartesian Circle,"
1984) 167-74.
[Descartes], Rose, Lynn E., "The Cartesian Circle," PHILOSOPHY
(See review by Kretzmann, N.)
[Descartes], Rose, Lynn E., "Reply to Mr. Kretzmann's 'On Rose's
"Cartesian Circle"'," PHILOSOPHY AND
[Descartes], Smith, Michael P. A VIRTUOUS CIRCLE: DESCARTES'
MEDITATIONS. Ph.D. Dissertation, University of North
Carolina at Chapel Hill, 1983.
[Descartes], Stubbs, A.C., "Bernard Williams and the Cartesian
Circle," ANALYSIS, 40 (1980) 103-08.
[Descartes], Tlumak, Jeffrey, "Squaring the Cartesian Circle,"
[Descartes], Van Cleve, James, "Foundationalism, Epistemic
Principles, and the Cartesian Circle," THE PHILOSOPHICAL
REVIEW,88,1 (Jan. 1979) 55-91.
[Descartes], Welch, Robert S. DOUBT, CERTAINTY, AND THE
CARTESIAN CIRCLE. Ph.D. Dissertation, University of
Massachusetts, 1983.
[Descartes], Zirlin, Robert B. THE CARTESIAN CIRCLE. Ph.D
Dissertation, Johns Hopkins University, 1980.
[Descartes], See: Prior, A.N.; Rorty, A.O.; Scheier, C.A.; Sibajiban
Desmonde, William H. [See work listed under GOdel.]
Detlefsen, Michael. [See work listed under GOdel.]
Devine, Phillip E. [See work listed under Anselm]
[Dilthey, Wilhelm], See: Maraldo, J.C.
Dingler, H., "On Definitions That Contain Themselves as
Determinants," METHODOS, 2 (1950-51) 296-98.

Suber A Bibliography of Works on Reflexivity

Dominian, Jack. CYCLES OF AFFIRMATION. London: Darton,

Longman and Todd, 1975.
Donceal, Joseph F. [See work listed under Aquinas]
Doney, Willis. [See work listed under Descartes]
Donnellan, Keith, "Categories, Negation, and the Liar Paradox," in
R.L. Martin (ed.), PARADOX OF THE LIAR (q.v.).
Donnellan, Keith, "A Note on the Liar Paradox," PHILOSOPHICAL
REVIEW, 66 (1957) 394-97. (See review by Bennet, J.)
PARADOX. Ph.D Dissertation, Stanford University, 1980.
Drange, Theodore, "The Paradox of the Non-Communicator,"
Drange, Theodore, "The Paradox Defended," PHILOSOPHICAL
STUDIES, 18 (1967) 1-12. (Paradox of the
non-communicator.) (See review by Dunn, M.J.)
Drange, Theodore, "Paradox Regained," PHILOSOPHICAL
STUDIES, 20 (June 1969) 61-64.
Dreisbach, Donald F. [See work listed under Descartes]
Driscoll, John A. [See work listed under Plato]
Duff-Forbes, D.R. [See work listed under Plato]
SPACES. Amsterdam: Mathematisch Centrum, 1978.
Dumitriu, Anton, "The Logico-Mathematical Antinomies:
Contemporary and Scholastic Solutions," INTERNATIONAL
PHILOSOPHICAL QUARTERLY, 14,3 (Sept. 1974) 309-28.
Dumitriu, Anton. [See work listed under Wittgentstein.]
Dunham, T.C. [See work listed under Grillparzer]
Dunn, MJ., "Drange's Paradox Lost," PHILOSOPHICAL STUDIES,
18 (1967) 94-95.
[Dworkin, Ronald], Farago, John, "Judicial Cybernetics: The Effects
of Self-Reference on Dworkin's Rights Thesis,"VALPARAISO
UNIVERSITY LAW REVIEW, 14 (1980) 371-425.
Dyson, M. [See work listed under Plato.]

Eckel, Michael D., "Bhavaviveka and the Early Madhyamika Theories
of Language," PHILOSOPHY EAST AND WEST, 28 (July
1978) 323-37.
Edidin, Aron, "Bootstrapping Without Boots," in John Earman (ed.),
Press, 1983. (A response to Glymour, C.)
Eigen, Manfred, and Peter Schuster. THE HYPERCYCLE: A

Suber A Bibliography of Works on Reflexivity


Springer-Verlag, 1979.
Elgin, C.Z., "The Impossibility of Saying What is Shown," THE
Eliade, Mircea. THE MYTH OF THE ETERNAL RETURN. Pantheon
Books, 1954.
Eliot, T.D., "Reactions to Predictive Assumptions," AMERICAN
SOCIOLOGICAL REVIEW, 2 (August 1937) 508-14.
Columbia University, 1980
Emerson, Ralph Waldo, "Circles," Essay X of his ESSAYS, FIRST
SERIES. Often reprinted, e.g. EMERSON'S ESSAYS; FIRST
212-28. Thomas Y. Crowell Co., 1926.
[Emerson], Witemeyer, H.H., "Line and Round in Emerson's Uriel,"
ASSOCIATION, 82 (March 1967) 98-103.
[Empedocles], Long, A.A., "Empedocles' Cosmic Cycle in the
'Sixties'," (pp. 397-425) in A.P.D. Mourelatos (ed.), THE
PRE-SOCRATICS, Anchor Books, 1974.
[Empedocles], O'Brien, D. EMPEDOCLES' COSMIC CYCLE.
Cambridge University Press, 1969.
Enc, Berent, "Spiral Dependence Between Theories and Taxonomy,"
INQUIRY, 19 (1976) 41-71.
Encarnacion, J. "On Ushenko's Version of the Liar Paradox," MIND,
64 (1955) 99-100. (See review by Bennet, J.)
Englebretsen, George, F., "The Incompatibility of God's Existence and
Omnipotence," SOPHIA, 10 (1971) 28-31. (See comments by
Londey, D., King-Farlow, J., and Miller, B.)
Epstein, David G., "Circuity of Priority Problems," in his
West Publishing Co., 2d ed. 1980.
[Escher, M.C.], Ernst, Bruno. THE MAGIC MIRROR OF M.C.
ESCHER. Random House, 1976.
[Escher, M.C.], See: Hofstadter, D.R.
[Euathlus], (Note: this name is sometimes spelled Eulathus) See:
Bryant, J.; Lenzen, W.; Goossens, W.K.
[Eudoxus], Maula, E., Kasanen, E., and Mattila, J., "The Spider in the
Sphere, Eudoxus' 'Arachne'," PHILOSOPHIA, 5 (1975-76)

Suber A Bibliography of Works on Reflexivity


Evans, E., "On Some Semantic Illusions," MIND, 63 (1954) 203-18.
(See review by Thomson, J.F.)
Everitt, N., "A Problem for the Eliminative Materialist," MIND, 40
(1981) 428-34. (See review by Smith, J.W.)

Falletta, Nicholas. THE PARADOXICON. Doubleday and Co., 1983.
Farago, John. [See work listed under Dworkin]
Farber, Edward, "Concentric Circles and Chemistry,"
Faris, William G. SELF-ADJOINT OPERATORS. Springer-Verlag.
Farley, B.G., "Self-Organizing Models for Learned Perception," in
Yovitts, M.C. and Cameron, S. (eds.), SELF-ORGANIZING
SYSTEMS, (q.v.), 1960.
Farley, B.G., and Clark, W.A., "Simulation of Self-Organizing System
by a Digital Computer," I.R.E. TRANSATIONS ON
INFORMATION THEORY, 4 (1955) 76-84.
Farrel, P.M., "Evil and Omnipotence" MIND, 67 (1958) 399-403.
Feldman, Fred. [See work listed under Descartes]
Filmer, Paul, "Sociology and Sound Stratification: Issues of Reflexivity
and Tradition," in Barry Sandywell et al. PROBLEMS OF
Routledge and Kegan Paul, 1975.
Fingarette, Herbert, "The Problem of the Self in the Analects,"
PHILOSOPHY EAST AND WEST, 29 (April 1979) 129-40.
Finkelstein, Maurice, "Further Notes on Judicial Self-Limitation,"
HARVARD LAW REVIEW, 39 (1925) 221.
Finkelstein, Maurice, "Judicial Self-Limitation," HARVARD LAW
REVIEW, 37 (1924) 338.
Finnis, J.M., "Scepticism, Self-Refutation, and the Good of Truth," in
Peter Hacker and Joseph Raz (eds.), LAW, MORALITY, AND
247-67. Oxford University Press, 1977.
LITERATURE. University of California Press, 1972.
Fishkin, James, "Justifying Liberty: The Self-Reflective Argument,"

Suber A Bibliography of Works on Reflexivity

paper delivered orally at a conference on Applied Philosophy,

April 13, 1985, Bowling Green State University.
Fitch, Frederic Brenton, "Comments and a Suggestion," in R.L. Martin
(ed.), THE PARADOX OF THE LIAR (q.v.).
Fitch, Frederic Brenton, "Excluded Middle and the Paradoxes," paper
delivered orally at a conference of the Association for Symbolic
Logic, New York City, 1975.
Fitch, Frederic Brenton, "Formalized Self-Reference," in Steven J.
Bartlett and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Fitch, Frederic Brenton, "On God and Immortality," PHILOSOPHY
Fitch, Frederic Brenton, "A Goedelized Formulation of the Prediction
(April 1964) 161-64.
Fitch, Frederic Brenton, "A Method for Avoiding the Curry Paradox,"
in Nicholas Rescher et al. (eds.), ESSAYS IN HONOR OF
CARL G. HEMPEL, pp. 255-65. D. Reidel Pub. Co., 1969.
Fitch, Frederic Brenton, "The Perfection of Perfection," THE
MONIST, 47 (1963) 466-71.
Fitch, Frederic Brenton, "Remarks on the Theory of Types," MIND,
56 (1946) 184.
Fitch, Frederic Brenton, "Self-Reference in Philosophy," MIND, 55
(1946) 64-73. Reprinted, slightly revised, as Appendix C to his
SYMBOLIC LOGIC, q.v., pp. 217-225. (See review by
Mostowski, A.)
Fitch, Frederic Brenton, "Self-Referential Relations," Actes du Xieme
congres international de philosophie, (Amsterdam), 14 (1953)
121-27. (See review by Prior, A.N.)
Fitch, Frederic Brenton. SYMBOLIC LOGIC. Ronald Press, 1952.
Fitch, Frederic Brenton, and R.J. Orgass, "A Theory of Computing
Machines," STUDIUM GENERALE, 22 (1969) 83-104.
Fitch, Frederic Brenton, and R.J. Orgass, "A Theory of Programming
Languages," STUDIUM GENERALE, 22 (1969) 113-36.
Fitch, Frederic Brenton, "Universal Metalanguages for Philosophy,"
REVIEW OF METAPHYSICS, 17 (1964) 369-402.
Fitzpatrick, PJ., "'Heterological' and Namely-Riders," ANALYSIS,
22 (1961) 18-22. (See review by Cargile, J.)
SOCIAL SYSTEMS. Transactions of the Conference on
Cybernetics. N.Y.: Josiah Macy, Jr. Foundation, 1949-55.
Foerster, Heinz von, "Cybernetics of Cybernetics," pp. 5-8 of Klaus

Suber A Bibliography of Works on Reflexivity


SOCIETY, Gordon and Breach, 1979.
Foerster, Heinz von, OBSERVING SYSTEMS. Intersystems, 1981.
(Includes a calculus of infinite regression to cope with certain
problems of self-reference.)
Fox, Richard, "Philosophy and Self-Reference," PHILOSOPHY IN
CONTEXT, 4 (1975) 28-36.
MODERN LITERATURE. Rutgers University Press, 1963.
Frankfurt, Harry G, "The Logic of Omnipotence," PHILOSOPHICAL
REVIEW, 73 (1964) 262-63.
Frankfurt, Harry G. [See works listed under Descartes]
[Frege, Gottlob], See: Hudson, J .L. Sloman, A.
Frutas, E. [See work listed under Guillen]
Fumerton, Richard A., "The Paradox of Analysis," PHILOSOPHY
Fiiredy, Viveca Y.G.M. THE PLAY WITH A PLAY
Dissertation, Hebrew University, 1983.

- G-

Gale, Richard M., "The Egocentric Particular and Token-Reflexive

Analyses of Tense," PHILOSOPHICAL REVIEW, 73 (April
1964) 213-28.
[Galileo], See: Mintz, S.I.
Gardner, Martin, "Logical Paradoxes," ANTIOCH REVIEW, 23
(1963) 172-77.
Garelick, William P. [See work listed under Blanshard]
Prentice-Hall, 1967. (Much on the reflexivity of the stance of
the sociologist.)
Garner, Richard T., "A Non-Paradoxical Paradox,"
PHILOSOPHICAL REVIEW, 78 (1969) 249-51.
Garvin, Lucius, "The Paradox of Aesthetic Meaning," PHILOSOPHY
1947) 99-106.
LITERATURB. Harvard University Press, 1972.
[Gass, William H.], See: McCaffery, L.
Geach, P.T., " Mr. Ill-named," ANALYSIS, 9 (1948) 14-16. (See

Suber A Bibliography of Works on Reflexivity

review by Church, A.)

Geach, P.T., "Namely-Riders Again," ANALYSIS, 22 (1962) 92-94.
(See review by Cargile, J.)
Geach, P.T., "A Note on the Reflexive Paradoxes," PHILOSOPHICAL
REVIEW, 62 (1953) 427-28. (A reply to Toms, E.)
Geach, P.T., "Omnipotence," PHILOSOPHY, 48 (1973) 7-20.
Geach, P.T., "On Insolubilia," ANALYSIS, 15 (1955) 71-72.
Geach, P.T., "Ryle on Namely-Riders," ANALYSIS, 21 (1961) 64-67.
(See review by Cargile, J.)
Geach, P.T., "Two Paradoxes of Russell's," JOURNAL OF
PHILOSOPHY, 67 (February 26, 1970) 89-97.
[Geach, P.T.], See: Woods, 1. and Walton, D.
Gellman, Jerome, "The Paradox of Omnipotence, and Perfection,"
SOPHIA, 14 (1975) 31-39.
AND TIME'. Harper & Row, 1970. Section on hermeneutic
circle, pp. 176-81.
Gendin, Sidney, "Omnidoing," SOPHIA, 6 (1967) 17-22.
George, L.C., "King Solomon's Judgment Expressing Principles of
Discretion and Feedback in Legal Rules and Reasoning,"
HASTINGS LAW JOURNAL, 30 (1979) 1549-75.
Gewirth, Alan, "Can Men Change Laws of Social Science?"
PHILOSOPHY OF SCIENCE, 21 (July 1954) 229-41.
CRITICAL RATIONALISM. Ph.D. Dissertation, Vanderbilt
University, 1979.
Gill, S. [See work listed under Wordsworth.]
Gilmore, Grant, "Circular Priority Systems," YALE LAW
JOURNAL, 71 (1961) 53-74.
Glymour, Clark, "Bootstraps and Probabilities," JOURNAL OF
PHILOSOPHY, 77, 11 (Nov. 1980) 691-99. (See replies by
Horwich, P. and Edidin, A.)
Goddard, Leonard, "True and Provable," MIND, 67 (1958) 13-31.
(See review by Resnick, M.D.)
Goddard, Leonard. [Co-authored work listed under Goldstein, L.]
Translated by B. Meltzer. Basic Books, 1962.
[Godel, Kurt], Barbo, Francesca Rivetti, "A Philosophical Remark on
Godel's Unprovability of Consistency Proof," NOTRE DAME
JOURNAL OF FORMAL LOGIC, 9 (1968) 67-74.
[GOdel, Kurt], Benacerraf, Paul, "God, the Devil, and GOdel," THE
MONIST, 51 (1967) 9-32.

Suber A Bibliography of Works on Reflexivity

[Go del, Kurt], Broyles, I.E., "Paradox and Argument,"

[Godel, Kurt], Chihara, Charles S., "On Alleged Refutations of
Mechanism Using Godel's Incompleteness Results," JOURNAL
OF PHILOSOPHY, 64 (1972) 507-26.
[Go del, Kurt], Coder, David, "GOdel's Theorem and Mechanism,"
PHILOSOPHY, 44 (1969) 234-37.
[Godel, Kurt], Desmonde, William R, "Godel, Non-deterministic
Systems and Hermetic Automata," INTERNATIONAL
[Go del, Kurt], Detlefsen, Michael, "On Interpreting Godel's Second
(1979) 297-313.
[GOdel, Kurt], Findlay, J., "Goedelian Sentences: A Non-Numerical
Approach," MIND, 51 (1942) 259-65.
[Godel, Kurt], Good, I.J., "Godel's Theorem is a Red Herring,"
(1969) 357-58.
[Go del, Kurt], Hanson, Norwood Russell, "The Godel Theorem,"
[Go del, Kurt], Hanson, W.H., "Mechanism and Godel's Theorems,"
(1971) 9-16.
[Godel, Kurt], Hutton, Anthony, "This Godel is Killing Me,"
PHILOSOPHIA, 6 (1976) 135-44. (See reply by Lucas, 1.R.)
[GOdel, Kurt], Nagel, Ernest, and James R. Newman. GODEL'S
PROOF. New York University Press, 1958.
[GOdel, Kurt], Nagel, Ernest, and James R. Newman, "GOdel's Proof,"
SCIENTIFIC AMERICAN, 194 (1956) 71-84.
[GOdel, Kurt], Safir, Orin, "Concrete Forms --Their Application to the
Logical Paradoxes and Godel's Theorem," JOURNAL OF
PHILOSOPHICAL LOGIC, 5 (1976) 133-54.
[Godel, Kurt], Slezak, Peter, "Godel's Theorem and the Mind,"
(1982) 41-52.
[Godel, Kurt], Yourgrau, Wolfgang, "Godel and Physical Theory,"
MIND, 78 (1969) 77-90.
[Godel, Kurt], See: Hofstadter, D.R.; Johnstone, A.A.; Lucas, J.R.;
Stone, R.; Stone-de Montpensier, R.; Tucker, 1.
Goldin, J., "On Honi the Circle-Maker: A Demanding Prayer,"

Suber A Bibliography of Works on Reflexivity


Goldstein, Gordon T. [Co-authored work listed under Yovitts, M.C.]
Goldstein, Laurence, "Four Alleged Paradoxes in Legal Reasoning,"
CAMBRIDGE LAW JOURNAL, 38 (1979) 373-91.
Goldstein, Laurence, and Leonard Goddard, "Strengthened
3 (Sept. 1980) 211-221.
Goldstein, Laurence. [See work listed under Wittgenstein.]
Go~dstick, D. [See work listed under Hume]
Good, I.J., "A Note on Richard's Paradox," MIND, 75 (1966) 431. (See
review by Tucker, J.)
Good, I.J. [See work listed under GOde1.]
Goossens, W.K., "Eulathus and Protagoras," LOGIQUE ET
ANALYSE 20 (1977) 67-75.
Gorsky, D.P., "Nonpredicative and Predicative Definitions," Section
PROBLEMS, pp. 75-78. Progress Publishers, 1974.
Gould, Carol Steinberg. [See work listed under Plato.]
Grambs, David. WORDS ABOUT WORDS. McGraw Hill, 1984. (A
dictionary of words used in writing about writing and
Grave, S.A., "On Evil and Omnipotence," MIND, 65 (1956) 249-62.
[Greenaigh], See:
Brown, T.M.
Greene, Murray. [See work listed under Aristotle]
Gregory, J., "Heterological and Homological," MIND, 61 (1952)
85-88. (See review by Baylis, C.)
Grelling, Kurt, "The Logical Paradoxes," MIND, 45 (1936) 481-86.
(See review by Langford, C.H.)
[Grillparzer], Dunham, T.C., "Circle Image in Grillparzer's Libussa,"
GERMANIC REVIEW, 36 (May 1961) 125-36.
Grimm, Ruediger Hermann. [See work listed under Nietzsche]
Grisez, Germain. [Co-authored work listed under Boyle, J.M.]
Grisez, Germain, "Sketch of a Future Metaphysics," NEW
SCHOLASTICISM, 38 (1964) 311-40.
Grossman, A. [See work listed under Milton]
Grover, Dorothy, "Inheritors and Paradox," JOURNAL OF
PHILOSOPHY, 74, 10 (Oct. 1977) 590-604.
Gruenberg, Barry, "The Problem of Reflexivity in the Sociology of
(Dec.1978) 321-43.
Griinbaum, Adolf. [See work listed under Buck, R.C.]
Grunberg, Emile, and Franco Modigliani, "Reflexive Predictions,"

Suber A Bibliography of Works on Reflexivity

PHILOSOPHY OF SCIENCE, 32 (Apri11965) 173-74.

Guaspari, Daud Solovay, R.M., "Rosser Sentences," ANNALS OF
MATHEMATICAL LOGIC, 16 (May 1979) 81-99.
[Guillen, Jorge], Frutas, E., "Circle and Its Rupture in the Poetry of
Jorge Guillen," BOOKS ABROAD, 42 (Winter 1968) 33-6.
Gupta, Anil, "Truth and Paradox," JOURNAL OF PHILOSOPHICAL
LOGIC, 11 (1982) 1-60.
[Gupta], Belnap, N., "Gupta's Theory of Truth," JOURNAL OF
PHILOSOPHICAL LOGIC, 11 (1982) 103-16.

- H-
Haack, Susan, "Paradoxes," Chapter 8 of her PHILOSOPHY OF
LOGICS, pp. 135-151. Cambridge University Press, 1978.
Haight, M.R. A STUDY OF SELF-DECEPTION. Harvester Press,
Johns Hopkins University, 1975.
Hanson, Norwood Russell. [See work listed under Godel]
Hanson, W.H. [See work listed under Godel]
Harel, David et aI., "Self-Reference Referenced, and Self-Referenced,"
COMMUNICATIONS OF THE ACM, 23, 12 (December 1980)
736-37. (Three letters to the editor on texts that cite
Harman, Gilbert, "Reasoning and Evidence One Does Not Possess,"
Harries, Karsten, "The Infinite Sphere: Comments on the History of a
13, 1 (Jan. 1975) 5-15.
Harris, R., "Self-Description and the Theory of Types," ANALYSIS,
28 (1968) 207-08.
Harris, R., "The Semantics of Self-Description," ANALYSIS, 27
(1967) 144.
Harrison, Craig, "Totalities and the Logic of First Cause Arguments,"
35, 1 (Sept. 1974) 1-19.
Hart, H.L.A., "Self-Referring Laws," in FESTSKRIFT TILLAGNAD
KARL OLIVECRONA, pp. 307-16. Stockholm: Kunglia
Boktryckeriet, P.A. Norstedt and Soner, 1964. Reprinted in
Oxford University Press, 1983, pp. 170-78.
Hart, John A. and J. Gregory Dees, "Paradox Regained: A Reply to
Meyers and Stem," JOURNAL OF PHILOSOPHY, 71, 12 (June

Suber A Bibliography of Works on Reflexivity

27, 1974) 367-72.

Hart, W.O., "Causation and Self-Reference," in Steven J. Bartlett and
Peter Suber (eds.), SELF-REFERENCE (q.v.).
Hart, W.O., "On Self-Reference," PHILOSOPHICAL REVIEW, 79
(1970) 523-28.
Hart, W.O., "Russell and Ramsey," PACIFIC PHILOSOPHICAL
QUARTERLY, 64 (1983) 193-210.
Hatab, Lawrence 1. [See work listed under Nietzsche]
TRUTH. Ph.D. Dissertation, McGill University, 1983.
Haynes, Richard Pierce. [See work listed under Plato]
Heap, James L., "Description in Ethnomethodology," HUMAN
STUDIES,3 (January 1980) 87-106.
[Hegel], Aboulafia, Mitchell. THE SELF-WINDING CIRCLE: A
STUDY OF HEGEL'S SYSTEM. St. Louis: Green, 1982.
[Hegel], Dahlstrom, Daniel 0., "Hegel's Principia," THE NEW
SCHOLASTICISM, 55, 4 (Autumn, 1981) 421-37. (That
Hegel's Logic is not viciously circular.)
[Hegel], Rockmore, Tom, "Hegel on Epistemological Circularity and
QUARTERLY, 21, 3 (Sept. 1981) 235-48.
Indiana University Press, 1986.
[Hegel], Yerkes, James, THE CHRISTOLOGY OF HEGEL, Section:
"The Hermenteutic Circle," State University of New York
Press, 1982.
[Hegel, G.W.F.], See: Scheier, C.-A. Shafer, I.H.
Heidegger, Martin. NIETZSCHE, vol. 2: THE ETERNAL
RECURRENCE OF THE SAME. Trans. David Farrell Krell.
Harper & Row, 1984.
[Heidegger], Norman, Bart Ross. METHODOLOGY,
Dissertation, University of New Mexico.
[Heidegger, Martin], See: Maraldo, J.-C.; Spanos, W.V.; Gelven, M.
Heidelberger, Herbert, "The Self-Presenting," GRAZER
Helm, Paul, "Omnipotence and Change," PHILOSOPHY, 51 (1976)
Henle, P., Review of Reach's "The Name Relation ... " (q.v.),
Hennessey, W.J. ARTISTS LOOK AT ART: University of Kansas

Suber A Bibliography of Works on Reflexivity

Press, 1978.
Henshel, Richard L., "Effects of Disciplinary Prestige on Predictive
Accuracy: Distortions from Feedback Loops," FUTURES, 7
(April, 1975) 92-106.
Bobbs-Merrill, 1976.
Henshel, Richard L., "Scientific Status and Boundaries of the
Self-Fulfilling Prophecy." Paper delivered orally at a
conference of the American Sociological Association, San
Francisco, 1975.
Henshel, Richard L., "Self-Altering Predictions," in 1. Fowles (ed.),
1978. Contains an excellent bibliography.
Henshel, Richard L. and Leslie W. Kennedy, "Self-Altering
Prophecies: Consequence for the Feasibility of Social
Prediction," GENERAL SYSTEMS, 18 (1973) 119-26.
Henshel, Richard L., "Sociology and Prediction," AMERICAN
SOCIOLOGIST, 6 (August, 1971) 213-220.
Herzberger, H.G., "Naive Semantics and the Liar Paradox,"
JOURNAL OF PHILOSOPHY, 79,9 (September 1982) 479-97.
Herzberger, H.G., "Notes on Naive Semantics," JOURNAL OF
PHILOSOPHICAL LOGIC, 11 (1982) 61-102.
Herzberger, H.G., "Paradoxes of Grounding in Semantics," JOURNAL
OF PHILOSOPHY, 67 (March 26, 1970) 145-67.
[Hesse, Herman], See: Serrano, M. Shafer, I.H.
Hicks, J.C., "The Liar Paradox in Legal Reasoning," CAMBRIDGE
LAW JOURNAL, 29 (1971) 275-91.
Hinman, Lawrence. [See work listed under Skinner, B.F.]
Hintikka, Jaakko, "Identity, Variables, and Impredicative Definitions,"
JOURNAL OF SYMBOLIC LOGIC, 21 (1958) 225-45. (See
review by Jensen, R.)
Hintikka, Jaakko, "Remarks on a Paradox," ARCHIV FOR-RECHTS
UND SOZIALPHILOSOPHIE, 44 (1958) 514-16. (On
parliamentary omnipotence.) (Reply to Tammelo, I.)
Hintikka, Jaakko, "Vicious Circle Principle and the Paradoxes,"
JOURNAL OF SYMBOLIC LOGIC, 22 (1957) 245-49. (See
review by Jensen, R.)
Hirsch, E.D., Jr., "The Self-Confirmability of Interpretations,"
Chapter 5, Section A, of his VALIDITY IN
INTERPRETATION,pp. 164-169. Yale University Press,
[Hobbes, Thomas], See: Mintz, S.1.
Hoerster, Norbert. [See work listed under Ross, A.]

Suber A Bibliography of Works on Reflexivity

Hoffman, Joshua. [See work listed under Mavrodes, G.I.]

Hoffman, Joshua. [Co-authored work listed under Rosenkrantz, G.]
Hoffman, Robert, "On Begging the Question At Any Time,"
ANALYSIS, 32 (Dec. 1971) 51.
[Hoffmann, E.T.A.], Daemmrich, Horst S., "Struggle For a New
Vision of the Infernal Circle," Chapter 4, of his THE
pp.39-46. Wayne State University Press, 1973. (On theme of
circle and confinement in the Johannes Kreisler stories.)
Hofstadter, A., "On Semantic Problems," JOURNAL OF
PHILOSOPHY, 35 (1938) 225-32. (See review by Nagel, E.)
Hofstadter, Douglas R, "Analogies and Metaphors to Explain Godel's
JOURNAL, 13,2 (March 1982) 98-114.
Hofstadter, Douglas R. GODEL, ESCHER, BACH: an ETERNAL
GOLDEN BRAID. New York: Basic Books, Inc., 1979.
Hofstadter, Douglas, R., "Metamagical Themas," SCIENTIFIC
AMERICAN,224, 1 (January 1981) 22-32. (On self-referential
sentences. )
Hofstadter, Douglas R "Metamagical Themas: About Nomic: A Heroic
Game That Explores the Reflexivity of the Law," SCIENTIFIC
AMERICAN, 246, 6 (June 1982) 16-28.
Hofstadter, Douglas R, "Metamagical Themas: Can Inspiration Be
Mechanized?" SCIENTIFIC AMERICAN, 247, 3 (September
1982) 18-34.
Hofstadter, Douglas R, "Metamagical Themas: A Self-Referential
Column About Last January's Column About Self-Reference,"
SCIENTIFIC AMERICAN, 246,1 (January 1982) 16-28.
Hofstadter, Douglas R, "Metamagical Themas: Tripping the Light
Recursive in LISP, the Language of Artificial Intelligence,"
SCIENTIFIC AMERICAN, 248 (March 1983) 22-29.
Hofstadter, Douglas R., "Metamagical Themas: On Viral Sentences and
Self-Replicating Structures," SCIENTIFIC AMERICAN,247, 1
(January 1983).
Hofstadter, Douglas R and Daniel Dennett (eds.). THE MIND'S I:
Basic Books, 1981.
Hofstadter, Douglas, R., "Self-Reference in Perception and
Cognition," Paper presented orally at Conference of the Society
for Philosophy and Psychology, Chicago, April 5, 1981.
Hofstadter, Douglas R, "Who Shoves Whom Around Inside the
Carenium? Or What is the Meaning of the Word 'I'?" Indiana
University Computer Science Department Technical Report No.

Suber A Bibliography of Works on Reflexivity

130, July 1982.

VERSE. Yale University Press, 1981. (Witty, self-exemplifying
descriptions of verse forms, genres, and tropes.)
Hollister, W.W., "Conduct and the Circle," JOURNAL OF
PHILOSOPHY, 50 (Jan. 29, 1953) 57-70.
Horwich, Paul, "The Dispensability of Bootstrap Conditions,"
JOURNAL OF PHILOSOPHY, 77, 11 (Nov. 1980) 699-702.
(Reply to Glymour, C.)
University of California Press, 1978.
Hudson, James L., "Frege's Way Out," [of Russell's Paradox],
Hugget, W. J., "Paradox Lost," ANALYSIS, 19 (1958) 21-23. (See
review by Bennet, J.)
Hughes, Patrick, and George Brecht. VICIOUS CIRCLES AND
Hugly, Philip, "A Semantical Account of the Vicious Circle Principle,"
1979) 595-98.
Hull, David L., "Certainty and Circularity in Evolutionary
Taxonomy," EVOLUTION, 21 (1967) 174-89.
[Hume], Goldstick, D., "Hume's 'Circularity' Charge Against
Inductive Reasoning," DIALOGUE, 11 (1972) 258-66.
[Hume], Kivy, Peter, "Hume's Standard of Taste: Breaking the Circle,"
[Hume, David], See: Rorty, A.O.
Hunter, John F.M., "Asking Oneself," PHILOSOPHICAL
INVESTIGATIONS, 1 (Summer 1978) 14-24.
Hupp, Jon A. [See co-authored work listed under Shoch, John.]
Hutton, Anthony. [See work listed under GDdel]

- 1-

[Ingarden, Roman], Cohen, Sybil, "Ingarden's Benign Circle,"

[Isaye, Gaston], See: Moleski, M.

Suber A Bibliography of Works on Reflexivity

Jackson, Frank, "Richard on Richard's Paradox," MIND, 80
(April,1971) 284-85.
Jacobi, George T. [Co-authored work listed under Yovitts, M.C.]
Jacobs, Francis, "Circularity and Responsibility," PHILOSOPHY, 39
(1964) 268-74.
Jacobs, Jaap. [See work listed under Suber, P.]
Illinois University Press, 1982.
History Publishers, 1974.
Jameson, G. [See work listed under Parmenides]
PARADIGM OF EVOLUTION. Pergamon Press, 1980.
Jaspers, Karl, "Exegesis in Circles," in his PHILOSOPHY, Vol. 3, pp.
129-130. University of Chicago Press, 1971.
University Press, 1985.
INVESTMENT MANAGMENT. Allyn and Bacon, 1974.
Jensen, R., Review of Hintikka's "Identity, Variables and ... ," (q.v.),
JOURNAL OF SYMBOLIC LOGIC, 32 (1967) 258-59.
Jensen, R., Review of Hintikka's "Vicious Circle Principle and ... ,"
(q.v.), JOURNAL OF SYMBOLIC LOGIC, 32 (1967) 258-59.
[Jimenez, Juan Ramon], Olson, Paul R .. CIRCLE OF PARADOX:
JIMENEZ. Johns Hopkins University Press, 1967.
[Jodelle, Estienne], Reiss, T.T., "Jodelle's Cleopatre and the Enchanted
Circle," YALE FRENCH STUDIES, 47 (1972) 199-210.
John, lA., and F.W. Wolock, H.A. David. CYCLIC DESIGNS. United
States Department of Commerce, Bureau of Standards,
Government Printing Office, 1972.
Johnson, Oliver A., "Begging the Question," DIALOGUE, 6 (1967)
135-50. (See review by Williams, M.E.)
Johnson, Oliver, A., "To Beg the Question: A Reply," DIALOGUE, 7
(Dec. 1968) 461-68.
Johnson, Oliver A., "Circular Arguments," Chapter VII, Section 2 of
his SKEPTICISM AND COGNITION, pp. 226-239. University

Suber A Bibliography of Works on Reflexivity

of California Press, 1978.

Johnstone, Albert A., "Self-Reference, The Double Life and G6del,"
LOGIQUE ET ANALYSE, 24 (March 1981) 35-47.
Johnstone, Henry W., Jr., "Argumentum ad Hominem with and without
Self-Reference," in Steven J. Bartlett and Peter Suber (eds.),
Johnstone, Henry W., Jr., "The Categorio-Centric Predicament,"
Johnstone, Henry W., Jr., "Persons and Self-Reference," JOURNAL
(Jan. 1970) 46-54. (See comment by Mays, W.)
Johnstone, Henry W., Jr., "Persons and Self-Reference: Reply to Mr.
PHENOMENOLOGY," 1 (Oct. 1970) 66.
Johnstone, Henry W., Jr. PHILOSOPHY AND ARGUMENT. The
Pennsylvnia State University Press, 1959.
Johnstone, Henry W., Jr. THE PROBLEM OF THE SELF. The
Pennsylvania State University Press, 1970.
Johnstone, Henry W., Jr., "Self-Refutation and Validity," THE
MONIST, 48, 4 (1964) 467-85; reprinted in his VALIDITY
TRANSITION. Dialogue Press of Man & World, 1978.
EXPECTANCIES. Hillsdale, N.J.: L. Erlbaum Associates
(Distributed by Halsted Press), 1977.
J!1lrgensen, J!1lrgen, "On Kattsoffs Reflexions on J!1lrgensen's Reflexions
on Reflexivity," MIND, 64 (1955) 542.
J!1lrgensen, J!1lrgen, "Some Reflections on Reflexivity," MIND, 62
(1953) 289-300. Also printed in the DANISH YEARBOOK OF
PHILOSOPHY, 6 (1969) 29-39.
[Joyce, James], See: Murillo, L.A.
[Jung, C.G.], See: Serrano, M.; Shafer, I.H.

Kalmer, L., Review of Curry's "The Paradox of Kleene and Rosser,"

(q.v.), JOURNAL OF SYMBOLIC LOGIC, 11 (1946) 136-37.
THEORY OF MEMORY. Ph.D. Dissertation, Stanford
University, 1984.
Suber A Bibliography of Works on Reflexivity

Kanger, S., Review of Prior's "Epimenides the Cretan" (q.v.), Cohen's

"Why do Cretan's Have to ... " (q.v.), and 7 others on indirect
speech, JOURNAL OF SYMBOLIC LOGIC, 32 (1967)
Kapadia, R, "Paradox Found," MIND, 83, 330 (April 1974) 296-97.
(Reply to T. Steel.)
Kaplan, David, and Montague, Richard, "A Paradox Regained,"
Kasanen, E. [Co-authored work listed under Eudoxus]
Kattsoff, L.O., "Some Reflections on Reflexivity," MIND, 64 (1955)
96-98. (See review by Jrgensen, 1.)
Katz, Bernard D., "Is the Causal Criterion of Event Identity Circular?"
1978) 225-29.
Kautsky, J.H., "Myth, Self-Fulfilling Prophecy and Symbolic
Reassurance in the East-West Conflict," JOURNAL OF
CONFLICT RESOLUTION, 9 (March 1965) 1-17.
Press, 1982.
Keene, G.B., "Capacity-Limiting Statements," MIND, 70 (1981)
Keene, G.B., "On the Logic of the Circularity of Logic," MIND, 84,
333 (Jan. 1975) 100-01.
Keene, G.B., "Self-Referent Inference and the Liar Paradox," MIND,
92,367 (July 1983) 430-33.
Keene, G.B., "A Simple Solution to the Paradox of Omnipotence,"
MIND, 67 (1980) 74-75. (See review by Mayo, B.)
Kellenberger, J., "The Ineffabilities of Mysticism," AMERICAN
PHILOSOPHICAL QUARTERLY, 16, 4 (Oct. 1979). (Four
species of ineffability and their paradoxes.)
Keller, James A., "Foundationalism, Circular Justification, and the
Levels Gambit," paper delivered orally at Convention of the
American Philosophical Association, April 27, 1984,
Kellman, Steven G. THE SELF-BEGETTING NOVEL. Columbia
University Press, 1980.
Kellogg, "Priorities Puzzle Under The Ship Mortgage Act,"
WASHINGTON LAW REVIEW, 2 (1927) 117. (On circular
Kelly, MJ. [See work listed under Descartes]
Kelly, RJ. [Co-authored work listed under Sagarin, E.]
Suber A Bibliogmphy of Works on Reflexivity

Kennedy, Leslie W. [Co-authored work listed under Henshel, R.L.]

Kester, S.W., and G.A. Letchworth, "Communication of Teacher
Expectations and Their Effects on Achievement and Attitudes,"
Khamara, Edward J., "In Defense of Omnipotence,"
[Kierkegaard, S0ren], See: Stack, G. Spanos, W.V.
Killalea, J.N., "Primeness and Heterologicality," ANALYSIS, 14
(1953) 20-24. (See review by Orey, S.)
Press, 1973.
King-Farlow, John, "God and the Stone Paradox: Comment III,"
SOPHIA, 10 (1971) 31-33. (Comment on Englebretsen, G.)
Kivy, Peter. [See work listed under Hume]
Kleene, S., and J. Rosser., "The Inconsistency of Certain Formal
Logics," ANNALS OF MATHEMATICS, 36 (1935) 630-36.
(See Curry, B.)
Press, 1984.
Klossowski, Pierre. [See work listed under Nietzsche]
Knaster, Stephen, "Is Determinism Self-Defeating?" Paper delivered
orally at Convention of American Philosophical Association,
Columbus, Ohio, April 30, 1982.
Kneale, William C., "Russell's Paradox and Some Others," in George
VOLUME. Humanities Press, 1978.
Kocourek, A., "A First Rate Legal Puzzle: A Problem in Priorities,"
ILLINOIS LAW REVIEW, 29 (1935) 852.
Kohl, H.R., and Parsons, C., "Self-Reference, Truth, and Provability,"
MIND, 69 (1960) 69-73.
Konyndyk, Kenneth, Jr., "Rational Affirmation and Free Choice: A
Study of 'Free Choice'," NEW SCHOLASTICISM, 53, 4
(Autumn 1979) 502-14. Review of Boyle, Grisez, Tollefsen
Kordig, Carl R., "Another Ethical Paradox," MIND, 78 (October
1969) 598-599.
Kordig, Carl R., "Evolutionary Epistemology is Self-Referentially
RESEARCH, 42, 3 (March 1982) 449-50.
Kordig, Carl R., "Objectivity, Scientific Change, and Self-Reference,"

Suber A Bibliography of Works on Reflexivity


vol. 8, pp. 519-23, D. Reidel Publishing Co., 1970.
Kordig, Carl R, "Some Statements are Immune to Revision," NEW
SCHOLASTICISM, 55, 1, (Winter 1981) 69-76.
Kosman, L.A, "Perceiving That We Perceive," PHILOSOPHICAL
REVIEW, 84, 4 (Oct. 1975) 499-519.
Kowzan, Tadeusz, "The Art of Underpinning," DIOGENES, (Winter
1976) 67-92.
Koyre, Alexander, "The Liar," PHILOSOPHY AND
review by Church, A)
RESEARCH, 8 (1947) 254-55.
Krippendorff, Klaus, "On the Ethics of Constructing Communication,"
delivered orally as Presidential Address at the International
Communication Association on Paradigm Dialogues, Honolulu,
May 23, 1985. (Long subsection on "The Self-Referential
Krippendorff, Klaus, "Paradox and Information," Chapter 2 (pp.
45-71) of Brenda Dervin and Melvin J. Voigt (eds.),
Krishna, Daya, "The Self-Fulfilling Prophecy and the Nature of
1971) 1104-07.
Krueger, Joe. [See work listed under Nietzsche]
Kvanvig, Jonathan L.,"The Confusion over Foundationalism," oral
presentation at convention of the American Philosophical
Association, March 1985, Chicago. (In part, on "self-warrant
foundationalism" .)

. L

Lambert, K., "On the Non-Communicator," PHILOSOPHICAL

STUDIES, 17 (1966) 27-30.
Landsberg, P.T., "On Heterological Paradoxes," MIND, 62 (1953)
379-81. (See review by Orey, S.)
Langford, C.H., "On Paradoxes of the Type of the Epimenides,"
MIND, 56 (1947) 350.
Langford, C.H., "The Paradoxes," JOURNAL OF PHILOSOPHY, 47
(1950) 777-78.
Langford, C.H., Review of Behmann's "The Paradoxes of Logic,"
(q.v.), JOURNAL OF SYMBOLIC LOGIC, 2 (1937) 92.

Suber A Bibliography of Works on Reflexivity

Langford, C.H., Review of Beth's "Some Remarks on Dr. Perelman's

Essay..." (q.v.), JOURNAL OF SYMBOLIC LOGIC, 2 (1937)
Langford, C.H., Review of Grelling's "The Logical Paradoxes," (q.v.),
Langford, C.H., Review of McIver's "More About Some Old Logical
Puzzles," (q.v.), JOURNAL OF SYMBOLIC LOGIC, 6 (1941)
Langford, C.H., Review of Ushenko's "A New Epimenides," (q.v.),
Langford, C.H., and Langford, M., "The Logical Paradoxes,"
21 (1959) 110-113.
Langford, M. [Co-authored work listed under Langford, C.H.]
Larson, Arthur, "The Self-Judging Clause and Self-Interest,"
(1960) 729-31.
Lawrence, Irene (ed.). SELF-DEFINITI~N EARLY
JANUARY, 6 1980. Center for Hermenenti 1 Studies in
Hellenistic and Modem Culture, The Graduate Theological
Union and the University of California, Berkeley, 1980.
Lawrence, N., "Heterology and Hierarchy," ANALYSIS, 10 (1950)
73-76. Reprinted in M. MacDonald (ed.) PHILOSOPHY AND
ANALYSIS, (q.v.). (See reviews by Church, A. and Orey, S.)
PREDICAMENT. Open Court Press, 1986.
Lee, Edward N. [See work listed under Plato]
[Leibniz], See: Benziger, J.; Sinisi, V.F.
Lenzen, Wolfgang, "Protagoras versus Euathlus: Reflections on a
So-Called Paradox," RATIO, 19 (Dec. 1977) 176-80.
[Lessing, Gotthold Ephraim], See: Higonnet, M.R.
Letchworth, G.A. [Co-authored work listed under Kester, S.W.]
Levi, Issac, "Induction as Self-Correcting According to Peirce," in
University Press, 1980.
Levine, Michael, "Can There Be Self-Authenticating Experiences of
God? A Reply to Robert Oakes," Paper delivered orally at
Convention of American Philosophical Association, Columbus,
Ohio, May 1, 1982. (With reply by Oakes.)
[Lewis, C.S.], See: Anscombe, G.E.M.
Lewis, David K. [See work listed under Scriven]

Suber A Bibliography of Works on Reflexivity


Dissertation, Case Western Reserve University. 1977.
Lipman, Jean, and Richard Marshall. ART ABOUT ART. E.P. Dutton,
PHILOSOPHY. Humanities Press, 1984.
Londey, David, "God and the Stone Paradox: Comment I," SOPHIA,
10 (1971) 23-25. (Comment on Englebretsen, G.)
[Lonergan, Bernard], See: Moleski, M.X.
Long, A.A., "The Stoics on World-Conflagration and Everlasting
Louisell, D.W., "Biology, Law, and Reason: Man as Self-Creator,"
Bibliographisches Institut, 1969.
L6with, Karl. [See work listed under Nietzsche]
Lucas, John Randolf. THE FREEDOM OF THE WILL. Oxford
University Press, 1970.
Lucas, John Randolf, "Mechanism: A Rejoinder," PHILOSOPHY, 45
(1970) 149-51.
Lucas, John Randolf, "Mind, Machines, and GOdel," PHILOSOPHY, 36
(1961) 112-27.
Lucas, John Randolf, "Satan Stultified," THE MONIST, 52 (1968)
Lucas, John Randolf, "This Godel is Killing Me: A Rejoinder,"
PHILOSOPHIA, 6 (1976) 145-48. (A rejoinder to Hutton, A.)


MacDonald, M. (ed.). PHILOSPHY AND ANALYSIS. Oxford

University Press, 1954. (Much on heterologicality.) (See
review by Orey, S.)
MacIver, A.M., "More About Some Old Logical Puzzles,"
ANALYSIS, 6 (1938) 63-68. (See review by Langford, C.H.)
Mackay, Alan, "Mandala Thinking," in David DeGrood, Dale Riepe,
John Somerville (eds.), RADICAL CURRENTS IN
Warren H. Green, 1971.
Mackenzie, J.D., "Begging the Question in Dialogue,"

Suber A Bibliography of Works on Reflexivity

Mackenzie, J.D., "Why Do We Number Theorems?,"
1980) 135-49. (Answer: to avoid circular proofs.)
Mackie, J.L., "Conditionally-Restricted Operations," NOTRE DAME
JOURNAL OF FORMAL LOGIC, 2 (1961) 236-43.
Mackie, J.L., "Evil and Omnipotence," MIND, 64 (1955) 200-12.
Mackie, lL., "Omnipotence," SOPHIA, 1 (1962) 13-25.
Mackie, J .L., "Self-Refutation -A Formal Analysis,"
PHILOSOPHICAL QUARTERLY, 14 (1964) 193-203. (See
review by Bennet, J.)
IN PHILOSOPHICAL LOGIC. Oxford University Press, 1973.
Mackie, J.L., "What Can We Learn From the Paradoxes?" CRITICA, 5
(Jan. 1971) 85-105.
Mackie, J.S., and J.J.C. Smart, "A Variant of the Heterological
Paradox," ANALYSIS, 13 (1953) 61-65. (See review by Orey,
Mackie, J.S., and J.lC. Smart, "A Variant of the Heterological Paradox
-A Further Note," ANALYSIS, 14 (1954) 146-49. (See review
by Orey, S.)
Madge, Nicole. [Co-authored work listed under Rutter, Michael]
[Malraux], Sonnenfeld, A., "Malraux and the Tyranny of Time: The
Circle and the Gesture," ROMANIC REVIEW, 54 (Oct. 1963)
Mandelbrot, Benoit. FRACTALS: FORM, CHANCE, AND
DIMENSION. W.H. Freeman, rev. ed. 1982. (On self-similar
Mann, William E. [See work listed under Plato]
Stanford University Press, 1965. Chapter I: "The Wicked Dance
in Circles... ", Chapter 3: "Ixion's Wheel".
Marans, David, "A Note on Reflexiveness," NOTRE DAME
JOURNAL OF FORMAL LOGIC, 16 (Oct. 1975) 501-06.
Mardiros, Anthony M., "A Circular Procedure in Ethics,"
PHILOSOPHICAL REVIEW, 61 (April 1952) 223-225.
Marino, Adrian, "Two Hermeneutical Circuits: Part/Whole and
Analysis/Synthesis," DIALECTICS AND HUMANISM, 3, 2
(1976) 125-34.
Markie, Peter. [See work listed under Descartes]
Marshall, Geoffrey. CONSTITUTIONAL THEORY. Oxford
University Press, 1971. (Much on legal omnipotence.)

Suber A Bibliography of Works on Reflexivity

COMMONWEALTH, Oxford University Press, 1957. (Much

on legal omnipotence.)
Marshall, Geoffrey, "Parliamentary Sovereignty and the Language of
Constitutional Limitation," JURIDICAL REVIEW, 67 (1955)
Marshall, Geoffrey, "Parliamentary Sovereignty: A Recent
Development," McGILL LAW JOURNAL, 12,4 (1966-67)
Marshall, Geoffrey, "What is Parliament? The Changing Concept of
Parliamentary Sovereignty," POLITICAL STUDIES, 2, 3
Marshall, James, Michael Peters, and Miles Shepheard,
"Self-Refutation: Arguments Against Young's Epistemology,"
1981) 43-50.
Marshall, Richard. [See co-authored work listed under Lipman, J.]
Martin, Richard M., "Self-Reference," Chapter Five of his
Reidel Pub. Co., 1979.
Martin, Robert L., "Are Natural Languages Universal?" SYNTHESIS,
32 (April 1976) 271-92.
Martin, Robert L., "A Category Solution to the Liar," in R.L. Martin
(ed.), PARADOX OF THE LIAR (q.v.).
Martin, Robert L., "On Grelling's Paradox," PHILOSOPHICAL
REVIEW, 77 (July 1968) 321-31.
Martin, Robert L. THE PARADOX OF THE LIAR. Yale University
Press, 1970. Contains excellent bibliography. (See review by
Ashworth, EJ.)
PARADOX. Oxford University Press, 1984.
Martin, Robert L., and Woodruff, Peter W., "On Representing
'True-in-L' in L," PHILOSOPHIA, 5 (July 1975) 213-17.
Martin, Robert L., "Sommers on Denial and Negation," NOUS, 3
(1969) 219-26.
Martin, Robert L., "Toward a Solution to the Liar Paradox,"
PHILOSOPHICAL REVIEW, 76 (July 1967) 279-311.
Martin, R. M., "The Pragmatics of Self-Reference," essay number 5 in
Publishing Co., 1979.
Martinich, A.P., "Conversational Maxims and Some Philosophical
(July 1980) 215-28. (Contains a solution to the liar.)
Marx, Wolfgang, "Epistemological Reflections on the Problem of

Suber A Bibliography of Works on Reflexivity

Impredicative Conceptual Structures," RATIO, 17,1 (June

1975) 35-48.
Mates, Benson, "Philosophical Skepticism and the Logical
Mates, Benson, "Two Antinomies," section of his SKEPTICAL
ESSAYS, pp. 15-57. University of Chicago Press, 1981.
Mattila, J. [Co-authored work listed under Eudoxus]
Maturana, H.R., and FJ. Varela. AUTOPOIESIS AND COGNITION.
Boston Studies in the Philosophy of Science, vol. 42. D. Reidel,
Maula, E. [Co-authored work listed under Eudoxus]
Mavrodes, George I., "Seif-Referential Incoherence," AMERICAN
Mavrodes, George I., "Some Puzzles Concerning Omnipotence,"
PHILOSOPHICAL REVIEW, 72 (1963) 221-23.
[Mavrodes], Hoffman, Joshua, "Mavrodes on Defining Omnipotence,"
PHILOSOPHICAL STUDIES, 35 (1979) 311-13.
[Mavrodes], Reichenbach, Bruce R., "Mavrodes on Omnipotence,"
PHILOSOPHICAL STUDIES, 37 (1980) 211-14.
Maynez, Eduardo Garcia, "Some Considerations on the Problem of
Antinomies in the Law," ARCHIC FOR RECHTS-UND
SOZIALPHILOSOPHIE,49,1 (1963) 1-14.
Mayo, Bernard, "Mr. Keene on Omnipotence," MIND, (1961) 249-50.
Mays, Wolfe, "Persons and Self-Reference: A Comment," JOURNAL
1970) 55-6. (See review by Johnstone, H.W.)
WILLIAM H. GASS. University of Pittsburgh Press, 1982.
McKinsey, J.C.C. Review of Ushenko's "Undecidable Statement
and ... " (q.v.), JOURNAL OF SYMBOLIC LOGIC, 9 (1944)
McLean, M.R., "The Unmakeable -Because-Unliftable Stone,"
McNaughton, R. Review of Wang's "The Irreducibility of
Impredicative Principles," (q.v.), JOURNAL OF SYMBOLIC
LOGIC, 18 (1953) 265-66.
[McTaggart, J.E.M.l, Roberts, Joy H., "Statements, Sentences and
States of Affairs in McTaggart and in General,"
ERKENNTNIS, 15 (March 1980) 73-89.
Meager, R., "Heterologicality and the Liar," ANALYSIS, 16 (1956)

Suber A Bibliography of Works on Reflexivity

Meierding, Loren, "The Impossibility of Necessary Omnitemporal

PHILOSOPHY OF RELIGION, 11 (1980) 21-16.
Meiland, Jack W., "On the Paradox of Cognitive Relativism,"
METAPHILOSOPHY, 11,2 (April 1980) 115-26. (See review
by Beach, E.)
Meiland, Jack W., "Is Protagorean Relativism Self-Refuting?"
Mele, A.R., "Self-Deception," PHILOSOPHICAL QUARTERLY, 33,
133 (Oct. 1983) 366-77.
[Melville, Herman], Wiley, Margaret L., "Melville and Circularity,"
Essay VIII in her CREATIVE SCEPTICS. London: George
Allen and Unwin, 1966.
Menger, K., "The New Logic," PHILOSOPHY OF SCIENCE, 4 (1937)
299-336. (See review by Quine, W.V.O.)
Menninger, Karl. LOVE AGAINST HATE. Chapter 5: Breaking the
Vicious Circle (pp. 122-33). Harcourt, Brace and World, 1942.
Merleau-Ponty, Maurice, "Circular Dialectic," in his THEMES FROM
1952-1980, pp. 57-60. Northwestern University Press, 1970.
Merton, Robert K., "The Self-Fulfilling Prophecy," ANTIOCH
REVIEW, 8 (1948) 193-210.
Merton, Robert K., "The Self-Fulfilling Prophecy," Chapter 13 of his
Press, 1968.
Merton, Robert K., "The Unanticipated Consequences of Purposive
(Dec. 1936) 894-904.
[Meyers, Robert G], See: Hart, lA. and Dees, J.G.
Michie, D., and Christopher Longuet-Higgens, "A Party-Game Model
of Biological Replication," NATURE, 212, 5057 (October
1966) 10-12. (Self-referential sentences modeling biological
replication. )
[Mill, John Stuart], See: Walton, D.
Miller, Arthur R., "Correct vs. 'Merely True' Act-Descriptions,"
INQUIRY, 17 (Winter 1974) 457-60.
[Miller, Arthur], Rayfield, David, "On Miller's Paradoxes and
Circles," INQUIRY, 17 (Winter 1974) 461-64.
Miller, Barry, "God and the Stone Paradox: Comment II," SOPHIA, 10
(1971) 26-31. (Comment to Englebretsen, G.)
KNOWLEDGE CLAIMS. Ph.D. Dissertation, New York
University, 1977.

Suber A Bibliography of Works on Reflexivity

Miller, Myron Michael, "The Pragmatic Paradox," in Steven J. Bartlett

and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Miller, Myron Michael, "Science and the Thumb-Chasing Game,"
paper delivered orally at the National Science Teachers
Association Convention, Washington, D.C., March 31,1968.
Miller, Richard W. [See work listed under Wittgenstein]
[Milton, John], Grossman, A., "Ring Pattern: Image, Structure and
Theme in Paradise Lost," STUDIES IN PHILOLOGY, 68 (July
1971) 326-39.
Mintz, S.I., "Galileo, Hobbes, and the Circle of Perfection," ISIS, 43,
pt.2, (1952) 98-100.
Mishkin, Paul J.,"Prophecy, Realism, and the Supreme Court,"
(Aug. 1954) 680-83 and 725-26.
Modigliani, Franco. [Co-authored work listed under Grunberg, E.]
Moleski, Martin X., S.J., "Retortion: the Method and Metaphysics of
QUARTERLY, 17, 1 (March 1977) 59-83.
Moleski, Martin X., S.1., "The Role of Retortion in the Cognitional
Analyses of Lonergan and Polanyi," in Steven J. Bartlett and
Peter Suber (eds.), SELF-REFERENCE (q.v.).
Molt, Peter, "Dates, Tenseless Verbs, and Token-Reflexivity," MIND,
82 (Jan. 1973) 73-85.
Montague, R. [Co-authored work listed under Kaplan, D.]
Morreall, John, "God as Self-Explanatory," THE PHILOSOPHICAL
QUARTERLY, 30, 120 (July 1980) 206-14.
Mortensen, C. [Co-authored work listed under Priest, G.]
Morton, John. "On Recursive Reference," COGNITION, 4 (1976) 309.
Mostowski, A., "Correction to the Paper 'Some Impredicative
Definitions in the Axiomatic Set-Theory'," FUNDAMENTA
MATHEMATICAE, 38 (1951-2) 238. (See review by Skolem,
Mostowski, A., Review of Fitch's "Self-Reference in Philosophy,"
(q.v.), JOURNAL OF SYMBOLIC LOGIC, 11, (1946) 95-96.
Mostowski, A., "Some Impredicative Definitions in the Axiomatic
(1950-51) 111-124. (See review by Skolem, T.)
Mott, Peter, "Dates, Tenseless Verbs and Token Reflexivity," MIND,
82 (January 1973) 73-85.
Moulder, James, "Is Russell's Paradox Genuine?" PHILOSOPHY, 49,
189 (July 1974) 295-302.
Muller, G.H. Review of Yuting's "Two Semantical Paradoxes," (q.v.),

Suber A Bibliography of Works on Reflexivity

Muller, Norbert, "Strategy and Reflexivity," in Leif Lewin (ed.),

POLITICS AS RATIONAL ACTION, pp. 121-32. Reidel Pub.
Co., 1980.
Munro, Colin R., "The Magical Roundabout of Conflict of Laws,"
JURIDICAL REVIEW, n.s. 23 (1978) 65-84.
Press, 1968.
Myers, C. Mason, "Circular Explication," METAPHILOSOPHY, 9, 1
(Jan. 1978) 1-13.
Myers, C. Mason, "The Circular Use of Metaphor," PHILOSOPHY
Myhill, John, "Paradoxes," Essay 8 in Hughes Leblanc, Elliott
Mendelson, and Alex Orenstein (eds.), FOUNDATIONS:
Myhill, John, "A System Which Can Define Its Own Truth,"

- N-

Naess, Arne, "Circularity of the Sceptic's Argument," Chapter 5,

Section 7 of his SCEPTICISM, pp. 126-27. Routledge and
Kegan Paul, 1968.
Nagel, Ernest, Review of Hofstadter's "On the Semantic Problems,"
(q.v.), JOURNAL OF SYMBOLIC LOGIC, 3 (1938) 90.
Nagel, Ernest, Review of Northrop's RIDDLES IN
LOGIC, 10 (1945) 21.
Nagel, Ernest. [See work listed under Godel]
Nagel, Ernest, and James R. Newman. [See works listed under Gode1.]
[Nagel], Nissen, Lowell, "Nagel's Self-Regulation Analysis of
Teleology," THE PHILOSOPHICAL FORUM, 12,2 (Winter
1980-81) 128-38.
Napalkov, A.V. [Co-authored works listed under Braines, S.N.]
Nehamas, Alexander. [See work listed under Plato]
Nell, E., "No Proposition Can Describe Itself," ANALYSIS, 26 (1966)
Nell, E., "Semantics and Self-Description," ANALYSIS, 18 (1967) 32.
Nelson. E., Review of Ushenko's THE PROBLEMS OF LOGIC (q.v.),
Nelson, John O. [See work listed under Descartes]

Suber A Bibliography of Works on Reflexivity


CONSCIOUSNESS. Princeton University Press, 1954. Chapter
A.I: The Creation Myth: The Uroboros. (On circle as symbol of
creation, development, perfection.)
[Neumann, J. Von], See: Harrison, C.
Newcomb, Theodore M. [Co-authored work listed under Chowdhry,
Newman, James R. [See work listed under GOdel]
THROUGH FLUCTUATIONS. N.Y.: John Wiley and Sons,
Nicolson, Marjorie Hope. THE BREAKING OF THE CIRCLE.
Northwestern University Press, 1950.
Ph.D. Dissertation, University of Michigan, 1970.
[Nietzsche], Altizer, Thomas J.1., "Eternal Recurrence and Kingdom of
God," in David B. Allison (ed.), THE NEW NIETZSCHE:
232-46. Dell Pub. Co., 1977.
[Nietzsche], Baker, Thomas A. THE ONTOLOGICAL CONNECTION
ETERNAL RETURN. Ph.D. Dissertation, Marquette
University, 1983. _
[Nietzsche], Brush, Stephen G., "Nietzsche's Recurrence Revisited: The
PHILOSOPHY, 19,2 (April 1981) 235-38.
[Nietzsche], Clegg, Jerry S., "Nietzsche and the Ascent of Man in a
PHILOSOPHY, 19, 1 (Jan 1981) 81-93.
[Nietzsche], Danto, Arthur, "The Eternal Recurrence," in Robert
CRITICAL ESSAYS, pp. 316-2l. Anchor Books, 1973.
[Nietzsche], Hatab, Lawrence J. NIETZSCHE AND ETERNAL
BECOMING. University Press of America, 1978.
[Nietzsche], Grimm, Ruediger Hermann, "Circularity and
Self-Reference in Nietzsche," METAPHILOSOPHY, 10 (1979)
[Nietzsche], Kain, Philip J., "Nietzsche, Skepticism, and Eternal
Recurrence," Paper delivered orally at convention of the
American Philosophical Association, Baltimore, December 28,

Suber A Bibliography of Works on Reflexivity

[Nietzsche], Klossowski, Pierre, "Nietzsche's Experience of the Eternal

Return," in David B. Allison (ed.), THE NEW NIETZSCHE:
107-20. Dell Pub. Co., 1977.
[Nietzsche], Krueger, Joe, "Nietzschean Recurrence as a Cosmological
PHILOSOPHY, 16 (1978) 435-44.
[Nietzsche], Lowith, Karl, "Nietzsche's Revival of the Doctrine of
Eternal Recurrence," Appendix II of his MEANING IN
HISTORY, pp. 214-22. University of Chicago Press, 1949.
POSSIBLE HYPOTHESES. Ph.D. Dissertation, University of
Missouri, Columbia, 1984.
[Nietzsche], Small, Robin, "Nietzsche and a Platonist Tradition of the
Cosmos: Center Everywhere and Circumference Nowhere,"
(January-March 1983) 89-104.
[Nietzsche], SolI, Ivan, "Reflections on Recurrence: A Reexamination
of Nietzsche's Doctrine, die Ewige Wiederkehr des Gleichen,"
in Robert Solomon (ed.), NIETZSCHE: A COLLECTION OF
CRITICAL ESSAYS, pp. 322-42. Anchor Books, 1973.
[Nietzsche], Stambaugh, Joan. Nietzsche'S THOUGHT OF ETERNAL
RETURN. Johns Hopkins University Press, 1972.
[Nietzsche], Williams, Robert. RECURRENCE, PARODY, AND
NIETZSCHE. Ph.D. Dissertation, Yale University, 1982.
[Nietzsche], Zuboff, Arnold, "Nietzsche and Eternal Recurrence," in
Robert Solomon (ed.), NIETZSCHE: A COLLECTION OF
CRITICAL ESSAYS, pp. 343-57. Anchor Books, 1973.
[Nietzsche, Friedrich], See: Delevsky, J.; Heidegger, M.; Jaki, S.C.
(chapter 13).
Institute of Technology Press, 1971.
Nissen, L. [See work listed under Nagel]
Noelle-Neumann, Elisabeth. THE SPIRAL OF SILENCE: PUBLIC
OPINION --OUR SOCIAL SKIN. University of Chicago Press,
Norman, B.R. [See work listed under Heidegger.]
Norman,1. [Co-edited work listed under Priest, G.]
PARADOXES. Van Norstrand Reihold Co., 1944. Reprinted,

Suber A Bibliography of Works on Reflexivity

R.E. Krieger Pub. Co., 1975. (See review by Nagel, E.)

University Press, 1981. Chapter 1, Section II: Reflexivity, pp.

Oakes, Robert A., "Religious Experience, Self-Authentication, and
Modality De Re: A Prolegomenon," AMERICAN
PHILOSOPHICAL QUARTERLY, 16, 3 (July 1979). (See
review by Levine, M.)
O'Brien, James F., "Teilhard and Aristotle: What is Radial and What is
Tangential?" NEW SCHOLASTICISM, 49 (1975) 486-91.
O'Carroll, M.J., "Improper Self-Reference in Classical Logic and the
Predication Paradox," LOGIQUE ET ANALYSE, 10 (June
1967) 167-72.
O'Carroll, M.J., "A Three-valued, Non-Levelled Logic Consistent For
All Self-Reference," LOGIQUE ET ANALYSE, 10 (June 1967)
[Ockham, William of], See: Spade, P.V.
O'Connor, John, "On Eliminating Self-Reference," ANALYSIS, 28
(1968) 131-32. (See review by Schlesinger, G.)
O'Connor, D.J., "Pragmatic Paradoxes," MIND, 57 (1948) 358-59.
(See review by Cohen, L.J.)
O'Connor, D.J., "Pragmatic Paradoxes and Fugitive Propositions,"
MIND, 60 (1951) 536-38.
Odegard, Douglas, "Classifying the Class-Membership Relation,"
LOGIQUE ET ANALYSE, 12 (1969) 221-24.
Odegard, Douglas, "The Irreflexivity of Knowledge," in Steven J.
Bartlett and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Odegard, Douglas, "Knowledge and Reflexivity," DIALOGUE, 15
(Apri11976) 226-40.
Odegard, Douglas, "Parasitical Reference and Paradox," AMERICAN
PHILSOPHICAL QUARTERLY, 13,4 (Oct. 1976) 295-301.
(Solutions to the liar, heterological, and Russell's paradox.)
Odegard, Douglas, "Truth," Chapter Four of his KNOWLEDGE AND
SCEPTICISM, pp. 40-55. Rowman and Littlefield, 1982. (On
paradox and its avoidance.)
Odegard, Douglas. [See work listed under Descartes.]
Olbrechts-Tyteca, L. [Co-authored work listed under Perelman, C.]
Oldenquist, Andrew, "Self-Prediction," in Paul Edwards (ed.),

Suber A Bibliography of Works on Reflexivity

[Oliver, Thomas], See: Thomas, I.

Olson, Paul R. [See work listed under Jimenez]
Onians, Richard Broxton. THE ORIGINS OF EUROPEAN
THOUGHT. Cambridge University Press, 1951. (Much on
theme of circle in re time, fate, telos.)
Orey, S. Review of Bowden's "Heterologicality" (q.v.), Landsberg's
"On Heterological Paradoxes" (q.v.), Mackie and Smart's "A
Variant of the ... " (q.v.), Killalea's "Primeness and
Heterologicality" (q.v.), Lawrence's "Heterology and
Hierarchy" (q.v.), MacDonald (ed.), PHILOSOPHY AND
ANALYSIS (q.v.), Ryle's "Heterologicality" (in MacDonald),
JOURNAL OF SYMBOLIC LOGIC, 20 (1955) 291-93.
Orgass, R.I. [Co-authored works listed under Fitch, F.B.]
Orr, William F., "A Minimal Self-Predicative Bibliography,"


[Paine, Thomas], See: Brown, T.M.

Palmer, Humphrey, "Do Circular Arguments Beg the Question?"
PHILOSOPHY, 56, 215 (July 1981) 387-94.
Pap, Arthur, "The Linguistic Hierarchy and the Vicious Circle
Principle," PHILOSOPHICAL STUDIES,S (1954) 49-53. (See
review by Bausch, A.)
Parker, DeWitt, H., "Reflexive Relations," PHILOSOPHICAL
REVIEW, 42 (1933) 303-311, and 43 (1934) 295-300.
[Parmenides], Jameson, G., "'Well-Rounded Truth' And Circular
Thought in Parmenides," PHRONESIS, 3 (1958) 15-30.
[Parmenides], See: Ballew, L.
Parsons, C., "The Impredicativity of Induction," in Leigh Cauman et
OF SIDNEY MORGENBESSER, Hackett Pub. Co., 1983.
Parsons, C. [Co-authored work listed under Kohl, H.R.]
[Pascal, Blaise], See: Borges, J.L.
Pask, G., and G. Von Foerster, "A Predictive Model for
Self-Organizing Systems," CYBERNETICA, 3, 4 (1960)
258-300; 4, 1 (1961) 20-55.
Passmore, John. PHILOSOPHICAL REASONING. Duckworth,
Pastin, Mark, "Modest Foundationalism and Self-Warrant,"
#4, pp. 141-49. Reprinted in G.S. Pappas and M. Swain (eds.),

Suber A Bibliography of Works on Reflexivity

University Press, 1978, pp. 279-88.

[Paul, St.], See: Anderson, A.R.
Paulos, John Allen, "Self-Reference and Paradox," Section of his
MATHEMATICS AND HUMOR, pp. 41-55. University of
Chicago Press, 1980.
Peck, "Federal Tax Liens -Their Removal or Foreclosure, Privity
Thereof, and the Problem of Circuity of Priorities,"
NEBRASKA LAW REVIEW, 38 (1959) 163, 170.
Pei, Mario, "Self-Designation," Chapter 10 of his THE STORY OF
LANGUAGE, pp. 394-96. J.B. Lippencott Co., revised edition,
[Peirce, Charles Sanders], See: Rivetti Barbo, F.; Thompson, M.H., Jr.;
Levi, I.
Penrose, L.S., "Automatic Mechanical Self-Reproduction," in NEW
BIOLOGY, pp. 92-117. Penguin, 1959.
Penrose, L.S., "Self-Reproducing Machines," SCIENTIFIC
AMERICAN, 200,6 (1959) 105-112, 114,202.
Pereleman, Chaim, and L. Olbrechts-Tyteca. THE NEW RHETORIC:
Dame Press, 1969. Section 28: Argumentation ad Hominem and
Begging the Question.
Peter, Rozsa. RECURSIVE FUNCTIONS. Academic Press, 1967.
Peters, Michael. [Co-authored work listed under Marshall, J.]
Peterson, Sandra. "A Reasonable Self-Predication Premise for the
(1973) 451-70.
Phillips, O. Hood, "Self-Limitation by the United Kingdom
QUARTERLY, 2 (1975) 443-78.
Pinto, W.L. [See work listed under Plato]
ASSUMPTION. Ph.D. Dissertation, University of Minnesota,
[Plato], Clegg, Jerry S., "Self-Predication and Linguistic Reference in
Plato's Theory of Forms," PHRONESIS, 18 (1973) 26-73.
[Plato], Driscoll, John A., "Self-Participation and the Non-Identity
Assumption in Plato's 'Sophist'," Paper delivered orally at
Convention of the American Philosophic Association, Pacific
Division, San Francisco, March 28, 1980.
[Plato], Duff-Forbes, D.R., "The Regress Arguments in the Republic,"
MIND, 87, 307 (July 1968) 406-10.
[Plato], Dyson, M., "Some Problems Concerning Knowledge In Plato's
'Charmides'," PHRONESIS, 19 (1974) 102-111.

Suber A Bibliography of Works on Reflexivity

[Plato], Gould, Carol Steinberg. EPONYMY AND SELF

FORMS. Ph.D. Dissertation, State University of New York,
Buffalo, 1984.
[Plato], Haynes, Richard Pierce. PLATO'S THEORY OF FORMS
Dissertation, University of Illinois, Urbana-Champagne, 1962.
[Plato], Lee, Edward N., "Reason and Rotation: Circular Movement as
the Model of Mind (nous) in the Later Plato," in W.H.
Supplementary Volume II of PHRONESIS: A JOURNAL FOR
ANCIENT PHILOSOPHY. Assen: Van Gorcum, 1976.
[Plato], Mann, William E., "The Third Man--The Man Who Never
(July 1979) 167-76.
[Plato], Nehamas, Alexander, "Self-Predication and Plato's Theory of
(April 1979) 93-103.
University of Pennsylvania, 1976.
[Plato], Savan, D. "Self-Predication in PROTAGORAS 330-331,"
PHRONESIS, 9 (1984) 130-35.
THE SOPHIST. Ph.D. Dissertation, University of Virginia,
[Plato], Spitzer, Adele, "The Self-Reference of the GORGIAS,"
PHILOSOPHY AND RHETORIC, 8 (Winter 1975) 1-22.
[Plato], Thomas, William J., "Platonism and the Skolem Paradox,"
ANALYSIS, 28 (June 1968) 193-96.
[Plato], Vlastos, Gregory, "On a Proposed Redefinition of
'Self-Predication' in Plato," PHRONESIS, 26, 1 (1981) 76-78.
[Plato], Yartz, Frank J., "Infinite Regress and the Sense World in
(1975) 17-28.
[Plato], See: Ballew, L.; Bumyeat, M.F.; Teloh, H.
Plommer, H. [See work listed under Vitruvius]
[Plotinus], Fiedler, John, "Plotinus and Self-Predication," in R. Baine
NEOPLATONIC APPROACH. State University of New York
Press, 1982.
[Polanyi, Michael], See: Moleski, M.X.

Suber A Bibliography of Works on Reflexivity

Pollock, John L., "The Liar Strikes Back," JOURNAL OF

PHILOSOPHY,74, 10 (Oct 1977).
Popper, Karl, "Is Determinism Self-Refuting?" MIND, 92, 365
(January 1983) 103-04.
Popper, Karl, "Self-Reference and Meaning in Ordinary Language,"
MIND, 63 (1954) 162-69. Reprinted in his CONJECTURES
KNOWLEDGE. Basic Books, 1962, pp. 304-11. (See reviews
by Thompson, J.F.; and Prior, A.N.)
Portmess, Lisa. TIME AND PARADOX. Ph.D. Dissertation, Queens
University, 1979.
Post, John F., "The Possible Liar," NOUS, 4, 4 (Nov. 1970) 405-09.
Post, John F., "Presupposition, Bivalence, and the Possible Liar,"
ISRAEL, 8, 4, (Dec. 1978).
Post, John F., "Shades of the Liar," JOURNAL OF PHILOSOPHIC
LOGIC, 2 (July 1973) 370-85.
THE LIAR. Ph.D. Dissertation, Brown University, 1980.
Hopkins University Press, 1966.
KNOWLEDGE. William Morrow and Co., 1985.
Prado, C.G., "Reflexive Awarness," THE NEW SCHOLASTICISM,
52, 3 (Summer 1978) 428-33.
Prado, C.G., "Reflexive Consciousness," DIALOGUE, 17 (1978)
Price, H.H., "Self-Verifying Beliefs," Lecture 6, Series II, of his
Gifford Lectures, 1959-60. Published in his BELIEF, pp.
349-75. Humanities Press, 1969.
Priest, Graham, "An Anti-Realist Account of Mathematical Truth,"
SYNTHESE, 57 (1983) 49-65.
Priest, Graham, "Classical Logic Aufgehoben," in Priest, Routley, and
Norman (eds.), PARACONSISTENT LOGIC, q.v.
Priest, Graham, "Contradiction, Belief and Rationality,"
Priest, Graham, "The Conventionalist Philosophy of Mathematics,"
CONVERENCE, ed. J. Bell et aI., Leeds, 1973, pp. 115-32.
Priest, Graham, "Hypercontradictions," LOGIQUE ET ANALYSE,
107 (1984) 237-43.

Suber A Bibliography of Works on Reflexivity

Priest, Graham, and R. Routley, "Introduction to Paraconsistent

Logic," STUDIA LOGICA, 44 (1983) 3-16.
Priest, Graham, and R. Routley, "Lessons from Pseudo-Scotus,"
PHILOSOPHICAL STUDIES, 42 (1982) 189-99.
Priest, Graham, "The Logical Paradoxes and the Law of Excluded
Middle," PHILOSOPHICAL QUARTERLY, 33 (1983) 160-65.
(See reply by Brady, R.T.)
Priest, Graham, "The Logic of Paradox," JOURNAL OF
PHILOSOPHIC LOGIC, 8 (May 1979) 219-41.
Priest, Graham, "The Logic of Paradox Revisisted," JOURNAL OF
PHILOSOPHICAL LOGIC, 12 (1984) 153-179.
Priest, Graham, "The Logical Paradoxes: A Unified Account," paper
delivered orally at a conference of the Australasian Association
of Logic, Melbourne University, 1977.
Priest, Graham, "The Logical Paradoxes," PHILOSOPHICAL
QUARTERLY, 33, (April 1983) 160-65.
Priest, Graham, and R. Routley. ON PARACONSISTENCY.
Research Report #13, Research School of Social Sciences,
Australian National University, 1983. Reprinted as
introductory chapters of Priest, Routley, and Norman (eds.),
Priest, Graham, R. Routley, and J. Norman (eds.).
PARACONSISTENT LOGIC. Philosophia Verlag, 1986.
Priest, Graham, "Principia Mathematica and the Vicious Circle
Principle," paper delivered orally at a conference of the British
Society for the Philosophy of Science, Warwick University,
Priest, Graham, "Reductio ad Absurdum et Modus Tollendo Ponens,"
in Priest, Routley, and Norman (eds.), PARACONSISTENT
LOGIC, q.v.
Priest, Graham, and J. Crosthwaite, "Relevance, Truth and Meaning,"
in R. Routley and J. Norman (eds.), DIRECTIONS OF
RELEV ANT LOGIC, Martinus Nijhoff, forthcoming.
Priest, Graham, "Semantic Closure," STUDIA LOGICA, 43 (1984)
Priest, Graham, "Sense, Entailment and Modus Ponens," JOURNAL
OF PHILOSOPHICAL LOGIC, 9 (1980) 415-35.
Priest, Graham, and C. Mortensen, "The Truth Teller Paradox,"
LOGIQUEET ANALYSE, 95-96 (1981) 381-88.
Priest, Graham, "Unstable Solutions to the Liar Paradox," in Steven J.
Bartlett and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Prior, A.N., "The Cogito of Descartes and the Concept of
Self-Confirmation," PAPERS IN LOGIC AND ETHICS, pp.

Suber A Bibliogmphy of Works on Reflexivity

165-75, P.T. Geach and A.J.P. Kenny (eds.) London:

Duckworth, 1976.
Prior, A.N., "Epimenides the Cretan," JOURNAL OF SYMBOLIC
LOGIC, 23 (1958) 261-66. (See review by Kanger, S.)
Prior, A.N., "On a Family of Pamdoxes," NOTRE DAME JOURNAL
OF FORMAL LOGIC, 2 (1961) 16-32.
Prior, A.N., "Symmetry, Transitivity, and Reflexivity," JOURNAL
Prior, A.N. Review of Fitch's "Self-Referential Relations," (q.v.),
Prior, A.N. Review of Popper's "Self-Reference and Meaning in
Ordinary Language," (q.v.), JOURNAL OF SYMBOLIC
LOGIC, 24, (1959) 240.
Prior, A.N., "Some Problems of Self-Reference in John Burdian,"
[Protagoras], See: Aqvist, L.; Bryant, J.; Bumyeat, M.F.; Lenzen, W.;
Goossens, W.K.; Meiland, J.W.
[Pseudo-Scotus], See: Priest, G. and RoutIey, R.
Pugmire, David, "'Strong' Self-Deception," INQUIRY, 12 (Fall 1969)
Putnam, Hilary. [Co-authored work listed under Smullyan, R.]


Quine, W.V.O., "On a So-Called Pamdox," MIND, 62 (1955) 65-67.

Quine, W.V.O., "Paradox," SCIENTIFIC AMERICAN, 206 (1962)
84-96. Reprinted in his THE WAYS OF PARADOX, (q.v.), pp.
Quine, W.V.O. Review of Menger's "The New Logic," (q.v.),
Random House, 1966.
[Quine], Roth, Paul A., "Paradox and Indeterminacy," JOURNAL OF
PHILOSOPHY, 75, 7 (July 1978) 345-67.
[Quine, W.V.O.], See: Swiggart, P.
Quiney, H.R., "A Non-Hierarchical Mathematical Logic,"

Radnitzky, G., "Toward a Theory of Research Which is Neither

Suber A Bibliography of Works on Reflexivity

Logical Reconstruction nor Psychology or Sociology of

Science," TEORIE E METODA, 3 (1973) 197-264. (On a
reflexive method.)
[Ramsey, Frank], See: Hart, W.D.
Ramsey, LT., "The Paradox of Omnipotence," MIND, 65 (1956)
Rankin, K.W., "Referential Identifiers," AMERICAN
PHILOSOPHICAL QUARTERLY, 1 (July 1964) 233-43.
Rayfield, David. [See work listed under Miller, A.R.]
Raz, Joseph. [See work listed under Ross, A.]
Reach, K., "The Name Relation and the Logical Antinomies,"
JOURNAL OF SYMBOLIC LOGIC, 3 (1938) 97-111. (See
review by Henle, P.)
Read, Stephen, "Self-Reference and Validity," SYNTHESIS, 42
(October 1979) 265-74.
Rechtin, Lisbeth, and William Todd, "Propositional Attitudes and
Self-Reference," PHILOSOPHIA, 4 (April-July 1974) 271-95.
Rees, W.J., "The Theory of Sovereignty Restated," MIND, 59 (Oct.
1950) 495-521. (Much on parliamentary omnipotence.)
Reeves, Sandra. [See work listed under Nietzsche.]
Regnell, Hans, "A New Approach to the Liar and Some Other
(1968) 7-30.
Regnell, Hans, "On Reflexivity," DANISH YEARBOOK OF
PHILOSOPHY, 1 (1964) 112-19.
Reichenbach, Bruce R. [See work listed under Mavrodes]
Reiss, T.T. [See work listed under Jodelle]
Rescher, Nicholas, "The 'No Criterion' Argument: 'The Wheel',"
pp. 11-15. Rowman and Littlefield, 1980. (Skeptical diallelus as
an alternative to infinite regress.)
Rescher, Nicholas, N., "A Note on Self-Referential Statements,"
Rescher, Nicholas, "Semantic Paradoxes and the Propositional
Analysis of Indirect Discourse," PHILOSOPHY OF SCIENCE,
28 (1961) 437-40.
Resnick, M.D., "Professor Goddard and the Simple Theory of Types,"
MIND, 77 (1968) 565-68.
[Richard, Thomas J.], See: Jackson, Frank
Richards, Thomas J., "Self-Referential Paradoxes," MIND, 76 (July
1967) 387-403.
Richardson, Jane Shelby. [Co-authored work listed under Scriven]

Suber A Bibliography of Works on Reflexivity

Richman, R.J., "On the Self-Reference of Meaning Theory,"

SCHOOL FOR AESTHETICS. Wayne State University Press,
1973. Section 46: Circular Wit.
Riguet, J. [Co-authored work listed under Ashby, W.R.]
Rinaldi, Fiori, "Dilemmas and Circles in the Law," ARCHIV FOR
Roberts, Joy H. [See work listed under McTaggart, lE.M.]
Roberts, Louise Nisbet, "Every Proposition Is False -A Medieval
Robinson, Richard, "Begging the Question," ANALYSIS, 31 (March
1971) 113-17.
Rockmore, Tom. [See works listed under Hege1.]
Romanos, George D., "Reflexive Predictions," PHILOSOPHY OF
SCIENCE, 40 (1973) 97-109.
Rorty, Amelie Oksenberg, "Reflexive Attitudes," Paper delivered
orally at Convention of American Philosophical Association,
N.Y., Dec. 29, 1979.
Rorty, :Arne lie Oksenberg. SELF-REFERENCE AND THE THEORY
PHILOSOPHIC METHOD, Ph.D. Dissertation, Yale
University, 1980.
Rorty, Richard, "The Limits of Reductionism," in I. Lieb (ed.),
Southern Illinois University Press, 1961. (Appeal to
self-referential consistency.)
Rorty, Richard, "Transcendental Arguments, Self-Reference, and
Pragmatism," in Rolf-Peter Horstmann (ed.),
77-103. Reidel Pub. Co., 1979.
Rosenkrantz, Gary, and Joshua Hoffman, "What an.Omnipotent Agent
OF RELIGION, 11 (1980) 1-19.
Ross, Alf. ON LAW AND JUSTICE. London: Stevens, 1958. (Section
16 on the paradox of self-amendment.)
Ross, Alf, "On Self-Reference and a Puzzle of Constitutional Law,"
MIND, 78 (1969) 1-24. (On parliamentary omnipotence and
self-amendment. )
[Ross, Alf], Hoerster, Norbert, "On Alf Ross's Alleged Puzzle in
Constitutional Law," MIND, 81 (1972) 422-26.
[Ross, Alf], Raz, Joseph, "Professor Ross and Some Legal Puzzles,"

Suber A Bibliography of Works on Reflexivity

MIND, 81 (1972) 415-21.

Rosser, J. [Co-authored work listed under Kleene, S.]
Rotenstreich, Nathan, "Self-Ascription and Objectivity,"
PHILOSOPHIA, 10,3-4, pp. 189-98. (See reply by Strawson,
P.F. at end of issue.)
Roth, G., and H. Schegler (eds.). SELF-ORGANIZING SYSTEMS.
Campus, 1981.
Roth, Paul A. [See work listed under Quine]
Rothschild, K.W., "Cobweb Cycles and Partially Correct Forecasting,"
Routley, R. [Co-authored work listed under Priest, G.]
Routley, R. [Co-edited work listed under Priest, G.]
Rowe, William L. [See work listed under Anselm]
Royce, Josiah, "Theory of the Sources and Consequences of Any
Recurrent Operation of Thought. The Nature of
Self-Representative Systems," Section III of the Supplementary
BEING, pp. 501-54. Dover Publications, 1959. (Original,
Macmillan, 1899.)
Pennsylvania Press, 1965.
Rozeboom, W.W., "Is Epimenides Still Lying?" ANALYSIS, 18
(1957-58) 105-13. (See review by Bennet, J.)
Russell, Bertrand, "Mathematical Logic as Based on the Theory of
(1908) 222-62.
Russell, Bertrand, "On 'Insolubilia' and Their Logical Solution by
Symbolic Logic," in Douglas P. Lackey (ed.), ESSAYS IN
ANALYSIS [all by Russell], pp. 190-214. Allen and Unwin,
Russell, Bertrand. [Co-authored work listed under Whitehead, A.N.]
[Russell], See: Harrison, Craig; Hart, W.D.;Moulder, J.
Rutter, Michael, and Nicole Madge. CYCLES OF DISADVANTAGE:
A REVIEW OF RESEARCH. London: Heinemann, 1976.
Ryle, Gilbert, "Heterologicality," ANALYSIS, 11 (1951) 61-69.
Reprinted in M. MacDonald, PHILOSOPHY AND ANLYSIS
(q.v.). (See reviews by Geach, P.T. and Orey, S.)

- S-

Suber A Bibliography of Works on Reflexivity

Sadovski, V.N. [Co-authored work listed under Blauberg, LV.]

Safir, Orin. [See work listed under GOdel]
Sagarin, Edward, and R.I. Kelly, "The Brewster Effect: Political
Trials and the Self-Defeating Prophecy," Paper presented at a
conference of the American Sociological Association, New
Orleans, 1972.
Saito, Setsuo, "Circular Definitions and Analyticity," INQUIRY, 5
(1962) 158-62.
Sandywell, Barry, et al. PROBLEMS OF REFLEXIVITY AND
THEORIZING DIFFERENCE. Routledge and Kegan Paul,
Sanford, David, "Begging the Question," ANALYSIS, 32 (1972)
197-99. (See review by Barker.)
Sanford, David, "The Fallacy of Begging the Question: A Reply to
Barker," DIALOGUE, 16 (1977) 485-98.
STOCHASTIC SYSTEMS. N.Y.: M. Dekker, 1977.
Sartre, Jean-Paul, "The Circularity of Dialectical Investigation,"
Chapter 7, Section 2 of his CRITIQUE OF DIALECTICAL
REASON. Verso/NLB, 1976.
Savage, L. Wade, "The Paradox of the Stone," PHILOSOPHICAL
REVIEW, 76 (1967) 74-79. (On God's omnipotence.)
Savan, D. [See work listed under Plato]
Schick, Frederic, "Consistency," PHILOSOPHICAL REVIEW, 75
(October 1966) 467-95.
[Schlegel, Friedrich], See: Higonnet, M.R.
[Schleiermacher, Friedrich Daniel], See: Maraldo, J.C.
Schlesinger, G., "Eliminating Self-Reference Once More,"
ANALYSIS, 29 (1969) 135-39. (A reply to O'Connor's reply.)
Schlesinger, G., "Elimination of Self-Reference," ANALYSIS, 27
(1967) 206-08. (See O'Conner, J.)
Schlesinger, G., "Omnipotence and Evil: An Incoherent Problem,"
SOPHIA, 4 (1965) 21-24.
DRAMA. Columbia University Press, 1979.
Schmidt, Paul F., "Self-Referential Justification," PHILOSOPHICAL
STUDIES, 8 (1957) 49-54.
Schoenberg, Judith, "Belief and Intention in the Epimenides,"
30 (December 1969) 270-78.
Scholes, Robert, "Imagination Dead Imagine: Reflections on

Suber A Bibliography of Works on Reflexivity

Self-Reflexive Fiction," in his FABULATION AND

METAFICfION, pp 210-18. University of Illinois Press, 1979.
Schuster, Peter. [Co-authored work listed under Eigen, M.]
Schwartz, Eugene M. "Self-Reference -Paradox . . . or Bluff?,"
published in form of xeroxed typescript sold for $1.00, by
Eugene M. Schwartz Association, 1160 Park Avenue, PH-W,
N.Y., N.Y. 10028.
Schwegler, H. [Co-edited work listed under Roth, G.]
Scriven, Michael, "An Essential Unpredictability in Human Behavior,"
in B.J. Wolman and E. Nagel (eds.), SCIENTIFIC
411-25. Basic Books, 1964.
[Scriven, Michael], Lewis, David K. and Jane Shelby Richardson,
"Scriven on Unpredictability," PHILOSOPHICAL STUDIES,
17 (Oct. 1966) 69-74.
Seebohm, Thomas, "Reflexion, Interpretation, Dialectics,"
Dissertation, Duke University, 1976.
Sextus Empiricus. AGAINST THE LOGICIANS. Loeb Classical
Library, Harvard University Press, 1961.
University of Oklahoma, 1984.
[Shakespeare], See: Calderwood, J.L.
Shaw-Kwei, Moh, "Logical Paradoxes for Many-Valued Systems,"
Shepard, Roger N., "Circularity in Judgments of Relative Pitch,"
36, 12 (Dec. 1964) 2346-53.
Shepheard, Miles. [Co-authored work listed under Marshall, J.]
AND BENIGN SPIRALS. Humanities Press, 1980.
Shiner, Roger A., "Self-Reference and the Third Man Argument,"
Shir, Jay, "Symbolism and Autosymbolism," JOURNAL OF
Suber A Bibliography of Works on Reflexivity

Shoch, John F. and Jon A. Hupp, "Notes on the 'Worm'
Programs--Some Early Experience With A Distributed
Shoemaker, Sidney, "Self-Reference and Self-Awareness," in J.G. Gill
(ed.), PHILOSOPHY TODAY No.3, pp. 32-90. Macmillan,
Shoenfield, J. R., Review of Smullyan's "Languages in which
Self-Reference is Possible," (q.v.), JOURNAL OF SYMBOLIC
LOGIC, 24 (1959) 228.
Shovokhova, E.B., "On the Reflexive Nature of Psychics," PAKISTAN
Shreider, LA. [Co-authored work listed under Braines, S.N.]
Sibajiban, "Can Doubt Be Doubted?," JOURNAL OF SYMBOLIC
LOGIC,27 (March 1962) 1114.
Sibajiban, "Mr. Eric Toms on the Liar Paradox," MIND, 74 (1965)
421-23. (See review by Bennet, J.)
Siegler, F.A., "On Lying to Oneself," JOURNAL OF PHILOSOPHY,
59 (August 1962) 469-74.
Sigler, J.A., "The Cybernetics Model of the Judicial System,"
TEMPLE LAW QUARTERLY, 41 (1968) 398.
Simon, Herbert, "Bandwagon ond Underdog Effects and the
Possibility of Election Predictions," PUBLIC OPINION
QUARTERLY, (Fa111954) 245-53.
Simos, Miriam (Starhawk). THE SPIRAL DANCE. Harper and Row,
PROBLEM. Ph.D. Dissertation, Yale University, 1979.
Sinisi, Vito F., "Leibniz's Law and the Antinomy of the Liar,"
30 (December 1969) 91-92.
[Skinner, B.F.], Hinman, Lawrence M., "Can Skinner Tell A Lie?
Notes on the Epistemological Nihilism of B. F. Skinner,"
1979) 47-60.
Skinner, John E. [See work listed under Royce]
Skinner, R. C., "The Paradox of the Liar," MIND, 68 (1959) 322-35.
Skolem, Thomas. Review of Mostwoski's "Some Impredicative
Definitions..." (q.v.), JOURNAL OF SYMBOLIC LOGIC, 16
(1951) 274-75.
Skolem, Thomas. Review of Mostowski's "Connection to the Paper... "

Suber A Bibliography of Works on Reflexivity

(q.v.), JOURNAL OF SYMBOLIC LOGIC, 18 (1953) 343.

Skyrms, Brian, "Definitions of Semantical Reference and
LOGIC, 17 (January 1976) 147-48.
Skyrms, Brian, "An Immaculate Conception of Modality," THE
JOURNAL OF PHILOSOPHY, 75,7 (July 1978) 368-387.
(Preferring a hierarchical to a self-referential interpretation of
metalinguistic modality.)
Skyrms, Brian, "Notes on Quantification and Self-Reference," in R. L.
Martin, PARADOX OF THE LIAR (q.v.).
Skyrms, Brian, "Return of the Liar: Three-Valued Logic and the
QUARTERLY, 7 (1970) 153-61.
Slezak, Peter. [See work listed under GOdel]
Sloman, Aaron, "Tarski, Frege and the Liar Paradox," PHILOSOPHY,
46 (April 1971) 133-47.
HELPING RELATIONSHIPS. Routledge and Kegan Paul,
Small, Robin. [See work listed under Nietzsche]
Smart, J,J.C. [Co-authored works listed under Mackie, J.S.]
Massachusetts Institute of Technology, 1982.
Smith, H. B., "The Theory of Multiple Implication and Its Application
to the Generalized Problem of Epimenides," BULLETIN OF
Smith, John E., "The Reflexive Turn, the Linguistic Turn, and the
Pragmatic Outcome," THE MONIST, 53 (Oct. 1969) 588-605.
Smith, John L. [See work listed under Plato]
Smith, Joseph Wayne, "Two Recent Self-Referential Arguments,"
AUSLEGUNG, 9, 3 (Winter 1982) 333-46. (Churchland, P.S.
on determinism, and Everitt, N. on materialism.)
Smith, Michael P. [See work listed under Descartes]
Smith, P. [Co-authored work listed under Buckner, D.]
Smith, S. E., "The Reflexive Turn, the Linguistic Turn, and the
Pragmatic Outcome," THE MONIST, 53, 4 (1969) 588-605.
Smorynski, Craig, "Calculating Self-Referential Statements, I: Explicit
Calculations," STUDIA LOGICA, 38 (1979) 17-36.
Smorynski, Craig, "Calculating Self-Referential Statements: Guaspari
Sentences of the First Kind," JOURNAL OF SYMBOLIC

Suber A Bibliography of Works on Reflexivity

LOGIC, 46 (June 1981) 329-44.

Smorynski, Craig, "Fifty Years of Self-Reference in Arithmetic,"
1981) 357-74.
Smorynski, Craig, "Modal Logic and Self-Reference," Essay 9 in D.
Gabbay and F. Guenther (eds.), HANDBOOK OF
CLASSICAL LOGIC, D. Reidel, 1984.
Smullyan, Raymond. ALICE IN PUZZLE-LAND. William Morrow
& Co., 1982.
Smullyan, Raymond, "Diagonalization and Self-Reference." Paper
delivered orally at conference of the Society for Philosophy and
Psychology, Chicago, April 5, 1981.
Smullyan, Raymond, and Hilary Putnam, "Exact Separation of
Recursively Enumerable Sets Within Theories,"
SOCIETY, 11,4 (1960) 574-77.
Smullyan, Raymond. 5000 B.C. --AND OTHER PHILOSOPHICAL
FANTASIES. St. Martin's Press, 1983.
Smullyan, Raymond, "Fixed Points and Self-Reference," JOURNAL
TO GODEL. Alfred A. Knopf, 1987.
ABILITY. Oxford University Press, 1987.
LOGICAL PUZZLES. Alfred A. Knopf, 1982.
Smullyan, Raymond, "Languages in Which Self-Reference is
Possible," JOURNAL OF SYMBOLIC LOGIC, 22, 1 (1957)
55-67. (See review by Shoenfield, J. R)
Smullyan, Raymond, "Logicians Who Reason About Themselves
--Theoretical Aspects of Reasoning About Knowledge,"
Kaufman Publishers, 1986, pp. 341-52.
Smullyan, Raymond. TO MOCK A MOCKINGBIRD. Alfred A.
Knopf, 1985.
Smullyan, Raymond, "Modality and Self-Reference --Intensional
Smullyan, Raymond, "Quotation and Self-Reference," in Steven J.
Bartlett and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Smullyan, Raymond. THEORY OF FORMAL SYSTEMS. Annals of
Mathematical Studies, 1961.

Suber A Bibliography of Works on Reflexivity


OF LIVING PARADOXES. Prentice-Hall, 1980.
Smullyan, Raymond, "Uniform Self-Reference," STUDIA LOGICA
(Tarski Symposium), 44, 4 (1985) 439-45.
PUZZLES. Prentice-Hall, 1978.
SolI, Ivan. [See work listed under Nietzsche]
Sommers, F., "On Concepts of Truth in Natural Languages," REVIEW
OF METAPHYSICS, 23 (1969) 259-86. (See Martin, R.L.)
Sommers, F., "Predictability," in Max Black (ed.), PHILOSOPHY IN
AMERICA. Cornell University Press, 1964. (See Martin, R.L.)
Sonnenfeld, A. [See work listed under Malraux]
Sorensen, Roy A., "Recalcitrant Variations of the Prediction Paradox,"
(December 1982) 355-62.
Spade, Paul Vincent, "General Semantic Closure," JOURNAL OF
PHILOSOPHIC LOGIC, 6 (June 1977) 209-21.
Institute of Medieval Studies, 1975.
Spade, Paul Vincent, "Ockham on Self-Reference," NOTRE DAME
JOURNAL OF FORMAL LOGIC, 15 (April 1974) 298-300.
Spanos, William V., "Heidegger, Kierkegaard, and the Hermeneutic
Circle: Towards a Postmodern Theory of Interpretation as
Dis-closure," in Spanos (ed.), MARTIN HEIDEGGER AND
Indiana University Press, 1976. Also printed in BOUNDARY 2,
4,2 (Winter 1976) 455-88.
Sparkes, A.W., "'Begging the Question'," JOURNAL OF THE
HISTORY OF IDEAS, 27 (1966) 462-63.
Spinelli, Aldo. LOOPINGS. Amsterdam: Multi Art Points Edition,
1976. (A self-describing text.)
Spitzer, A. [See work listed under Plato]
Stack, George J., "Reflexivity, Prediction and Paradox," DIALOGOS,
13 (1978) 91-101.
Stack, Michael, "Self-Refuting Arguments," METAPHILOSOPHY, 14,
3-4 (July-Oct. 1983) 327-34.
Stambaugh, Joan. [See work listed under Nietzsche]
Steel, T., "In Reply to a Paradox," MIND, 80 (Oct. 1971) 616.
Steiner, Mark. Review of Chihara's ONTOLOGY AND THE

Suber A Bibliogmphy of Works on Reflexivity

PHILOSOPHY, 72 (Apri110,1975) 184-96.

Stenius, Erik, "Semantic Antinomies and the Theory of Well-Formed
Rules," THEORIA, 36 (1970) 142-60.
[Stem, Kenneth], See: Hart, J.A. and Dees, lG.
Stem, R. H., "Priority Paradoxes in Patent Law," VANDERBILT
LAW REVIEW, 16 (1962) 131.
[Sterne, Laurence], Diamond, R.I., "Resolution of the Paradox of
Tristram Shandy," PHILOSOPHY OF SCIENCE, 31 (1964)
[Sterne, Laurence], Swearingen, James E. REFLEXIVITY IN
"TRISTRAM SHANDY". Yale University Press, 1977.
LANGUAGES. Ph.D. Dissertation, University of Texas at
Austin, 1976.
Stone, Roy, "Affinities and Antinomies in Jurisprudence,"
CAMBRIDGE LAW JOURNAL, (1964) 266-85. (Inter alia,
that GDdel's theorem applies to legal systems.)
Stone-de Montpensier, Roy L., "The Compleat Wrangler,"
MINNESOTA LAW REVIEW, 50 (1966) 1001-25. (Same
author as previous; also defends thesis that Godel's theorem
applies to law.)
Stone-de Montpensier, Roy L., "Logic and Law: The Precedence of
Precedents," MINNESOTA LAW REVIEW, 51 (1967) 655-74.
Strawson, P. F., "Paradoxes, Posits, and Propositions,"
PHILOSOPHICAL REVIEW, 76 (1967) 214-19.
[Stmwson, P. F.], See: Rotenstreich, N.
Stroll, A. "Is Everyday Language Inconsistent?," MIND, 63 (1954)
219-25. (See review by Thomson, J.F.)
Strom, James J., "On Squaring the Circles of Logic," ANALYSIS, 37
(1977) 127-29.
Stubbs, A.C. [See work listed under Descartes]
Suber, Peter, "Nomic," in Douglas R. Hofstadter, "Metamagical
Themas: About Nomic: A Heroic Game That Explores the
Reflexivity of the Law," SCIENTIFIC AMERICAN, 246, 6
(June 1982) 16-28. Reprinted, e.g., in OXYMORON 12, 12
(August 10, 1982) unpaginated, with comments by Jaap Jacobs;
slightly revised, in Douglas R. Hofstadter, METAMAGICAL
THEMAS, Basic Books, 1985.
Suber, Peter, "A Bibliography of Works on Reflexivity," in Steven J.
Bartlett and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Suber, Peter, "Logical Rudeness," in Steven J. Bartlett and Peter Suber
(eds.), SELF-REFERENCE (q.v.).
Suber, Peter. [Co-edited work listed under Bartlett, S.J.]

Suber A Bibliogmphy of Works on Reflexivity

Suits, Bernard, "Games and Pamdox," PHILOSOPHY OF SCIENCE,

36 (September 1969) 316-21.
Swearingen, James E., "Reflexivity and the Decentered Self," in Steven
J. Bartlett and Peter Suber (eds.), SELF-REFERENCE (q.v.).
Swearingen, James E. [See work listed under Sterne, L.]
Swiggert, Peter, "Quine's Logic and the Class Paradox," MIND, 84,
335 (July 1975) 321-37.
Swiggert, Peter, "Self-Reference in Formal Lanaguages," NOTRE
Swineburne, Richard, "Omnipotence," AMERICAN


Talland, G. A., "The Assessment of Group Opinion by Leaders and

Their Influence on Its Formation," JOURNAL OF
Tammelo, Ilmar, "The Antinomy of Parliamentary Sovereignty,"
(1958) 495-516. (See review by Hintikka, 1.)
Tarski, A. "The Concept of Truth in Formalized Languages," in
1923-38, pp 152-278. Oxford University Press, 1956.
Tarski, A., "The Semantic Conception of Truth and the Foundations of
RESEARCH, 4 (1944) 341-76.
Tarski, A., "Truth and Proof," SCIENTIFIC AMERICAN, 220 (1969)
[Tarski, A.], See: Stegmuller, W.; Rivetti Barbo, F.; Sloman, A.
Taubes, Jacob, "The Realm of Paradox," THE REVIEW OF
METAPHYSICS, 7 (1954) 482-91.
Taylor, Charles, "Man as Self-Interpreting Anima!." Paper delivereed
orally at Convention of American Philosophical Association,
University of California at Berkeley, March 22, 1983.
PARADOXES. Ph.D. Dissertation, Columbia University,
Teensma, E. THE PARADOXES. Assen: Van Gorcum, 1965.
Teloh, Henry, "Self-Predication of Anaxagorean Causation in Plato,"
APEIRON, 9 (1975) 15-23.
Thomas, Ivo, "The Written Liar and Thomas Oliver," NOTRE DAME
Suber A Biblio.graphy o.fWo.rks o.n Reflexivity

JOURNAL OF FORMAL LOGIC, 6 (1965) 201-08.

Tho.mas, William J. [See wo.rk listed under Plato.]
Tho.mpso.n, M. H., Jr., "The Lo.gical Parado.xes and Pierce's [sic]
Semio.tic," JOURNAL OF PHILOSOPHY, 46 (1949) 513-36.
(See review by Turquette, A.P.)
Tho.msen, Dietrick E., "Self-Assembly: <--> Physics:
and Physics Can Illuminate Fundamental Questio.ns," SCIENCE
NEWS, 121,20 (May 15, 1982) 334.
Tho.mso.n, J. F., "On Parado.xes," in R. J. Butler (ed.),
ANALYTIC PHILOSOPHY, pp. 104-19. Oxfo.rd: Basil
Blackwell, 1962. (See Anderso.n, A.R.)
Tho.mso.n, J. F., Review o.f Po.pper's "Self-Reference and Meaning ..."
(q.v.), Evans's "On Semantic Illusio.ns" (q.v.), Stro.ll's "Is
Everyday Language Inco.nsistent?" (q.v.), JOURNAL OF
SYMBOLIC LOGIC, 21 (1956) 381.
[Tho.reau], Tuerk, Richard Carl. CENTRAL STILL: CIRCLE AND
SPHERE IN THOREAU'S PROSE. The Hague: Mo.uto.n, 1975.
Tipler, Frank J., "General Relativity and the Eternal Return," in F. J.
Academic Press, 1980.
Tlumak, Jeffrey. [See wo.rk listed under Descartes]
To.dd, William. [ wo.rk listed under Rechtin, L.]
To.llefsen, Olaf, "The Equivo.catio.n Defense o.f Co.gnitive Relativism,"
in Steven 1. Bartlett and Peter Suber (eds.),
To.llefsen, Olaf. [ wo.rks listed under Bo.yle, J.M.] E., "The Liar Parado.x," PHILOSOPHICAL REVIEW, 65
(1956) 542-47. (See reviews by Bennet, J. and Sibajiban.), E., "The Reflexive Parado.xes," PHILOSOPHICAL REVIEW,
61 (1952) 557-67. (See reply by Geach.), E., "Reply to. a No.te o.n the Liar Parado.x," PHILOSOPHICAL
REVIEW, 67 (1958) 101-05. (See review by Bennet, J.)
Trigg, Geo.rge L. et at., "In Yo.ur Own Write," PHYSICAL REVIEW
LETTERS, 42 (1979) 748. (Co.llectio.n o.f self-vio.lating rules
and maxims.)
REFORMATION. University o.f Califo.rnia Press, 1979.
Tucker,, A co.mment o.n I. J. Go.o.d's "No.te o.n Richard's
Parado.x," MIND, 78 (1969) 272.
Tucker,, "Godel and Epimenides," PROCEEDINGS OF THE
ARISTOTELIAN SOCIETY, 59 (1958-59) 25-48.
Tucker,, "An Outline fo.r a New Pro.gram fo.r the Fo.undatio.ns o.f

Suber A Bibliography of Works on Reflexivity


(June-December 1969) 28-37.
Tucker, John, "Philosophical Arguments," PROCEEDINGS OF THE
ARISTOTELIAN SOCIETY, supp. vol. 39 (1965) 47-64.
Tuerk, Richard Carl. [See work listed under Thoreau]
Turquette, A. R. Review of Thompson's "The Logical Paradoxes and
Pierce's Semiotic," (q.v.), JOURNAL OF SYMBOLIC LOGIC,
16 (1951) 214-15.
Tustin, Arnold, "Feedback," SCIENTIFIC AMERICAN, 187,3 (Sept.
1952) 48-55, reprinted in SCIENCE, CONFLICT AND
H. Freeman and Co., 1969.
Tymoczko, Thomas, "An Unsolved Puzzle About Knowledge," THE
PHILOSOPHICAL QUARTERLY, 34, 137 (Oct. 1984). (On
the sentence, "No one knows that this sentence is true.")
Tyree, Alan L., "Circular Priorities in Secured Transactions,"

Ukraintsev, B. S., "Self-Controlling Processes and Causality,"
VOPROSY FILOSOFII, no. 4 (1968).
Ushenko, A. P., "An Addendum to the Note on the Liar Paradox,"
MIND, 66 (1957) 98. (See review by Bennet, J.)
Ushenko, A. P., "A Modification of the Theory of Types," THE
MONIST, 44 (1934) 147-49.
Ushenko, A. P., "A New Epimenides," MIND, 46 (1937) 549-50. (See
reviews by Encarnacion, J., and Langford, C. H.)
Ushenko, A. P., "A Note on the Liar Paradox," MIND 64 (1955) 543.
(See review by Bennet, J.)
Ushenko, A. P., "A Note on the Semantic Conception of Truth,"
(1944) 104-07.
Ushenko, A. P. THE PROBLEMS OF LOGIC. [especially Ch. 2].
Princeton University Press, 1941. (See reviews by Nelson, E.,
and Wedberg, A.)
Ushenko, A. P., "Undecidable Statements and Metalanguage," MIND,
53 (1944) 258-62. (See review by McKinsey, J.C.C.)


[Valle-Inclan], Boudreau, H. L., "Circular Structure of Va1le-Inc1an's

Suber A Bibliography of Works on Reflexivity


LANGUAGE ASSOCIATION, 82 (March 1967) 128-35.
VanCleve, J. [See work listed under Descartes]
van Fraassen, Bas C., "Inference and Self-Refereence," SYNTHESE,
21 (Oct. 1970) 425-38.
van Fraassen, Bas C., "Presupposition, Implication, and
Self-Reference," JOURNAL OF PHILOSOPHY, 65 (March
1968) 136-51.
van Fraassen, Bas C., "Presuppositions, Supervaluations, and Free
Logic," in K. Lambert (ed.), THE LOGICAL WAY OF DOING
THINGS. Yale University Press, 1969.
van Fraassen, Bas C., "Truth and Paradoxical Consequences," in R. L.
Martin, PARADOX OF THE LIAR (q.v.).
Van Heijenoort, John, "Logical Paradoxes," in Paul Edwards (ed.),
ENCYCLOPEDIA OF PHILOSOPHY, 5:45-51. Macmillan and
the Free Press, 1967.
Van Heijenoort, John (ed.). FROM FREGE TO GODEL: A SOURCE
University Press, 1967.
Varela, Francisco, "A Calculus for Self-Reference,"
(1975) 1-25.
D. Reidel, 1979.
Varela, Francisco. [Co-authored work listed under Maturana, H.R.]
Veatch, H. B., and Young, T., "Metaphysics and the Paradoxes,"
REVIEW OF METAPHYSICS, 6 (1952) 199-218.
Vetterling, Mary K., "Discussion: More on Reflexive Predictions,"
PHILOSOPHY OF SCIENCE, 43 (1976) 278-82.
Visader, John, "The Use of Paradox in Urboric Philosophies,"
PHILOSOPHY EAST AND WEST, 28 (October 1978) 455-67.
[Vitruvius], Plommer, H., "Circle of the Winds in Vitruvius i.6,"
CLASSICAL REVIEW, n.s. 21 (June 1971) 159-62.
Vlastos, Gregory. [See work listed under Plato]
Von Foerster, H., "On Self-Organizing Systems and Their
Environments," in Yovitts, M. and Cameron, S. (eds.),
SELF-ORGANIZING SYSTEMS (q.v.), 1960, pp. 31-50.
Von Foerster, H., and Zopf, G.W. Jr. (eds.). PRINCIPLES OF
SELF-ORGANIZATION. Pergamon, 1960.
Von Foerster, H. [Co-authored work listed under Pask, G.]
von Wright, G. H., "The Heterological Paradox," SOCIETAS

Suber A Bibliography of Works on Reflexivity

[von Wright, G. H.], See: Woods, 1. and Walton, D.

University of Chicago Press, 1986. (SYll'.bols in the creation of
culture and meaning systems.)
Walton, Douglas, "Are Circular Arguments Necessarily Vicious?"
1985) 263-74.
Walton, Douglas and Lynn M. Batten, "Games, Graphs and Circular
Arguments," LOGIQUE ET ANALYSE, 106 (1984) 133-164.
Walton, Douglas, "Mill and DeMorgan on Whether the Syllogism is a
FALLACIES. University Press of America, 1984.
Walton, Douglas, "The Omnipotence Paradox," CANADIAN
JOURNAL OF PHILOSOPHY, 4 (1975) 705-15.
Walton, Douglas, "Petitio Principii and Argument Analysis," in R. H.
Johnson and J. A. Blair (eds.), INFORMAL LOGIC, pp. 40-54.
Point Reyes, California: Edgepress, 1980.
Walton, Douglas. [Co-authored works listed under Woods, J.]
Wand-Schneider, Dieter, "Concerning the Antinomy Problem of
Logic," RATIO, 16, I (June, 1974) 82-102.
Wang, H., "The Irreducibility of Impredicative Principles,"
MATHEMATISCHE ANNALEN, 125 (1952) 56-66. (See
review by McNaughton, R.)
Wang, H., "Undecidable Sentences Generated By Semantic
Paradoxes," JOURNAL OF SYMBOLIC LOGIC, 20 (1955)
Webb, Judson C., "The Mechanization of Self-Reference." Paper
delivered orally at Conference of the Society for Philosophy and
Psychology, Chicago, April 5, 1981.
Wedberg, A., Review of Ushenko's THE PROBLEMS OF LOGIC.
(q.v.), PHILOSOPHICAL REVIEW, 52 (1943) 208-11.
Weinberg, J., "A Possible Solution of the Heterological Paradox,"
PHILOSOPHICAL REVIEW, 46 (1937) 657-59. (See review
by Church, A.)
Welch, Robert S. [See work listed under Descartes]
Wheeler, Samuel C. III, "Megarian Paradoxes as Eleatic Arguments,"

Suber A Bibliography of Works on Reflexivity


1983) 287-96.
Whewell, David, "Self-Reference and Meaning in a Natural Language,"
in Steven J. Bartlett and Peter Suber (eds.),
Whewell, David, "Transcendental Deductions and the A Priori
Legitimation of Knowledge Claims," PROCEEDINGS OF THE
Verlag Herbert Grundmann, 1981. (In part devoted to
self-guaranteeing and self-refuting statements.)
White, "A Problem in Priorities," OHIO LAW REPORTER, 25 (1926)
116. (On circular priorities.)
Whitehead, Alfred North, and Bertrand Russell. PRINCIPIA
MATHEMATICA. Volume I, Chapter II, Section I: "The
Vicious Circle Principle." Section II: "The Contradictions."
Cambridge University Press, 1910.
[Whitehead, A.N., and Russell, B.], Brown, James M., "Bernays's
Non-Circular Proof of the Non-Independence of the Fourth
Axiom of 'Principia Mathematica'," ANALYSIS, 36 (1976)
[Whitehead, A.N., and Russell, B.], Danquah, Joseph, "The Circularity
of the Proof of the Non-Independence of the Fourth Axiom of
'Principia Mathematica'," ANALYSIS, 36 (1976) 110-11.
Whitely, C. H., "Let Epimenides Lie!," ANALYSIS, 19 (1958) 23-24.
(See review by Bennet, J.)
Wiener, Norbert, "Cybernetics," SCIENTIFIC AMERICAN, 179, 5
(Nov. 1948) 14-19. Reprinted in SCIENCE, CONFLICT AND
W. H. Freeman and Co., 1969.
MACHINE. Massachusetts Institute of Technology Press, 2d ed.
Wiley, Margaret L. [See work listed under Melville]
Williams, John N., "Justified Belief and the Infinite Regress
18,1 (Jan. 1981).
Williams, M.E., "Begging the Question," DIALOGUE, 6 (1968)
567-70. (Reply to Johnson, O.A.)
Wilson, R.R., "This Is Not a Meta-Review of Three Books on
Metafiction, But What Account Should Be Given of a
Self-Referential Title," CANADIAN REVIEW OF
COMPARATIVE LITERATURE, 12,2 (June 1985) 292,.305.

Suber A Bibliography of Works on Reflexivity

Witemeyer, H.H. [See work listed under Emerson]

[Wittgenstein], Dumitriu, Anton, "Wittgenstein's Solution of the
12 (1974) 227-37.
[Wittgenstein], Goldstein, Laurence, "Wittgenstein and the
Logico-Semantical Paradoxes," RATIO, 25, 2 (Dec. 1983).
[Wittgenstein], Miller, Richard W., "Solipsism in the TRACfATUS,"
Section: "From the Denial of Self-Reference to the Denial of
Other Minds," pp. 60-65; Section: "Psychic Representation and
the Denial of Self-Reference," pp. 65-70, JOURNAL OF THE
HISTORY OF PHILOSOPHY, 18, 1 (Jan. 1980) 57-74.
[Wittgenstein], Worthington, B.A., "Ethics and the Limits of Language
in Wittgenstein's TRACTATUS," Section 4: "Self-Reference
and Self-Consciousness," pp. 494-96, JOURNAL OF THE
HISTORY OF PHILOSOPHY, 19,4 (1981) 481-96.
[Wittgenstein, Ludwig], See: Chihara, C.S.
Wolf, Eleanor, "The Invasion-Succession Sequence as a Self-fulfilling
Prophecy," JOURNAL OF SOCIAL ISSUES, 13 (Oct. 1957)
Wolfe, Julian, "Omnipotence," CANADIAN JOURNAL OF
PHILOSOPHY, 1 (1971) 245-48.
Wolock, F.W. [Co-authored work listed under John, J.A.]
University Press, 1941. (Discusses self-referentially
inconsistent arguments, pp. 193-203.)
Woodruff, Peter W., "Paradox and Truth Value Gluts," paper
delivered orally at Convention of the American Philosophical
Convention, March 22, 1984, Long Beach.
Woodruff, Peter W., "Paradox, Truth, and Logic. Part I: Paradox and
Woodruff, Peter W. [