Professional Documents
Culture Documents
Episteme
A SERIES IN THE FOUNDATIONAL,
METHODOLOGICAL, PHILOSOPHICAL, PSYCHOLOGICAL,
SOCIOLOGICAL, AND POLITICAL ASPECTS OF THE SCIENCES, PURE AND APPLIED
VOLUME 16
The titles published in this series are listed at the end of this volume.
FRANK PLUMPTON RAMSEY
On Truth
Original Manuscript Materials (1927-1929)
from the Ramsey Collection
at the University of Pittsburgh
edited by
NICHOLAS RESCHER
University of Pittsburgh, U.S.A.
and
ULRICH MAJER
Georg-August University, Gottingen, Germany
..
SPRINGER SCIENCE+BUSINESS MEDIA, B.V
'Library of Congress Cataloging-in-Publication Data
ISBN 978-94-010-5662-5
Preface Vll
Editor's Introduction IX
ON TRUTH
Intrc:1uctory 81
I. The Nature of Truth 84
II. The Coherence Theory of Truth 95
III. Judgment 98
V. Judgment and Time 103
Frank Plump ton Ramsey (22 February 1903 -19 January 1930) was an extra-
ordinary scholarly phenomenon. Son of a distinguished mathematician and
President of Magdalene College, Cambridge and brother of Arthur Michael,
eventual Archbishop of Canterbury, Ramsey was closely connected with
Cambridge throughout his life, ultimately becoming lecturer in Mathematics
in the University. Notwithstanding his great mathematical talent, it was
primarily logic and philosophy that engaged his interests, and he wrote original
and important contributions to logic, semantics, epistomology, probability
theory, philosophy of science, and economics, in addition to seminal work
in the foundations of mathematics. His original editor spoke the unvarnished
truth in saying that Ramsey's premature death "deprives Cambridge of one
of its intellectual glories and contemporary philosophy of one of its profoundest
thinkers,"l and J.M. Keynes characterized one of his papers as "one of the
most remarkable contributions to mathematical economics ever made."2
Considering the scope and variety of his achievements, it is astonishing to
contemplate that Ramsey died Uust) before attaining his 27th birthday.
In 1982 the University of Pittsburgh acquired through the mediation of
Nicholas Rescher a substantial collection of Ramsey's manuscripts, consisting
of notes, lectures, and various unfinished writings. The collection was
augmented in 1986 by a gift of further autograph material from Ramsey's
daughter, Mrs. Jane Burch, who generously vested the publication rights to
the Ramsey manuscripts in the University of Pittsburgh. With these subsequent
additions, the collection currently (1988) comes to some 120 items amounting
to some 1200 holograph pages. This Ramsey Collection forms part of the
Archives of Scientific Philosophy in the Twentieth Century housed in the
University's Hillman Library, and bears an impressive silent witness to the
astonishing versatility and intellectual power of this English prodigy.
The Ramsey Collection consists of manuscripts of variable time of origin,
x EDITOR'S INTRODUCTION
length, and value. The earliest manuscripts are notes and essays dating [rom
his undergraduate days; the latest represent substantial writings on which
he was still working at the time of his death. The shortest manuscripts are
brief notations and memoranda of only a few lines; the longest are substantial
chapters or essays ranging in length up to a manuscript of 45 pages devoted
to issues on The Foundations of Mathematics, which was published under that
title as the lead essay in the Braithwaite anthology of Ramsey's work. 3 The
least significant papers are notes and abstracts on books Ramsey was reading;
the most valuable are drafts or even virtually completed versions of researches
that Ramsey himself destined for publication.
Topically, the collection covers the whole diverse range of Ramsey's
extensive interests, ranging from logic, probability, and the foundations of
mathematics to the theory of knowledge and the philosophy of language
and beyond to issues of economics and social philosophy. The overall spectrum
has a rather wider range than is represented in the Ramsey material published
to date (there is, for example, an essay on sex, sexuality, and perversion).
But even when the same ground is covered, the contentions and views put
forward in the manuscript often supplement and sometimes differ significantly
from those that have seen the light of print.
Although the manuscripts only rarely carry notations of date, they can
in most cases be dated at least approximately, even without using physical
evidence of paper or ink. Internal evidence apart, Ramsey often worked in
interaction with the published literature and treats issues whose topicality
can be dated from other sources.
The manuscripts present no graphological difficulties. They are written
in a good English school hand, without use of special notations or abbreviations.
The distribution and ordering of the manuscripts in the elegant cloth-bound
boxes in which most of them reached the University of Pittsburgh is neither
chronological or topical. To some extent, the material was arranged by genre,
but no very systematic order was established. Ramsey himself was certainly
not responsible for this arrangement; such order as there is is due to those
who collected his scattered papers together after his death.
deal with a connected and clearly interrelated way with one overarching
question: the nature of truth. Unlike the other materials in the collection,
they constitute a single large-scale thematic unit.
The box contains eighteen folders of manuscripts in all. Two of them
(nos. 10 and 16) consist of brief memoranda. The rest, however, constitute
a unifed whole. The thematic of the material and its complex interrelationship
is set out in Appendix A to this Introduction.
The overall project bears the title "Truth and Probability," as Ramsey
himself indicated in the Alpha table of contents. For Ramsey presented several
somewhat divergent table-of-contents sketches for the project, which are set
out in Appendix B. The latest of them (Delta) agrees with the manuscript
material as we have it, except for one missing final chapter on "The Object
of Judgment. "
The development of Ramsey's overall project was foreshadowed in the
1926 essay "Truth and Probability," subsequently published by Braithwaite,
whose table of contents stood as follows:
1. The Frequency Theory
2. Mr. Keynes's Theory
3. Degrees of Belief
4. The Logic of Consistency [= Probability Logic]
5. The Logic of Truth [= Inductive Logic]
As these rubrics indicate, this range of themes largely coincides with the
second (probabilistic) part of the projected work on Truth and Probability in
the version of the Alpha table of contents, which is roughly contemporary
(1926127). The book was presumably planned to extend the materials of this
essay into new areas (chance, causality), as well as adding a wholly new
initial treatment of truth, judgment, and logic.
It is thus clear that Ramsey initially contemplated a very extensive project
- a substantial book on Truth and Probability with an initial section on semantics
and epistemology, a midsection on logic, and a final section on probability
and induction. However, with the passage of time, and the unfolding realization
of the initial parts of the project, its character changed. The initial, semantic-
epistemological section expanded from three chapters into six. The mid-section
on logic also expanded substantially (compare Alpha with Gamma) and
increasingly assumed a life of its own as Ramsey came to project a logic
book for which we also possess a projected table of contents (see MS 002-
22-03). And with the expansion of the first two sections, the third section
on probability, which had given the initial impetus to the whole project,
was left aside. 4 The newer interests became absorbing and displaced the old.
Ramsey's original project On Truth and Probability thus evolved into two: a
book On Truth and a treatise on Logic.
XlI EDITOR'S INTRODUCTION
3. INTELLECTUAL DEVELOPMENT
In point of time, Ramsey's work on the project On Truth fell during the
period from 1927 to early 1929. The manuscript itself makes various references
to material published in 1927 and 1928, and has for its latest citation the
paper on "The Growth of the Perception of the External World" by H.W.B.
Joseph in the January 1929 issue of Mind (see Chapter IV, footnote 15). Further
evidence for this dating is given by the close kinship between the issues
treated here and those treated in the "Fact and Propositions" paper of 1927
on the one hand and the so-called "Last Papers" of 1929 on the other (both
published by Braithwaite).
In both On Truth and in "Facts and Propositions" Ramsey defends his
redundancy theory of truth: "It is true that Caesar was murdered" means
no more than that Caesar was murdered, and "It is false that Caesar was
murdered" means that Caesar was not murdered" ("Facts and Propositions,"
p. 42). Indeed, Ramsey's commitment to this theory dates from his under-
graduate days and forms an ongoing Leitmotiv of his thought. 5 But in On
Truth Ramsey went on to maintain that the redundancy theory is not only
compatible with a correspondence theory of truth but actually constitutes
the heart and core of such a theory.
Internal evidence marks that "On Truth" as something of a way-station
in Ramsey's transit from his early logicism via a pragmatic position towards
the version of intuitionism he favored at the end of his brief life. For despite
its close relationship to the 1926 "Truth and Probability" essay published
by Braithwaite, the 1927-28 project On Truth betokens a significant shift
in Ramsey's thought. In "Truth and Probability" Ramsey viewed induction
as a useful habit which 'is reasonable because the world is so constituted
that inductive arguments lead on the whole to true opinions" (p. 197). (Note
that Ramsey thus foreshadows Braithwaite's own inductive justification or
vindication.) In On Truth, however, he accepts the pragmatic standpoint more
extensively and sees his own theory of truth as supporting a Peircean
justification of induction. Thus On Truth stands far closer to Ramsey's 1927
publication "Facts and Prospositions" than to his 1926 essay on "Truth and
Probability.' ,
The project On Truth shows the beginnings of Ramsey's conversion to
finitism and intuitionism as manifested in the "Last Papers." In "The Foundation
of Mathematics" (1925) and "Mathematical Logic" (1926), Ramsey endorsed
the logicist standpoint of Russell and Wittgenstein which regarded universal
and existential statements as propositions and specifically as infinite conjunctions
or disjunctions of particular statements. This is a view with which Ramsey
is now no longer satisfied, and which he later rejects explicitly in "General
Propositions and Causality" (1929). In On Truth Ramsey inclined to the
XIV EDITOR'S INTRODUCTION
"holistic" view of Norman Campbell and Heinrich Hertz that theories (and
the universal and existential statements they comprise) are not to be seen
as propositions subject to the tertium non datur principle of having to be either
true or false. Rather, they are "secondary systems" (to use Campbell's expression),6
propositional superstructures that are only capable of highly indirect veri-
fication, and are best characterized in terms of probability rather than truth-
status. In this respect, then, On Truth marks a break between the earlier
logicist Ramsey of "Facts and Propositions" and the later intuitionist Ramsey
of the "Last Papers. "
4. HISTORICAL CONTEXT
The 1927-1929 period when Ramsey wrote the material On Truth was not
a propitious time for a philosophical treatment of this subject. Between the
work of Russell and Wittgenstein in the period up to the close of W orId
War I and the publication in 1935 of Alfred Tarski's monumental "Der
Wahrheitsbegriff in den formalisierten Sprachen," the field belonged pred-
ominantly to the representatives of neo-Hegelian idealism. In the 1920's, truth
was simply not a theme that engaged the best philosophical minds of the
day. This circumstance, combined with the difficulty of producing, and indeed
even of seeing the need for constructing, a conceptually appropriate definition
of the expression "true statement" such as Tarski was to produce in 1935
was not as yet appreciated. People were still inclined to regard as mere
puzzles the semantic antinomies for whose resolution such a definition was
needed. Those few theorists who appreciated the significance of the issues
were inclined to see them as insuperably difficult. In Cambridge, a G.E.
Moorean attitude prevailed, anticipated by the thesis of Frege's Logical
Investigations: "It thus appears likely that the content of the word "true" is
altogether unique and indefinable." The new-model mathematical philosophers
did not yet see the elucidation of truth as a pressing issue and inclined to
leave the matters to the more traditional philosophers (then principally
idealists). Against this background it is less surprising that Braithwaite did
not appreciate the importance of Ramsey's project than that Ramsey himself
was impelled to pursue it with such vigor and commitment.
It is in fact remarkable how close Ramsey came to anticipating Tarski's
theory of truth - in spirit if not in letter:
We can say that a belief is true if it is a belief that p, and p. This definition
sounds odd because we do not at first realize that "p" is a variable sentence
and so should be regarded as containing a verb; "and p" sounds nonsense
because it seems to have no verb and we are apt to supply a verb such
EDITOR'S INTRODUCTION xv
5. EDITORIAL PRELIMINARIES
NICHOLAS RESCHER
ULRICH MAJER
XVlll EDITOR'S INTRODUCTION
NOTES
Lastly we come to truth, which I shall deal with as briefly as possible. The most certain
thing about truth is that 'p is true' and 'p', if not identical, are equivalent. This enables
us to rule out at once some theories of truth such as that "to be true" means "to
work" or "to cohere" since clearly 'p works' and 'p coheres' are not equivalent to
'p '. There are I think only three sensible theories of truth (1) that a true belief is defined
to be one which has a certain relation with a fact, (2) truth is indefinable and has
no connection with a relation between belief and fact, (3) truth is indefinable but as
a matter of fact true beliefs do have a certain relation to facts which false beliefs do
not have.
(3) I think we can dismiss; we have seen no reason to suppose there are facts and
if truth be indefinable, I think none can be drawn from the nature of truth; so if
truth be indefinable we have no reason to suppose there are facts and therefore no
reason for thinking true beliefs are related to facts in a way which false ones are not.
Of course in the special case of events we have seen that some true propositions and
therefore true beliefs are related to events in a way false ones are not, since false
beliefs have no quasi-subjects.
When he wrote this, Ramsey was still an undergraduate. (He took the Tripos examination
in June of 1923.)
6. Ramsey does not actually use Campbell's term until his 1929 paper on "Theories," but
already here he is drawn into the range of the "holistic" consideration of Campbell and
Hertz that constitute its natural habitat.
7. Frege's seeming self-contradiction in maintaining both the redundancy thesis and the
indefinability of truth is removed by the consideration that a merely contextual definition
is at issue which, as such, explains the use of the term rather than establishing its conceptual
content.
S. Introduction, 001-1S version, ad fin.
EDITOR'S INTRODUCTION XlX
APPENDIX A
001-09 Older version of Chap. II Chapter II. The Coherence Theory of Truth
001-10 Assorted notes and jottings
001-12 Chap. IV in its sole version Chapter IV. Knowledge and Opinion
001-13 Newer version of Chap. V Chapter V. Judgment and Time
001-14 Older Version of Chap. V Judgment and Time (The Problem of
Judgment)
001-15 A misconceived defense of the
coherence theory from the
philosophy of science
001-16 Assorted notes and jottings
001-17 Introduction to a treatise on Chapter I. Logical Values
logic
001-18 Older version of the preceding Chapter II. Introductory
introduction
xx EDITOR'S INTRODUCTION
APPENDIX B
Appendix B (continued)
Generality further
considered
Implication
Causality Causal Laws
Chance
Attempts to Justify
Induction
Nature of Degrees of Belief
Knowledge and of
Rational Belief
Scope of Inductive
Logic
[Theory of Statistics1
Yet another table of contents, which stands close to Alpha in its concern with probability,
is given in MS 002-22-01. It runs as follows:
1. Belief and Truth 6. Meaning of If
2. Terms 7. Causality
3. Logic and Consistency 8. Chance
4. Partial Belief 9. Justification of Induction
5. Keynes' Probability Relations 10. Rational Belief
ON TRUTH
001-05--0/
1""~ 96v11TVV.> u.~"", Iv. c(1i,;. ... fi"3 rA.~ Ilk it, llii 10 111";;6
f/;,/Lt }lAtf~_o.iJ 1""-y.4. va., ..... to. rk 1___ MId}- Iv"Y ..... ~ "' .... ,u
we I<o~ r. ~l~$ ~I>\ _ H-!J>v .........1 I(e..ir.1 "'~I-',< ,,,.,, 4 ......
....r It" lie t. .. ltf"? ~.. ". e~F';" Ii ........ ,je.. ... l.e{ ' / t""", ""<.
'" )..J,?/ .. ~ Ik"f Iu: u i ... /~ (~J~rl,~ ... ;- "'''Y I<! SO va.)<&( ...,
hoI- 6.. tfA.;'IALIi~ <..t4,. fi......, <11 ;"'"H)'. B. ... r Ii... (,<, .. ,,1-1(-
)C_ F", ~ tt..... tt....'? cI)- v..~ '" ut.U H.I!. t.. (--- .A
~ W.,.....,6' ~l".",ko{ tf) (J&i~,-I7"'l,sl"~ ... -sy.tlt ... DJ' ""'~ J",dy'",,,tI'"
Iv"",Lc! ~c. l..jJ. v'<)"f. t.f.." it.. IIU. ... , , " ' , Ht{ .." II...)' " .. t""II-,<t
It,
/A,<f~ ~kL4 C\. lYl1it ... t.... ).. ..til. ") 1;.../1 I....... I..,~ cl.'-)t-~.Q-14,. ... ,'I.
~ ..... J~ ~ o~st. .,-~ Itt; J:~ Ir; 1t6$,,1.. lQ:" t;: ... t( c.~~ 'f"r/~i"
-.' . "'-')/oo.o.t--~
)< (',9 -Ail )'K.I't kO(":I)''' Utw,;dii..~ '1 '''-t./.. ')'-"-fAa.1 A ... s ~
'A-it 6Q.ft/ ~", ~ ~ /,=(';'4,"
V'Q.'~IUI$
<INTRODUCTION>!
and the truth of opinions, but it is not easy to say exactly what the relation
is. It is not so simple as it would be if true opinions were never based on
unsound arguments, and inferences of high probability never led to error.
But whatever the relation may be, the fact remains that the primary subject
of the logician is inferences or arguments, not opinions or statements, and
his predicate of value is rationality not truth.
Nevertheless, before coming to his real point the logician is bound to
begin by preliminary investigations into the nature and forms of opinions
and statements, which must be conceded to belong properly to psychology
since they are concerned not with values but with the actual characters of
mental processes. Since, however, psychologists grossly neglect the aspects
of their subject which are most important to the logician, they are commonly
regarded as belonging to logic, and logic as the term is ordinarily used consists
to a great extent of psychology. In the same way, students of ethics and
aesthetics are obliged to undertake for themselves all sorts of psychological
preliminaries.
The three normative sciences: Ethics, Aesthetics and Logic begin, then,
with psychological investigations which lead up, in each case, to a valuation,
an attribution of one of the three values: good, beautiful, or rational, predicates
which appear not to be definable in terms of any of the concepts used in
psychology or positive science. I say 'appear' because it is one of the principal
problems of philosophy to discover whether this is really the case [whether,
that is to say, 'good', 'beautiful', 'rational' (or for that matter 'true') represent
undefinable qualities which ... ].
It is, of course, possible to take one view in regard to one kind of value
and the other view with regard to the other kinds; it could be held, for
instance, that whereas goodness and beauty could be defined in terms of
our desires and admirations, rationality introduced some new element peculiar
to logic, such as indefinable probability relations. But the arguments that
can be used are so much the same, that when the alternatives that can be
used are clearly stated, any normal mind is likely to make the same choice
in all three cases. It would be out of place to discuss goodness and beauty
in a book on 10gic,3 but it will be one of my chief objects to show that
the view, which I take of them, that they are definable in [ordinary factual]
natural terms, is also true of rationality and truth: so that just as ethics and
aesthetics are really branches of psychology, so also logic is part, not exactly
of psychology, but of natural science in its widest sense, in which it includes
psychology and all the problems of the relations between man and his
environment. 4 But this is not a matter which can be settled in advance: logic
tries to discover what inferences are rational; we all have some idea as to
what this means, but we cannot analyse it exactly until we have made
considerable investigations, which are commonly regarded as belonging to
5 INTRODUCTION
logic which is expected not merely to determine the application but also
the analysis of its standard of value.
NOTES
1. <Ramsey himself gave the rubric: Chapter I; Logical Values and in the earlier version Chapter I:
Introductory. That it was intended as the opening of the Logic book is evidenced by Ramsey's
table of contents for this project.>
2. Besides this we can of course criticize its verbal expression as felicitous or otherwise,
and we can say that it is important or trivial.
3. For an excellent treatment of these ideas see Mr LA. Richard's Principles of Literary Criticism
(London, 1924). See also Professor G.E. Moore's Principia Ethica (Cambridge, 1903).
4. [Logic cannot be altogether contained in psychology, because the soundness of our thought
depends on that thought agreeing with its object, and hence in part on the properties
of the object, to an extent not in general regarded as belonging to psychology.J
(001-02)
CHAPTER I
we can take it without analysis as something with which we are all familiar.
Propositional reference is not, of course, confined to beliefs; my knowledge
that the earth is round, my opinion that free trade is superior to protection,
any form of thinking, knowing, or being under the impression that - has
a propositional reference, and it is only such states of mind that can be either
true or false. Merely thinking of Napoleon cannot be true or false, unless
it is thinking that he was or did so and so; for if the reference is not propositional,
if it is not the sort of reference which it takes a sentence to express, there
can be neither truth nor falsity. On the other hand not all states which have
propositional reference are either true or false; I can hope it will be fine
to-morrow, wonder whether it will be fine to-morrow, and finally believe
it will be fine to-morrow. These three states all have the same propositional
reference but only the belief can be called true or false. We do not call
wishes, desires or wonderings true, not because they have no propositional
reference, but because they lack what may be called an affirmative or assertive
character, the element that is present in thinking that, but absent in wondering
whether. In the absence of some degree of this character we never use the
words true or false, though the degree need be only of the slightest and
we can speak of an assumption as true, even when it is only made in order
to discover its consequences. For states with the opposite character of denial
we do not naturally use the words true or false, though we can call them
correct or incorrect according as beliefs with the same propositional reference
would be false or true.
The mental states, [then], with which we are concerned, those, namely,
with propositional reference and some degree of the affirmative character,
have unfortunately no common name in ordinary language. There is no term
applicable to the whole range from mere conjecture to certain knowledge,
and I propose to meet this deficiency4 by using the terms belief and judgment
as synonyms to cover the whole range of [mental] states in question [although
this involves a great widening of their ordinary meanings] and not in their
ordinary narrower meanings.
It is, then, in regard to beliefs or judgments that we ask for the meaning
of truth and falsity, and it seems advisable to begin by explaining that these
are not just vague terms indicating praise or blame of any kind, but have
a quite definite meaning. There are various respects in which a belief can
be regarded as good or bad; it can be true or false, it can be held with
a higher or a low degree of confidence, for good or for bad reasons, in
isolation or as part of a coherent system of thought, and for any clear discussion
to be possible it is essential to keep those forms of merit distinct from one
another, and not to confuse them by using the word "true" in a vague way
first for one and then for another. This is a point on which ordinary speech
is sounder than the philosophers; to take an example of Mr Russell's, some-
THE NATURE OF TRUTH 9
one who thought that the present Prime Minister's name began with B would
think so truly, even if he had derived his opinion from the mistaken idea
that the Prime Minister was Lord Birkenhead; and it is clear that by calling
a belief true, we neither mean nor imply that it is either well-grounded
or comprehensive and that if these qualities are confused with truth as they
are, for instance, by Bosanquet,5 any profitable discussion of the subject becomes
impossible. The kind of merit in a belief to which we refer in calling it
true can be easily seen to be something which depends only on its propositional
reference;6 if one man's belief that the earth is round is true so is anyone
else's belief that the earth is round, however little reason he may have for
thinking so.
After these preliminaries we must come to the point: what is the meaning
of 'true'? It seems to me that the answer is really perfectly obvious, that
anyone can see what it is and that difficulty only arise when we try to say
what it is, because it is something which ordinary language is rather ill-
adapted to express.
Suppose a man believes that the earth is round; then his belief is true
because the earth is round; or generalising this, if he believes that A is B
his belief will be true if A is B and false otherwise.
It is, I think, clear that in this last sentence we have the meaning of
truth explained, and that the only difficulty is to formulate this explanation
strictly as a definition. If we try to do this, the obstacle we encounter is
that we cannot describe all beliefs as beliefs that A is B since the propositional
reference of a belief may have any number of different more complicated
forms. A man may be believing that all A are not B, or that if all A are
B, either all Care D or some E are F, or something still more complicated.
We cannot, in fact, assign any limit to the number of forms which may
occur, and must therefore be comprehended in a definition of truth; so that
if we try to make a definition to cover them all it will have to go on forever,
since we must say that a belief is true, if supposing it to be a belief that
A is B, A is B, or if supposing it to be a belief that A is not B, A is not
B, or if supposing it to be a belief that either A is B or C is D, either
A is Bore is D, and so on ad infinitum.
In order to avoid this infinity we must consider the general form of a
propositional reference of which all these forms are species; any belief whatever
we may symbolise as a belief that p, where 'p' is a variable sentence just
as 'A' and 'B' are variable words or phrases (or terms as they are called
in logic). We can then say that a belief is true if it is a belief that p, and
pJ This definition sounds odd because we do not at first realize that 'p'
is a variable sentence and so should be regarded as containing a verb; "and
p" sounds nonsense because it seems to have no verb and we are apt to
supply a verb such as "is true" which would of course make nonsense of
10 CHAPTER I
It is really obvious that these statements are all equivalent, in the sense that
it is not possible to affirm one of them and deny another without patent
contradiction; to say, for instance, that it is true that the earth is round
but that the earth is not round is plainly absurd.
Now the first statement of the three does not involve the idea of truth
in any way, it says simply that the earth is round. [In the second we have
to prefix "It is true that" which is generally added not to alter the meaning
but for what in a wide sense are reasons of style [and does not affect the
meaning of the statements J. Thus we can use it rather like "although" in
conceding a point but denying a supposed consequence, "It is true that the
earth is round, but still ... ," or again we often use it when what we say
has been questioned: "Is that true?" "Yes, it is perfectly true." But in the
last case the phrase "it is true that the earth is round" is changing from
simply meaning that the earth is round ... ]
The meaning of the second, on the other hand, is less clear: it may be
a mere synonym for the first, but more often contains some reference to
the possibility of someone believing or saying that the earth is round. We
are thinking not merely that the earth is round, but that because it is round
anyone ll who believes or says that it is round believes or says truly. We
have passed from the first of our statements to the third. But the third amounts
in a sense to no more than the first, and it is merely the first thought of
THE NATURE OF TRUTH 13
find out how and in what sense those images or ideas in Mr Jones' mind
at 10 o'clock constitute or express a "belief that the earth is flat." Truth,
it will be said, consists in a relation between ideas and reality, and the use
without analysis of the term propositional reference simply conceals and shirks
all the real problems that this relation involves.
This charge must be admitted to be just, and an account of truth which
accepts the notion of propositional reference without analysis cannot possibly
be regarded as complete. For all the many difficulties connected with that
notion are really involved in truth which depends on it: if, for instance,
"propositional reference" has quite different meanings in relation to different
kinds of belief (as many people think) then a similar ambiguity is latent
in 'truth' also, and it is obvious that we shall not have got our idea of truth
really clear until this and all similar problems are settled.
But though the reduction of truth to propositional reference is a very
small part and much the easiest part of its analysis, it is not therefore one
which we can afford to neglect. [Not only is it essential to realize that truth
and propositional reference are not independent notions requiring separate
analysis, and that it is truth that depends on and must be defined via reference
not reference via truth. ]14 For not only is it in any event essential to realise
that the problem falls in this way into two parts,15 the reduction of truth
to reference and the analysis of reference itself, and to be clear which part
of the problem is at any time being tackled, but for many purposes it is
only the first and easiest part of the solution that we required; we are often
concerned not with beliefs or judgments as occurences at particular times
in particular men's minds, but with, for instance, the belief or judgment "all
men are mortal"; in such case the only definition of truth we can possibly
need is one in terms of propositional reference, which is presupposed in the
very notion of the judgment "all men are mortal"; for when we speak of
the judgment "all men are mortal" what <we> are really dealing with is
any particular judgement on any particular occasion which has that pro-
positional reference, which is a judgment "that all men are mortal." Thus,
though the psychological difficulties involved in this notion of reference must
be faced in any complete treatment of truth, it is well to begin with a definition
which is sufficient for a great many purposes and depends only on the simplest
considerations.
And whatever the complete definition may be, it must preserve the evident
connection between truth and reference, that a belief "that p" is true if
and only <if> p. We may deride this as trivial formalism, but since we cannot
contradict it without absurdity, it provides a slight check on any deeper
investigations that they must square with this obvious truism.
THE NATURE OF TRUTH
NOTES
1. How difficult the problem is may be judged from the fact that in the years 1904-25
Mr Bertrand Russell has adopted in succession five different solutions of it.
2. I use "state" as the widest possible term, not wishing to express any opinion as to
the nature of beliefs etc.
3. Or, of course, that something is not so and so, or that if something is so and so, something
is not such and such, and so on through all the possible forms.
4. [It should perhaps be remarked that the late Professor Cook Wilson held that these
mental states do not in fact belong ... J It should, however, be remarked that according
to one theory this is not really a deficiency at all, since the states in question have
nothing important in common. Knowledge and opinion have propositional reference in
quite different senses and are not species of a common genus. This view, put forward
most clearly by J. Cook Wilson, (but also implied by others, e.g., Edmund Husserl)
is explained and considered below.
5. Bernard Bosanquet, Logic, 2nd ed., Vol II (Oxford, 1911), pp. 282 ff. Of course he sees
the distinction but he deliberately blurs it, arguing that an account of truth which enables
an ill-grounded statement to be true, cannot be right. His example of the man who
makes a true statement believing it to be false, reveals an even grosser confusion. He
asks why such a statement is a lie, and answers this by saying that "it was contrary
to the system of his knowledge as determined by his whole experience at the time."
Granting this, it would at most follow that coherence with the man's system of his
knowledge is a mark not of truth (for ex hypothesi such a statement would have been
false) but of good faith; and this is brought in as an argument in favour of a coherence
theory of truth!
6. It has been suggested by Professor Moore ("Facts and Propositions," Proceedings of the
Aristotelian Society, Supplementary Volume VII (1927), pp. 171-206; see p. 178) that the
same entity may be both a belief that (say) the earth is round and a belief that something
else; in this case it will have two propositional references and may be true in respect
of one and false in respect of the other. It is not to my mind a real possibility, but
everything in the present chapter could easily be altered so as to allow for it, though
the complication of language which would result seems to me far to outweight the
possible gain in accuracy. See ...
7. In Mr Russell's symbolism
8. In a sentence like this "in fact" serves simply to show that the oratio obliqua introduced
by "he believed" has now come to an end. It does not mean a new notion to be analyzed,
but is simple a connecting particle.
9. See below. <Presumably this is a reference to the unwritten chapter on negation.>
10. Metaphysics, Gamma, 6 1011b25, Mr Ross' translation.
11. For instance the man we are talking to may have just made the point and we concede
it. "Yes, it's true, as you say, that the earth is round, but -" or we may have made
it and be questioned "Is that true, what you were saying, that the earth is round?"
"Yes, it's perfectly true."
12. Thus according to William James a pragmatist could think both that Shakespeare's plays
were written by Bacon and that someone else's opinion that Shakespeare wrote them
might be perfectly true "for him." ("The Meaning of Truth," p. 274.) On the idea
that what is true for one person may not be true for another see below.
16 CHAPTER I
13. See Kritik der reinen Vernunft, "Die transzendentale Logik." Einleitung III (A57=B82): "Die
alte und beriihmte Frage ... Was ist Wahrheit? Die Namenerklarung der Wahrheit, dass
sie namlich die Obereinstimmung der Erkenntnis mit ihrem Gegenstande sei, wird hier
geschenkt und vorausgesetzt; man verlangt aber zu wissen, welches das allgemeine und
sichere Kriterium der Wahrheit einer jedem Erkenntnis sei." The reason why there can
be no such criterion is that every object is distinguishable and therefore has something
true of it which is true of no other object. Hence there can be no guarantee of truth
irrespective of the object in question.
14. [This might perhaps be denied if reference were something essentially different in the
cases of true and of false beliefs; e.g., if the precise way in which a man's belief today
that it will be wet tomorrow was a belief "that it will be wet to-morrow" depended
on how to-morrow's weather actually turned out. But this is absurd for it would allow
us to settle the weather in advance by simply considering the nature of the prophet's
expectation and seeing whether it had true-reference or false-reference.]
15. It might possibly be questioned whether this division of the problem is sound, not because
the truth of a belief does not obviously depend on its reference, i.e., on what is believed,
but because reference might be essentially different in the two cases of truth and falsity,
so that there were really two primitive ideas, true-reference and false-reference, which
had to be separately analysed. In this case, however, we could tell whether a belief
that A is B were true or false, without looking at A by simply seeing whether the
manner in which the belief was a "belief that A is B" was that of true-reference or
false-reference, and infer with certainty that to-morrow would be fine from the fact
that someone believed in a particular way, the way of false-reference, that it would
be wet. See below.
APPENDIX: TWO VARIANT ELABORATIONS
(001-04)
Objections to such a definition are of two kinds, [merely] formal and serious.
By a formal objection I mean one which does not deny that a belief that
A is B is true when and only when A is B, but says that this, though a
correct statement, is not the right definition or explanation of truth. For
instance it might be argued that a true belief is a belief in a true proposition,
and that though the proposition 'A is B' is true, when and only when A
is B, yet for the proposition 'A is B' to be true and for A to be B, are
two different (though equivalent) facts, and that the former, not the latter,
is the one which should be used in defining true belief. [This sort of niggling]
These formal questions I propose to leave till later, and devote the present
chapter to examining the objections of a more serious sort, which deny that
a belief is true if and only if it is a belief "that p" and p, and propose
instead definitions such as that a belief is true if it is useful, or if it is part
of a coherent system, definitions which are not merely formally but materially
at variance with ours.
I must confess I find it hard to see how anyone can dispute that a belief
that A is B is true if and only if A is B. Let us take the three sentences.
It is clear that the first two of these are equivalent; anyone who said "The
earth is round but it is not true that it is," or "It is true that the earth
is round but the earth is not round" would be contradicting himself. I think
they have, in general, the same meaning, and that when we use "It is true
that the earth is round" in preference to the shorter formulation we do
so for what may be called in a broad sense reasons of style; for instance
*< Editor's note: This discussion elaborates matters treated in the second half of Chapter I.>
18 CHAPTER I
about redness) in the one being fulfilled by green in the other; so that if
one belief has the same structure as the fact so also does the other, and
the difference between truth and falsity clearly cannot be simply a difference
of structure.
But to knock down such men of straw is not to give any serious reason
for abandoning the obvious truism that a "belief that p" is true if and only
if p [which is what our opponents propose, and it behooves us to consider
what positive reasons they can have for their theories] and the motive forces
behind the Coherence and Pragmatist theories must be found elsewhere.
NOTES
*<Editor's note: This discussion elaborates matters treated towards the end of Chapter 1. >
22 CHAPTER I
that gives our meaning when we say that the man's belief is true. [Since
by hypothesis they must all happen together always, I know not how one
can decide which we mean by ... ] It is surely a futile question, like asking
whether "great uncle" really means parent's uncle or grandparent's brother;
but if we have to choose there is every advantage in choosing the definition
which says that the belief is true simply if A is B, and avoids all the psychological
and metaphysical difficulties involved in the correspondence of beliefs with
facts or the existence of 'propositions.'
Not that these difficulties will not have to be faced. [If we wish really
to get to the bottom of the notion of truth we must not be content merely
to reduce it to propositional reference but must proceed with the analysis
of that in turn. But for many philosophical purposes such a reduction to
propositional reference will suffice; we often do not need to .... ] To give
a complete analysis of truth it is not enough merely to reduce it to propositional
reference but this too must be analysed [in turn J. So long as we accept it
[that notion] without analysis there is [still] an element of obscurity in talking
of "belief that p," a feeling of lurking difficulties which must not be shirked.
And since truth depends on and is defined in terms of [propositional reference]
this notion of "believing that," obscurity [and ignorance] with regard to
propositional reference means obscurity with regard to truth. For instance
we do not yet know whether there really is a general notion of propositional
reference so wide as to embrace all cases of believing, or whether the term
"believing that" may not really be ambiguous and its different meanings
have nothing important in common. Since in this latter case 'truth' too would
be ambiguous and the 'correspondence' between 'true' beliefs and facts would
be quite different for different kinds of beliefs, we shall clearly not have
finished with truth until we have got to the bottom of propositional reference.
[Nor till then can we conveniently answer a certain sort of objection
to our definition that a belief that A is B is true if A is B. I mean the
kind of criticism which allows that this is a correct statement about truth but
objects to it as a definition [of truth J. For instance, some people hold that
a true belief is a belief in a true proposition and that though the proposition
"A is B" is true when and only when A is B yet for the proposition A
is B to be true and for A to be B are two different (though equivalent)
facts; and, so they think, it is the former and not the latter of these facts
that should be used in defining true belief. Clearly this objection can only,
or at any rate can much more easily be disposed of when we have decided
whether propositional <reference exists> J.
But for many philosophical purposes it is not necessary to go so deep:
we need not go into the analysis of the notion of a "belief that p" and
can be content with a definition of truth which tells us what is meant by
calling a belief true when its propositional reference is given. At all events,
APPENDIX: TWO VARIANT ELABORATIONS 23
the first step in the analysis of the idea of truth is to show that it depends
on propositional reference in such a way that when we know what a belief
is <as> a "belief that," we know in what case it is true, i.e., if it is a "belief
that p" it is true if p.2
This may seem such a truism as hardly to be worth enunciating, and
indeed a great many philosophers in discussing "what is truth?" take this
first step for granted and tackel straightaway the next and much more difficult
problem of what is propositional reference. Thus William James often3 insists
that truth means the agreement of our ideas with reality, and that the problem
is what this agreement means, i.e. in our terminology, a belief is true if
it is a belief that A is B and A is B. But what is this connection between
the belief and A's being B that makes it a belief that A is B?
Yet, surprising though it may seem, many philosophers have explicitly
or by implication denied this truism, and constructed theories completely
at variance with it. One suspects at first, that there is some misunderstanding,
that in asking "what is truth?" they are not, like us, seeking merely for
the meaning of the word, but for some criterion for truth, some way of
telling what is true. 4 But this hope can not survive an examination of their
works, and the number of philosophers who are clear on the essential point
is surprisingly small.
[By the essential point I mean this; when we are asking what is meant
by calling a belief true, we must be clear whether or not we are taking
its propositional reference for granted, whether we are asking merely what
is meant by calling true a belief that so-and-so ("so and so" being either
explicitly given or at least assumed to be theoretically definite) or taking,
say, "what is passing in Mr Smith's mind at 10 o'clock" <to be at issue>.]
NOTES
1. [But we shall come later to the conclusion that they are not legitimate. See ... ]
2. <The first page of the miscellaneous jottings of folder 001-10 reads as follows:>
And secondly, even if, however, we agree that our problem is to define truth in the
sense of explaining its meaning, there are still two different ways of taking the problem
according as we do or do not accept the notion of propositional reference as sufficiently
[precise, clear, and intelligible] understood to be used in our definition of truth.
If for instance we are discussing a belief identified as what Mr Smith was thinking at
ten o'clock in the morning, our definition of truth only tells us what is meant by calling
such a belief true, if we already know what is meant by saying that his belief was a
belief that so-and-so, perhaps that the earth is flat. His thoughts at that time were (at
least on one view) images and ideas of some kind whose connection with what they are
said to refer to, namely that the earth is flat, is by no means easy to describe.
Secondly, <truth> may be taken either as defined in terms of reference or in terms of
more primitive notions. We have only done the former, but the latter is more complete.
24 CHAPTER!
Nevertheless the former does for many purposes, and is all many have attempted to give.
A definition of truth in terms of reference must not bring in any extraneous idea, like
Peirce, Meinong, Hofler. The idea of this former <approach> might be questioned as not
on <the right> way, but is must be <accepted>. Reference cannot depend on truth or
truth <be> intrinsic <to it>. James agrees, but gets quite tied up: he gets a fragmentary
account of reference which makes him get completely <muddled>. Also he does not understand
<its> propositional nature. Russell sees <the> point but gets words involved and confuses
this muddle with C<orrespondence>.
3. E.g., William James, The Meaning of Truth (New York and London, 1906), Preface, p. v.
4. $0 Kant. "Die alte und beriihmte Frage ... Was ist Wahrheit? Die NamenerkIarung der
Wahrheit, dass sie namlich die Ubereinstimming der Erkenntnis mit ihrem Gegenstande
sei, wird hier geschenkt und vorausgesetzt; man verlangt aber zu wissen, welches das
allgemeine und sichere Kriterium der Wahrheit einer jeden Erkenntnis sei." Kritik der reinen
VernunJt, "Die transzendentale Logik," Einleitung III.
(001-05)
<CHAPTER II>
Let us take the Coherence Theory first; this holds that the truth of a belief
that A is B depends not on whether A is in fact B but on how far the
belief that it is forms part of a coherent system. It is a theory which is
very easy to reduce to absurdity and after Mr Russell's amusing essay on
"The Monistic Theory of Truth"! it is difficult to see how anyone can still
cling to it; but the defect of all refutations by reductio ad absurdum is that
they do not reveal where the line of thought which leads to the absurdity
first goes astray. According to Mr Russell the first mistake of the advocates
of the Coherence Theory lies in their assuming an abstract metaphysical axiom
called the Axiom of Internal Relations, but some of their arguments seem
to me not to depend on any such axiom but to arise from confusions of
a simpler sort such as ensnare not only the abstruse metaphysician but also
the common man.
The lines of argument in question are two; in the first place it is argued2
that truth is a property not of single judgments but of systems of knowledge,
so that no judgment can be wholly or absolutely true but "the truth of
'true' judgments is essentially the truth of a system of knowledge and ...
the 'truth' of systems of knowledge is borrowed from the Ideal Experience
which is struggling for self-fulfilment in them."3 And, secondly, it is argued
that truth cannot be a relation between our beliefs and facts, for it must
be something which we can test, and all we can test are the relations between
our beliefs. Truth must therefore lie within the circle of our beliefs and
not pass outside them to an unknowable reality.4
In giving our account of truth we have assumed that any belief or judgment
has a definite propositional reference or meaningS in virtue of which it is
absolutely true or false. The obvious reason for disputing this lies in the
possibility that judgments may be vague in the same sort of way as the
words we use to express them. How many hairs must a man have not to
26 CHAPTER II
identified with the concrete thinking of those who learn or teach geometry.
The judgments which enter into such a system are not yours [Mr Smith's]
and mine [Mr Jones's], but, for instance, the judgment that the angles at
the base of an isosceles triangle are equal, and the judgment that all men
are mortal, which are abstractions from the actual judgments of men. We
do not [want] need here to take any particular philosophical view of abstraction,
for on all such views this at least is clear: in order that we can talk significantly
of the judgment that all men are mortal, be it abstraction, universal or what,
your, my and other people's judgments that all men are mortal must have
something in common in order that they may be instances of the same universal
or type; and this that they have in common, is not the words we use to
express them, but their meaning. If single judgments had not definite meanings,
to talk of the, or even of a judgment that all men are mortal, i.e., of a
judgment with a certain meaning would be nonsense. A "system of judgments "
must really mean a system of meanings of judgments, and if judgments had
not their own meanings the phrase would be sheer nonsense.!!
(c) The third argument used against absolute truth turns simply on confusing
two uses of words like "condition" or "basis"; thus it is argued by Prof.
Joachim!2 that a proposition of elementary geometry such as that the angles
of a triangle add up to two right angles is only true if we assume the basis
of Euclidean Space, and that similarly no scientific judgment is true unless
regarded as "the abbreviated statement of a meaning which would require
a whole system of knowledge for its adequate expression." Now it may
well be that the geometrical proposition is not true of physical space but
merely in abstract Euclidean geometry; i.e., that all that is really true is
that the proposition is a consequence of certain definite assumptions, and
this, which is a single hypothetical judgment, is an absolute truth. But this
is not, I fancy, what Prof. Joachim means: he does not doubt, or at least
did not then doubt, that physical space13 (so far as anything is true of it)
is Euclidean, and what he intended to say was that the proposition about
the angles of a triangle was only true because space is Euclidean. 14 But to
argue from this that the meaning of the proposition must be expanded so
as to include the whole nature of Euclidean space, would, I think, be a
sheer confusion, for just because space is (on this view) Euclidean the proposition
is true in its literal meaning, Euclidean space being a kind of space in which
the angles of a triangle are together equal to two right angles.
Precisely the same argument occurs at the crucial point in Bradley's Logic
when he is dealing with the singular judgment of sense,15 of which he claims
that "what it says is true, if it is true at all, because of something else ...
so we have a judgment which is really conditioned, and which is false if
you take it as categorical." He sees, however, that this argument confuses
two senses of absolute or categorical truth; first that in which absolutely
28 CHAPTER II
true means true, and not merely the consequence of a certain hypothesis,
in which, that is to say, absolute is opposed to conditional; and secondly
that in which absolutely true means true for no reason, absolute being opposed
to conditioned. That a judgment is true because of something else does not
mean that it is not true. [He nevertheless thinks his argument can be preserved
by complicated considerations about the nature of conditions, which seem
to me to have no cogency.16 His answer is to compare the series of phenomena
to a chain <with> each link fast to the next, but no fast link fastened to
firm ground to hold it. I do not see why this should be thought absurd
and why the links of such an endless chain should not be real. He has, however,
other reasons of a metaphysical character. In any event it is the sort of a
metaphor which only confuses the issue.]
His answer to this is to say that "the condition, on which the judgment
holds, is unknown, and it admits also the opposite of what is asserted. The
judgment therefore, in its present form, is at once both true and false. "17
This answer seems to depend simply on confusing a necessary with a sufficient
condition; our judgment p was said to be true only because of q; i.e., if
q were not true, p would not be; i.e., q is a necessary condition of p. But
this is in no way incompatible with q admitting also the opposite of p.
Thus suppose p is my reading Bradley: this only came about because of
q, my hearing him praised. But of course, q is perfectly compatible with
the opposite of p, I might have heard him praised without reading him,
and this is supposed to prove that I never have read him! At least, this is
the only sense I can attach to the argument; for if condition in the first
place meant sufficient condition, it is self-contradictory to say that it is
compatible with the opposite of that for which it is a condition. Nor is
this conclusion escaped by saying that the condition is unknown or unspecified:
for even the unknown cannot have self-contradictory properties. But from
Essays on Truth and Reality, p. 232, it is clear that Bradley means that two
sets of conditions are involved, but unknown, one sufficient for p, the other
for not-po This means that he proposes to argue: "For reasons unknown,
you read my book." (Remember we are starting from the hypothesis that
p is conditioned and trying to prove it conditional, not assuming it conditional
to begin with). "But under other unknown circumstances you would not
have read it. Therefore, it is not absolutely true that you did read it! "18
NOTES
4. See Bernard Bosanquet, Logic, 2nd ed. (London, 1911), Vol. II, pp. 265-267.
5. 'Judgment' and 'meaning' are Prof. Joachim's terms corresponding to our 'belief and
'propositional reference.'
6. For a theory of vagueness see below, p .....
7. E.g., a system of 'psychology' consisting of such general laws as "All bald men are
foolish. "
8. It is difficult to see that Mr Bradley's argument in Essays on Truth and Reality (Oxford,
1914), pp. 262-265, has any plausibility except on the basis of this confusion. He suggests
that there might be many worlds in some of which a Caesar crossed the Rubicon, in
others of which he did not. Even so, "at least one Caesar crossed the Rubicon" would
be an absolute truth (if true at all).
9. Cf. Joachim, op. cit., pp. 92-95.
10. Joachim, op. cit., p. 28. But he surely should have said Euclidean geometry.
11. The point is also clear if we take Prof. Joachim's own comparison of a system of knowledge
and a symphony; the possibility of a symphony being performed depends on its consisting
of notes with assigned pitches producible at will by sufficiently expert players in that
or any other context. And so too a science consists of judgments with definite meanings
which can be grasped by any sufficiently intelligent student and reproduced by him
in any other context. But cf. below pp .....
12. Joachim, op. cit., pp. 95-99.
13. Of course, physical space must be more precisely defined, by determining what is to
be meant by "straight line" etc., but these definitions only fix what we are to mean
by our terms, they do not form a "system of knowledge."
14. This is the only sense I can give to the phrase categorical basis.
15. F. H. Bradley, Principles of Logic, 2nd ed. (London, 1922), vol. I, pp. 97-98.
16. <In margin:> Confusion of if and because.
17. Bradley, op. cit., p. 113, note 54. In the text p. 100 there is an even less satisfactory
metaphorical argument.
18. [In the last resort his argument reduces to his theory that in judgment we ascribe predicates
to Reality. This seems to me clearly false for reasons to be given in Chap ...... But
even if it were granted his argument still requires us to suppose that Reality is a must
mean that a is its whole nature, and that is merely a. (Bradley, Essays, pp. 226-9.) This
is simply the elementary confusion of the "is" of predication and with the "is" of identity.
See below .... J <Against this footnote Ramsey first placed question marks, and subsequently
struck it out entirely.>
CHAPTER II
(001-06)
"a nucleus of 'brute fact' encased, solid and distinct, within a surrounding
complex of conditions. It was Caesar at the head of his army and animated
by conflicting motives of patriotism and ambition, who crossed. And he
crossed the Rubicon at this determinate political juncture, with a full
consciousness of the effect of his action on the political crisis at Rome.
This - and more - is the meaning of the historical judgment in its proper
context, its definite meaning. This concrete happening is 'the fact' affirmed
in the judgment if indeed you can arrest the expansion of its meaning
even here. We can be sure, at any rate, that the actual happening contains
no bare crossing of a stream by a man in the abstract as a solid grain of fact,
separable from a complicated setting which particularises it. "1
Now it is obvious that "Caesar crossed the Rubicon" is not the whole
truth about that event, but I cannot see that Prof. Joachim's arguments have
any tendency to show that it is not part of the whole truth; because he
was at the head of his army was he not still Caesar, because he was led
to cross by motives of ambition did he any the less cross? When we say
simply that he crossed, we do not particularise where, when and from what
motives, but we are not denying that he must have crossed at some definite
place and time for definite reasons. Because we call him simply a man and
say no more about him we do not mean he as a "man in the abstract"
without body parts or passions. If I asked someone whether there was a
dog in the house would it be reasonable for him to reply "No; there isn't
an abstract dog, but only a poodle "?
Nor do I see how a complete life of Caesar could be written which did
not say either that he crossed the Rubicon, or something else from which
that fact could at once be extracted, i.e. something which contains that fact
in its meaning. Professor Joachim is, indeed, prepared to allow that "Caesar
THE COHERENCE THEORY OF TRUTH 31
crossed the Rubicon" is not wholly false, but since he thinks that every
judgment, e.g. "Caesar did not cross the Rubicon," involves some truth,
this concession seems hardly to do justice to the facts.
It seems to me that the refusal to admit that anything but the whole
truth can be wholly true comes from the confusion I mentioned before between
truth and intellectual merit in general, and that this in turn comes from
an emotional attitude towards the word truth which does not allow her
votaries to admit that she can ever be tedious or trivial. Thus Professor
Joachim speaks somewhere of "truth so trifling that serious falsehood is involved
in it"2 and Bosanquet asks, "could we seriously say that a judgment about
it is true in which its full significance and implications ae ignored?"3 I say
this is reverence for the word "truth" advisedly; for the idea of truth is one
which no one could reverence who clearly apprehended it; knowledge and
ingenuity one can admire, but mere truth can be attained by a fool by accident
and has no moral value.
Another argument sometimes used is that a historical work, which is accurate
in every detail, may be less true than one which although less accurate gives
a better idea of the period. The truth of the work cannot therefore consist
in the truth of the separate statements. The answer to this is simple; by
saying one work is truer than another we mean something about the relative
importance of the true and false judgments they express, including not only
judgments about details but also those about the relative importance of different
causal factors, etc., which are sometimes not expressed explicitly in statements,
but by the proportionate space given to the different topics, and if these
are misleading we get the paradox of a work not being regarded as wholly
true in spite of the accuracy of every statement.
Bradley argues4 that "Caesar crossed the Rubicon" is not absolutely true
because there may be many worlds, in some of which he crossed it and
in others of which he didn't, or again because Caesar might recur in this
present world. The possibility of identifying which Caesar you mean he rejects,
because he links that <ostensive> designations ('this', 'now', 'here', etc.) have
no place in a genuine judgment. This view seems to me plainly contrary
to fact [and it is defended by the merest rhetoric J.5 But even if it were
granted, the argument in no way shows that 'At least one Caesar crossed
a Rubicon' is not an absolute truth. Nor does it even show that "The one
and only Caesar crossed the one and only Rubicon" is not absolutely true
or absolutely false according as there are or are not many Caesars. In fact,
the argument's whole plausibility rests on the confusion between truth and
certainty.
32 CHAPTER II
NOTES
(001-15)
NOTES
1. See especially Norman Campbell, Physics: The Elements (New York, 1919) and also Heinrich
Hertz, Prinzipien der Mechanik <Leipzig, 1894>, Introduction.
2. Many scientists wish wherever possible to eliminate unobservable entities, such as the orbit
of the electron in a hydrogen atom, but this is so far no more than an ideal.
3. Why they should be so manipulated is, on this view of science, rather a mystery.
4. <Ramsey first placed a question mark alongside the opening sentence of this paragraph
and subsequently struck the paragraph out entirely.>
CHAPTER II
We must now return to the Coherence Theory. We have seen that the
contention that truth is an attribute of systems of knowledge, not of single
judgments cannot be sustained, but there is still the possibility that the truth
of single judgments is to be defined by coherence. The arguments advanced
in favour of this, mainly consist of refutations of alternative doctrines. Thus
Professor Joachim disposes of the theories that truth is a quality of independent
entities and that it consists in correspondence "for a mind" and shows that
each of them leads ultimately to coherence.
We need not linger long on Prof. Joachim's refutation of the doctrine,
then held by Mr Russell and prof. Moore, that truth and falsity are qualities
of independent entities. This is not my view, and it has, I think, been abandoned
even by its authors. Nor can we defend the "fundamental postulate of all
logic" that "Experiencing makes no difference to the facts," in which, if
it is to be relevant to the present discussion, "experiencing" must be taken
widely enough to include judgment and belief. For the success of M. Cone
is alone enough to show that believing that a thing will happen can often
make it happen, at any rate in the medical field.!
It is, I feel, unfortunate that in Prof. Joachim's discussion this view that
truth is a quality of independent entities is the only one he considers on
which truth is not essentially "for a mind." Truth is, we may agree, an
attribute of mental states consisting, as I believe, in correspondence to fact;
but this does not make it in Prof. Joachim's sense 'for' a mind. As is shown
clearly by his example from Leibniz, by saying that a correspondence is 'for'
a mind he means not merely (or perhaps not at all) that one of the corresponding
entities is a mental state, but that some mind is aware of or recognizes these
entities as corresponding. Against the position which I adopt, that a belief
may be true by correspondence to fact, even when no one recognizes it
as corresponding, he says nothing no doubt because it had not then2 sufficient
authority to deserve his attention. [He dismisses <this view> in one place3
with a reference to his discussion of the quite different view that truth is
a quality of non-mental propositions, and in another with the dogmatic assertion
that the angles of a triangle would not be equal to two right angles if no
one had ever thought they were!]4
His refutation of the correspondence theory does not, therefore, claim
to apply to the version of that theory which I advocate, but it is nevertheless
worth considering as showing the sort of objections we have to meet. In
the first place we may grant that truth could not possibly consist in the
THE COHERENCE THEORY OF TRUTH 37
only kind of correspondence which Prof. Joachimn considers [and that for
much simpler and more conclusive reasons than any he advances]. "Two
different factors," he says, "'correspond' when each of them is a whole
whose inner structure is teleological, when that structure is identical as the
explication of the same idea of purpose, and when, finally, for every distinctive
part fulfilling a determinate function within the one factor there is a part
fulfilling the same function within the other."5 It does not really seem to
me that the view that truth is constituted by a correspondence of this sort
deserves a moment's consideration. A belief that grass is green is true, a
belief that grass is red is not: could anyone seriously suggest that this was
because the former belief was a structure "explicating the same idea or purpose"
as is explicated by the greenness of the grass, and the second belief was
not such a structure? The structures of the two beliefs could reasonably be
regarded as identical, the function fulfilled by the part red6 in the one being
fulfilled by the part green in the other. But it would surely be absurd to
suppose that of these two beliefs of identical structure one had the same
structure as the fact but the other had not.
The question can be made clearer if we talk not of beliefs but of statements
or sentences, whose structures are more obvious. It is clear that there is
a sort of correspondence between the sentence "grass is green" and the fact
that grass is green, in as much as the elements of the fact are the meanings
of the terms in the sentence. But this correspondence does not consist in
identity of structure, for the sentence "grass is red" has precisely the same
structure as the sentence "grass is green" but does not correspond to the
fact in the same way since the word "red" does not mean green. Whether
or not structural identity is necessary for truth, it is clearly not sufficient for
it, and cannot be supposed to be the kind of correspondence which constitutes
it. The kind of correspondence which might be sufficient is that arising from
the relation of meaning. The kind of correspondence based on meaning does
not, as we shall see, make error the same as not thinking, as does the kind
considered by Professor JoachimJ
This is presumably dismissed by Prof. Joachim when he says that he assumes
that a purely external relation is in the end meaningless and impossible. What
mayor may not be the case "in the end" I do not know, but in the middle
or wherever it is that this life is situated, there is a relation between words
and what they stand for, capable of giving rise to a sort of correspondence,
and determining or at least helping to determine, the truth of statements.
That this may not be so "in the end", is to my mind irrelevant. In the
end, I understand, there is no time, and no events in time. Consequently,
since judgments are events in time there are no judgments, and no true
judgments. The problem of truth as an attribute ofjudgments cannot, therefore,
I conceive, be a problem as to what happens in the end, and arguments
38 CHAPTER II
drawn for this are no more relevant here than they are in aeronautics.
I have spoken above as if truth were constituted in part at least by a
kind of correspondence based on the meaning of words, but I do not want
to assert this at the present stage [nor is it exactly true]. It belongs also
to what we have called Question II, the question of what constitutes
propositional reference, not to Question I, what, given propositional reference,
<MS 001 - 08>
constitutes truth, over which we are primarily at issue with the Coherence
Theory. I introduced it as a type of correspondence, not considered by Professor
Joachim, and relevant here as the type [of correspondence] which would
constitute truth, if Question II were answered in a certain way. For it will
be remembered that although we decided that truth depended on correspon-
dence, we postponed the decision as to what type of correspondence it was,
saying merely that it was the kind [of correspondence] there evidently is
between believing that a thing is so and its actually being SO.8
Nor can I believe that this is not what Prof. Joachims means at any rate
when he pronounces anyone else's opinion true: suppose he reads Bosanquet's
logic and wonders whether Bosanquet's view that every hypothetical implies
a categorical ground is true or false. Surely he is comparing Bosanquet's
opinion with the state of the case: if the fact is as Bosanquet says, if every
hypothetical does imply a categorical ground, then and only then is what
he says true [and otherwise false].
To this I suppose it might be replied that in such a case when A is considering
whether B's opinion is true, it is wrong to say that A compares B's opinion
with the fact, since he himself has no sure knowledge of the fact but only
his own opinion. Consequently, we should say that A compares B's opinion
not with the fact but merely with his own opinion, and the relation which
he pronounces is not one between an opinion and a fact but between two
opinions. But such an objection would be invalid, as it is of the first importance
to see. According to the objector, A has no sure knowledge of the fact,
but then he equally has no sure knowledge as to what is really B's opinion,
and if through ignorance of the fact he can compare nothing to the fact,
then through ignorance as to B's opinion he can not compare that but merely
his own opinion about it; so that what he really does is to compare his
opinion about B's opinion with his opinion as to the fact. We can now see
that the point has no special relation to truth, but arises in connexion with
any kind of comparison, for instance, comparison of colour. Suppose the
question arises, whether two regiments of soldiers which I have seen or read
about have uniforms of the same colour, and I think "The uniforms of the
Northshires are red and the uniforms of the Southshires are red, so they
have the same colour". Then the relation which I affirm in so concluding
THE COHERENCE THEORY OF TRUTH 39
is one between the actual colours or uniforms, not one between my opinions
about them. My conclusion is founded no doubt on my opinions as to what
each colour is, but [is not about those opinions but] states a relation which
I believe to hold not between my opinions (or at least not merely between
them) but between the real colours. 9 So also if A says B's opinion agrees
with the fact, the relation he assumes to hold is one between B's opinion
and the fact, not one between his own opinions about them: although, of
course, his making the assertion is the result of his holding these opinions.
I do not therefore see how it can be denied that when we say that other
people's opinions are true or false we mean that they do or do not correspond
to the facts; but how is it, we shall be asked, when we come to consider
the truth of our own opinions. For this is the case from which the advocates
of the Coherence Theory draw their strongest arguments, and the source,
as it seems to me, of all its plausibility.
On the correspondence theory, it is argued,lO truth means a correspondence
between our beliefs and a reality outside them, an external standard to which
we claim to conform. But we cannot put before ourselves such a standard
with any hope of attaining it; nor can we use it as a criterion. If I make
a judgment, and claim that is is true, this cannot be because I see that it
corresponds to a fact other than itself; I do not look at the judgment and
the fact and compare them; the fact as I see it is simply my judgment, and
all I can test it by is my other judgments. Wherever I turn I am bound
by the circle of my own ideas, and the truth I seek [cannot consist in a
relation between them and something outside but] must lie inside the system
of my beliefs, not in a relation between them and an unknowable reality.ll
The second line of reasoning employed by supporters of The Coherence
Theory is important as an example of a general fallacy, which is by no
means confined to this school of thought, and lies at the root of a lot of
mistakes in regard to truth and knowledge. In the present context the argument
runs as follows.
On The Correspondence Theory, it is said, truth means a correspondence
between our beliefs and a reality outside them, an external standard to which
we claim to conform. But we cannot put before ourselves such a standard
with any hope of attaining it; when, for instance, I say that someone else's
opinion is false I mean merely that it does not agree with my own, and
when I make a judgment myself and claim that it is true this cannot be
because I see that it corresponds to a fact other than itself; I do not look
at the judgment and the fact and compare them, but the fact (as I see it)
is simply my judgment, and all I can test it by is my other judgments. Wherever
I turn I am bound by the circle of my own ideas and the truth I seek must
lie inside the system of my beliefs and not in a relation to an unknowable
reality.1 2
40 CHAPTER II
NOTES
that one simple [explanation] kind of analysis must be given for all kinds of belief; as
we shall see, copying and pragmatism are both elements in the true analysis which is
exceedingly complicated, too complicated for us to hope to give it accurately.
9. By "real colours" can be meant here either their real colours in a conventional sense,
the colours the uniforms would have in normal light etc., or the colours they really
appeared to me to have when I saw them. The contrast is not between real colour
and apparent colour, but between the colour as it really is or really appeared and the
colour as it is thought to be or remembered to have appeared.
10. See especially B. Bosanquet Logic, 2nd ed. (Oxford, 1911), Vol. II, pp. 265-267. [Also
F.H. Bradley, Essays on Truth and Reality (Oxford, 1914), p. 108; "If truth is to copy
fact, then truth at least seems to be in fact unattainable." But he is really discussing
a view that is not in question here, that truth should copy given facts.
11. <The following two paragraphs are taken from a misplaced MS page (numbered 34)
from MS 001-05-01. The continuity is guaranteed by the identity of the ensuing paragraph.>
12. <The next bit comes from 001-05-01/45! The page missing in the MS 001-08, 45, was
misplaced there.>
13. <The next sentence overlaps between the misplaced page 45 (see footnote 12) and the
continuation of the present manuscript MS 001-08, p. 46.>
14. This is virtually admitted by Bosauquet iu Mind, vol. 31 (1922), pp. 335-336. He admits
that "Caesar was murdered" is true, if the murder really took place but argues that
we are not entitled to say it is true without proof. But this is quite irrelevant to the
meaning of truth; we are equally not entitled to say a thing is useful without proof
that it is, but this does not [illuminate] bear on the meaning of the word "useful".
To use "true" to mean "proved true" is an atrocious misuse of language and leads
to the kind of confusion which brings philosophy into merited disrepute.
15. Joachim, op. cit., Ch. I, passim.
16. The question in what senses a judgment can and in what senses it cannot be about itself
is a difficult one, but in the sense required here it clearly cannot. On the whole question
see ....
(001-11)
CHAPTER III
JUDGMENT
The truth and falsity of mental states we found to depend upon their
propositional references and our next task is to analyse this characteristic,
to answer, in fact, our Question II, what is propositional reference? But
before starting on this it is best to make certain preliminary distinctions between
the states of mind with which we have to deal.
To begin with there is a very elementary point about the way in which
we use the words knowledge, belief and opinion. In general, these refer
not to definite acts of thinking but to the persistent background of the mind.
If we say that a boy knows the date of the Norman Conquest, we do not
mean that he is at the moment thinking "1066" but that he would be able
to give that answer if required. Weare talking not about what he is actually
thinking, but about what he would think or say in certain circumstances.
of course, he would not know it if he had not learnt it and so thought
about it at least once; but we say he still knows it even when he hasn't
thought about it for years provided he has not forgotten and could still answer
our question.
By "knowledge," then, we usually mean something mere potential, which
would be manifested if occasion arose but may still be there without being
manifested, and the same is true of "belief' and "opinion." An opinion is
not an act of thinking; when we "form" or "change" an opinion we are
thinking, but the opinion so formed remains ours and is carried round with
us ready to be given to anyone who asks for it without our needing to
think of it at all in the mean time.
Knowledge, belief and opinion when the terms are used in this way [in
this ordinary sense], we shall call dispositional, because in their capacity for
remaining latent they resemble qualities of disposition or character in the
ordinary sense, for a man may be called brave or irascible without any
suggestion that he is at the moment actually displaying these qualities.
It must not be thought that these dispositional characteristics are only
to be found in connection with mind; for they are common enough in the
44 CHAPTER III
material world. When, for instance we say that a poker is strong, we mean
that it would support a considerable strain without breaking. This fact is
something purely hypothetical; for the poker in question may never be subjected
to such a strain, but still it is true of it, that if it were, it would not break,
and this is, in our sense, a dispositional characteristic.
In the case of the poker we suppose that whether it would or would
not break depends on the nature and arrangement of the particles of which
it is composed; and that its strength (which is a dispositional characteristic)
depends on some positive (i.e., non-dispositional) characteristics of the
component particles. But most of us have no idea what those positive
characteristics are, and we can perfectly well talk of the dispositional quality
strength, without knowing the positive qualities on which it depends. So
also in the case of the boy who knows the date of the Conquest, we must
suppose his knowledge to depend on some arrangement, 'trace' or 'record'
in his mind or brain, which is formed when he learns the date and persists
until he forgets it, his forgetting being simply the disappearance of this trace. 1
These traces, or in different cases other formations, constitute the positive
qualities from which dispositional knowledge and beliefs are derived, but
most of us have no idea as to what sort of structures or modifications the
traces are, and take them simply as unknown causes which bring it about
that if for instance we ask the boy for the date he tells us correctly. So
when we are trying to explain as at present what we mean by knowledge,
etc. we have no concern with the real nature of these traces but merely
with the kind of thoughts or actions which they are supposed to cause. Just
as in explaining the meaning of strength, we have only to explain what
is meant by supporting a strain without breaking, not what properties of
a body they are which enable it to support a strain.
To say a man has such and such knowledge, beliefs and opinions means
then generally something hypothetical, something about what he would think,
say,2 or do in suitable circumstances. It is, in my view, important to realise
that it is not only a question of what he would think or say but also of
what he would do, for many of our dispositional beliefs are manifested far
more in our actions than in our thoughts. For instance, I have a dispositional
belief (or perhaps I should rather say knowledge) that the Cambridge Union
is in Bridge Street; but this belief is very rarely manifested in an act of
thought: I do not often have occasion to judge that that is where the Union
is: I only do this when I have to inform a stranger, or just now when I
thought of it for an example. On the other hand, this belief of mine is frequently
manifested by my turning my steps that way when I want a book from
the Union Library, which I do without any process of thought which could
properly be called thinking that the Union is in Bridge Street. In Oxford,
I should have to think where the Union was, but in Cambridge, where I
JUDGMENT 45
"We see at a little distance a person whom 'we mistake for an acquaintance'
and without hesitation perform some act which it would be a liberty
to take with anyone but an acquaintance, do something in fact which
we rightly say we should not have done if we had ever suspected he
was not an acquaintance. "10
"It is true that, if asked, we might say 'I thought it was my friend' -
'I believed it was my friend' - 'I was sure it was my friend', but," says
Cook Wilson, "these expressions are all inaccurate. The truth is, as will
be admitted, that in the given case, when I perceive the familiar characteristic
of my friend, it never 'enters into my head' that they could belong to
anyone else. I don't think about that at all and so the processes of judgment
belief and opinion are impossible."12
for the fact that if asked "who's that?" he would undoubtedly have replied
"Jones" (unless something made him pause to reflect). Indeed we might even
say that this "thought" was a dispositional belief, and that by saying he
thought it was Jones' back, he meant that he would have behaved (in this
case actually did behave) as if it was, and if asked have said it was. [The
only difficulty in identifying it with a dispositional belief lies in its temporary
nature; for he has never seen this particular back before.]
There is, however, a certain difficulty in identifying this so-called thought
with a dispositional belief, for what exactly is to be the propositional reference
of this belief? On the one hand it appears to be about just this particular
back which he treated as Jones', but clearly the same mental disposition or
habit was manifested the day before when he slapped what was really Jones'
back, and the disposition has no particular reference to this back but to any
back of a certain sort. But on the other hand, we cannot say that it is a
dispositional belief that all backs of this kind are Jones' for though on each
occasion that we met a back like that he would probably jump to the conclusion
that it was Jones', he might easily shy at saying that all such backs were
Jones', since the form of this question probably make him reflect. The best
thing to say seems to be that strictly speaking we have not a dispositional
belief that all such backs are Jones', but a disposition to treat both in thought
and action any such back that may turn up as Jones', and this we might
call a dispositional belief function (by analogy with propositional function)
which is not in itself either true or false, since it has no determinate reference,
but in each manifestation is either true or false according to whose back
occasions it to be manifested.
The same phenomenon occurs in many other cases in which we are tempted
to speak loosely of a dispositional belief; thus in the example given above
of my belief that the Union is in Bridge Street, manifested by my turning
my own steps or directing a stranger's thither, what is relevant is everyday
something different namely than on that day the Union is in Bridge Street.
Here again we cannot get out of the difficulty by making it a belief that
the Union will always be in Bridge Street, since I am not at all sure this
might not cease to be true, and we must say that we are strictly concerned
with a dispositional belief-functionY
But we must return to our problem, of whether the man can properly
be said to have judged that it was Jones' back that he saw. The conclusion
we have come to is this: if his seeing the back led either to his saying to
himself "Hallo, there's Jones" or to his having an image of Jones' face of
such a kind or with such accompaniments that it issues in action, then we
must say he made a judgment. If on the other hand he acted directly on
seeing the back without any such intermediary process, then there was no
judgment, although we might perhaps say that his response manifested or
50 CHAPTER III
"consists in the fact that certain specific bodily feelings (connected with
the automatic adjustment of the body), certain emotions, and certain feelings
of expectation, are related in an unique way to the apprehended sensum.
These are causally dependent on the traces left by past experience. When
a sensum of a sepecific kind is intuitively apprehended certain traces are
excited; there arouse certain emotions and induce certain bodily adjustments
which are accompanied by specific bodily feelings. They may in addition
call up certain images; and, even if they do not do this, they may evoke
a more or less vague feeling of 'familiarity.' These 'mnemic consequences'
of the apprehension of the sensum do not just coexist with it; they
immediately enter into a specific kind of relation to it, which I do not
know how to analyse further. And these 'mnemic consequences' in this
specific relation to this intuitively apprehended sensum constitute the quasi-
belief about the sensum, which gives the situation its specific External
Reference. Any situation constructed of such materials in such relations,
ipso facto, has such and such an External Reference. "18
On this view, then, that two mental acts are expressed by the same form
of words only shows that they are concerned with the same object or objects
not that they have anything further in common. The obvious answer to this
is the one contained virtually in our definition of judgment by reference
to an "affirmative attitude"; namely, that there must be something in common
to those cases in which we express our attitude by "The earth is round"
which is not common to those in which we ask "Is the earth round?". The
fact with which we are concerned is in both sets of cases the roundness
of the earth; but in the one case we adopt to it an affirmative attitude,
in the other a merely interrogative one. It is the obvious difference between
thinking that and wonder whether, and it is absurd to deny that there is
something in common to the affirmative acts in virtue of which they are
not merely wonderings whether but thinkings that, capable of truth and falsity.
Nevertheless it can still be argued that this common character of affirmative
acts is only superficially so important, and for deeper investigations is trifling
in comparison with the differences between them. In grouping a prejudice
and reasoned conclusion together as judgment, we are recognising their
resemblance in two respects. In the first place they both have the felt quality
or absence of felt quality22 characteristic of assertion as opposed to doubt
or inquiry; and in the second place they have similiar effects on subsequent
thought and conduct. If we judge that A is B, then until we reconsider
the matter we take A to be B both in practice and in any subsequent theoretical
investigation, and this happens whether our judgment was based on good
reasons or on no reasons at all. Any two judgments that A is B have in
common, therefore, a certain felt quality and certain effects, and also, of
course, each of them is in some way concerned with A's being B, but this
need not prevent the relation between the judging self23 and A's being B
from being quite a different one in the two cases, so different that the two
relations are not properly speaking species of the same genus at all.
It is argued that this is actually the case when one of the judgments is
knowledge and the other merely opinion;24 that in knowing or apprehending
that A is B, the mind has quite a different relation to A's being B from
JUDGMENT S3
that which it has in opmmg that A is B. If this is right, then our term
"propositional reference" is essentially ambiguous and has at least two quite
different meanings; for ajudgment's propositional reference is its characteristic
of being a judgment, say, that A is B; and to say that it has this characteristic
is a way of saying that either the judgment or25 the judging self has a certain
relation or relations to A and B or to A's being B; so that if in two cases
these relations are different, so must the two characteristics called "pro-
positional reference" be different. 26 Since our present inquiry is into the meaning
of "propositional reference" this possibility of an essential ambiguity is a
most important one for us, and we must examine carefully the arguments
used to support it. As regards terminology, however, the most convenient
plan seems to be to continue to use our terms 'judgment' and 'propositional
reference' in the wide senses we have given them, but to keep in mind their
possible ambiguity.
NOTES
1. This view is disputed; but even if there are ultimately no such things we can regard
traces as "logical constructions" and talk about them e.g. as in the text without serious
error. In the whole subject see C.D. Broad, Mind and its place in Nature (London, 1925),
Ch.X.
2. We neglect there and elsewhere the possibility of his lying in saying the opposite of
what he thinks (not necessarily saying what is false); it seems necessary to include "say"
because it may not always be possible to distinguish any process of thought apart from
his saying it either to himself or aloud. [But if he is lying there will be such a thought,
and what is relevant there will be that thought and not his words.]
3. See below.
4. See below.
5. This formula obviously requires modification to include the case of partial beliefs, and
is anyhow inexact as it takes no account of the impulsive element in action.
6. [Weare neglecting throughout the possibility of dying> ... ]
7. Although it is often hard to say in a particular case whether we are or are not confronted
with an act of thought, the distinction between such acts and mere dispositions is in
principle a perfectly clear one, since the former are real events and the latter mere
capacities or potentialities. The practical difficulty is not to distinguish an act of thought
from a disposition, but to distinguish it from some other occurence such as a mental
image or a bodily action. The difficulty is increased by the fact that a bodily action
may, as we have seen, resemble a thought in being a manifestation of the same dispositional
belief.
8. It is sometimes denied that there is any thinking which is not thinking that; if so there
is no need to exclude it, and in any case it is sufficiently excluded by the qualification
that the acts must have an affirmative character, since it is obviously impossible to affirm
without affirming that something or other is so and so.
9. J. Cook Wilson, Statement and Inference, Vol. I (Oxford, 1926), p. 92. His criticism on
<sic for of> the wider use we considered below.
54 CHAPTER III
CHAPTER IV
We saw above that judgment in the wide sense in which we use the term
was held by some not to be a real genus, but to comprehend two essentially
different processes, knowledge and opinion; and our next step must be to
examine the meaning and validity of this distinction.
The words "knowledge" and "opinion", we have also seen are most
commonly used not of judgments but of dispositions; but since the distinction
which we are investigating is primarily one between different kinds of
judgments, (which can be extended to the dispositions arising from or
manifested by these two kinds of judgments) we shall use the words knowledge
and opinion in the present chapter to mean judgments and not dispositions.
It is not held that the two categories of knowledge and opinion, exhaust
the class of judgments, thus belief may be admitted as a third category, and
there may be others such as "being under the impression that" which according
to Cook Wilson! is neither knowledge, belief nor opinion. But the fundamental
distinction proposed is that between judgments which are cases of knowing,
and those which are not, opinions being taken as representative of all those
judgments which are not knowledge.
The importance attached to this distinction may perhaps be explained in
the following way. What we know, it is said, are facts, for instance the
fact that I am sitting at a table or that the earth is round, so that knowledge
is a relation between a mind and a fact. 2 On the other hand a false judgment3
cannot be a relation between a mind and a fact, for if the judgment is false
there is no fact to which the mind could be related. If a man judges that
the earth is flat, his judgment cannot be a relation between his mind and
the fact that the earth is flat, for there is no such fact as that the earth
is flat. What false judgment is, is a difficult question, but it is certain that
a false judgment that A is B cannot be a relation between the mind and
the fact that A is B. Hence it appears that knowledge and false judgment
are two utterly different things, and that the characteristics we have called
propositional reference must be quite different in the two cases, so different
S6 CHAPTER IV
that it is hardly reasonable to call them by the same name. For if a man
judges that A is B and his judgment is a case of knowledge, its propositional
reference consists in a relation between the mind and the fact that A is
B; but if his judgment is false the characteristic that makes it a judgment
that A is B cannot be any relation to the fact that A is B, since there is
no such fact.
It might be thought that this difference was not one between knowledge
and other judgments, but between true judgments and false ones. Such a
view would, however, be very strange: take, for instance, a man who believes
that it is going to rain next day; can we really suppose that the nature
of his belief, the sense in which it is true to say that he believes it will
rain, depends on whether it actually rains or not? So that by analyzing his
mental processes today we could tell what the weather would be like tomorrow?
Since this seems absurd, everyone agrees that believing is the same kind of
process whether the belief is true or false, and since the false belief that
A is B cannot be a relation between the mind and the fact that A is B,
a true belief cannot be such a relation either. Similarly no other species
of judgment, which is even false, can consist in a relation to the fact judged:
the objects of belief, opinion and conjecture cannot be facts.
It is however, supposed that in knowledge we have a kind of judgment,
which is infallible. This means not merely that we only call a judgment
knowledge when it is true, but that there is a class ofjudgments, distinguishable
otherwise than as true ones, which are not merely all in fact true, owing
to some law of nature, but are in the nature of the case incapable of being
false, since in them the mind is apprehending a fact, and were there no
such fact, there would be not a false judgment but no judgment at all. Let
us call judgments of this kind, if such there be, apprehensions,4 without however
committing ourselves to the view that there are any such things at all. The
theory that besides the fallible species of judgment, there is this infallible
species called apprehension arises from the use (in ordinary language) of the
term knowledge, a use which we must now investigate.
When we say that a judgment is knowledge, we imply at least two things;
that the judgment in question is true and that it is made with complete
conviction. I should not apply the term knowledge to a judgment that the
earth is round, unless I myself thought the earth was round, i.e. unless in
my view the judgment was true; nor could the term knowledge be properly
used, if the man who judged were not absolutely certain that the earth was
round, but merely regarded that as the more probable alternative.
The term knowledge, then, implies both truth and certainty,S and on the
other hand belief and opinion imply neither truth nor falsehood, but generally
imply at least a slight degree of uncertainty. Thus none of these terms are
applicable to the common case in which a man is certain of something false;
KNOWLEDGE AND OPINION 57
for then we cannot say he knows it, because it is false, nor can we say
that he believes it because he is certain. We have, in fact, to say that he
is sure of it, or certain of it, or else to use the misleading expression that
he thinks he knows it, which suggests quite falsely that he is thinking about
is his own thinking or knowing and not the matter in question.
But although truth and certainty are part of what we mean by knowledge
they are rarely, if ever, the whole of it for in using the term knowledge
we nearly always imply that the thinker has sufficient reason or evidence
for his judgment: not merely that it is true and that he is certain of it,
but that in some sense he has a right to be certain of it. A man's hereditary
political or religious convictions may be true, but if he has never thought
the matter out, and seen that there is conclusive evidence for their truth,
they would generally be regarded not as knowledge, but merely as prejudices
which happened to be true;-in order to be knowledge, a judgment must
be not merely true and certain but also well grounded. 6
In what this well groundedness consists is a very difficult problem which
properly belongs to a later stage of logical inquiry: but something must be
said about it here, because it is one possible interpretation of the distinction
between well grounded judgments and others, which leads to the view that
we are here considering that judgments and propositional references are of
two essentially different kinds.
The usual account of knowledge is that it is of two kinds, direct and
indirect. Perception, memory and insight into self-evident truths are forms
of direct knowledge; they are not required to be and could not be justified
by argument. [I perceive that I am sitting at a table, I remember that I
had coffee for breakfast and I see that two straight lines cannot enclose
space. In none of these cases is my judgment founded on arguments, our
confidence in such judgments does not require arguments to justify it, nor
could any such arguments be found.] But the remainder of our knowledge
is indirect and obtained by inference or argument from what we know already,
and so ultimately from what we know directly; and for these judgments
to rank as knowledge, for our confidence in them to be justified, the arguments
on which they are founded must have conclusive or demonstrative force.
[Refer to Cook Wilson if possible.]
It is, however, clear that most of what we ordinarily call knowledge
does not satisfy this severe standard: we speak of historical or scientific
knowledge, although neither in history nor in any of the sciences except
pure mathematics can the truth of our judgments ever be rigidly demonstrated.
[All the arguments on which they are founded are of an inconclusive character,
being drawn from testimony or experience. It is very unlikely that we are
mistaken in such things, but it cannot be proved impossible [that we are
not.] We say a man knows, if his judgment is <appropriately> founded. The
58 CHAPTER IV
arguments in which they are founded all depend on testimony and experience
and so are necessarily inconclusive. ] However many witnesses assert that
something happened, it cannot be proved that they are not lying, and however
many experiments confirm a scientific hypothesis it may yet break down
later. By knowledge we mean justified confidence, but we regard ourselves
as justified in placing complete confidence in arguments that do not begin
to amount to strict demonstration. It is even clear that arguments of the
sort usually accepted as a sufficient foundation for knowledge not only may
but sometimes actually do lead to erroneous conclusions. For what scientific
proposition is better established than that a man once dead cannot come
to life again? And for what fact of ancient history is there better evidence
than for the Resurrection of Our Lord?
Yet one or other of these strong lines of argument must in this instance
lead to a false conclusion; the" contest of opposite improbabilities," whichever
way we resolve it, shows that it is on improbabilities and not on impossibilities
that our knowledge is founded. The truth is that we accept as giving knowledge
any argument of sufficiently high probability:7 a confident judgment based
on such an argument from known premises is regarded as knowledge when,
as is usually the case, it is true. But in the rare cases when the argument
leads to a false conclusion, the judgment being false can no longer be called
knowledge, although it is just the same sort of mental act as it would be
if it were true and were called knowledge. The process of judging on such
evidence is a fallible one and cannot even when true be supposed to be
the apprehension of a fact.
Knowledge, therefore, as we ordinarily use the word does not mean
apprehension; science and history are not forms of apprehension, and we
do not pursue them in the hope of apprehending, for no such hope could
possibly be justified.
The sphere of apprehension cannot therefore be so wide as that ofknowledge
in the ordinary sense, but its friends may still claim that it includes [all]
our direct knowledge and all that we have rigidly deduced from it. But
such a claim cannot be sustained, if rigid deduction or demonstration is used
in the ordinary sense in which anyone who constructs or follows a correct
mathematical proof is held to have demonstrated his conclusion; for although
the conclusion must be true if the premisses are true and the proof is really
correct, the proof may nevertheless seem correct to its author and yet its
conclusion be utterly mistaken; so that the final judgment that the conclusion
is true, depending as it does on the argument merely seeming and not actually
being correct, is not infallible and even when true is not a case of apprehension.
For instance a man may correctly conclude by simple addition that if the
inhabitants of the four quarters of a town number 2037, 3156, 1432, and
614 respectively, then the total population is 7239, but a judgment obtained
KNOWLEDGE AND OPINION 59
in such a way would not be a case of apprehension, for a similar but false
state of mind could perfectly well arise from a mistake in arithmetic. If
any knowledge founded on argument is apprehension, it can only be that
founded on an argument which is not merely in fact valid, but whose validity
is itself apprehended. The apprehension of the validity of an argument would
be a case of direct knowledge to which we shall come shortly; it is obviously
something we can never achieve except perhaps in the simplest cases; in
no argument of any [interest or] complexity can we hold the whole process
before our mind at once and see that it is right with an infallible insight.
I at least cannot even see that 57 and 47 make 104. I can only work it out.
We come now to direct knowledge, of which the principal kinds 8 are
perception, memory and insight into abstract truths. The easiest one to begin
with is memory.
I remember having coffee for breakfast this morning: my memory is a
judgment, and according to the ordinary use of words it is knowledge; and
also it is direct knowledge in the sense that it is not inferred from any other
judgment. If I had forgotten it, I could [no doubt] have inferred it from
[many] other things which I remember, but if I had lost all my memories
I should have no way of inferring to the past at all, and memory must be
regarded as an independent form of knowledge. The question is whether
in such a case a memory can be an apprehension, for instance can I apprehend
the past fact that I had coffee for breakfast? It seems to me clear that memory
cannot be apprehension, simply because it is fallible: not merely do we often
fail to remember (which has no bearing on the present question), but we
sometimes remember wrong. If now remembering right were apprehension,
it would have to be a mental process of a quite different kind from remembering
wrong, and it would be impossible to explain why it is so hard to tell one
from the other. For the man who confidently remembers wrong always thinks
that he is remembering right; and on the view in question this means that
he not only misremembers, but also misintrospects, since by examining the
nature of his present experience he ought to be able to tell that it is not
a memory at all, but merely a mis-memory.
The alternative view is that the experience of remembering that you had
coffee for breakfast is one which could conceivably occur when you had
not really had it, although the constitution of the mind or brain is such
that this very rarely happens. The clearer the memory and the smaller the
time interval between the event remembered and the remembering the less
often do the two fail to correspond. It may be the case that sufficiently
clear memories of sufficiently recent events are never mistaken, but even
so this would be merely a law of nature depending on empirical evidence
not a logical necessity. Those who hold that such memory is apprehension
imply not merely that it is never wrong, but that it could not be wrong
60 CHAPTER IV
from the nature of the case, the experience of remembering being something
that could not happen without the existence of the fact remembered, in the
same sort of way as there could not be a battle without combatants.
The only support for the view that memory is apprehension is derived
from the way we use the word "remember," as implying truth. So that if
a man remembers wrong, we generally say that he doesn't really remember
but merely thinks he does. The word remember thus means to make a certain
sort of judgment truly, and we have no word for the sort of judgment itself:
a peculiarity of language which makes us think that truth belongs essentially
to such judgments, and that if the judgment is false, some quite different
mental process must be occurring. It is however obvious that in this case
language cannot possibly be a safe guide, since the phrase we are driven
to, "he thinks he remembers,"9 does not properly express what we want
at all: it refers strictly not to the primary memory-like (or on the apprehension
view memory-"simulating") judgment that such and such happened in the
past but to a subsequent introspective thought about this first judgment. We
have remarked on this difficulty before in connection with the words know,
perceive, etc., which also imply truth, but in the case of know we avoided
it by substituting "be sure (or certain) of' which has no such implication.
With regard to most other kinds of direct knowledge we can argue similarly:
perception in the ordinary sense of knowledge of the external world is evidently
a fallible process. Although I am completely certain that I am now sitting
at a table and writing with a pen, it is clear that the mental processes which
constitute or produce this knowledge are of a kind which sometimes lead
astray. There are many illusions which show that such perceptions are fallible
and can not be apprehensions. But when we come to what Dr Broad calls
"inspection," our knowledge of our own sense data and bodily feelings, the
case may well be different. Thus Dr Broad comes to the conclusion that
there is no reason to doubt that inspection so far as it goes is infallible10
and if so, it would seem that in inspection we have at last found a process
of apprehension. But the value of Dr Broad's arguments on this point must
be considerably discounted by his subsequent assertion that what is called
Stumpfs argument that it is possible to have a series of colours or other
qualities each indistinguishable from its neighbors, though the two ends of
the series are evidently different "proves with almost complete certainty that
some pairs of sensa, which seem to us to be exactly alike in quality or intensity
must really differ in these respects."l1 For this seems at complete variance
with his previous conclusion since it follows that judgments of inspection
relating to pairs of sensa may perfectly well be mistaken, and inspection
at least of pairs of sensa, but if so surely of single sensa also, cannot be
an infallible process. [Without wishing to be dogmatic] It seems to me that
we here have a perfectly good reason for rejecting, let alone doubting, the
KNOWLEDGE AND OPINION 61
infallibility of inspection, and deciding that here agam our search for
apprehension is fruitless.
The last possibility of our finding it seems to be in our insight into self-
evident truths; but here again the prospect seems unfavourable: the feeling
of self-evidence can be no guarantee of the truth of a judgment, for the
progress of science is continually obliging mankind to abandon what it took
for self-evident truths. [The axiom of parallels and the uniqueness of the
time system are sufficient examples from our own time to illustrate the recent
destruction of them.]
The Theory of Relatively has overthrown those "necessary forms of
thought" Euclidean Geometry and the single time series, and the Quantum
Theory has thrown doubt on the Law of Causation itself. It is true that
no similar revolution has subverted the laws of logic and arithmetic but that
is because these are not genuine judgments but mere tautologies.1 2
But though all judgments which feel self-evident cannot be true, it might
still be argued that some of them were real acts of insight and in our sense
apprehensions; others mere pretenders simulating generalizations from insight
or self-evidence which could naturally be prejudices or results of experience
sometimes true, but sometimes false. This hypothesis has, however, nothing
to recommend it; for the two classes of really self-evident acts of insight
and apparently self-evident judgments are not really distinguisable. The man
who makes the judgment cannot tell which it is, or he would not be liable
to think it an act of insight when in fact it is false, and the outside observer
can indeed tell that if it is false it must be a pretender and not really an
insight, but if it is true he has no way of telling one from the other. There
seems therefore no reason to hold that there is this separate class of really
self-evident judgments and not simply [true and false prejudices,] judgments
which feel self-evident <but> which (like all other kinds of judgments) are
sometimes true and sometimes false.
We have now been through the principal kinds of knowledge and in none
of them have we found a plausible place for the apprehension of facts, so
that the conclusion is suggested that in spite of its importance in some
epistemologies there is simply no such thing as this apprehension. This is
not to deny that there may be apprehension of terms, whether particular
or universal. We can (perhaps) apprehend the colour red and a particular
place in our visual field; against such apprehensious our arguments have no
force since in these cases there is no possibility of error: were there (per
impossible) no such colour as red, we should not be thinking falsely of it;
we simply couldn't think of it at all.!3 But when we come to the judgment
that a certain place is red, the possibility of falsehood shows that our judgment
cannot (even if true) be a relation between us and the fact that the place
is red, for to give this account of it is to make <no sense of> thinking of
62 CHAPTER IV
nothing or not thinking at all. What other account we should give of judgment
is a difficult problem, with which we must shortly deal; but it is a problem
for everyone even those who believe in apprehension. By separating off
apprehensions or infallible judgments, they may narrow the scope of the
difficulty to fallible judgments (since for infallible judgments they suppose
the answer is easy), but the difficulty still remains.
In our view no judgments are apprehensions and we shall argue later14
that this is not merely a limitation of the human mind, that it is not merely
beyond our capacity to apprehend facts, but that such a thing is in the nature
of the case impossible. But even if we are wrong in this, if there <indeed>
are apprehensions, their relative importance can easily be exaggerated. Even
if inspection or perception, for instance, are infallible the knowledge we
obtain by them must be remembered, if we are to use it for inference or
any other process of thought; and with memory the door is opened to fallibility,
and apprehension must give place to ordinary judgment. The typical judgment,
the premisses and conclusion of the typical inference; made in ordinary life,
and not apprehensions and a logic which regards inference as an advance
from apprehension to apprehension15 cannot <should?> have much bearing
on most ordinary thought. For our part we shall dismiss this apprehension,
and confine ourselves to ordinary fallible judgments. These, various as they
are, can be regarded as species of one genus: they do not fall into two classes,
knowledge and opinion whose propositional references are of quite different
kinds.
But before passing on we must consider for a moment the sort of objection
which would be made to our conclusion by some philosophers; if there is
no apprehension, no infallible mode of knowledge, they would ask, what
right we can have to be certain of anything? Thus Cook Wilson16 argues
that it is impossible for a man to decide falsely that the evidence for a proposition
is conclusive, for if such a false judgment were possible "we should never
be sure that any demonstration was true and therefore there would be no
such thing as demonstrative knowledge."
This kind of argument seems to me entirely mistaken. In the first place,
there is no doubt that people do make mistakes as to the conclusiveness of
evidence: and not merely stupid people but even the cleverest, as any teacher
or mathematical analysis knows. This fact, with the risk of error which it
involves, must simply be faced; it cannot be denied; nor can it be escaped
by making a distinction between the genuine apprehension of the conclusiveness
of evidence and the mere fallible belief in its conclusiveness. For even if the
genuine apprehension were infallible that would not banish the risk of error,
unless we had an infallible way of telling whether our conviction were a
genuine apprehension or a mere belief, and the facts show that we have no
such way. We sometimes make mistakes and it's no use pretending we don't.
KNOWLEDGE AND OPINION 63
Our fallibility cannot, therefore, be explained away; but are its consequences
really so disastrous as Cook Wilson seems to suppose? I see clearly before
me a book-case; does the fact that men are occasionally the victims of illusions
mean that I have no right to be certain that what I see is indeed a bookcase?
Does the fact that men sometimes make mistakes in addition mean that if
I and an opponent have arrived independently at the same totals for one
bridge scores, we may not be certain that we are right? To such questions
common sense gives us a perfectly clear answer; illusions are so infrequent
that it is far best for men to be certain that their judgments of perception
are true, and to act accordingly. The possibility of illusion might be allowed
for by their having not a complete conviction but a conviction just a minute
fraction, say one part in a million, short of certainty; and in a sense that
is what they all have. I say, for instance, that I am certain that what I
see is a book case, but I am not so absolutely certain that I could not conceivably
be persuaded that I was wrong, if anyone could bring forward sufficiently
strong arguments to that effect. If one of my friends explained to me that
he had been conducting experiments on optical illusions at my expense, I
might be convinced by him that what I had seen was not a book case at
all. If, however, I put an absolute trust in my judgments of perception, I
could only come to the conclusion that my friend was a liar, and I should
have to take the <b>ally tricks of conjurers for demonstrations of <the> most
extraordinary physical phenomena. So also my arithmetical fallibility means
that r should not be wise to put too much trust in an unchecked addition,
but if two people get the same answer the chance that they got the same
wrong answer by coincidence is negligible and we may be reasonably certain
that they are right. But again not so certain as to refuse to listen to anyone
who claimed to have discovered the contrary.
Our so called direct knowledge, perception, memory, etc. is all of this
nature; we make such judgments with practical certainty which is not however
so complete, that we might not be brought to abandon them if they came
into conflict with other beliefs. In general I trust my memory, but if I
remembered writing something yesterday and found today that there was
something different in my m.s., I should be more likely to conclude that
my memory was defective than that the m.s. has changed in the night. Such
cases, however, are rare and the mass of our perceptions and memories are
never called into question. They are true and made with practical certainty,
a certainty which is justified in the sense that the instinct to judge in this
way leads to truth in the overwhelming majority of cases, and is therefore
one which it is advisable or rather essential for men to trustY Such judgments
being (nearly always) true, certain and justified may properly be called
knowledge, even though the processes which lead to them are not infallible.
Of course, on this view on the rare occasions when they lead to false judgments,
64 CHAPTER IV
those judgments will be justified even though false; but this is not really
a paradox, for everyone must admit that there are circumstances in which
error is humanly speaking unavoidable.
Some of our abstract or universal judgments may seem to be more certain
than this: nothing, we feel, could make us abandon them: but we must not
forget that some of the greatest revolutions in science are made by those
who manage to doubt what everyone has previously thought unquestionable.
The laws of logic are, in our view, in a different position: they are not
genuine judgment and offer no scope for doubt. The same is true of pure
mathematics, but mathematical propositions soon become too complicated
for us fully to appreciate and we easily make mistakes about the properties
of symbols when we cannot clearly conceive their meaning?18
It is essential in the theory of knowledge to realize that our ultimate
premisses are not absolutely certain but liable to be abandoned or modified
in the course of the inquiry. Some have denied this because it conflicts with
their narrow theory of inference as a passage from apprehension to appre-
hension. The syllogism, they say, would be impossible if our conviction of
the truth of one premiss could be upset by our subsequent conviction of
the truth of the second. 19 Quite so; but in fact we sometimes cannot draw
the conclusion of our syllogism, because we are sure that it is false, and
are driven back on our premisses, one of which must be abandoned. More
fundamental than the syllogism is what Mr Johnson calls the antiiogism20
which forbids us to conjoin in thought the three propositions all A are B,
S is A, S is not B. Generally we start by believing two of the three and
use the antilogism to reject the third; but sometimes this third proposition
too is something we believe, and we are given a hard choice as to which
to abandon. But unless we are willing sometimes to reconsider our premisses,
we are in the plight described by Hobbes as that of those who trust to
books, who "do as they, that cast up may little summs into a greater, without
considering whether those little summes were rightly cast up or not; and
at last finding the error visible, and not mistrusting their first grounds, know
not which way to clear themselves; but spend time in fluttering over their
books; as birds that entering by the chimney, and finding themselves inclosed
in a chamber, flutter at the false light of a glass window, for want of wit
to consider which way they came in. "21
Equally serious is the opposite error made by supporters of the Coherence
Theory of Truth, who think that there are no ultimate premisses but only
the need for a coherent system. But evidently we require of our coherent
system that it should embody as many as possible of the things we instinctively
believe. We should not regard as plausible a historical system, however
coherent, which contradicted all our memories: and held that memory was
an illusory faculty or even that it "went by contraries."
KNOWLEDGE AND OPINION 65
NOTES
the coherence theory of truth, but the argument seems to apply as well against our
theory of certainty.
20. Cf. W.F. Johnson, Logic (3 Parts; Cambridge, 1921122124). Part II, p. 78.
21. Thomas Hobbes, Leviathan, Part I, ch. 4.
(001-13)
CHAPTER V
is not so clear is that this analysis, which makes them consist in the holding
of relations between substances and times, is a complete one; that substances
and times are among the ultimate constituents of the world. Of course, no
one would think that the bodies of ordinary life, cows and chairs and socks,
could not be analysed further; or that the fact that a sock is on a chair
was a simple relationship between two entities (and a time) and not a
complicated fact about the spatial relationships of millions of particles. 4
But it might be held, indeed at sometime it used frequently to be held,
that some of the ultimate physical facts were that such and such particles
(atoms or what not) had such and such spatial relations to one another at
such and such times, and that physical events consisted in the holding of
these relations. This view was compatible both with the relative and with
the absolute theory of space, (since the spatial relations between particles
might either be direct or derived indirectly from the spatial relations of the
points in absolute space which the particles occupied), but it presupposed
inevitably an absolute theory of time, although this was not always realised.
For if time is relative, the relations of earlier and later hold primarily
between events, and to say that an event happened at a certain time means
that it happened simultaneously with or before or after certain other events,
times being defined in terms of events. But on the view we are considering,
events are facts about times, and times cannot be defined in terms of events
without circularity. Besides this it can be seen that time cannot be derived
from relations between (or qualities of) substances; for the time series is
essentially asymmetrical: if tl is before t2, t2 is not before t 1, but after it.
But if one set of relationships between substances occurs before another,
there is nothing to prevent it also occurring again after the other. 5
The absolute theory of time is, however, no longer tenable after Einstein's
discoveries, and we have to replace it by a conception of the world as a
four-dimensional manifold or 'space-time'. The elements of this manifold,
which are called point instants, are generally taken to be derived by a process
which Dr Whitehead has called "extensive abstraction" from events, which
are, as it were, regions or volumes of space time, and are taken to be the
ultimate constituents of the world. These events are the terms of spatio-
temporal relations, and instead of substances we have long events or series
of events with certain persistent characters, which belong to each event of
the series or each temporal slice of the long event. It may seem that this
is a mere change of name, for what other difference is there between a
"long event" and a "substance"? But the change of name involves a change
of conception corresponding to the modern assimilation of time and space;
for the old theory of substance treats time and space differently, by allowing
a substance spatial but not temporal parts. If a wall is red in one place and
blue in another, that is because one part of the wall is red and another
70 CHAPTER V
blue; but if the wall is red at one time and blue at another we do not
think that this means that one temporal part of it is red and another blue,
for we suppose, or at any rate talk as if, it were the very same thing which
was both red (at one time) and blue (at another). We have seen how this
refusal of temporal parts to substances involves us in [the absolute theory
of time] the difficult view that the fact that a thing is red at a time is
an ultimate kind of triple relationship between a thing, a colour, and a time;
and so in the absolute theory of time. Now this <is a> difficulty which simply
does not arise in the case of space, since that a wall is red at a certain
place, obviously means that part of the wall is red, and that that part is
at the place, i.e. (on the relative view) has certain spatial relations to other
things. No one would postulate an ultimate kind of relationship between
the whole wall, the colour and the place; the colour so obviously belongs
not to the whole wall but only to part of it.
The change from "substance" to "event" means that we can adopt the
same analysis for time as for space. The new term is chosen because an
"event" can be supposed to have temporal parts, without any straining of
language; for instance, though we cannot speak of the earlier part of the
wall we can perfectly well speak of the earlier part of its history; and the
new "event" which we substitute for the wall is supposed to have temporal
parts like the wall's history as well as spatial parts like the wall. For a wall
to be red at one time and blue at another, means that one part event in
its history is blue and another red; and to say that these events happened
at such and such times is to say that they have certain temporal (strictly
spatio-temporal, since space and time cannot be separated) relations to certain
other events. Time is now defined in terms of events, for events are temporal
parts of substances supplying the terms required for the relations of a relational
theory of time to hold between. Before this was impossible, since substances
had no temporal parts and what were called events were facts about substances
and times, and times could not be defined in terms of such events without
circulari ty.
On this view the identity of a substance through time is constituted by
spatio-temporal continuity and the permanence of certain characters (or
sometimes their systematic variation) throughout its history. What makes
the different temporal parts of a fountain pen all parts of the same thing,
is much what makes the spatial parts of a lawn all parts of the same lawn,
namely continuity and permanence of character, in the one case through
time and in the other through space. 6
The belief in the abstract identity of a substance through time has been
the source of many fallacious arguments purporting to prove the immortality
of the soul, which as Mr Russell has remarked would ... J
Dr Broad in Mind and its place in Nature (pp. 588-598) argues that the
JUDGMENT AND TIME 71
theory that the notion of "material event" can be taken as fundamental,
and the notion of "material substance" as derivative, owes its plausibility
to the tacit assumption of Absolute Space; and he shows, as if it were a
matter of some difficulty, how it can be maintained without assuming
"substantival" but only "adjectival" Absolute Space and Time. Since his view
appears to be contrary of that taken in the present work according to which
the theory that events are primary does not, and the theory that substance
is primary does require Absolute Time, some consideration of his arguments
seems to be required.
As regards time nothing that he says show the slightest need for an Absolute
Theory, whether substantival or adjectival, since nothing is easier than to
replace the temporal qualities which Dr Broad ascribes to events by the
relations of simultaneity, before and after. (The last two he must have in
any case in some form or other.)
As regards space the position is more complicated: the difficulty about
the motion of a homogeneous liquid is a genuine one but easily overcome
in the "Field Theory of Matter" in which the motion is represented by
appropriate terms in the matter-energy tensor.
Besides this, there is a difficulty which only occurs if you start with
"instantaneous punctiform particulars" and not with extended events. This
difficulty is as follows; if you reject both persistent substances and absolute
space, it may seem hard to see how two instantaneous states of the world
are in any way connected together,s though the difficulty is at once removed
if we postulate direct spatial relations between events at different times. But
even the resemblance of a difficulty vanishes when we adopt the method
of extensive abstraction (called by Dr Broad himself "the Prologomena to
every future Philosophy of Nature").9 For we can now define the different
events which form [part of] the history of a thing as a series whose terms
overlap, instead of a series with continuously varying spatial qualities,lO and
as this is the only connexion in which these absolute spatial qualities are
required, they can evidently be dispensed with altogether.
We must now consider the relation of the events, which, on the new
view, are the ultimate constituents of nature to what we previously called
events when we were discussing the substance theory. An event we then
decided, was a fact about the relations (or qualities) of substances at a time:
for instance, an eclipse was the fact that at a certain time the sun, moon
and earth were in line. For such facts, let us quite arbitrarily use the word
"occurrence", in order to distinguish them from the events of the new theory,
which we will continue to call "events".
Of course, the abandonment of the substance theory does not mean that
there are no longer any such things as occurrences; no theory can deny that
there are eclipses, i.e. that sometimes the moon comes between the earth
72 CHAPTER V
and the sun. What has changed is our view as to the ultimate analysis of
such facts; for instance the eclipse is no longer held to consist in a relationship
between substances and a time, but in there being events in those series
of events known as the histories of the sun, moon and earth, which have
certain spatio-temporal relations both to one another, and to other events
by reference to which they are dated. The occurrence is, thus, not an event,
but a fact about events, and when, in questions of philosophical analysis, we
are concerned with something which we naturally call an event, we must
consider carefully whether it is in our special terminology, an event or an
occurrence, for these are fundamentally different categories.
In the case of an eclipse the occurrence is a fact about several events,
which makes the distinction clearer than when, as often happens, the occurrence
is a fact about one event only. In such cases, we are accustomed to use
the same phrase for both event and occurrence without any consciousness
of ambiguity. For instance, "the murder of Caesar" often stands for an
occurrence, and could be replaced by "the fact that Caesar was murdered."
Thus in, "he was aware of the murder of Caesar," or "the murder of Caesar
was due to his ambition," what the man was aware of, and what was caused
by Caesar's ambition, was not an event, i.e. a region of space-time, but
a fact, the fact that Caesar was murdered. This fact, properly analysed, consists
in their <sic for there> being an event at a certain time and place having
certain characteristics and being part of such and such series or long events.
And this event is also called "the murder of Caesar" when we say "the
murder of Caesar preceded the accession of Augustus by - years", for an
interval of time is a relation between two events not one between facts
or occurrences.
It should be realized that our distinction between the occurrence or fact
that there is an event of a certain sort, and the event itself, is not in any
way pedantic or unnecessary; in other cases similar distinctions are so obvious
that no one would dream of bothering to point them out. It requires, for
instance, little subtlety to distinguish between Signor Mussolini and the fact
that Italy has a Duce, and events and occurrences are really as distinct as
the dictator and his dictatorship.
We must now return to the mental events such as judgment with which
we are primarily concerned. Here again no philosophical system can deny
the existence of what we may call mental occurrences. People at various
times judge that various things are the case, and such a fact as the fact
that so-and-so judges such-and-such at a certain time we may call a mental
occurrence, just as the fact that the moon was between the earth and the
sun at a certain time is called a physical occurrence.
The existence of such occurrences is therefore, beyond dispute, but as
soon as we try to analyse them, and ask what are the terms and relations
JUDGMENT AND TIME 73
plausible, for it can hardly arise simply from the ambiguity of 'now' though
this may well be a contributing factor. It is due mainly, I think, to the
inevitable difficulty of imagining a temporal series of events. One way of
doing this is to go through the events one after another in the order in
which they happened, as when one mentally rehearses a tune. But this method
is often unsatisfactory as we may want to have the events in our mind not
one after another, but all at once so as better to see their relations. We
then imagine them spread out before us along a line, replacing the time
order by a space order. This method of representing a temporal series of
events by a spatial series of images or symbols correctly reproduces the
properties of the time series with the exception of its 'sense', the difference
between 'before' and 'after'. This can only be represented by the <sic for
an> arbitrary convention such as that "to the left of' is to stand for 'before',
and not for 'after'. The representation being thus imperfect, forfeits to
convenience much of the essence of the time series, and we try to correct
this by combining it with the other mode of representation (a representation
by a time series) and imagine the events spread out before us being let up
in turn by the bullseye lantern of presentness. We then suppose presentness
to be a real quality which moves down the series, and invent with Mr Dunne!7
another time series for it to move in, or adopt the less logical hypotheses
of McTaggart or Broad. But the whole difficulty comes from combining
two disparate modes of representation. We can represent the time series
of events, by a time series of images which come one by one in succession;
this gives the time quality correct but is inconvenient for purposes of study.
Or we can represent the time series by a space series which we can contemplate
at our leisure, but which is after all spatial and not temporal. But if we
combine the two ways, if we both have all the images spread before us and
light them up one by one, we are representing one series by two and langing
ourselves in a hopeless fog if we use such an inaccurate symbolisation in
any philosophical discussion;!8 but in practical matters the double representation
may be the most convenient, since we can attend to either aspect at will,
and enjoy now the sense of succession, now the repose which enables us
to look backwards and forwards till we find what we want. So long as
we merely alternate the two methods no harm is done, but ideas based on
their combined and simultaneous use are inevitably nonsensical.
NOTES
1. Cf. G.E. Moore, "Facts and Propositions," in "Mind, Objectivity and Fact," Proceedings
of the Aristotelian Society, Supplementary Volume VII (1927), p. 176.
2. See for an instance of this W.E. Johnson, "Analysis of Seeing," Mind, vol. 27 (1918),
p. 142 and G.E. Moore, Proceedings of the Aristotelian Society, vol. 21 (1920-21), p. 136.
But reference above shows that he has changed his view since then.
3. But not necessarily exactly the fact mentioned, which is to some extent indeterminate
or general, but the absolutely specific fact that the particles of the sun, moon and earth
are in such and such definite relations.
4. It will, no doubt, be objected that since we can know that the sock is on the chair
without knowing anything about the particles, these cannot come into the analysis of
the fact we know, the fact, namely, that the sock is on the chair. This is perfectly
true, and what is said refers strictly not to the fact about actual or possible sensations
which a man knows when he knows that a sock is on a chair, but to the specific physical
fact, which is the possibly unknown necessary and sufficient condition for what he knows
to be true.
JUDGMENT AND TIME 77
5. Cf. the first part of Mr Russell's article on "Is Position in Time and Space Absolute
or Relative," in Mind, vol. 10 (1901), pp. 293-317. It need hardly be mentioned that
this article does not give his present views.
6. For an admirable discussion of the relations of events and objects see A.N. Whitehead's
An Enquiry Concerning the Principles of Natural Knowledge (Cambridge, 1919) and Concept
of Nature (Cambridge, 1920).
7. <Ramsey broke off at this point.>
8. This difficulty seems to be the one exploited by Kant in the proof of the First Analogy.
9. C.D. Broad, Scientific Thought (London, 1923), p. 4.
10. Cf. C.D. Broad, Mind and its Place in Nature (London, 1925), p. 597.
11. Broad, Mind and Its Place in Nature (op. cit.), p. 559. Of his two forms of the Pure Ego
theory the first is in our sense a substance theory, the second an event theory.
12. Cf. G.E. Moore, "Facts and Propositions," in "Mind, Objectivity, and Fact," Proceedings
of the Aristotelean Society, Supplementary Volume VII (1927), pp. 176-7.
13. A fact of Prof. Moore's first class. Op. cit., p. 174.
14. McTaggart, The Nature of Existence (Cambridge, 1927), Vo!' 2, Ch XXXIII.
15. Broad, Scientific Thought (London, 1923), Ch. 2. See also R.B. Braithwaite, Proceedings of
the Aristotelian Society, Supplementary Volume VIII (1928), pp. 162-174.
16. Dr Broad's theory that the future is unreal is, of course, no solution at all, supposing,
that is, that the difficulty is geuuine. For the contradiction that the death of Queen
Anne was both future (up to 1714) and past (after 1714), he merely substitutes the
contradiction that there both was no such event as Queen Anne's death (up to 1714)
and was such an event (after 1714).
17. J.W. Dunne, An Experiment with Time. (London, 1927).
18. Russell's argument against Neutral Monism based on Emphatic Particulars (Monist, July
1914) seems to involve just this muddle. Since his conversion to Neutral Monism no
more has been heard of it, although in 1919, it was the argument on which he placed
most reliance (Monist, 1919, p. 55).
19. MacTaggart, Nature of Existence, Vo!' 2, Ch. XXXIII.
20. Broad, Scientific Thought, Ch. 2.
21. R.B. Braithwaite, Proceedings of the Aristotelian Society, Supplementary Volume VIII (1928),
pp.162-174.
22. Bertrand Russell, Principles of Mathematics, sect. 442; quoted by McTaggart, op. cit., p. 14.
APPENDIX I
OLDER DRAFTS
(001-1S)
CHAPTER I
INTRODUCTORY
lead us to hold them; the man who is led to a true opinion by a bad argument
finds no favour with the logician, whose criticism is directed not to the
primary value of truth in judgment, but to the secondary value of reasonableness
in argument. 1
That the value of truth is primary and of reasonableness secondary is
a point on which we must insist; for it follows from the fact that the whole
purpose of argument or inference is to arrive at true judgments, that the
reasonableness of an argument must be connected with some tendency in
it to lead to true conclusions. In fact, the validity of arguments is [derived
from] related to the truth of opinions, in somewhat the same way as the
rightness of actions is related to the intrinsic value of their consequences.
But the precise nature of the connection must be left to be determined later
when we come to the actual analysis of the concept 'reasonable'.
With regard to the meaning of any concept of value such as goodness,
[beauty,] truth or validity there are three main schools of opinion which
may, perhaps, be called idealist, realist and naturalist. With the idealists such
as Green, Bradley and Bosanquet I shall not deal; their writings seem to
me to be almost entirely nonsense; the living issue is between the realists
and the naturalists. [It is, of course, possible to take different views of two
different kinds of value, but the arguments in all cases are so much the same
that if they were set out clearly, this would only rarely happen.]
Some philosophers, of course, are realists for one sort of value and naturalists
for another, but this is generally due to confusion, since the arguments for
realism are so similar in all three fields, that if they were clearly apprehended
they would be found convincing either in all cases or in none.
In regard to ethics the opposition has been clearly set out in the writings
of Dr Moore; in his Principia Ethica he convicted of "the naturalistic fallacy"
all those who sought to define 'good' in terms of any natural attribute such
as the pleasurable, the approved or the desired. He argued that 'good' stood
for none of these things but was the name of a simple [indefinable] unanalysable
quality, which was very different from all ordinary qualities, stood in a sense
'outside' the natural world, and formed the distinctive subject matter of ethics.
Dr Moore's attitude, in contrast to the naturalistic views which he condemns,
I shall call, for want of a better word, realistic. [Its essence does not lie
exactly in the declaring that good is unanalysable, because, for instance, another
ethical predicate, right (as applied to actions) was said by Dr Moore to be
analysable in terms of good, an action being right if it has the best possible
consequences. ]
By this term <'realism'> I shall not mean his general tendency to declare
<as> unanalysable concepts which others have claimed to be able to analyse
[although that is certainly a general tendency of this school of thought];
but [simply] his belief that ethical valuation involves a distinctively ethical
INTRODUCTORY 83
predicate not belonging to the natural world. It is this rather than the simplicity
of 'good' that I take to be characteristic; thus I shall call realists in logic
not only people who hold that truth or validity are unanalysable, but also
those who think that they involve in their ultimate analysis one or more
distinctively logical predicates or relations, such as a unique kind of cor-
respondence or indefinable probability relations.
[The principal upholders of naturalism in logic in recent years have been
the pragmatists. ]
Opposed to this almost universal view, is the position defended by the
pragmatists and behaviourists, and in his recent works by Mr Russell2 , that
all the notions used in logical criticism can be defined in terms of natural
qualities and relations, such as are involved in ordinary psychological in-
vestigations. This naturalistic view of logic will be taken in the present work,
in which I hope to show how truth and reasonableness can be defined without
assuming any unique unanalysable logical relations, and further to show that
the current modes of explaining logic by means of such relations are altogether
untenable.
NOTES
1. [It remains to be examined later whether the reasonableness of arguments can be determined,
independently of the knowledge of actual fact provided by actual fact. J It remains however
to be seen whether, as the above paragraph suggests, the validity of arguments is really
independent of the positive knowledge provided by science. We shall show later that
certain types of valid arguments are in this sense a priori while others are not.
2. The position of Mr Russell, as we shall see later, is peculiar; in general he is a naturalist
but he appears to accept Mr Keynes' probability relations in a way altogether at variance
with his <OWll> main position.
(001-01)
CHAPTER I
dispositions. 2 But, of course, not all states of mind are true or false but only
such as have what I shall call a propositional reference.
As typical of a mental state with a propositional reference we may take
a belief; a belief is necessarily a belief that something or other is so and
so, for instance that the earth is flat, and it is this aspect of it, its being
"that the earth is flat", that I propose to call its propositional reference.
So important is this character of propositional reference, that we are even
apt to forget that belief has any other aspects or characters at all. Thus
if two men both believe that the earth is flat, we say that they have the
same belief in spite of the fact that they may believe it at different times
for different reasons and with different degrees of conviction, and express
their beliefs in different languages, words or images; if the propositional
references are the same, if they are both beliefs that the same thing, we
commonly ignore all other differences between them and call them the same
belief. It is usual in philosophy to express this resemblance between two
men's beliefs, not by saying, as I do, that they have the same propositional
reference, but by calling them beliefs in the same object, objective or
proposition, and there have been endless disputes about what kind of entity
or fiction this object or proposition is, as it is only the hardiest verbalists
who can persuade themselves that "that the earth is flat" is the name of
something real. Our change of language wi11 not, of course, enable us to
escape these difficulties, but it will make it possible to postpone them for
a while and to conduct our present discussion without presupposing any
particular solution of them. Whatever the objective of a belief may be, its
character of reference is undoubtedly real, and by basing our treatment on
this character, as something with which we are all familiar, we can [postpone
the difficulty until later when the character of reference must be analysed
<and so>] simplify the problem considerably.
Propositional reference is not, of course, confined to beliefs; my knowledge
that the earth is round and my opinion that free trade is superior to protection,
any form [in fact] of thinking of knowing or being under the impression
that - [has a propositional reference, and all of these are either true or
false, except knowing which is ... ], any mental state in fact which is either
true or false has a propositional reference, and in discussing truth and falsity
it is with such states alone that we sha11 be concerned. On the other hand
it is not the case that any state with propositional reference must be either
true or false. I can wish it would be fine to-morrow, wonder whether it
will be, and finally believe that it wi11 be; these three states a11 have the
same propositional reference but only the belief can be called true or false.
We do not call wishes, desires or wonderings true; not because they have
no propositional reference, but because they lack what may perhaps be ca11ed
an affirmative or assertive character, the element that is present in thinking
86 APPENDIX I: OLDER DRAFT VERSIONS
that, but absent in wondering whether. At the opposite pole to this affirmative
character, there is perhaps a negative character present in acts of denial or
disbelief; these we do not usually call true or false, though we can call them
correct or incorrect, words which can for the present be regarded as definable
in terms of true and false as applied to affirmative states.
Such an account may not be ultimately correct and will have to be
reconsidered when we come to deal with negation, but for our present
preliminary purposes we can say that denying that so and so is the case
is correct or incorrect, according as asserting that it is the case is false or
true, and confining our attention to affirmative states with propositional
references, say that it is these and these alone which are either true or false. 3
And of these we shall take, for convenience, beliefs as typical.
[If, then, a state of mind has an affirmative character and a propositional
reference it will be either true or false; and we have to discover the meaning
of the ... J
It is important to realise that the words "true" and "false," when properly
used, are not vague terms indicating praise or blame of any kind, but have
a quite definite meaning. There are various respects in which a belief can
be regarded as good or bad: for instance it can be held for good or for
bad reasons, and it can either be isolated or part of a coherent system of
knowledge, but it is not to these merits or defects that we refer when we
call the belief true or false. To take an old example of Mr Russell's,4 someone
who thought that the present Prime Minister's name began with B would
think so truly, even if he had derived his opinion from the mistaken idea
that the Prime Minister was Lord Birkenhead; and it is clear that by calling
a belief true we neither mean nor imply that it is either well grounded,
or comprehensive, and that if these qualities are confused with truth, as they
are by Bosanquet,S any profitable discussion of the subject becomes quite
impossible. The kind of merit in a belief to which we refer in calling it
true can be easily seen to be something which depends only on the beliefs
propositional reference;6 if one man's belief that the earth is round is true
so is anyone else's belief that the earth is round, however little reason he
may be able to give for thinking so.
After these preliminaries we must come to the point: what is the meaning
of 'true '? It seems to me that the answer is really perfectly obvious, that
anyone can see what it is, and that difficulty only arises when we try to
say what it is, because it is something which ordinary language is rather
ill-adapted to express.
Suppose a man believes that the earth is round; then his belief is true
because the earth is round; or generalizing this, if he believes that A is B,
his belief will be true if A is B, and false otherwise, [and it is clear, that
this connection is not a mere coincidence but arises from and constitutes
THE NATURE OF TRUTH 87
the nature <of truth> and it is clear that what is meant by 'true' is simply
this - "believing that A is B when A is B." To believe truly is to believe
that A is B, when A is B, or to believe that A is not B, when A is not
B, or to believe that either A is B or C is D, when either A is B or C
is D, and so on, through all the possible forms of belief.]
It is, I think, clear that in this last sentence we have the meaning of truth
explained, and that the only difficulty is to formulate this explanation strictly
as a definition. If we try to do this the obstacle we encounter is that we
cannot describe all beliefs as beliefs that A is B since the propositional reference
of a belief may have any number of different more complicated forms. A
man may believe that all A are not B, or that if all A are B, either all
Care D or some E are F, or something still more complicated. We cannot,
in fact, assign any limit to the number of forms which may occur, and must
therefore be comprehended in a definition of truth; so that if we try to
make a definition to cover them all it will have to go on forever, since
we must say that a belief is true, if supposing it to be a belief that A is
B, A is B, or if supposing it is to be a belief that A is not B, A is not
B, or if supposing it to be a belief that either A is B or C is D, either
A is B or Cis D, and so on ad infinitum.
In order to avoid this infinity we must consider the general form of a
propositional reference of which all these forms are species; any belief whatever
we may symbolise as a belief that p, where "p" is a variable sentence just
as "A" and "B" are variable words or phrases (or terms as they are called
in logic ). We can then say that a belief is true if it is a belief that p, and
p.7 This definition sounds odd because we do not at first realize that "p"
is a variable sentence and so should be regarded as containing a verb; "and
p" sounds nonsense because it seems to have no verb and we are apt to
supply a verb such as "is true" which would of course make nonsense of
our definition by apparently reintroducing what was to be defined. But "p"
really contains a verb <already>; for instance, it might be "A is B" and
in this case we should end up "and A is B" which can as a matter of ordinary
grammar stand perfectly well by itself.
The same point exactly arises if we take, not the illustrative symbol 'p',
but the relative pronoun which replaces it in ordinary language. Take for
example "what he believed was true." Here what he believed was, of course,
something expressed by a sentence containing a verb. But when we represent
it by the pronoun 'what', the verb which is really contained in the 'what'
has, as a matter of language, to be supplemented again by "was true." If
88 APPENDIX I: OLDER DRAFT VERSIONS
however we particularize the form of belief in question all need for the
words "was true" disappears as before and we can say "the things he believed
to be connected by a certain relation were, in fact,S connected by that relation."
As we claim to have defined truth we ought to be able to substitute
our definition for the word 'true' wherever it occurs. But the difficulty we
have mentioned renders this impossible in ordinary language which treats
what should really be called pro-sentences as if they were pro-nouns. The only
pro-sentences admitted by ordinary language are "yes" and "no", which are
regarded as by themselves expressing a complete sense, whereas "that" and
"what" even when functioning as short for sentences always require to be
supplied with a verb: this verb is often "is true" and this peculiarity of
language gives rise to artificial problems as to the nature of truth, which
disappear at once when they are expressed in logical symbolism, in which we
can render "what he believed is true" by "if P was what he believed, p".
So far we have dealt only with truth; what about falsity? The answer
is again simply expressible in logical symbolism, but difficult to explain in
ordinary language. There is not only the same difficulty that there is with
truth but an additional difficulty due to the absence in ordinary language
of any simple uniform expression for negation. In logical symbolism, from
any proposition symbol p (corresponding to a sentence), we form the
contradictory -Ip (or-p in Principia Mathematica): but in English we often
have no similar way of reversing the sense of a sentence without considerable
circumlocution. We cannot do it merely by putting in a "not" except in
the simplest cases; thus "The King of France is not clever" is ambiguous,
but on its most natural interpretation means "There is a King of France
but he is not clever" and so is not what we get by simply denying "The
King of France is clever"; and in more complicated sentences such as "if
he comes, she will come with him" we can only deny either by a method
special to the particular form of proposition, like "if he comes, she will
not necessarily come with him" or by the general method of prefixing "It
is not true that ... ", "It is false that - " or "It is not the case that ... ", where
[ again] it looks as if two new ideas, "truth" and "falsity," were involved,
but in reality we are simply adopting a round-about way of applying not
to the sentence as a whole.
Consequently our definition of falsity (to believe falsely is to believe p,
when -Ip) is doubly difficult to put into words; but to argue that it is circular,
because it defines falsity in terms of the operation of negation which cannot
always be rendered in language without using the word "false", would simply
be a confusion. "False" is used in ordinary language in two ways: first as
part of a way of expressing negation, correlative to the use of "true" as
a purely stylistic addition (as when "it is true that the earth is round" means
no more than that the earth is round); and secondly as equivalent to not
THE NATURE OF TRUTH 89
In the last pages we have been defining truth and falsity in terms of propositional
reference, which we have taken for granted. For instance, our analysis of
what is meant by "At 10 o'clock Mr. Jones had a true belief' is roughly
"At 10 o'clock Mr. Jones was believing that certain things were related in
a certain way, and they were related in that way," in which we are leaving
altogether unanalyzed the notion of believing that - , i.e., of propositional
reference. The analysis of this will be our task later, and for the present
all I wish to do is to mention the possibility that this analysis will throw
the nature of truth in a new light; we have seen, we may say, how any
analysis of believing that can be extended at once to give an analysis of
believing truly; but it may turn out that we have really been putting the
cart before the horse and that we can only analyse believing that by reference
to believing truly. I do not think myself that this is so, but since it is the
view of some philosophers 1o it cannot be ruled out in advance.
[But while we must for this reason regard our definition of truth as to
some extent merely provisional, it ... ]
But this doubt does not take us very far. At worst it means that our
definition of truth is redundant and capable of simplification, for nothing
can make us doubt that a true belief is one which is a belief that p, when
p. Even if this is not the correct definition, it still expresses something which
arises necessarily from the nature of truth, something with which the proper
definition is certainly compatible. And so this is all we require to take our
bearings with regard to some famous doctrines, such as Pragmatism and the
Correspondence and Coherence Theories. For this purpose we may sum up
our own discussion of Truth as follows: If with regard to a belief we know
what it is a belief that, for instance, that it is a belief that the earth is
round, then we know in what case it is true and in what false, in the instance
given it is true if the earth is round, false otherwise. This is absolutely obvious
and a theory of truth which contradicts it is quite certainly wrong.
On the other hand, there is another question, the question as to what
constitutes propositional reference (i.e., what is meant by saying a belief
is a belief that so-and-so), which we have not discussed, so that we are
not yet in a position to reject any answer to it which may be suggested.
89
90 APPENDIX I: OLDER DRAFT VERSIONS
Let us call these two questions I and II, Question I being "Given propositional
reference, what determines truth?" and Question II "What constitutes
propositional reference?"
The first main type of theories of truth are Correspondence Theories;
and they have primarily to be divided according to whether it is Question I
or Question II that they hold should be answered by reference to cor-
respondence. Thus Mr Russell in Philosophical Essays and his Home University
Library volume <on The Problems of Philosophy> only uses correspondence to
answer Question I, but in his The Analysis of Mind, in definite disagreement
with his earlier view, he introduces correspondence in his answer to Question II.
To Question I the correspondence theorists give substantially the same
answer as ours. They too think that Mr Smith's belief that A is B is true
if A is B, and they say that in such a case the truth of the belief is constituted
by a relation between the fact that Mr Smith thinks that A is B on the
one hand, and the fact that A is B on the other.!1 This relation they call
"correspondence," but nothing precise can be said about its nature until we
know how to analyse the fact that Mr Smith thinks that A is B, i.e., until
we have answered question II. It is clear that this theory amounts in a sense
to the same thing as ours, and we shall have to discuss later whether [it
is a more correct way to word it] and how it is preferable.
We shall find that ours is a superior method of statement in that we
are able to avoid mentioning either correspondence or facts. One advantage
of this can be easily seen: suppose what is believed is a disjunction, as when
Smith thinks that Jones is either a liar or a fool: this belief, we say, is true
if Jones is either a liar or a fool, and surely this is a better way to explain
truth in such a case than one which introduces "the fact that Jones is either
a liar or a fool", since although such a phrase is permissible as a matter
of language, there is no such fact, no such either-or in reality but either
the fact that he is a liar or else the fact that he is a foo1. 12
The correspondence answer to Question I we shall, therefore, regard as
a cumbrous and possibly inexact method of expressing our answer, but it
is a useful phrase, which leads to very little harm, to speak of a true belief
as one which corresponds to a fact, [leaving vague] defining the type of
correspondence no further than as being that type (or types, since there may
be different with different forms of belief) of relation that there [obviously]
must be between thinking that such and such is the case and such and such
actually being the case.
But the correspondence theory is as we have said, [also] sometimes put
forward as an answer to Question II; it is held that to believe that A is
B means to have in your mind an entity or entites corresponding to or
representing A being B; by "correspondence" being here meant some specific
relation such as resemblance. When we come to discuss it in Ch. II we
THE NATURE OF TRUTH 91
shall find that in spite of its unpopularity this view is part, though not the
whole, of the truth.
[As regards Question I we can, then, say that we agree with the
correspondence answer, but propose so to simplify it as to eliminate all
mentioning of correspondence, but as regards Question II we must postpone
the discussion of it; we shall find later that the true answer is very complicated
and contains correspondence only as one of its elements. [If further we are
asked what kind of correspondence it is that occurs is referred to in ... ]]
The next theory of truth to be considered is Pragmatism, the view that
by a true belief is meant one which is useful. By the inventor of Pragmatism,
C.S. Peirce, this was, I think, given primarily as an answer to Question II,
a belief that A is B, being roughly a belief leading to such actions as will
be useful if A is B, but not otherwise.!3 Although this is not the whole
analysis of propositional reference, I think that like correspondence it gives
one element in it, and is not simply to be mocked at. And when stated
in the above crude form it leads to the conclusion that a belief is true if
and only if it is useful, and this conclusion is by no means ludicrous but
in complete harmony with my account of truth. For a belief that A is B,
means on this view, a belief which is useful if and only if A is B; such
a belief will therefore be useful if and only if A is B, i.e., if <and only
if> it is true; and so conversely it will be true if and only if it is useful.
What is ludicrous, is not the general idea, although in such a crude form
this is obviously incorrect, but the way in which William James confused
it especially in its application to religious beliefs. This application depended
on using the theory in just those cases in which its crudeness and need for
complication and refinement are most glaring and so turning a sensible sort
of answer to Question II into an absurd answer to Question 1.
Let us take, for example, the belief in hell. To say a man believes in
hell means, according to the pragmatists that he avoids doing those things
which would result in his being thrown into hell, and which he would not
avoid on account of any other consequences they may have. Such conduct
will be useful to the man if it really saves him from hell, but if there is
no such place it will be a mere waste of opportunities for enjoyment. But
besides this primary utility there are other ways in which such conduct may
or may not be useful to the man or others; the actions from which a belief
in hell would cause him to abstain might bring disasters in their train either
for him or for others even in this present life. But these other consequences
of the belief, whether useful or not, are clearly not relevant to its propositional
reference or truth, and if the pragmatist definition makes them relevant this
only shows that the definition must be corrected by qualifying clauses or
even abandoned altogether. William James, on the other hand, included
explicitly these further kinds of utility and disutility, which must obviously
92 APPENDIX I: OLDER DRAFT VERSIONS
NOTES
1. This may need some qualification in view of the possibility that when we think in words,
the words may be not so much an expression of our thought as a part or even the
whole of the thought itself.
2. See infra, Ch. dII "Judgment''>.
3. Assumptions entertained without belief are also called true or false by an extension of
meaning easy to explain. For refutations of other theories on which truth is ultimately
independent of the affirmative character of mental acts, see below.
4. <Bertrand Russell, The Problems of Philosophy (Oxford, 1912), p. 205.>
5. [Bernard Bosanquet, Logic, Vol. II, 2nd ed. (Oxford, 1911), pp. 282 ff. He even confuses
a true statement with one made in good faith.] Footnote to go in here on next page.
d t is given in Chapter I as note 5 thereof.>
6. It has been suggested by Professor Moore that the same entity may be both a belief
that (say) the earth is round and a belief that something else; in this case it will have
two propositional references and may be true in respect of one and false in respect
of the other. I think that everything I say could easily be altered so as to allow for
this possibility, but the complication of language which would result seems to me to
outweight the possible gain in accuracy.
7. In Mr Russell's symbolism
8. In a sentence like this "in fact" serves simply to show that the ratio obliqua introduced
by "he believed" has now come to an end. It does not mean a new notion to be analyzed,
but is simply a connecting particle.
9. See below. <Presumably a reference to the projected chapter on negation is intended.>
10. E.g., the late Dr McTaggart (The Nature of Existence, Vol. I (Cambridge, 1921), p. 11,
sect. 10), who held that a true belief was one which corresponded to a fact, and a
belief that p was one which either corresponded (if true) or merely "professed to
94 APPENDIX I: OLDER DRAFT VERSIONS
correspond" (if false) to a certain fact. On this view believing truly that p is evidently
a simpler notion than believing that p; [but in the absence of any account of the meaning
of "professing" it is not one which need be taken very seriously] and much the same
would follow from some of Mr Russell's language in the The Analysis of Mind (London,
1921), pp. 272-273.
11. The fact that A is B, is something that exists if and only if A is B. Hence if the belief
is false there is no such corresponding fact.
12. <The paragraph originally reads as follows:> [<It is clear> that this is decisive in its
favour. One simple consideration in this sense can be urged at once: suppose that what
is believed is a disjunction as when Smith thinks that Jones is either a liar or a fool;
this belief as such is true if Jones is a liar or if he is a fool, and surely this is a better
way to explain its truth in such a case, than one which introduces "the fact that Jones
is either a liar or a fool," since although such a phrase is permissible as a matter of
language, there is a sense in which there is [evidently1 no such [disjunctive1 fact, no
such mere either-or in reality, but <only> either the fact that he is a liar or else the
fact that he is a fool.]
13. Wanted: Note on Peirce. <Ramsey's principal access to Peirce was via Morris R. Cohen
(ed.), C.S. Peirce: Chance, Love, and Logic (New York, 1923).>
14. <At this point, Ramsey initially proposed to close the chapter with the following,
subsequently deleted, passage: [The last well known theory of truth is the Coherence
Theory; unlike the others this affords us no assistance whatever in answering either
of our questions but derives its plausibility entirely from [certain] confusions. It is evident
that coherence does not in general determine either the propositional reference or the
truth of beliefs. The beliefs of a man suffering from persecution mania may rival in
coherence those of many sane men but that does not make them true; besides, coherence
is a character [property] of a system, and truth and reference are characters of single
beliefs.]>
15. William James, Pragmatism: A New Name for Some Old Ways of Thinking (New York and
London, 1907), p. 222.
(001-09)
CHAPTER II
This contention that no single judgment can be wholly true must, of course,
be distinguished from the much more plausible contention that no single
judgment can be made with complete certainty. It may be, that I cannot
be really certain that Charles I was beheaded, but this does not prove that
that he was beheaded is not wholly true and part of the whole truth.
Nevertheless truth and certainty are often confused and those different persons
who "make no claim to absolute truth" frequently mean no more than that
they are not absolutely certain.
The contention, then, that we have to consider is that no single judgment
can be absolutely true, and if this connection is right all that we have hitherto
said must be wrong, or at the least, require serious modification, for we
have assumed that any belief or judgment has a definite propositional reference
or meaning in virtue of which it is without qualification true or false. The
most obvious reason for doubting this assumption lies in the possibility that
judgments may be vague in the same sort of way as the words we use to
express them. How many hairs must a man have not to be bald? There
is no definite number, and if someone is judging that he is bald, his judgment
may be vague and not definitely either true or false. 5 But this cannot be
our opponents' ground of objection, since a system of vague judgments would
be no less vague than its members; for their contention is that such a system
is capable of truth in a higher degree than its members, and the obstacle
they see to absolute truth can not lie in vagueness. 6
It will be replied no doubt? that single judgments have some meaning,
but not the full meaning that they have in a system of knowledge: and so
to assess the value of this reply we may use the comparison made by Prof.
Joachim between a system of knowledge and a symphony. The notes in a
symphony, it is contended, are not what they would be apart from the
symphony, but are determined by their mutual relations. This is true in the
sense that what note comes next depends on what would best continue the
melody etc., but it is not true if it means that each note has not its own
definite pitch. If pitch were not a property of single notes we could not
talk of a symphony at all in the sense in which we speak of Beethoven's
Fifth Symphony, as opposed to certain particular performances, for the Fifth
Symphony is essentially a system of notes each of a definite pitch, which
could occur also in other symphonies or even in tuning up.s [Similarly, if
each judgment had not its own meaning. To say that the pitch, a note has
in the symphony, was different from that which it would have outside is
absurd; for what is it?] So also each judgment in a system of knowledge
must have a definite meaning, which could also occur outside that system.
Otherwise "system of knowledge" would be just a meaningless phrase.
THE COHERENCE THEORY OF TRUTH 97
NOTES
1. <Bertrand Russell Philosophical Essays (London, 1910), pp. 150-169; see esp. pp. 156-160.
See also C.D. Broad "Mr Bradley on Truth and Reality," Mind, vol. 23 (1914), pp. 349-
370; see esp. pp. 361-363.
2. See especially Harold H. Joachim's The Nature of Truth; (Oxford 1906), and Bernard
Bosanquet's Logic (2nd ed.; Oxford, 1911), especially Vol. II, Chap. IX, which will be
cited as Joachim and Bosanquet respectively. Also see F.H. Bradley, Essays on Truth and
Reality (Oxford, 1914).
3. 'Judgment' is the word used by these authors, where we have been using 'belief, and
its 'meaning' is what we have called "propositional reference".
4. Joachim, op. cit., p. 89. For his defense of the position see p. 85-113.
5. Vagueness must not be confused with indeterminateness. "He has fewer than 50 hairs"
is less determinate than "He has exactly 37 hairs," but not vague in the way that "He
is bald" is.
6. The following analogy may perhaps be helpful, though it should not be pressed too far.
The dependence of the meaning of a judgment on its context may be compared to the
way in which the colour of one part of the visual field may be affected through contrast
by the colour of a neighboring part. And just as each part of the field has its own colour,
whatever may be the cause of its having that colour, so each judgment has its own meaning
and according to that meaning is true or false.
7. Joachim, op. cit., p. 102.
8. This is not to say that a symphony is a collection of notes; what kind of a system it is
is irrelevant; all we require is the plain truth that the elements of the system are notes.
(001-03)
<CHAPTER lIb
<JUDGMENT>
The truth or falsity of a mental state we found to depend upon its propositional
reference, and our next problem is to analyse or explain this characteristic;
but before starting on this, it is as well to consider certain preliminary
distinctions between the states of mind in question.
The most important of these is that between acts and dispositions. In ordinary
language we use knowledge, belief, and opinion primarily of dispositions;
"he knows he's got to leave at the end of the term," "he believes that when
he's dead he'll go to heaven," "in his opinion Protection would cure
unemployement"; in all these cases we are not speaking of an act in the
man's mind with a definite date, but of a general state or disposition. When
we say he knows he's got to leave or he knows his multiplication table,
we are talking of enduring dispositions of his mind, manifested at times in
particular acts of knowing, but conceived as existing even when not so
manifested, just as a man is called courageous even when not at the moment
displaying his courage. Again by an act or acts of thinking at a certain date
we "form an opinion," and the opinion so formed remains our opinion until
we 'change' it; although we may never think about the matter in the meantime,
our opinion is supposed still to be there and ready to be given to anyone
who happens to ask for it.
But we also have other words which refer not to dispositions but to definite
dateable (but not necessarily instantaneous) acts of mind. Thinking, as in
"I was just thinking that its going to rain," (but we should never say "I
was just knowing that 2 + 2 = 4"), judging, inferring, asserting, perceiving,
discovering and learning all refer to acts not to dispositions. And we can
also use knowledge and opinion in this sense in the combinations "coming
to know" and "forming the opinion."
It is clear that in common language both acts and dispositions can be
called true or false, and that both have in some sense propositional references.
But it seems also clear that the fundamental use of truth and reference is
that in which they are applied to acts, for whatever is the correct account
JUDGMENT 99
mistaken. In general the beliefs on which we act are true, but when just
one of them turns out to be false, as for instance when the Union has moved,
our attention is fixed on that one and our conduct condemned as erroneous
in one particular respect.
Weare thus led to the following conception; we have on the one hand
particular acts of thinking, knowing or judging that, and on the other our
dispositional beliefs or knowledge. These dispositional beliefs manifest them-
selves in two ways: firstly by giving rise to corresponding judgments when
occasion arises for making them, and secondly by governing our actions,
roughly according to the rule that we perform those actions which if our
beliefs were true would have the most satisfactory consequences.
Dispositional beliefs themselves can be caused by acts of thought, as when
we form an opinion by thinking the matter over; but they can also be caused
in other ways. Thus if a man sees what looks like a lump of iron and stoops
to pick it up, he makes the bodily adjustments necessary for lifting a heavy
weight, and if it turns out to be a light shell he is surprised and says he
thought it was a solid lump of iron. But save in exceptional cases he didn't
really think about it at all, his bodily adjustments being due to 3 a merely
dispositional belief that it was solid, which caused him to behave as if it
were solid, and would have caused him if asked to say it was. (At least
that would be his first impulse, though on reflection he may have become
doubtful.) Now this dispositional belief, or as it is called by Dr Broad4 quasi-
belief, is not caused by a corresponding judgment or other thinking in the
past. If the man has never seen this lump before he has never thought about
this lump before, and his quasi-belief about it cannot be caused by any previous
thought about it. It is obviously due to his experience of similar lumps in
the past, but it is not even necessary that he should have thought about
those, since he might evidently get into the way of making such adjustments
without ever thinking at all. In view of its temporary nature, it may seem
absurd to call such a belief or quasi-belief a disposition at all, especially
as a way we suggested of distinguishing acts from dispositions was by their
definite dates. But it is evident that this is not the real ground of distinction
and the man's belief that the lump is solid is fundamental[ly], analogous to
my belief that the Union is in Bridge Street. Each of them is known by
its influence on actions and neither of them is an act of thinking; nor is
my belief about the Union strictly always the same enduring disposition.
For what is relevant is every day something different, namely that on that
day the Union is in Bridge Street, and the belief which primarily influences
my action is every day a new one.
To give an adequate account of dispositional beliefs or dispositional
knowledge would be a very difficult task. Their logical status is far from
clear. Dr Broad speaks as if they were identical with the actions to which
JUDGMENT 101
they lead. This cannot be quite right even in the simplest cases. But it can
well be held that in speaking of them we are really talking about actions
and thoughts, a statement about a dispositional belief having some complicated
meaning of that sort. In this case they would be what Mr Russell calls "logical
constructions"; alternatively they may be entities or characters inferred or
hypothecated to explain our actions and thoughts; [and] the problem of their
status is very analogous to that of the unobservable entitites in physics, and
I may perhaps be forgiven for not discussing it here, especially as our primary
concern is with acts of thought, and not with dispositions. The reason I
have lingered so long on the latter is that I do not think it is possible to
understand the propositional references or truth or falsity of thoughts without
considering the effects they have on our acting either directly or indirectly
through dispositional beliefs. In particular, the phrase "degree of belief' as
it occurs in the theory of probability can only be understood by taking into
account the effects of belief on action.
There is, however, an objection that may be made somewhat in the following
terms. "The distinction you make is radically unsound. The bodily adjustments
of the man lifting the weight you ascribe to a dispositional belief, on the
ground that he did not think about it.S But suppose he had thought about
it, how would the case have been different? When you say he thought about
it all you mean is that he said to himself 'Its solid', and his saying this to
himself is simply a movement of the muscles of the throat not radically different
from the movements of the muscles in his arms?" One reply I could make
to this would be to deny that a thought could possibly be merely a movement
of the throat, or even of the brain, but I do not wish to assume even this,
and prefer to try to find a reply which might be convincing even to such
an objector. Let us, therefore, admit for the purpose of argument that thought
is only a movement of the throat; still, I would contend that although there
may possibly be cases which fall between our two alternatives the distinction
is valid and important.
If the man says to himself "The lump is solid," all that takes place may
be movements of his throat, but these words or movements have a fairly
definite meaning or propositional reference in virtue of which thay are true
or false; whereas the movements in his arms have no such meaning; they
have in a vague way reference to the environment, he would not make
them if he knew how light the lump really was or if he thought it was
too heavy to lift or fastened to the floor, but they have not the same precise
or relatively precise meaning as words have. The distinction is not necessarily
that between thoughts and bodily movements, but between actions, whether
of mind or body, which have definite or fairly definite propositional references
or meanings and those which have no such references.
After this preliminary discussion, let us set aside dispositional beliefs and
102 APPENDIX I: OLDER DRAFT VERSIONS
NOTES
CHAPTER V
this is part of its nature which requires careful discussion, for the harmless
looking word 'event' conceals an ambiguity which has been the source of
considerable confusion.
[The problems to which this chapter is devoted concern not only judgment
but many other kinds of events both mental and physical. The treatment
of questions of such generality may seem out of place in a discussion of
the particular nature of judgment, but it seems indispensable [if certain
confusions are to be avoided] and <if we are to get> rid of certain general
misconceptions which frequently prevent the recognition of the truth.]
In order to get a true view of judgment it seems necessary to get rid
of certain general misconceptions, which concern not merely judgment but
many, if not all, other kinds of events both mental and physical. Such a
discussion takes us rather far afield from our proper subject, but cannot be
avoided since the points in question have not to my knowledge been set
out fully by any other author,l and it will help us later too when we come
to deal with the distinction between substance and accident2 and the connotation
or non-connotation of proper names. 3
NOTES
1. The best discussion I know is by C.D. Broad in Mind and Its place in Nature (London,
1925), pp. 588 ff., but I cannot altogether agree with his argument.
2. Vide infra, p ... ..
3. Vide infra, p... ..
APPENDIX II
SUPPLEMENTAL MATERIAL
(006-05-01 )
Mr. Russell in The Analysis of Mind divides propositions into two main kinds,
word propositions and image propositions. I want it to be clear that I do
not use propositions in this sense; what Mr. Russell calls word propositions
I shall call sentences, reserving proposition for those things, if any, which
sentences stand for. This paper is divided into four parts; first I shall examine
the view that the phrases which express propositions are incomplete symbols,
which is put forward in those words in the Introduction to Principia Mathematica,
and is clearly also held by Dr. McTaggart in The Nature of Existence, although
he expressed it differently. I shall give reasons for rejecting this view and
in the second part of the paper put forward the alternative in which I believe.
Then I shall discuss the relations between propositions and facts, and lastly
truth.
We may notice first that in Principia Mathematica it is said, not that sentences,
but that phrases expressing propositions are incomplete symbols. It is not
asserted that when I say, for example, "Socrates is mortal," that sentence
in that case is an incomplete symbol; such a view would be absurd. What
we are concerned with are propositions about propositions, or sentences
containing propositional phrases such as "Aristotle believed that plato was
mortal," where "that plato was mortal" is a propositional phrase. The Principia
view holds that in this sentence the propositional phrase "that plato was
mortal" stands for nothing. The proposition, Aristotle believed that plato
was mortal, is apparently of the form aRb, asserting that the relation of
belief holds between Aristotle and the proposition "plato was mortal";
Whitehead and Russell assert that it is not really of this form but of some
other, perhaps the triply relational; that is, it asserts a three term belief relation
between Aristotle, Plato and mortality. These three with the belief relation
are its only constituents and none of them are expressed by "plato was mortal."
So we see that what is meant by the assertion that propositional phrases
108 APPENDIX II: SUPPLEMENTAL MATERIAL
is the analysis of this proposition? Take first the simplest case in which p,
q, r are all subject-predicate propositions, e.g. "If he comes, I go and if
I go, he dies; imply Ifhe comes, he dies." The first of the constituent propositions
"If he comes, I go" expresses on this view the assertion of a relation which
we may call 14 between the four terms "him coming, I, going" and similarly
for the other constituent propositions; so the whole is to be interpreted as
a relation which we may call 115 between the fifteen terms He, coming,
14, I, going, I, going, 14, he, dying, he, coming, 14, he, dying.
The point is that this analysis, which in the simplest case is so complicated,
will vary with the forms of p, q, r; so that the analysis of the general proposition
"for all p, q, r, p:J q p r: :J p:J r" will be of the form for all a, b, c, P, Q,
R, 115 holds of a, P, 14, b, Q, b, Q, 14, c, R, a, P, 14, c, R, and a similar
thing for the next form in which p, say, is relational, and similar things
for all the infinity of other forms possible for p, q, r; so that this general
proposition would be seen on analysis to be infinitely complex. Now it is
generally agreed, and is, I think, evident, that no proposition entertainable
by us can be infinitely complex. Such infinite complexity is, in fact, what
is generally meant by a vicious infinite regress, which I have thus shown
to occur in the only analysis of this proposition possible on the incomplete
symbol vi~w. It may be remarked that this whole argument from the analysis
of implication, can be applied still more forcibly if based on the analysis
of probability relations; if, instead of considering "p implies q" we had
considerer "q has to p the probability relation of degree d," i.e., "the probability
of q on hypothesis p is d," we should have been still more astonished at
the bewildering multitude of probability relations, not dual, but of indefinite
multiplicity, infinite in number for each degree of probability, and possessing
no obvious properties whereby we could collect together those for each degree
of probability.
The reasoning, by disproving one of its necessary consequences, has I consider
shown that the incomplete symbol view of propositions must be rejected,
and we have now to consider possible alternatives to it.
II
All accounts of propositions and judgment which I have read are variations
of either the propositional theory of judgment or the incomplete symbol
theory of the propositions. The propositional theory of judgment asserts that
belief is a dual relation between something mental, perhaps a self, perhaps
a mental event, and a proposition. This view avoids the difficulties due to
supposing propositional phrases incomplete symbols; but its opponents doubt
whether there is any reason to believe that there are such things as propositions
110 APPENDIX II: SUPPLEMENTAL MA TERrAL
and no reason can be given except that no one has seen any alternative
except the incomplete symbol theory which I think I have disproved. But
it seems to me easy to construct a mixed theory, which satisfies the two
apparently incompatible requisites of a really satisfactory theory, which are
that it should hold: (1) that propositional phrases are not incomplete symbols,
(2) that there are no such things as propositions in the usual sense. That
is, (1) propositional phrases stand for something, but (2) they do not stand
for propositions in the usual sense since there are no such things. Therefore
they stand for things other than propositions. It may sound very paradoxical
to say that propositional phrases such as P, q, "p implies q" stand for things
other than propositions, but when you hear what I suppose it is they stand
for, you will probably say that it would have been misusing words to call
them propositions.
Consider, for a moment, a somewhat similar case, the discovery that numbers
are classes of classes; which always seems absurd until you are accustomed
to it. People naturally say that it is a misuse of language to call classes of
classes numbers; but what Frege discovered was that classes of similar classes
had all the arithmetical properties of numbers, so that it was reasonable to
suppose that in such statements as 2+2=4 the symbols 2, 4 stood for such
classes of classes. It was then possible to put his discovery in two ways;
he could either say "I have discovered what numbers are, they are classes
of classes"; or in order to avoid being told he was misusing language he
could have said, "I have discovered the meaning of 2+2=4; 2, 4, etcetera, do
not stand for numbers; they stand for classes of classes; I see no reason to
suppose that these strange metaphysical things called numbers exist." This
second way is the way I prefer to put my theory; it seems to me that in
such statements as "p implies q" p, q stand for things as certainly existent
as classes of classes seemed. These can be called propositions when you are
sufficiently accustomed to the theory not to be misled by the old associations
of the word, but till then had better be given some other name, since they
are not at all the kind of thing logicians suppose propositions to be; it would
be ridiculous to call them unities or constructs as propositions are called
by Mr. Russell in the Principles of Mathematics and Mr. Johnson respectively.
If we consider those mental events we call thoughts, we see that there
are two kinds of characters which they can have which are of great importance.
A thought may be a belief, a doubt, or an assumption, etc. These I call
pistic characters. I think that they consist in the presence or absence of feelings,
such as Mr. Russell calls belief feelings, in connection with the thought; but
their analysis is not very relevant to the matter in hand and I mention them
to distinguish them clearly from the other fundamental kind of characters
of thoughts, which I call referential characters or references. A referential
character is a character like that which we assert of a belief in saying that
THE NATURE OF PROPOSITIONS 111
"1 have a p-belief," thus making p what is, 1 believe, called a cognate accusative.
So just as Mr. Johnson said that "I experience red" means "I have a red
experience," I say that "I believe p" means "I have a belief with the reference
or referential character p." So that once the theory is understood, it is better
stated as that propositions are characters of beliefs, just as Frege's theory
of numbers is best stated as that numbers are classes of classes.
The conclusions reached may be summarised thus; when I believe p, my
belief is multiply related to the things ordinarily called the constituents of
p; if then 1 write 'p', the symbol stands not for a proposition which is before
my mind when I write, but for the property of my belief, expressed by
saying that it is a belief that p. Similarly "p implies q", "p or q" assert
relations between these referential properties of beliefs.
III
Hence forward I shall use proposition to mean what I have hitherto meant
by reference.
We have now to consider truth and facts. It is generally agreed that there
are facts; dispute is mainly about whether there is a different fact associated
with each true proposition; we can in English put 'the fact' before any
propositional phrase so that "the fact that 2+3 >4" and "the fact that 2+3=5"
appear to be different, but it is held by Mr. Johnson that they are identical,
the same fact being characterised determinately by 2+3=5 and less deter-
minately by 2+3>4. It may be remarked that on my view of propositions
namely that they are references, it is impossible to hold that they literally
characterise facts. For references are characters of beliefs and therefore the
relation between proposition and belief is characterisation; it is impossible that
this should be also the relation between proposition and fact; for if that
were the case we should be unable to distinguish the evidently different
statements "my belief is a belief that p" and "p is about my belief' since
both would be analysed as "p characterises my belief. "
Let us consider what arguments there are which can throw light on the
question of facts? They are of two kinds. Some are based on an analysis
or the supposed need for an analysis of truth; these 1 postpone; others are
based on supposed implications of our use oflanguage. We often make remarks
apparently about facts "it is a fact that King Charles was executed," "Newton's
Theory of Gravitation is contrary to the facts," "this view is forced upon
us by the facts." And "The severity of the Puritans caused the licentiousness
of the Restoration" apparently asserts a causal relation between two facts:
the fact that the Puritans were severe and the fact that the restoration was
licentious.
THE NATURE OF PROPOSITIONS 115
events are individuals. In order to explain this alternative I must first explain
what I mean by subjects and quasi-subjects.
When a proposition is of the subject predicate form, i.e. when to assert
it is to assert of one thing that it has a certain property, then I call the
thing of which the property is asserted the subject of the proposition. If
the proposition is of a relational form, i.e. if to assert it is to assert that
a certain relation holds between certain things, then those things are the
subjects of the proposition. It is easily seen that if an individual is a constituent
of a proposition it must be a subject. The definition of quasi-subject is more
difficult. The only propositions which have quasi-subjects are those which
assert existence, i.e. those which assert that there is a thing or things having
a certain property. Thus including such [singular] propositions as the Author
of Waverley was Scotch which asserts that there is a thing which has a
property of being the only writer of Waverly and of being Scotch. Thus
"Lions Exist," "Some animals are lions," "The most interesting animal in
the zoo is a lion," all assert existence. We then define "the class of quasi-
subjects of the proposition (Ex)P(x) is the class defined by the property p"
i.e. the quasi-subjects of "there are lions" are the lions themselves; the quasi-
subject of "The Author of Waverley is Scotch" is Scott. And it is an immediate
consequence of the definition that a false proposition has no quasi-subjects.
The view about events which I wish to advocate is that they are individuals,
and that the relation between an event and a time proposition which is,
as one might say, 'about' the event, is simply that the event is the only
subject or quasi-subject of the proposition. This is often obscured in language.
For example, neither "The table fell over just now" nor "Old St. Paul's
was then standing" seem to have events as subjects or quasi subjects. But
when they are analysed this is seen to be the case; the first should be analysed
as "There was an event x such that x shortly preceded this present and
was a situation of this table in which the table was in motion in a certain
manner." This has no subjects and if it is true only one quasi-subject, the
event x, (unless of course the table fell over twice). "Old St. Paul's was
then standing" should be analysed "There was an event x such that x had
the temporal relations indicated by 'then' and was a situation of Old St.
Paul's." Again "Yesterday I believed that" should be analysed "There was
an event with the temporal relations indicated by yesterday which was a
belief was mine and had the reference indicated by 'that'. It might be asked;
"why if events are individuals can we only identify them by means of
propositions?" The alternative would be to name them but we can only
name them when they are immediately before us by such a name as 'this';
we can name objects otherwise because they recur in experience and can
be recognised. Being unable to identify an event by a name we do so by
means of a proposition of which it is the only quasi-subject. There is no
118 APPENDIX II: SUPPLEMENTAL MATERIAL
real necessity for this; we could equally identify events by descriptive phrases;
instead of talking of the event which is the existence of the Great Pyramid
through a certain day, we could talk of the event extending through that
day in which the Great Pyramid was situated.
The only remaining objection to this theory that I can think of is that
though it explains how some time propositions are about events, and in
particular how a true proposition with only one event as subject or quasi-
subject is about one event; it does not explain what would be meant by
saying that a false proposition such as "the bicycle went 50 mph" when
it really only went 20 mph is about an event, in fact, the same event as
"the bicycle went 20 mph" is about. But this explanation can easily be given.
Let us recall the definition of a quasi-subject "The class of quasi-subjects
of the proposition (Ex) Px is the class defined by the property P." If (Ex) Px
is false as I said before there will be no quasi-subjects. Suppose now P is
a compound property, i.e. to have P is to have Pj and P2 and P3 etc. Then
in such a case the event or events which we should say (Ex) Px was about
would be the quasi subject or subjects of (Ex)Pjx or (Ex)Pzx or etc. one
of these at least being supposed true. In particular if we require a unique
event which (Ex) Px is about we choose one of (Ex) PjX, (Ex) Pzx, etc. which
is true and has a unique quasi-subject. Thus in our example the unique quasi-
event which "The bicycle went 50 mph" is about is the unique quasi-subject
of the "The bicycle went" namely the event in which the bicycle was located.
To sum up this account of facts and events; we have seen that there
are events and that they are individuals; we have seen so far no reason to
suppose that there are any such things as facts other than events or any
things such as complexes which are neither individuals nor properties nor
relations.
IV
Lastly we come to truth, which I shall deal with as briefly as possible. The
most certain thing about truth is that "p is true" and 'p', if not identical,
are equivalent. This enables us to rule out at once some theories of truth
such as that "to be true" means "to work" or "to cohere" since clearly
"p works" and "p coheres" are not equivalent to 'p'. There are I think
only three sensible theories of truth (1) that a true belief is defined to be
one which has a certain relation with a fact, (2) truth is indefinable and
has no connection with a relation between belief and fact, (3) truth is
indefinable; but as a matter of fact true beliefs do have a certain relation
to facts which false beliefs do not have. (3) I think we can dismiss; we
have seen no reason to suppose there are facts and if truth be indefinable,
THE NATURE OF PROPOSITIONS 119
I think none can be drawn from the nature of truth; so if truth be indefinable
we have no reason to suppose these are facts and therefore no reason for
thinking true beliefs are related to facts in a way which false ones are not.
Of course in the special case of events we have seen that some true propositions
and therefore true beliefs are related to events in a way false ones are not;
since false beliefs have no quasi-subjects.
(007-06-02 )
any such hypotheses; nor are they the sort of things of which we could
be supposed to have a priori knowledge, for they are complicated generalisations
about the world which evidently may not be true. To this it may be answered
that it is not necessary to suppose that we know the hypothesis for certain
a priori, but only that is has a finite a priori probability. (It may be explained
that the alternative to a finite probability is not an infinite, but an infinitesimal
one, like the probability that a cushion has one definite shade of colour out
of an infinite number of possible ones.) For inductive arguments would have
some force if the initial probability of the hypothesis were finite, and could
then be applied to the hypothesis itself, and so the probability of the hypothesis
would be increased which would again increase the force of inductive argument.
This would correspond to our feeling that induction derived its validity in
part at least from our experience of its success.
So we have to consider the a priori probability of an hypothesis of limited
variety; how are we to determine whether it is finite? Presumably, by direct
inspection; but owing to the abstract nature of the hypothesis this is difficult;
it seems to me easier to approach the question if we notice, that if the
hypothesis has a finite a priori probability, so has any generalisation such as
"all swans are white"; this indeed is the sole point in introducing the hypothesis.
If then we can see, as I think we can, that the a priori probability of "all
swans are white" is infinitesimal, so must be that of the hypothesis of limited
variety.
I do not see how to proceed any further without discussing the general
nature of probability; according to Keynes there holds between any two
propositions some objective logical relation, upon which depends the degree
of belief which it is relational to have in the one proposition, if the other
is what is known already. This clear objective theory is however blurred
by his saying in one or two passages that "probability is relative in a sense
to the principles of human reason." With the word human we pass from
a purely logical notion to one which is in part, at least, psychological, and
in consequence the theory becomes vague and muddled. Keynes tries to identify
the logical relations between propositions, with the psychological ones which
express the degree of belief which is rational for men to entertain, with
the result that the probability relations of which he speaks cannot be clearly
identified, whence the difficulty in deciding whether or not they can always
be compared with one another or measured by numbers.
Between propositions there are indeed logical, or as I should prefer to
say formal relations; some of these enable us to infer one proposition from
the other with certainty, others only with probability. For example if p,
q are two elementary propositions, i.e. such as assert atomic facts, we can
see what probability the proposition p V q gives to the proposition p in the
following way:
122 APPENDIX II: SUPPLEMENTAL MATERIAL
* Paper first published in the American Philosophical Quarterly, vol. 24 [1987], pp. 357-359.
"THE LONG AND SHORT OF IT" 125
is not a member of itself, then it is not one of the classes which are not
members of themselves, and so must be a member of itself. We have thus
proved both that it is and that it is not a member of itself.
About a dozen such contradictions have been found of various kinds, from
the absolutely non-mathematical one about "heterological" to an undeniably
mathematical one about infinite ordinal numbers. And it is probable that
only a little ingenuity is required to invent as many more. The result is
a serious crisis for mathematical and philosophical thought, which has not
however received the attention it deserves, owing to the scarcity of philosophers
who are good enough mathematicians, and of mathematicians who are good
enough philosophers to get a firm grasp of the situation. Nevertheless some
of the greatest mathematicians in the world have been led to investigate
anew the fundamental principles of their science in order to avoid not merely
the possibility but the actuality of contradiction with which they are threatened.
The situation is not altogether a new one in the history of thought, for
at least twice before the intellectual situation has been over-shadowed by
apparently insoluble contradictions. In ancient Greece, Zeno of Elea pro-
pounded his paradoxes about motion, which cannot be regarded as having
been finally cleared up until the latter half of the last century. The most
famous of these is the race between Achilles and the tortoise. Achilles, who
runs ten times as fast as the tortoise, gives the tortoise ten yards start. By
the time Achilles has run this ten yards the tortoise is a yard further on;
by the time Achilles has run this yard, the tortoise has gone 1 yard further.
By the time Achilles has gone this .1 yard, the tortoise will have gone another
.01 yard; and so on, the tortoise being always ahead and Achilles never catching
him up: The solution of this paradox requires an adequate theory of the
infinite divisibility of space which was only achieved in recent times.
Fortunately, however, men were prepared to trust their common sense before
they were in possession of a theoretical justification for their conclusion.
The second great thinker who invented contradictions of this sort was
the philosopher Kant, whose "antinomies" were an essential part of his system.
He produced proofs that, for instance space is infinite in extent, and again
that it is only finite: that it is infinitely divisible and again only finitely
divisible. From this possibility of proving contradictory results he deduced
that the sphere of human reason was limited, and that contradiction was
the natural consequence of arguing about things beyond the range of possible
experience. His position has, however, been overthrown by the general
recognition that the arguments by which he reached his contradictions are
fallacious, and that so far from it being possible to prove both that space
is infinite and that it is finite, it is not possible to prove by a priori reasoning
either of these results.
Now again we are faced with contradictions, but the present situation
126 APPENDIX II: SUPPLEMENTAL MATERIAL
is different in several ways from either of the preceding ones; then the
contradictions were invented by philosophers to support their systems, whereas
the modern ones were, to begin with, stumbled upon by mathematicians
who had nothing to gain for their theories from them and could only regard
them with dismay. The modern contradictions, especially the more mathe-
matical ones, therefore appear as obstacles inevitably encountered in the pursuit
of science, not as the artificial creations of philosophers. Another difference
is that the paradoxes of Zeno and the antinomies of Kant related principally
to space, time and motion, whereas the modern "paradoxes of the Theory
of Aggregates," as they are called, are matters of pure logic, and demand
for their solution not merely a new theory of space and time, but a reform
of logic; that is, some new idea as to what is a good argument and what
a fallacy. Hence although history teaches us that such paradoxes will be
solved in time, we must expect that in this case the task will be decidedly
more difficult than it was in the previous ones.
Although there is a considerable literature about the contradictions of the
theory of aggregates, none of it can be said to contribute materially towards
a solution except the work of Mr. Bertrand Russell, which although it is
not universally accepted, and does not profess to be completely satisfactory,
is by many regarded as giving an approximately correct account of the matter.
His solution is given by his "Theory of Types." The most important idea
in this is that a series of words which form a grammatical sentence may
nevertheless be nonsense. For instance "virtue is triangular" is a grammatical
sentence, but according to Mr. Russell nonsense. It is not merely that it
is false to say that virtue is not triangular, but that this statement is literally
nonsense, for it is only things and not qualities like virtue that can significantly
be said to be triangular. It is therefore necessary to distinguish different "types"
of things, and to declare that the result of applying an adjective to a thing
of the wrong type is not falsehood but nonsense. In particular, a class is
always of a different type from its members, so that if the members are
men it is nonsense to say that the class either is or is not a man; and it
is nonsense to say that a class is or is not a member of itself, so that the
arguments used in establishing the contradiction about the class of classes
which are not members of themselves are all literally nonsense and there
is not really a contradiction but simply a nonsensical arrangement of words
which looks like a contradiction.
In this way, by the principle that the rules of grammar do not guarantee
significance, we can escape about one half the contradictions of which the
one about classes is typical. But the other contradictions like the one about
"heterological' are not so easy, because "heterological" is not a nonsensical
word, but expresses a real property of an adjective word, and since it is
itself an adjective word it is the right type of thing to be autological or
"THE LONG AND SHORT OF IT" 127
Aristotle 11, 107 Joachim, H. H. 19, 20n, 27, 28n, 29n, 30,
31 , 32n, 36, 37, 38,41, 41n, 42n, 96,97n
Bacon, Francis 15n, 92 Johnson, W. E. 66n, 76n, 110, 111, 114, 115
Baldwin, Stanley 3 Joseph, H. W. B. xiii, 65n
Beethoven, Ludwig van 96
Birkenhead, Lord 9, 86 Kant, Immanuel 20n, 24n, 77n, 125, 126
Bosanquet, Bernard 9, 15n, 29n, 31, 32n, 38, Keynes, J. M. ix, 83n, 120, 121
40, 42n, 82, 93n, 97n
Bradley, F. H. 27,28, 29n, 31, 32n, 41n, 42n, Leibnitz, G. W . von 36
82,97n Lloyd George, David 108
Braithwaite, R. B. x , xi, xii, xiii, xiv, xxin,
41n, 76, 77n McTaggert, J. M. E. 74, 76, 77n, 93n, 107
Broad, C. D. 50, 51, 53n, 54n, 60, 65n, 70, Mellor, D. H. xxin
71, 73, 74, 75, 76, 77n, 97n, 100, 102n, Moore, G. E. xiv, 5n, 15n, 36, 41n, 65n, 76n,
l04n 77n, 82, 93, 115
Buonaparte, Napoleon 8
Peirce, C. S. 91, 94n
Caesar, Augustus 72 Plato 107
Caesar, Julius xiii, 26, 30, 31, 32, 42n, 72
Campbell, Norman xiii, xiv, xv, xxin, 35n Richard, l. A. 5n
Cohen, Morris R. 94n Ross, W. D. 15n
Cook, Wilson J. 15n, 46, 47, 48, 50, 53n, Russell, Bertrand xiii, xiv, xvi, 8, 15n, 24n,
54n, 55, 57, 62,63, 65n,99, 102n 25, 28n, 36, 41n, 54n, 70, 76, 77, 77n, 83,
Cone, M. 36 83n, 86, 90, 93n, 94n, 95, 97n, 101, 102n,
107,110,112,113,126,127
Dunne, J. W. 75,77n
Socrates 107, 110
Frege, Gottlob xiv, xv, xxin 114 Stumpf, Kar! 60
Hertz, Heinrich xiii, xv, xxin, 35n Whitehead, A. N. 69, 77n, 112, 115
Hobbes, Thomas 66n Wittgenstein, Ludwig xiii, xiv, 120
Hume, David 123
Husser!, Edmund 15n Zeno 125, 126