MENTAL ACTS

THEIR CONTENT AND THEIR OBJECTS

STUDIES IN PHILOSOPHICAL PSYCHOLOGY

Edited by

R. F.

HOLLAND
P. T.

Mental Acts

Geach

The Psychology of Perception

D.

W. Hamlyn

MENTAL ACTS
THEIR CONTENT AND THEIR OBJECTS
BY

PETER GEACH

ROUTLEDGE

&

KEGAN PAUL

LONDON

First published in Great Britain

by Routledge & Kegan Paul Ltd Broadway House, 68-74 Carter Lane London, EC4 Printed in Great Britain

By

Tanner Ltd Frome and London

Butler

&

CONTENTS
Explanation of the way I use the terms "mental act", "content", "object-expression"
2

The occurrence of mental acts, in my sense of the word, is not controversial. Wittgenstein did not, as some people think, wish to controvert it
Reasons for rejecting Ryle's view that 'reports of mental acts', as I should call them,
are really hypothetical or semi-hypothetical statements about overt behaviour

Reasons

for beginning

with a discussion of

concepts and acts of judgment. There

are acts of judgment, not merely disepisodic positions of a certain sort (beliefs)

Concepts are for me specific mental abilities exercised in acts of judgment, and expressed in the intelligent use of words
is

(though not exclusively in such use). There no reason to ascribe concepts (in this
11

sense) to brutes

"Abstractionism" defined. I hold that no

acquired by the supposed process of abstraction. Abstractionism, being a muddled theory, cannot be established ex-

concept

is

perimentally. How abstractionists regard (a) concepts applying to sensible objects, (b) psychological concepts

18

vi

CONTENTS
Abstractionism cannot account for logical concepts like those of not, or, every; though some abstractionists have tried to say that such concepts are derived from characteristic 'feelings' evoked by the logical words. It cannot be said, either, that logical words are mere signals for special exercises of nonlogical concepts

22
is

Abstractionism

an unworkable theory

for

arithmetical concepts, even at the level of school arithmetic. 'Abstract' counting developing the series of numerals is in fact temporally prior to counting objects: contrary to what abstractionists ought to

expect
9

27

The

10

marks of relational conby abstractionists; they would not be able to explain them Abstractionism does not in fact work even
characteristic cepts are ignored
for concepts of simple sensible characteristics, e.g. that of red. The abstractionist argument about men born blind, refuted. difficulty for abstractionists over the con-

32

cepts red

11

and (chromatic) colour The mind makes concepts, and neither

33
this

nor the exercise of concepts consists in finding a characteristic repeated in our experience; but conceptual thought does not
falsify reality.

Psychologists' attempts to abstractionism establish experimentally either do not treat of what I should recognize as concept-formation, or fail to show that concepts are acquired by a process of

abstraction

38

12

Russell's theory of judgment: judgment consists in some psychical entity's coming

CONTENTS
into

vii

a many-termed relationship, whose other terms are the particular and universal

entities being

judged about

45
47

13 14

Logical inadequacies of Russell's theory

Sketch of a new theory: a judgment to the stand in an n-termed relation R itself consists of Ideas standing in a derivative n- termed relation (R) (where "( )" is a special, undefined, operator).
effect that things

Russell's objections to Ideas' entering into judgment; reply to these objections

52

15

In contrast to Russell, I started by considering judgments wholly expressible in general terms. Judgments about directly sensible particulars are to be analysed as judgments of general content standing in some special connexion with certain sense-experiences

ad phantasmata). conversio the actual time-reference of a Similarly, tensed judgment is no part of that which is judged, but comes from the sensory context in which the act of judging occurs
(Aquinas's

61

16

Proper names express identifications e.g. "the Thames" expresses the identification of something as one and the same river. Identification involves an unsolved problem which arises also for the use of "this" to signify not "what is now before me" but "what I had in mind just then" (its use for demonstratio not ad sensum but ad intellectum). The reference of a proper name to one outside object rather than another is not a
genuinely psychological problem, and is to be explained only by bringing in the physical context of the utterance

66

viii

CONTENTS
The theory that the concept judging
is

17

an

analogical extension of the concept saying. General remarks on analogy-theories of

psychological concepts

75

18

I shall consider psychological uses of the

oratio recta construction (as in "The fool hath said in his heart "There is no God" "),

rather than those of oratio obliqua. of oratio recta

The

logic

79

19

arguments against the posof using oratio recta in psychological sibility descriptions
fallacious

Some

87

20

Discussion of the sort of psychological statelogically involves 'quantifying into' oratio obliqua: e.g. "More people came to Smith's party than James thought would come". Our oratio recta paraphrases are adequate for these

ment that

92

21

The predicate "true" primarily

expressions in a language;

applies to

here as elsewhere language about thought is an analogical extension of language about language

96 98

22

The theory sketched in 14 can be developed, interpreted, in terms of our present theory
Like all analogies, the analogy of judgment to spoken language must not be carried too far. Ockham's idea of an inner language certainly goes

23

wrong

in this respect. Again,

spoken language has a definite position in the physical time-series, and spoken words occur in a definite order; but it would be quite wrong to ask about the time-order of Ideas in an act of judging, or how long an act of judging goes on

101

CONTENTS
24
Psychological

ix

judgments

have

been

re-

garded as based on the deliverances of an 'inner sense', which affords the materials from which by abstraction we get psychological concepts. But 'inner sense' is a
chimera
25
Psychological concepts relating to 'sensuous' experiences like seeing, hearing, fearing, pain, are applicable only as part of a complicated web of concepts, others of which relate to external physical objects and to behaviour. This is not behaviourism; but it does involve that, if we try to apply these psychological concepts otherwise than to living organisms (e.g. to disembodied spirits or machines), we do not know what we are doing
fallacy of "Cogito ergo sum" as a proof that each of us is directly aware of an

iu?

111

26

The

27

immaterial agent There are genuine descriptions of sensations (besides Aeusserungen in Wittgenstein's sense); we have here conversio ad phantasmata, the phantasmata being the sensations themselves and not 'inner-sense' representations of them. (E.g. if I judge that my
pain waxes and wanes regularly,

117

my thought

of pain waxing and waning regularly stands in the conversio-ad-phantasmata relation to the pain, not to an inner-sense representation of it)

121

28

is the difference between the judgments expressed by "I see a round yellow disc a foot across" and "there is a round yellow disc a foot across"? The former is an analogical development of the system of

What

CONTENTS
description used in the latter; the analogy is like the analogy whereby we come to call pictures by the names of the things pictured. can indeed say that sensations really last a certain time and really have parts (formaliter, as Descartes would say, not just obiective); but even here the concepts we use are used analogically

We

124

Appendix Aquinas used the jargon of "abstraction", but his maturest thought at any rate (in the Summa Theologica) is explicitly opposed to what I call "abstractionism"
Bibliography

180

182

Index

188

ACT, CONTENT,

AND OBJECT

label for a set of problems; the controversial views that have historically been expresssed by using the terms "act", "content", and "object" ought not at the outset to be borne in mind. My own use of the term "mental act" may be explained, sufficiently for present purposes, as follows. In historical or fictional narrative there occur reports, not only of what human beings overtly said and did, but also of what they thought,

THE

TITLE I have chosen for this work is a mere

how they

so on; I shall call the latter kind of reports "reports of mental acts". The psychological character or, as I shall say,
felt,

what they saw and heard, and

the content of mental acts is expressed by the use of various psychological verbs, such as "see", "hear", "hope", "think". Many of these psychological verbs
require a grammatical object a noun, noun-phrase, or noun-clause to complete their sense. The hero sees
the postman coming, hears his knock, hopes (has a sudden pang of hope) for a letter from his sweetheart, thinks that

his letter to her

may

have been opened by her guardian,

and
call

of such expressions is thus essential in stating the content of many mental acts; we may
so on.

The use

them

"object-expressions".

I shall avoid such uses of the

letter

term "object" as "A

from

his sweetheart
1

is

an object of the hero's

MENTAL ACTS
possibility of his letters' being interher guardian is an object of his thought"

hope" or "The

cepted by F&miliar and understandable as such uses are, they raise logical difficulties that I prefer to avoid. (One of these difficulties is that the so-called object may not exist; in my examples, there may be no letter from the hero's sweetheart such as he hopes for, and no such possibility as he thinks, of her guardian's intercepting his letters; so the use of "object" that I shall avoid might easily lead us on to such odd statements as "some objects of mental acts do not exist".) I shall accordingly state my problems not in the form "What sort of objects do these mental acts have?" but rather in the form "Suchand-such object-expressions are used in describing these mental acts; what is the logical role of these
expressions?"

WITTGENSTEIN'S ALLEGED REJECTION OF MENTAL ACTS

THE
mental

of mental acts, as I have exthe term, ought not to be a matter of conplained troversy. According to the Gospel story, something happened to St Peter when the cock crew; he heard it crow, and he remembered Christ's prophecy of his

EXISTENCE

three denials; his hearing

acts.

and

his

remembering were

There

is

indeed a danger that

when we

speak of mental acts or mental events or what happened in a person's mind, we may be led to an illegitimate assimilation of psychological to physical reports.

WITTGENSTEIN ON MENTAL ACTS

There are logical similarities between the two kinds of reports, but there are no less important logical differences; these differences, on which Wittgenstein cofitinually insisted, were already noticed by Aquinas, who remarked that when people speak of the mind as
a 'subject of change' the meaning (ratio) of "subject" and "change" is quite other than it is in discourse about physical processes (la q. 75 art. 5 ad 2 um). But since such logical differences are just what we are looking out for in our enquiry, there is little risk of our forgetting that they may exist. Wittgenstein has been understood as denying the
existence of mental acts; and certain remarks of his about 'private objects' (WITTGENSTEIN, Part I, 293; Part II, p. 207) are very easily taken this way. I am
sure,

however, that I have not so far maintained any-

thing Wittgenstein would have attacked. Of course Wittgenstein did not want to deny the obvious truth that people have a 'private' mental life, in the sense that they have for example thoughts they do not utter and pains they do not show; nor did he try to analyse away this truth in a neo-behaviouristic fashion. In one
of his lectures he mentioned Lytton Strachey's imaginative description of Queen Victoria's dying thoughts.

He

expressly repudiated the view that such a descripis

tion

meaningless because 'unverifiable'; it has meana ing, he said, but only through its connexion with wider, public, 'language-game' of describing people's thoughts; he used the simile that a chess-move worked out in a sketch of a few squares on a scrap of paper has significance through its connexion with the whole practice of playing chess. It is useful to observe Frege's
distinction of sense (Sinn)

and reference (Bedeutung) in stating Wittgenstein's position; what Wittgenstein

MENTAL ACTS

wanted to deny was not the private reference of psychological expressions e.g. that "pain" stands for a kind of experience that may be quite 'private' but the possibility of giving them a private sense e.g. of giving sense to the word "pain" by just attending to one's

pain-experiences, a performance that would be private and uncheckable. The view that psychological
this way is part of a theory as I shall call it) which Wittgenstein (abstractionism, rejected in toto and not only as regards psychological terms. Another view he rejected (one closely bound up,

own

words are given a sense in

with an abstractionist view of psychological the view that such relations as that between terms) is a pain and its symptoms or a thought and the words
I believe,

expressing it have nothing to do with our concept of pain or thought and are just established inductively.

We

shall discuss

both these topics later on.

KYLE'S REJECTION OF MENTAL ACTS

A I SHALL try to show that reports jLJLof mental acts are logically different from reports of physical events, I hold that reports of both kinds are categorical. This view is sharply opposed to Ryle's view that psychological statements are not reports of private events (mental acts) but are hypothetical or semi-hypothetical statements about overt behaviour. (RYLE, pp. 33, 46.) I shall not criticize Ryle's attempted analyses in detail; the entire programme seems to me misconceived. In the first place, he makes no serious

LTHOUGH

RYLE'S REJECTION OF MENTAL ACTS

attempt to carry out
leaves
his

programme

consistently; he

some reports of mental acts standing, without offering any analysis of them into hypothetical or semthypothetical statements about behaviour. The mental
acts in question are indeed referred to throughout in a highly depreciatory style, as "itches", "tingles", "tweaks", "agitations", etc.; but this rhetorical trick proved nothing. If reports of these mental acts cannot be reduced to hypothetical or semi-hypothetical statements about overt behaviour, then the view that the distinction between categoricals and hypotheticals is the logical distinction between physical and psychological statements must be completely wrong. A logical principle allows of no exceptions not even if the ex-

ceptions are events that only into a novel.

James Joyce would put

Secondly,

when Ryle

actual difference

explains a statement of an between two men's mental states as

really asserting only that there are circumstances in

which one would act

there
is

differently apparently holds that this could

from the other, and be all the difference

between the two, he

is

running counter to a

very deep-rooted way of thinking. When two agents differ in their behaviour, we look for some actual, not merely hypothetical, difference between them to

account for
in quantum

this; as
est

the scholastics said, omne agens agit

in actu. Ryle explicitly and repeatedly compares psychological accounts of behaviour to saying that a glass broke because it was brittle (had the dispositional property of brittleness); in so doing, howis setting them on a level with the statement that opium puts people to sleep because it has a dormitive power which I believe was not his intention.

ever, he

(RYLE, pp. 50, 86-7, 88-9,

et al.)

MENTAL ACTS
The
is

cited

principle expressed in the scholastic tag just of importance in scientific investigation.

physicist would be merely impatient to him: "Why look for, or postulate,

somebody any actual differ-

if

said

ence between a magnetized and an unmagnetized bit of iron? Why not just say that if certain things are done to a bit of iron certain hypothetical become true of it?" He would be still more impatient at being told that his enquiries were vitiated by the logical mistake of treating "x is magnetized" as categorical, whereas it is really hypothetical or semi-hypothetical. Of course there

may

be people prepared to say that, although men of science regularly look for differences already existing between the agents in order to explain differences of
behaviour, there is no reason to expect that such differences always do exist; the principle on which men of
science proceed might be as unsound as any gambling system, and their success up to now mere luck. I shall

not argue the point. We may well be reluctant, then, to expound in Ryle's fashion even psychological accounts of actual differences in behaviour. Our reluctance should be still greater when we are invited to regard a statement that two men, whose overt behaviour was not actually different, were in different states of mind (e.g. that one checked a column of figures with his mind on the job, the other with his mind on his domestic troubles) as being really a statement that the behaviour of one man would have been different from that of the other in hypothetical circumstances that never arose. It ought to be, but plainly is not, generally known to philosophers that the logic of counterfactual conditionals is a very ill-explored territory; no adequate formal logic for them has yet been devised, and there is an exten-

ACTS OP JUDGMENT
sive literature
is

on the thorny problems that crop up. It a scandal that people should count it a philoreally sophical advance to adopt a programme of analysir^
ostensible categoricals into unfulfilled conditionals, like

the programmes of phenomenalists with regard to 'physical-object' statements and of neo-behaviourists with regard to psychological statements. Wii!h specious lucidity, Ryle describes the counterfactual conditionals he uses as having the 'sentencejob' not of stating facts but of licensing inferences.

The terms "inference-licence" and "inference-ticket" seem to be explanatory because we associate them with
very familiar institutions. In order to see that nothing has been explained at all, we need only ask the question what inferences from fact to fact a counterfactual conditional does give us licence to perform. I need hardly comment on Ryle's view that "the rubber has begun to lose its elasticity" has to do not with a change in the rubber but with the (incipient?) expiry of an
inference-ticket. (See RYLE, pp. 120-5.)

ACTS OF JUDGMENT
of mental acts that 1 shall chiefly investigate are acts of judgment. This is not an arbitrary choice. Concepts, as we shall see, are capacities exercised in acts of

SORT

judgment

psychological con-

cepts, in psychological judgments about oneself and others; and one of the chief obstacles we shall en-

counter

is

a tangle of mistakes and confusions about

MENTAL ACTS

judgments and concepts generally, by reason of which people go widely astray in their account of psychological judgments and concepts in particular. I hope my work on the nature of judgment will do something
to clear

away the

tangle.

Many readers may wish to object in limine that even

there are mental acts, there certainly are not acts of judgment. However difficult the logic of disposltional
if

words may be, it is surely clear both that "believe" is such a word, and that the acts which make it true to
say that So-and-so believes such-and-such are not only, perhaps not at all, acts of judgment. A gardener's belief that it will rain is a disposition exercised in suitable acts of preparation rather than in intermittent mental
acts of judgment. (RYLE, pp. 175-6.) But is there in fact any behaviour characteristic of

a given belief? Can action be described as "acting as if you held such-and-such a belief" unless we take for granted, or are somehow specially informed about, the needs and wants of the agent? In Ryle's example this information is smuggled in by his speaking of a gardener's rain-expecting behaviour (and tacitly assuming that the gardener is not e.g. a discontented or corrupt servant who wants the garden to be ruined). When Dr Johnson did penance in Uttoxeter market-place, he may have begun by standing around bareheaded until the threatened shower should fall; this would not be recognizable as rain-expecting behaviour without a knowledge of Johnson's wish to do penance. There is indeed one sort of behaviour that does characterize a
belief

namely, putting it into words; lying is necesthe exception, not the rule. But putting your sarily belief into words is one of those intellectualistic' activities that Ryle seeks to depreciate. In this instance, his
c

ACTS OF JUDGMENT
doing so
is

very understandable; for on the face of it if somebody puts his belief into words, not parrotwise but with consideration, then there occurs a mental act,
of the sort that I call an act of judgment. Anybody performs an act of judgment at least as often as he

of

makes up his mind how to answer a question; and acts judgment in this sense are plainly episodic have a
position in a time-series. (We shall see, indeed, that we cannot cannot in principle, not for lack of information or technique assign positions in time to acts of judgment in the same way as we do to physical events, nor even in the same way as we do to sensations. But
this
is

In very

ment
tion

irrelevant to our present discussion.) many languages, a report of an act of judgis expressed by using the oratio obliqua construc-

i.e. the same construction as is used with 'verbs of saying' to report the gist or upshot of somebody's remark rather than the actual words he used. Thus, in

English and Latin and many other languages (including, as a pupil once told me, some African languages), there need be no difference except in the main verb between a statement that James said it was time for everybody to go home and a statement that James judged it was time for everybody to go home. And for every judgment whose content can thus be given by means of a piece of oratio obliqua, there is some form of words, in the language considered, that would be suitable to express the act of judgment namely, some form of words whose gist is given in the piece of oratio obliqua. (We must of course distinguish between expressing a judgment and expressing a report of a judgment. "James judged that the crevasse was too wide for him to jump" expresses a report of James's judgment; a form of words suitable to express this act of

10

MENTAI^ ACTS
is

judgment would be "the crevasse

too wide for

me to

jump".) " Now given a statement made by James in a language L, the grammatical rules of L generally allow us to construct a piece of oratio obliqua that preserves the gist of the statement. (Schoolboys often do this as an exercise in Latin grammar.) Having done this, we need only tack on the expression in L for "James judged (that .)", and we have a report in L of the judgment expressed in James's original statement. Moreover, if we have sufficient mastery of L to understand both the
.

.)" and the piece expression for "James judged (that of oratio obliqua, we shall understand the report so constructed, no matter what the actual sense of the re.
.

mark done

into oratio obliqua

may

be.

Nonsense cannot be turned into oratio obliqua; for oratio obliqua, which serves to report the upshot of what is said, cannot be used to report nonsense. So an at-

tempt to report that somebody judges nonsense is itself nonsense. A British pupil of Heidegger might say "Nothing noths" in a tone of conviction; assuming that this is nonsense, "he judged that Nothing nothed" would also be nonsense. (We ought rather to say: "he judged that "Nothing noths" was the expression of a
truth".)

These remarks about the expression of judgments and the reports of judgments may appear very obvious;
but in
fact, as

we

shall see,

they impose severe logical

restrictions

upon

analyses of judgment.

THE NATURE OF CONCEPTS

/^<ONCEPTS,the old-fashioned logic-books
tell us,

are presupposed to, and exercised in, acts of judgment.* Of course this is psychology, not logic; but I

V^

think that it is correct psychology. It has been argued, however, that, so far from concepts' being presupposed to judgments, the common feature of judgments in which one and the same concept is supposed to be exercised is something that appears on its own account only in subsequent reflection on the judgments. I do not, it is said, first acquire a concept of man and then put it to work in forming judgments about man (judgments whose natural expression in English would contain a use of the word "man"); on the contrary, I first of all form certain judgments and then, reflecting upon them, bring them under the head "judgments about man"; this

now attend to on its own account did not exist as an independent factor in the original judgments. My reply is as follows. Any reportable act of judgment is apt for verbal expression, in a form of words whose gist is contained in the oratio obliqua of the report. (I am not asserting, nor do I think, that an act of judgment is always put into words, even in the head'). Let us concentrate, for the moment, on judgments that are verbally expressed. The sentence expressing a judgment will consist of a number of words; and the significant utterance of the sentence therefore
feature that I
c

brings into exercise various relatively simple skills

11

12

MENTAL ACTS

the abilities to use the various expressions that occur in the sentence. It is important to notice that as applied tl> whole sentences the philosophical slogan "the meaning of an expression is its use" is seriously mistaken; at any rate if "use" is here taken to mean "established usage". For in general there is no established usage for a sentence as a whole; e.g. the sentences in this book do not get their meaning from any established usage, nor have I learned to use them in circles where they were commonly used. It is words and phrases that have an established usage; a language is mastered not by learning whole sentences out of a guidebook but by learning to make up sentences in it and to understand sentences not previously heard. That is how we learn foreign
languages, and how children come to understand and 1 speak their native language. The ability to express a

judgment in words thus presupposes a number of capapreviously acquired, for intelligently using the and phrases that make up the sentence. I shall apply the old term "concepts" to these special an application which I think lies fairly capacities close to the historic use of the term. It will be a sufficicities,

several words

having the concept of so-andhe should have mastered the intelligent use so that (including the use in made-up sentences) of a word for so-and-so in some language. Thus: if somebody knows how to use the English word "red", he has a concept of red; if he knows how to use the first-person pronoun, he has a concept of self; if he knows how to use the nega1 Psychologists appear not to take this fact sufficiently into account. Thus, Humphrey takes as his example of the use of language, not a sentence devised for a new situation, but an ejaculation "Rain!" when it is raining; not surprisingly he concludes that the problem of meaning is just a case of the problem of learned responses.

ent condition for James's

(HUMPHREY, pp. 227-34.)

THE NATURE OF CONCEPTS

tive construction in

18

some language, he has a concept

of negation.

An analogy from playing chess may help to sho*v how concepts and acts of judgment are related. Making an appropriate move from a certain position may be,
and at the opening of the game very likely will be, a learned response; but in the middle game it will certainl/ not be so, for the position may well occur only once in a life-time of play. On the other hand, the ability to make an appropriate move from a given position always presupposes a number of simple, previously acquired, skills the capacities to carry out the

moves and captures that are lawful for the pawns and the various pieces. As these skills are related to the chessmove, so concepts are related to the act of judgment. I have stated only a sufficient, not a necessary, condition for somebody's having a given concept. If a man struck with aphasia can still play bridge or chess, I certainly wish to say he still has the concepts involved in
the game, although he can no longer exercise them verbally. But it would be hard to devise non-verbal criteria for the patient's having retained a concept of
the

day

tions

and typical applicaof the term "having a concept" are those in which
after tomorrow.

The

central

man is master of a bit of linguistic usage; we can then

reasonably extend the term to cases sufficiently like these, e.g. where the man can play 'intellectual' games like bridge and chess. I shall not try to draw any sharp line between what is 'sufficiently like' the central and typical cases and what is not; I do not think we shall go far wrong if we concentrate henceforward on concepts exercised linguistically. it concept, as I am using the term, is subjective is a mental capacity belonging to a particular person.

14

MENTAL ACTS

use of "concept" is thus to be contrasted e.g. with Russell's use of it in The Principles of Mathematics aftid again with the use of it to translate Frege's "Be-

(My

gnjfjf";

Russell's 'concepts' and Frege's Begriffe were supposed to be objective entities, not belonging to a

The subjective nature of concepts does not however imply that it is improper to speak of two people as "having the same concepts"; conformably to my explanation of the term "concept", this will mean that they have the same mental capacity, i.e. can do essentially the same things. Thus, if each of two men has mastered the intelligent use of the negative construction in his own language, we may say that they have the same mental capacity, the same concept; they both have the concept of negation. There are of course all degrees of transition between cases where we should say unhesitatingly "they have different words, but they use them in the same way; they have the same concept", and where we should rather say "it's not just a verbal difference; they have different concepts". I thus accept the psychology of the old logic-books, to the extent of recognizing the possession of concepts
particular mind.)

as presupposed to acts of judgment, and regarding a judgment as the exercise of a number of concepts. feature of the traditional theory that I find unaccept-

able

is

hension', in

the doctrine that there are acts of 'simple apprewhich concepts are somehow exercised

singly without being applied to anything. I think these acts are mythical; in chess one can practise the moves of a solitary man on the board, but I think there is no

analogous exercise of a single unapplied concept. Having criticized Ryle for his reckless use of the term "dispositions", I may myself be criticized for

THE NATURE OF CONCEPTS

15

introducing concepts, which are supposed to be capacities. But to say that a man has a certain concept is to say that he can perform, because he sometimes d<9es perform, mental exercises of a specifiable sort. This way of using the modal word "can" is a minimal use, confined to a region where the logic of the word is as
clear as possible.

Ab

esse

ad posse

valet consequentia

what'is can be, what a man does he can do; that is clear if anything in modal logic is clear, and no more than
this is involved in

tempt to

expound

talking of concepts. Ryle's atostensible statements about actual

my

mental states in terms of complicated sets of subjunctive conditionals is a far more dubious undertaking. The exercise of a given concept in an act of judgment is not in general a definite, uniform sort of mental act; it does not even make sense to ask just how many concepts are exercised in a given judgment. Our chess analogy may here again be of service, in showing why this question is unreasonable. Playing chess involves a number of abilities, which are not only distinguishable but can actually exist separately; for one way of teaching chess would be to play first just with the kings and the pawns and then add the other pieces successively in later games. It would, however, be absurd to ask
just

how many of these abilities there were, or just how many were exercised in a particular move; although one might perfectly well say that somebody knew the
knight's move, and that this knowledge was or was not exercised in a particular move. Our language about the concepts exercised in a given act of judgment makes sense or does not make sense in much the same way.
shall) concepts that are involved in the intelligent use of words, I shall often have to use the phrases "to know the use of the

In discussing (as for the most part I

16

MENTAL ACTS
",
like.

ambiguous, btit practically indispensable. One ambiguity of them is notorious: ambiguity as between being able to use a word intelligently and being able to give an account of its use between knowing how and knowing that; I shall
irritatingly

word and the

"to

know

the

way

to use the

word

",

These phrases are

consistently keep to the knowing

how sense unless the

contrary
guity

is

may man who knows no English knows the use of the French

expressly stated. Another, trickier, ambibe brought out by this question: if a French-

adjective "rouge", does he know the use of the English adjective "red"? In one sense of "knowing the use of

the word" he must know the use of the word "red", because it is the same as the (relevant) use of the word "rouge", which ex hypothesi he knows. In another sense, he does not 'know the use of any English word, because he knows no English. The former sense of the " phrase "know the use of the word "red" might be out by saying: "know how to use some word brought in the way that the word "red" is in fact used in English"; but to avoid clumsily long clauses I shall express this sense just by saying: "know the use of the word "red" ", and shall avoid using such expressions in the sense in which they are truly predicable only of English-speaking persons. (I am afraid some recent philosophers have often been ensnared by this latter ambiguity of "knowing the use of a word"; and so they give us tedious and inaccurate supplements to Modern English Usage, instead of philosophical discussion of a 'way of using a word' which could be found in many
languages.)

Many

cept" far

psychologists, wishing to use the term "conmore widely than 1 do, would strongly object

to my concentrating on concepts expressed in language;

THE NATURE OF CONCEPTS

IT

they would say that an animal has acquired a concept if it has learned a discriminative response to some feature of its environment. If a rat or a dog is trained to react in a certain way whenever it has a triangle shown to it (rather than some other shape), then they would say it has acquired a concept of triangle. (Earth-

worms, we are told, regularly pull bits of paper into their 'burrows by the sharpest angles; I do not know whether any psychologist would be prepared on this account to say earthworms have an innate concept of
angular magnitude.) What is at issue here is not just the way the term "concept" is to be used, but the desirability of comparing these achievements of rats and dogs with the performances of human beings who possess a concept of triangle; the psychologists I am criticizing want to play down the differences between human and animal performances, and I want to stress them. The life of brutes lacks so much that is integral to human life that it can only be misleading to say that they have concepts like us as misleading as it would be to say that men have tails and women lay eggs (as I have actually heard someone of strong 'evolutionist' opinions say in discussion) or to call the noises made by brutes "language". Experience in training dogs to 'recognize' triangles can be no guide in (let us say) teaching geometry. It is scholasticism in the worst sense of the word to argue that if men are descended from brutes they cannot have anything that the brutes did not already possess (nihil dat quod non habet!); the argument is just as bad whether it is used to disprove evolution (as it actually once used to be JAMES, vol. II, pp. 670-1), or to prove that brutes, from whom man is descended, must have concepts at least in a 'rudimentary' form.

ABSTRACTIONISM
SHALL USE
"abstractionism" as a
is

name for the

acquired by a process of out in attention some one feature given in singling direct experience abstracting it and ignoring the other features simultaneously given abstracting from them. The abstractionist would wish to maintain that all acts of judgment are to be accounted for as exercises of concepts got by abstraction; but he is often driven into allowing exceptions for certain kinds of concepts, because in their case abstractionism is palpably unreasonable. My own view is that abstractionism is wholly mistaken; that no concept at all is acquired by the supposed process of abstraction. My refutation of abstractionism will make no appeal to the empirical data collected by professional child psychologists. My procedure may thus appear to many readers arrogantly a priori. Surely the origins of anything have to be discovered empirically; what is the use of arguing without information? But since acquiring a concept is a process of becoming able to do something, an enquiry as to the origin of a concept may be vitiated from the start by wrong analysis of what is done when the concept is exercised; and such an error may be shown without any empirical data as to concept-formation in children. Again, as Wittgenstein remarked, the connexion between learning and being able to do something is not just an empirical one; for we should plainly not be willing to call any process that

I doctrine that a concept

ended

in ability to

do something 'learning" to do
18

'

it.

ABSTRACTIONISM
Of course we cannot work out
ties are acquired;

19

in our

heads

how

abili-

but there are conceivable ways of which we should unhesitatingly refuse to apply the term "learning". If, as in a story of Stephen Leacock's, a boy could come to know Latin by submitting to a brain operation, he would not have learned Latin from the surgeon. Now abstractionism is a the*ory of how certain mental performances are
acquiring them
to

am entering upon is designed show that the processes alleged by abstractionists would not be a learning of how to make these mental acts. Moreover, I shall try to show that the whole idea of abstraction of discriminative attention to some
learned; the discussion I

to

feature given in experience is thoroughly incoherent; if I succeed, no notice need any longer be taken of claims that abstractionism has experimental support.

No

experiment can either justify or straighten out a confusion of thought; if we are in a muddle when we design an experiment, it is only to be expected that we should ask Nature cross questions and she return crooked answers. Abstractionism has played an important role in the history of philosophy; but it is not this that concerns us the doctrine is very much alive. (See PRICE passim
ix.) Two applications of abstractionism are of special interest: to concepts of sensible things, and to psychological concepts. The first application came into vogue during the decadence of scholasticism: concepts of sensible things have to be formed by picking out features directly given in sense-experience (nihil in intellectu nisi prius in sensu), so that we cannot form any 'proper' concepts of kinds of substance, like gold and water, nor of substance as such; the word "substance" will stand for 'something we

and HUMPHREY, chap,

20

MENTAL ACTS

know not

what', and kinds of substance will be conceived only by piecing together 'proper' concepts of

their characteristic sensible features. (It is mere ignorance to make Locke the originator of this style of talk;

day it was the commonest scholastic claptrap, and this bad tradition still survives in certain quarters.) Such a view as to our knowledge of substance is grossly incoherent; what ought to be inferred on these premises is not that "substance" means something, we know not what, but that it means nothing; for assuredly we can find no discriminable feature of our senseexperience answering to the word "substance", so there ought not, on the view we are considering, to be any concept answering to it either. And similarly an
in his

abstractionist ought to hold that when we think of gold or water our thoughts must somehow be fadged up out

of sense-given elements; the concepts exercised must be concepts of features given in sense-experience like

wetness and yellowness, and will not include a special concept of substance. In fact this consistently abstractionist view of substance-concepts is now pretty
generally held.

As regards psychological concepts,

tic tradition in

there

is

a scholas-

nisi prius in sensu"

which the maxim "nihil in intellects, was strictly applied to them; they were supposed to be analogical developments of concepts acquired in the first instance by performing abstraction upon sense-experience. I shall return to this doctrine of analogical development of concepts at a

not necessarily connected with abstractionism, and is indeed open to serious objection on abstractionist premises. Let us suppose that we use a word C originally to express our concept of a recognizlater stage; it
is

able characteristic c found in sense-experience,

and

ABSTRACTIONISM
sense.

21

then go on to use C in a transferred, psychological,

The new use of the word C will express recogninot of the sensible characteristic c, but of some psychological characteristic y\ only because we do recognize y as a feature of our mental life can we see the suitability of expressing this recognition by the metaphorical use of the term C. But if we do thus recognize y, then we can in principle have a 'proper'
tion,

concept of y, independent of any 'analogy' between y and c. This objection seems to me pretty conclusive

from an abstractionist point of view;

be

if

any sense

is

to

of the idea that psychological concepts are developed by 'analogy', it must be freed from its con-

made

nexion with abstractionism.

characteristic abstractionist view of psychological concepts is that they are abstractively derived from inner experience; we are supposed to possess a

The

quasi-sense that

is

related to psychical occurrences in

much the same way as our senses are to physical occurrences; and just as we form a concept of red by discriminatively attending to, and performing abstraction upon, visual experience, so we form a concept of judging or desiring by discriminatively attending to, and performing abstraction upon, the deliverances of this
'introspective' quasi-sense. The classical statement of this doctrine is to be found in Locke (Essay, II. i. 4);
it is
still

a doctrine
talk

still

widely held

at least,

many people

and argue as if it were true. (The quasi-sense is compared sometimes to seeing intTo-spection sometimes to feeling; sometimes we are supposed to learn what judging and desiring are by taking a quick inward look and catching ourselves in the act of judging or desiring, and sometimes, by the way we feel

when performing

these acts.)

22

MENTAL ACTS
I shall discuss later

on the peculiar difficulties of 'inner sense'; if this doctrine cannot be maintained, then abstractionism must be abandoned also, at least as regards psychological concepts. (We have seen
already that a serious attempt to make out that such concepts are developed 'analogically' from concepts applicable in sense-experience would oblige us to go outside the limits of abstractionism.) Conversely, if we see reason to reject abstractionism altogether, even for concepts like that of red, then we shall have overcome the strongest inducement to describe self-knowledge in terms of inner 'sense'; for there will no longer be any tendency to think that psychological concepts

must be got
experience.

'abstractively'

from something

like sense-

ABSTRACTIONISM AND LOGICAL CONCEPTS

rarely attempt an abstracjLJLtionist account of logical concepts, like those of

ABSTRACTIONISTS
or,

some,

and not; they will usually admit that such conare a special case to which abstractionism does cepts not apply. There have, however, been some attempts to explain even logical concepts as obtained by abstraction (as in RUSSELL (4) and PRICE). These attempts appeal to 'inner sense' as the material from which logical concepts are to be extracted. Such an appeal could hardly be avoided; for logical concepts are not to be

explained as the result of performing abstraction upon any sense-experience. In the sensible world you will

LOGICAL CONCEPTS
find

23

no specimens of alternativeness and negativeness from which you could form by abstraction the concept
of or or of not.
sible

The

world that

is

sort of concept applicable in the SCAplausibly representable as got by

abstraction

is the sort that could in favourable circumstances be conveyed by ostensive definition; the word answering to the concept could be used to label a specimen o*f what the concept applies to, or a picture of such

a specimen. But nowhere in the sensible world could you find anything, nor could you draw any picture, that could suitably be labelled "or" or "not". Hence the abstractionist appeals to 'inner sense' as the source of logical concepts: "or" gets its meaning through our per-

forming abstraction upon experiences of hesitation, and "not" is similarly related to experiences of frustration or inhibition (stopping oneself from doing things), and so on. So far as I can tell, these abstractionist views as to logical concepts being got out of inner experiences are based upon the feelings that happen to be aroused in a particular writer when he says a logical word over and over to himself a magical rite of evoking its meaning. (For a conspicuous example of this, see JAMES, vol. I, pp. 245-6, 252-3.) To many people, such recitation of the word "or" suggests a feeling of dithering between alternatives; to me, on the other hand, it " !" or else naturally suggests a threat Would an abstractionist be willing to say on that account that my concept of or was different from the others'? He surely ought to say so, on his own premises. But in fact the coherency of other men's language, and the possibility of communicating with them, should be conclusive proof that all have the
,

logical concepts (in the sense of "same concepts" explained in the last Section but one); idiosyncratic

same

24

MENTAL ACTS

feelings produced by logical words are neither here nor there. Of course in the living use of such words,

opposed to the recitation of them, such feelings may be wholly absent, without the words' in any
at,

way

being deprived of meaning by their absence. I very much doubt whether anybody either dithers or feels threatened when he contemplates the statement that every number is odd or even. Russell's attempted account of "every S is P" is highly instructive. A judgment that a definite individual, A, is a swan and is not white would consist in my applying my concept swan to A but stopping myself from applying my concept white to A. I can now abstract what would be common to the possible judgments that A is a swan and not white, that B is a swan and not white, and so on; this gives me the content of a possible judgment that something is a swan and not
white. If I

now

inhibit myself
is

from making

this judg-

from the contents of certain possible judgments that would themselves involve my inhibiting myself from doing something, then I am judging that nothing is a swan without being white, i.e. that every swan is white.
ment, whose content
derived,

by

abstraction,

it is quite fair play to use inhibition of inhibition as part of his apparatus; has not Pavlov an studied pre-verbal forms of this operation in dogs?

Russell thinks

(RUSSELL

(4), p.

255;

cf.

also pp.

88-90 on "some".)

professional psychologist may well protest that Russell's account just exhibits the characteristic weaknesses of a logician trying to do psychology, and is too

obviously wrong to be of any interest. But the logical part of Russell's analysis is right: "every S is P" does mean the same as "no S is not P", and this is the negation of "some S is not P". The trouble arises rather

LOGICAL CONCEPTS

25

from Russell's abstractionist accounts of negation and "some"; since his account as a whole is clearly wrong, and the logical part of it is right, we ought to reject abstractionism about logical concepts. The attempt to relate logical words to inner experiences and feelings thus breaks down. For a convinced abstractionist, though, a possible alternative would be
to

deny that there are

special logical concepts corre-

sponding to the logical words. This is especially plausible as regards negation. Is it not a mere blind copying
of language-structures to suppose that use of the term "not red" brings into play two distinct concepts the

concept not and the concept red*! Surely what I exercise in using the term "not red" is simply the concept red; knowing what is red and knowing what is not red
are inseparable eadem est scientia oppositorum. The abstractionist might thus say that the problem how we get a concept of not
in positive
is spurious; only concepts expressed terms need to be explained, and for these abstractionism has not been shown to be inadequate. If concepts were fundamentally capacities for recognition, as abstractionists think they are, this argument would be decisive; for clearly we recognize the presence and the absence of a characteristic by one and the same l capacity. (See PRICE, p. 123. ) But even apart from this special view as to the nature of concepts, it seems un-

deniable that in some sense the concept red is the very same mental capacity as the concept not red; and thus there seems to be no place for a special concept of negation putting "not" before "red" would simply
1 Oddly enough, on the very next page of his work Price slides into the view that negation is understood through abstraction performed on 'inner' experiences of frustration: "Disappointed expectation", he says, "is what brings NOT into our lives".

26

MENTAL ACTS

serve to distinguish a special exercise of the concept

red.

however, we relate the concept not to judgments expressed in language, then I think we can see a justification for saying that there is such a distinct concept, since ability to use the word "not" intelligently is certainly a distinct mental ability. Again, students of elementary logic show ability to understand the word "not" in logical schemata, such as: "if every P is M, and some S is not M, then some S is not P". If "not" in "not red" were merely a signal for a special exercise of the concept red, our grasp of its meaning in such schemata would be inexplicable. We cannot say that "not" in "not M" is merely a signal for a special use of the concept expressed by "M"; for the schematic letter "M" does not express a concept, as "red" does.
ft

If,

Even if we consider non-linguistic performances in which concepts are exercised, we can still, I think, distinguish between exercise of the concept red and not red and exercise of the concept not as such. To sort a lot of pebbles into red ones and ones that are not red could be simply an exercise of the concept red and not red; and here it would be plainly silly to distinguish a concept of red and one of not red as distinct abilities. But now suppose that the sorter goes over his two heaps
when the
is done, to see if there are any misthe act of transferring a pebble to the placed pebbles: other heap is an exercise of the concept of negation.

sorting

way that the concept not would the sorter's behaviour, I found it natural Latin "after", not "a/ws": to use the word "other" ( "the other of two"). This concept of other is near akin
In describing the
itself in

show

to the concept of negation; and is equally inexplicable on abstractionist lines. There does not even seem to be

NUMBER-CONCEPTS

27

a characteristic feeling of otherness; and there is certainly no feature common to all the things that are
other.

Yet this concept is easily and early arrived at; "a child understands the use of "other hand" and "other
understands "right" and left", for which an abstractionist account is plausible. The logical concepts must then, I think, be recognized as distinct mental abilities; and if so they do not
foot" before
it

tion of

admit of any abstractionist explanation. Such recognithem does not commit us to trying to answer
the question how many distinct logical concepts there are, or how many are exercised in a given act of judgment; we saw in the last Section but one that such questions are unreasonable.

ABSTRACTIONISM AND ARITHMETICAL CONCEPTS

very many concepts, of a more homely appearance than the logical concepts, for which abstractionism is no less unworkable. Numberconcepts are one example. We are indeed often taught arithmetic in childhood by such procedures as putting three apples with two apples; and then later on we =5" hear of the addition of 'abstract units' "3

ARE

+2

means, we

are told, that three abstract units

five in all.

and two

more make

This

is

surely because abstrac-

tionism has invaded pedagogic theory and practice; the talk about 'abstract units' presumably means that is taken as expressing the result of

"3+2=5"

28

MENTAL ACTS

abstraction from the nature of the concrete units added, e.g. from their being apples. But this is one of the

Contexts in which the abstractionist talk of "discriminative attention" is most obviously a muddled and indefensible way of talking. It may well be all right to say: "Don't bother about the shapes of the things I show you, just attend to their colours"; this is the sort of case that makes abstractionism plausible. *But it is certainly a self-contradictory instruction to say: "Don't bother about what kind of things they are, just attend to their number". For a number is essentially a number of a kind of things; things are numerable 1 only as belonging to a kind of things. What number I find may vary, without my observations' varying, because I am considering a different kind of things; the same auditory experience may give me the number 2 if what I have in mind is heroic couplets, 4 if it is lines of
verse,

40 if it is syllables, 25 if it is words; if I have no special kind of thing in mind, no number will suggest itself to me at all. (Of course, if I am shown a lot of apples and just told to count, the kind of things I shall most likely think of will be apples; but it need not be.) Thus number-concepts just cannot be got by concentrating on the number and abstracting from the kind of things that are being counted. Before applying a number-concept we must first apply the concept of some kind of things to the things that we count. This shows why number-concepts cannot be among the
first

to be acquired,

and why there are languages of

This truism is certainly what Frege intended to bring out in saying that a number attaches to a Begriff. What makes the truism worth enunciating is that people who would verbally assent to it are often sufficiently muddle-headed to say other things that are incom1

patible with

it.

NUMBER-CONCEPTS

29

primitive peoples that are rich in words for kinds of things (corresponding to common nouns and adjectives
in English)

but very poor in numeral words. Moreover, the elementary arithmetical operations cannot be coherently explained as being performed on
collections of abstract units, or as what operations on concrete collections look like when we abstract from

the special natures of such collections. Such an account superficially appears adequate for addition and subtraction; for multiplication and division at all. It sounds plausible to say that
it

will

not do

"3
of

+2 =5"

puttingtogether and taking-away operations performable on pennies and conkers; features that we apprehend on their own account by abstracting from the pennies and the conkers. It is perhaps a bit less plausible to say

and

"5

=3"

describe

features

by taking an operation performed on concrete and abstracting from the concrete units, we can conceive an operation upon that many abstract units;
that, units

whatever "abstract unit" means, surely "three conkers and two more are five conkers" remains true when for "conkers" we substitute "abstract units"! But it does not in the same way look like sense to talk of getting six abstract units by multiplying two abstract units into three abstract units, or of getting two abstract units by dividing three abstract units into six abstract units; nor are there any operations with pennies that would enable us to give such talk a sense when once we had abstracted from the pennies. To talk of raising two abstract units to the power of three abstract units is, if possible, even worse nonsense; and no manipulations of three pennies will teach a child what the exponent "3" means. In teaching arithmetic, the idea that the number n consists of n abstract units is
still,

80

MENTAL ACTS

not just logically open to exception, or lacking in rigour; it is quite unusable in practice. Even at the level of school arithmetic, a child needs to master a way of talking about the number n which has nothing to do with groups of n concrete or abstract units viz. in expressions of the form "doing a thing n times over". Exponentiation has to be explained this way; 2 to the power of 3 is 1 multiplied 3 timeS over by 2 (1 x 2 x 2 x 2). The same thing is clear for multiplication and division. "2 x 3 = 6" means that to add 2 to anything 3 times over is to add 6; "13 divided by 5 gives 2, remainder 3" means that from 13 you can subtract 5 twice over, the result being 3. Now this account could be used, instead of the nonsense about concrete and abstract units, for addition and subtraction too. Adding 5 to 7 can be explained as performing 5 times over upon the number 7 the operation of going on to the next number (7), 8, 9, 10, 11, 12. Subtraction is similarly explicable in terms of counting backwards. "This may be more correct logically than talking about abstract units", someone will proit is psychotest, "but it is too difficult for children unsound". But if a child finds "doing a thing logically so many times over" too difficult to understand, it will never acquire genuine number-concepts and never be able to do arithmetic at all. And learning to say number words when presented with groups of apples varying in number is not even to start learning what "to do a thing so many times over" means. To teach a child this, one would count, or get the child to count, while

some action was repeatedly done.

Now the way counting is learned is wholly contrary to abstractionist preconceptions. What an abstrac* tionist might call 'abstract counting" i.e. recitation

NUMBER-CONCEPTS
of the
successive numerals
is

31

without counting any objects logically and temporally prior to counting objects; the series of numerals has to be learned before it can be applied. The first numerals ("one" to "twenty" in English) must indeed be learned parrotwise to begin with, like "eany-meany-miny-mo"; but
success in counting
child' s'practical

beyond

this point

grasp of a rule

forming numerals indefinitely. a rule of course does not imply ability to formulate it.) Young children sometimes amuse themselves by counting up to large numbers (several thousands); they have never heard their elders do this, and quite likely they will never in their lives have occasion to count so high for any practical purpose; the exercise of going on in the series of numerals is one that they find pleasurable on its own account. The ability to do this is not yet possession of full-blown number-concepts, because the series of numerals has not yet any application; but the performance is a characteristic intellectual performance. Moreover, the pattern of the numeral series that is thus grasped by the child exists nowhere in nature outside human languages; so the human race cannot
possibly have discerned this pattern by abstracting it from some natural context. The practices of counting

depends on the which it could go on by (Such practical grasp of

and again of counting the repeated performances of some action or operation, are developed on the basis of this 'abstract counting'; what has to be
objects,

mastered here is the establishment of one-one correspondence between the numerals and the things or performances being counted e.g. one gets the child to go on to the next numeral when and only when it goes up another stair of the staircase. The abstractionist doctrine that 'abstract numbers' are understood by

32

MENTAL ACTS
num-

abstracting from the special nature of the things bered puts the cart before the horse.

ABSTRACTIONISM AND RELATIONAL CONCEPTS'

near akin to logical are, turn out to resist abstractionist explanations. Consider the concepts big and small. Is there a common identifiable feature shared by a big elephant, a big rat, and a big flea, and which they do not share with a small elephant, a small rat, and a small flea? Can I pick this feature out by discriminative attention, abstracting from what kind of thing it is that I am calling big or small? Certainly not; I cannot rightly apply the term "big" or "small" unless I am meaning a big or small thing of a certain kind. A big flea or rat is a small animal, and a small elephant is a big animal. There can be no question of ignoring the kind of thing to which "big" or "small" is referred and forming a concept of big or small by abstraction. It has indeed been pointed out that I can recognize a thing as big or small of its kind without any explicit mental reference to the standard size of that kind of thing (PRICE, p. 65); but though I need not have the standard size in mind, reference to the kind of thing is inherent in any use of

OTHER than number-concepts concepts

CONCEPTS,

less

"big" and "small". Relational concepts in general raise

difficulties

of

which

lump

abstractionists are blithely they relations together with attributes as 'recurrent

unaware

COLOUR-CONCEPTS
characteristics' that

83

The

trouble

is

we can recognize (cf. PRICE, p. 10). that a relation neither exists nor can be

observed apart from its converse relation; what is more, the concept of a relation and of its converse is one and the same indivisible mental capacity, and we cannot exercise this capacity without actually thinking of both relations together; relativa sunt simul natura et intellectu. It is difficult, in view of this, to give any account of our understanding the difference between "the knife is to the left of the book" and "the knife is to the right of the book". But I do not see how an abstractionist account of it is even possible; for all that is obtainable by abstraction, one would think, is
ability to recognize the 'recurrent characteristic' of right-left ordering, not ability to tell which thing is to

to the right. If we are to concenthe 'recurrent characteristic', and 'abstract upon from' the terms between which the relations hold, what else can be the result? Yet surely it is only if there is at least an ability to tell which thing is to the right and which to the left, that we can say there exists a con-

the

left

and which

trate

cept of right and

left.

10

ABSTRACTIONISM AND COLOUR- CONCEPTS

EXAMPLES
so far

show only that ab-

stractionism does not cover the whole field of concepts. I wish to maintain a far stronger conclusion that there is no concept at all of which an abstractionist account is adequate. If there were any truth in

84

MENTAL ACTS

abstractionism, it would at any rate be adequate for concepts of simple sensible qualities, for concepts like red and round. Now if I possess the concept red, then
I can

tences containing the

perform acts of judgment expressible in senword "red". This ability, however, certainly cannot be learned by any kind of attention to red patches for any length of time; even
if

after a course of attending to red patches the kbility turned out to be present, we should still be justified in

refusing to say

it

had been learned that way.

We

can

say this quite as confidently as

we can say that the ordinary use of the word "red" cannot be learned by hearing the word "red" uttered ceremonially in the
presence of a red object simply on this account, that such a ceremony is not the ordinary use of the word "red". Price has the rare merit among abstractionists of having pointed out that ceremonious ostensive definition normally plays a rather small part in the learning of language. His own theory is that we learn the sense of words like "cat" and "black" by a double process of abstraction; that

the

common feature,

e.g.

"cat", in these otherwise un-

with a common environmental situations which are factor in observed otherwise unlike. Similarly "black" gradually sorts itself out from another range of utterances which are otherwise unlike, and is correlated with a visible
like utterances is gradually correlated

quality experienced in otherwise unlike situations. (PRICE, p. 215.)

This

is much more plausible than the usual stuff about ostensive definition, but I think it is still open to two fatal objections. First, it is integral to the use of a

COLOUR-CONCEPTS
general term that

35
it

we

are not confined to using

in

some object to which the term applies; we can use the terms "black" and "cat" in situations not including any black object or any cat. How could this part of the use be got by abstraction?
situations including
is part of the very beginnings of lanthe child calls out "pot" in an 'environmental guage; situation' in which the pot is conspicuous by its ab-

And

such use

of course not enough, even when used to describe the immediate situalanguage being tion, that we should utter a lot of words corresponding to several features of the situation; but the abstraction that Price appeals to could scarcely account for our doing anything more than this. It is indeed hard to see at first sight what account other than an abstractionist account could possibly be given for such concepts as colour-concepts; and for this very reason such concepts are the abstractionists' favoured examples. The reason, surely, why men born blind cannot have colour-concepts is that they have no colour-sensations; and does not this imply that men who can see form their colour-concepts by
sence. Secondly,
is

it is

attending to certain features given in their visual experiences? In the first place, it is not true that men born blind can form no colour-concepts of any sort. A man born blind can use the word "red" with a considerable

measure of

intelligence;

of the logic of the word. 1

he can show a practical grasp What is more, he may, if he

1 "Red" has, for example, a very considerable logical resemblance to "hot"; both redness and heat admit of degrees, a surface can be red all over and hot all over, there can be redness or heat throughout a volume, etc., etc. A blind man could show that he had a concept of red akin to our own by using "red" and "hot" in logically similar ways; he need not be able to formulate the logical similarities.

36
is

MENTAL ACTS

intelligent, grasp something of the aesthetic significance of red and its place in human life like Locke's blind man who said he thought scarlet must be like the sound of a trumpet. Of course a man born blind cannot have our colour-concepts; but this, it turns out, proves nothing whatsoever. If it is logically impossible that a man blind from birth should have colour- concepts quite like those of a sighted man, then the unlikeness cannot be regarded as a remarkable empirical fact, throwing light upon the origin of concepts. Now concepts are mental capacities; somebody who is enabled

by

his colour-concepts to

do what somebody

else can-

not do has a specifically different capacity and thus a different concept. A part of a sighted man's colourconcepts is his ability to apply them to visual experiences; and a man blind from birth necessarily lacks this ability. There is sufficient similarity between the
abilities of

men

blind from birth and of sighted men,

alike use colour-words intelligently, for us to be justified in speaking of "colour-concepts" in both

cases;

who

what

dissimilarity there

is

is

a logical conse-

quence of the blind man's having always been blind, and calls for no abstractionist explanation. If we supposed that
the same
blind from birth could (logically) have colour-concepts as sighted men, their not

men

having them would call for explanation; but such a supposition would be just a muddle. Different men
tools that are quite similar in dissimilar ways; but a concept is not externally related to its applications, as a tool is to its uses; a man who can do different things with his concept has a concept that is to some extent different from the other fellow's. In point of fact, abstractionism goes to wreck even over colour-concepts, as may be shown by a further

may use

COLOUR-CONCEPTS

37

argument. Let us consider chromatic colour, i.e. colour other than white, grey, and black. The concept
chromatic colour, in spite of my having had to explain the term by using a string of words, is surely no less 'simple', no less a concept of something 'given in
direct experience', than the concept red. (One common meaning of the adjective "coloured" is the possession

of chromatic colour
glass",

"coloured chalks", "coloured

twopence coloured".) Now it is that I should form this concept, quite impossible chromatic colour, by discriminative attention to a feature given in my visual experience. In looking at a red window-pane I have not two sensations, one of redness and one barely of chromatic colour; there are not, for that matter, two distinct sense-given features, one of them making my sensation to be barely a sensation of chromatic colour, the other making it a sensation of redness. If I abstract from what differentiates red from other chromatic colours I am abstracting from red itself; red is not chromatic colour plus a differentia, so that we can concentrate our attention upon chromatic colour and abstract from the differentia.

"penny

plain,

The difficulty I have just raised resembles one raised by Locke (Essay, III. iv. 6) about the relation between the concepts red and colour "colour" having for him
the wider sense of the word, covering also white, black, and grey. Locke observed that "red" cannot be defined by adding a differentia to "coloured", and inferred that the concept coloured cannot be reached by abstracting from the distinctive marks of the various colours. His solution is that the word "coloured" stands for 'the way' that all the 'ideas' of the several colours 'got into the mind'; "it signifies no more but such ideas as are produced in the mind only by the sight and have

38

MENTAL ACTS

is

entrance only through the eyes". Presumably his view that all 'ideas' with this common origin have a pecuor look, which is manifest to 'inner sense' and can be singled out abstractively; at least, I can think of no other interpretation that makes his view plausible. Even so, this is a pretty obvious makeshift. But I would rather not discuss Locke's actual problem, because (using the word "colour" in his way) we should encounter special, and for the moment irrelevant, difficulties about the concept colour. It is clear, however,

liar feel

that the concept chromatic colour stands on the same footing as the concept red, and does not raise any of these peculiar difficulties; clear, too, that no such

account as Locke gives for the concept colour will fit. Failing to account for so simple a concept as chromatic colour, abstractionism must be pronounced finally
bankrupt.

11

MAKING CONCEPTS IS NOT A FINDING OF RECURRENT FEATURES

THIS POINT we feel inclined indeed, almost jL\.foYced to say that the mind makes a distinction between redness and chromatic colour, although there are not two distinct features to be found in the red
glass or in

AT

my visual sensation. And here, I think, we are approaching the true solution. "If the glass has not two distinct attributes, redness and chromatic colour, how can these attributes be dis-

CONCEPT-FORMATION
tinct

80

anywhere? and then how can anything be

chromatically coloured without being red?" This objection, natural as it is, rests upon a false Platonistic

an attribute is being thought of as an identifiable object. It would be better to follow Frege's example and compare attributes to mathematical functions. 1 "The square of" and "the double of" stand for distinct functions, and in general the square of x is different from the double of x; but if a? is 2 then the square of x is the double of #, and the two are not in this case distinguishable even in thought. Similarly, since what
logic;
is chromatically coloured is often not red, "the chromatic colour of #" or "#'s being chromatically coloured" does not in general stand for the same thing as "the redness of a?"; but if x is red, then it is by one and the same feature of x that x is made red and chromatically coloured; the two functions, so to speak, assume the

same value

for the

argument

x.

"If there are two distinct features to be found in my mental representation, and only one in the pane of glass, then my mind represents things otherwise than they are, i.e. represents them falsely. It would follow

that the conceptual way of thinking distorts reality." can we get over this difficulty? The statement "An understanding which understands a thing otherwise than the thing is, is wrong" is ambiguous; the qualification of the verb "understands" by the adverb "otherwise" may relate either to the thing that is understood or to the person who understands. If it relates to the thing that is

How

1 The use of this analogy does not commit us to the whole of Frege's apparatus. Readers who are interested hi the development of the analogy may care to refer to my paper "Form and Existence"

(Proc. Arist. Soc., 1954-5).

40

MENTAL ACTS

understood, then the statement is true, its meaning being this: Any understanding that understands a thing to be otherwise than it is, is wrong. But this meaning is inapplicable to the case we are considering. For when our understanding frames a proposition about the colour of the glass, it does not assert that this colour is comIf, howplex, but on the contrary that it is simple. the qualification is taken to refer to the person ever, who understands, then the statement is false; for the way it is with our understanding when we understand is different from the way it is with the thing we understand, in its actual existence. When our understanding understands things that are simple, it may understand them in its own complex fashion without understanding

them to be complex. The last paragraph is a

translation

from Aquinas (la

q. 13 art. 12 ad 3 um), modified so as to fit my chosen 1 example. It appears to me to be a decisive solution to the problem of conceptual thought's 'falsifying' reality. We can now say something that goes for all concepts without exception: Having a concept never means

being able to recognize some feature we have found in direct experience; the mind makes concepts, and this concept-formation and the subsequent use of the concepts formed never is a mere recognition or finding; but this does not in the least prevent us from applying concepts in our sense-experience and knowing sometimes that we apply them rightly. In all cases it is a matter of fitting a concept to my experience, not of picking out the feature I am interested in from among other features given simultaneously. Suppose I look at a lot of billiard balls on a table, and form the judgment that some of them are red and some are not. If I state
1

For Aquinas's attitude to abstractionism, see the Appendix.

CONCEPT-FORMATION
this

41

judgment in words, "red" may plausibly be taken to report a feature of what I see, but "some" and "not" certainly cannot. But it would be perverse to infer that

distorting conceptual thought represents the reality as exhibiting features, somehood and nottish-

my

and no less perverse to my judgment is correct, there must be somehood and nottishness in rebus. We must resist
ness,

which are not really

there;

argue that, since

the perennial philosophical temptation to think that if a thought is to be true of reality, then it must copy it feature by feature, like a map. Even the use of the word "red" is not to be explained in terms of reporting or copy ing real features; for, as we saw, the terms "red" and "(chromatically) coloured" may answer to the same feature of the same thing; and again, what is logically distinctive in the use of colour- words is certainly not to be reached, by an act of abstraction, from the seeing of red things. In rejecting abstractionism, we deny a privileged
position to 'sensory' concepts, e.g. colour-concepts. These concepts are indeed specially involved in the

procedure of describing one's sensations; geometrical concepts and colour-concepts enter into the description of visual sensations in a way that other concepts do not. If I look at a tomato, there is some justification for saying: "what I actually see can be completely described in terms of colour, shape, and size, without mentioning any other characteristics". There is however no reason to think that this gives 'sensory' concepts an epistemological primacy over others; for the description of sensations is a highly sophisticated exercise of concepts, and is secondary to the application of 'sensory' concepts to the material environment. In
this primary,

outward-looking application, 'sensory'

42

MENTAL ACTS

concepts have not in fact any privileged position; a child with only a few concepts and only a small understanding of language may easily possess concepts like door and book (and even the logical concept other, as in "other hand", "other foot") before it has any colourconcepts at all. About psychological investigations designed to prove that concepts are formed by abstraction, I think it is safe to say that either what is studied is not the formation of concepts, or it is manifest on a little consideration that the concepts were not formed by abstraction. The first criticism applies to such experiments
as getting rats to 'recognize triangles as such', i.e. training them to react similarly to triangles of various

shapes and sizes. The psychologist may well say: "the rats are able to perfect a type of behaviour which is fully described by the implications in our use of the

term "concept" "; but 'our' (the experimenters') use of the term is then wholly different from its use in this book. (See HUMPHREY, p. 254.) This is not merely a verbal point; the psychologist's explanation of such a

now

phrase as "the concept of triangularity" is what it is fashionable to call a persuasive definition, not a mere Humpty-Dumpty pronunciamento. He can scarcely deny that human beings show their possession of the concept triangle by behaviour very different from rats'; but he invites us to regard this difference as

unimportant.
Consider, on the other side, a study of genuine concept-formation by Smoke (HUMPHREY, pp. 252-3, 275-6). On cards three inches square, figures of various
sorts

were drawn; some of these were pogs, a pog being

a blue rectangle enclosing a blue circle that touches only one long side of the rectangle. (Humphrey des-

CONCEPT-FORMATION
cribes

43

"pog" as a "nonsense name"; this is a selfcontradictory expression, and is clearly wrong; "pog" is merely not part of ordinary language.) The subject was shown a number of pogs, was given the verbal instruction "Try to find out everything a figure must be if it is to be called a pog", and was told to raise his hand when he thought he knew what a pog was. His claimto knowledge was then tested by getting him to try (i) to pick out further pogs, (ii) to draw a pog, (iii) to
define "pog" in words. Many subjects who failed at test (iii) succeeded at test (i). Humphrey thinks that this

experiment shows "that a subject can form and utilize a concept without verbalization". What Humphrey appears not to notice is that the verbal instruction given to the subject requires for its understanding some highly sophisticated concepts (the concept of name, the concept implicit in "everything that a thing must be if ."); such concepts are not the first to be acquired, and the experiment among does nothing to show that they are acquired 'without
. .

verbalization'. Again, the verbal instructions, and the whole set-up of the experiment, already delineate

the a mere nonsense-syllable at the start of the word is not experiment (as Humphrey's tendentious term "nonsense name" suggests). "Pog" is to be a common noun, used to refer to a certain design of colours and shapes; a subject who has grasped this much already has an inchoate concept of a pog, and with one who had not grasped this much the experiment would not succeed. (The 'grasp' that I mean is a matter of partly knowing

many of the logical peculiarities of the word "pog";

how to use the word "pog", not of knowing, explicitly judging, that the word is to be a common noun, etc.) So the experiment does not show that the subject can

44

MENTAL ACTS

abstractively form the concept pog from scratch, independently of 'verbalization'.

Are we to say that subjects who can recognize pogs but cannot verbally define the term "pog" possess the concept pog? This certainly seems reasonable; how many of us could give a water-tight definition of "chair" or "money", words that we should certainly wish to say express concepts? Defining a term is normally a particular exercise of the corresponding 1 concept rather than a way of getting the concept, and performance of this exercise is not a necessary con-

But though the subject the concept pog without being able to give may acquire a verbal definition of "pog", in the experimental set-up described the concept is acquired only against the background of a considerable command of language; to speak of "forming concepts without verbalization" merely obscures the matter. As I remarked before in reference to aphasia, I do not wish so to use the term "concept" that every condition of having the concept.

cept necessarily involves some linguistic capacity.

But

Smoke's experiments certainly do not show that concepts can be formed from scratch by abstraction by
discriminative attention to features given in senseexperience and that language is of no importance in
their formation. Linguistic capacities, for

an abstrac-

tionist, are necessarily an external, adventitious aspect of the possession of concepts. 'Fundamentally a con-

a recognitional capacity' for the abstractionist (PRICE, p. 355); so a depreciation of linguistic performances is an essential part of the abstractionist programme. Smoke has supplied no experimental support for such depreciation.
cept
is
1

As Aquinas already

recognized: la q. 85 art. 2 ad 8

um.

RUSSELL'S THEORY OF JUDGMENT

to the analysis of the act of judgment, in which concepts are exercised. I shall begin by considering Russell's account of judgment (RUSSELL (1), (2), (3) ).

WE
(i)

MAY NOW RETURN

The preambles to the act of judgment. Russell from unexceptionable assertions: that we cannot judge if we do not know what we are judging about, and that we cannot understand the sentence in which a judgment is expressed unless we know the meanings of the several words. He then interprets 'knowing the meaning of a word" as 'standing in a direct cognitive relation to the thing that the word
starts
4
'

means". This 'direct cognitive relation' is called "acquaintance" and "awareness". When we know the meaning of a general term, we are acquainted with a universal entity that the general term signifies; awareness itself is one of the universal entities that we are

aware of or acquainted with. As regards particular entities, we must not be misled by the fact that we understand proper names of people, and that we speak of "being acquainted" with them, into holding that people, or for that matter material things, can be objects of acquaintance. For Russell, only a man's sensedata and some psychical states of his own 1 can fall
within his direct awareness;
are
all

other particular entities

known only by description. A grammatically proper

1 Russell tentatively allows the possibility that the pronoun "I" (used as in the Cartesian "I think") also stands for an object with which the user is acquainted.

45

46

MENTAL ACTS
is

name

but is short for a definite description; and in order to understand a definite description I need only be acquainted with the universal and particular entities that are meant by the terms of the description I need not be acquainted with the thing described. (E.g. in order to understand the definite description "the oldest of men" I need to be acquainted with the universal entities that constitute the meanings of "older than" and "man"; but I clearly need not be acquainted even in the ordinary sense, still less in Russell's, with the oldest of men.) Russell's restricting our acquaintance to sense-data, rather than material objects and other people, is partly a matter of Cartesian doubt; I cannot doubt the existence of the sense-data, whereas I can doubt the existnot logically
so,

ence of material objects and other people, so it is only to the sense-data that I can stand in a 'direct cognitive relation'. Another reason for the restriction is that Russell regards the judgment as a complex, of which the things really judged about (sc. the objects of
acquaintance) are actual constituents. Now if I judged that the present King of France is fat and bald, it would be hard to suppose that a bald fat man was bodily a constituent of my judgment, even if my judgment were true and even if I had the honour to be acquainted (in the ordinary sense) with the King. Sense-data, on the other hand, are sufficiently private, fleeting, and insubstantial to be plausibly taken as constituents of a private mental act. As for the universal entities that are judged about, I do not think
their being constituents of judgments raises any worse problems than their being constituents of any other particular entities (bad
(ii)

enough cf. Plato's Parmenides}. Analysis of the act of judgment. Russell recants

INADEQUACIES OF RUSSELL'S THEORY

his earlier

47

view that when we judge we are acquainted with a single entity which, if the judgment were put into words, would be the meaning of the sentence
expressing it. He holds that when James judges that p, there is an irreducible many-termed relation whose terms are, on the one hand, James's Cartesian ego, or at any rate some psychical element in James, and on the othsr hand those entities, universal and particular,

with which James has to be acquainted in order to understand the sentence represented by "p". Let us take the simple case in which James judges that a is larger than 6, a and b being objects of his awareness: the terms of the judging relation will then be: (1) a psychical element in James, j\ (2) the object a; (3) the
relation larger than] (4) the object b. may symbolize the relation by "B (j a, larger than, &)"; if we had
9

We

instead

"B

(j, b,

James judged

larger than, a)", this would that b was larger than a.

mean

that

13

INADEQUACIES OF
RUSSELL'S THEORY
RUSSELL'S
of our having direct acquaintance with universal entities is one which

NOTION

anybody who has been convinced by my argument that the formation of concepts is in no case a matter of finding something will be disposed to reject; and I shall
not here discuss his notion of sense-data. Let us rather observe a peculiar twist of Russell's argument. He begins, I said, with the unexceptionable premise that in

48

MENTAL ACTS

order to understand a sentence you must understand several words; then he takes the meaning of a word to be an object to which the user of the word has a
its

about what sorts of

'direct cognitive relation'; then he forms his doctrine entities we can stand in direct

in ordinary sentences these sorts of entities;

cognitive relation to; then he notices that many words do not have as their meanings and then we are told thst ordin-

ary language is logically misleading, that ordinary proper names are not logically such, etc., etc. But this conclusion wrecks the inference with which the whole theory began. For then knowing the meaning of a word in ordinary language need not be acquaintance (in Russell's sense) with an object that the word stands for; and if it need not be, why should we suppose that it ever is so? Russell seems to me to be certainly right in denying that the meaning of a sentence expressing James's judgment is a complex object (or 'objective', as Meinong would call it) with which James is acquainted. Here we need to bring into play that 'sense of reality' to which Russell rightly attaches importance in logical investigations; we ought to find it incredible on the face of it that if James believes something false, then there is a falsehood in the nature of things that James

contemplates and, in his judgment, accepts. We may indeed be disposed to attach some weight to this type of argument: "Smith and James both judge that the prisoner is guilty; so they both judge the same thing; so there is a thing that they both judge". But whatever force this argument has is only what it shares with the argument: "Smith and James are both bald; so there is a thing that they both are". If the common predicate "bald" answers to something in rebus, so does the com-

INADEQUACIES OF RUSSELL'S THEORY 49 mon predicate ". judges that the prisoner is guilty";
. .

this is no ground for supposing that a fragment of the latter predicate, the oratio obliqua clause, has a

but

complex entity as its reference. Anyhow, analysing judgments in terms of 'objectives' is a lazy
is treated as a simple twothe complexity of the judgment is just relation, transferred in its entirety to that which is judged, the

special sort of

analysis;

when judging

termed

'objective Russell's positive account of the judging relation very sketchy, and it is easily shown that as it stands
.

is
it

the logical conditions for an account of judgment, given in 4. In particular, it does not show how it is that we should be able to understand the statement abbreviated as "James judges that p" pro" vided that we understand both the expression James judges (that) ..." and the sentence (whose oratio obliqua form is) abbreviated to "p". If we use the

does not

fulfil

symbolism "B

(j, a, /?,

&)"

(cf.

p. 47),

where "12"

schematically represents a relation- word, then we can give an analysis of "James judges that p" for all cases

which "p" is interpreted as (the oratio obliqua form of) a statement "oR6"; but this does not give us a clue to the analysis in cases where "p" must bear a different sort of interpretation; Russell's theory would here rein

quire different relations of judging (differing as to the number and the logical types of the terms between which they hold) for every different logical form of

sentences expressing judgments. This objection could be overcome if we were given some recursive procedure whereby the judging relations for more complex cases could be defined in terms of those for simpler cases; but in Russell there is no hint even of the need for such

a procedure,

still less

of

its

actual form.

50

MENTAL ACTS

Again, Russell holds that if James judges that a is larger than 6, then the relation larger than is one of the things between which the judging relation obtains. But
this idea, of a relation's being itself

that are related,

like "larger

is

by no means
incomplete;

clear.
it

than"

is

one among things A relative term carries with it, so to

is larger say, two blanks that need filling up (" The blanks need not, of course, fce filled than "). with singular terms: "some cat is larger than any rat"

the requirement just as well as "Jemima is than Joey". Or again, the blanks may be filled larger up with variables, explicitly or implicitly; "larger than is converse to smaller than" is exponible as "for any x and any y, x is larger than y if and only if y is smaller than #". But how are the blanks to be filled up in "James judges that a is larger than 6"? Russell would maintain that, logically speaking, "a" and "6" do not fill up what is lacking to the sense of "larger than" a, 6, and the relation larger than would on his view enter into the judging relation as separate terms, not as forming a complex on their own account. He holds this with some reason; he desires to give an account of false judgments without bringing in false 'objectives'
satisfies

as complexes composed of the entities judged about; and if a, 6, and the relation larger than entered into the judging relation otherwise than as separate terms, it
looks as though they might form such an objectionable complex. This feature of his theory was what I meant
to suggest
tion

by separating the
9

variables with

commas

in

my notation "B (j

a, JR, &)".

But

failing

some

specifica-

how the double incompleteness of "larger than" is to be made good; failing some such development of "B (j, a, larger than &)" as I gave just now for "larger
9

than

is

converse to smaller than"; Russell's theory re-

INADEQUACIES OF RUSSELL'S THEORY

51

mains very obscure and unsatisfactory. Russell seems not to have noticed the difficulty.
connected difficulty is this: since "larger than" is logically not to be read in "B (j a, larger than, &)" as having its first blank filled up with "a" and its second with "fe", how are we to distinguish this from "B (j ft,
9 9

larger than a)"?

9

i.e.

how

are

we

to distinguish between

is larger than 6" and "James than a"? Russell did notice this larger difficulty, but his attempts to deal with it are quite unavailing. To say that "the relation of judging places the constituents in a different order" (RUSSELL (1), p. 198) is no explanation, but just a way of stating the difference that has to be explained. In RUSSELL (2) on

"James judges that a

is

judges that b

the other hand

we

read:

"the relation must not be abstractly before the mind, but must be before it as proceeding from a to b rather than from 6 to a ... We may distinguish two 'senses' of a relation, according as it goes from a to b or from
b to a.

must have a

judgment Thus the judgment that two terms have a certain relation R is a relation of the mind to the two terms and the relation R with the appro'sense'
.
. .

Then the

relation as

it

enters into the

priate sense" (pp. 183-4).

Yet if the relation R is before the mind, not as relating a and 6, but only as a term of a judging relation that holds between the mind, a, the relation R, and 6, how
can there be any talk of the relation J?'s 'proceeding' from a to b rather than from b to a? How can a relation that occurs not as relating things, but as one of the things related by another relation, occur with one or other 'sense'? This difficulty looks even worse if we
consider

how

in fact the relation

R does enter into the

52

MENTAL ACTS

judgment, namely, as an object of thought. For not only is the concept of a pair of converse relations a single and indivisible mental capacity (eadem est
scientia oppositorum); the exercise of that concept in judgment also brings in the two relations equally and

tion to 6

simultaneously, for to judge that a bears the one relais the very same act as judging bears that^fc the converse relation to a. Thus both relations of a
pair of converses, or, in Russell's language, the relation both its 'senses', must enter into the act of judgment equally and simultaneously; his solution then
collapses,

R in

and indeed the problem looks

intractable.

14

A REVISION OF RUSSELL'S

THEORY
admissible solution will jTjL sketched. Its interpretation must for the

A FORMALLY
left aside; it is

now be moment

'formally admissible' in the sense of the logical objections to Russell's theory. avoiding We begin by introducing an operator "( )"; if a relational expression is written between the brackets, we shall get a new relational expression of the same logical type as the original one. If "R" is dyadic, so is

be

"(R)"; if "R" is triadic, so is "(R)"; and so on. In this respect the operator "( )" resembles negation. Given any relational expression, we can form a negation of

same
is

be a relational expression of the is an important difference polyadicity. between the negation-operator and "( )". Negation
it,

which

will

But there

what

is

called extensional; that

is

to say,

if

two

A REVISION OF RUSSELL'S THEORY

relative terms

58

are truly applicable to precisely the same objects, then their negations also are applicable to precisely the same objects. But our analysis in terms of "( )" would break down if we allowed the inference that when the relative terms "R"

"R" and "S"

and "S" are applicable

also the relative terms

same objects, then and "(S)" are applic"(R)" able to the same objects. Many logicians would object
to just the

introducing a non-extensional operator; but a plausible analysis of descriptions of judgments has not yet been given within a purely extensional logic. We next introduce the term "idea", in an arbitrary and artificial sense; to distinguish this sense I shall write the word with a capital letter. In contemporary psychology "idea" is not in general use as a technical term, so there is no confusion with a competing technical sense to be feared; but it is well to guard against possible confusion with my own non-technical use of the word (as roughly equivalent to "view", "doctrine", "theory", etc.). By "an Idea" I shall mean the exercise of a concept in judgment; and I shall refer to particular Ideas in much the same way as I refer to the correto

sponding concepts; the exercise of James's concept man in a judgment will be an Idea of man. I deliberately leave this explanation vague and sketchy, because I do not want to commit my self to too many assumptions. I shall speak hereafter of Ideas such as those of evert/ knife and some spoon; this presupposes an understanding of the corresponding expressions "concept of every knife", "concept of some spoon". I wish to say that the concept every knife is exercised in any judgment to the
is . ., and is in fact the ability this form; similarly, the concept to frame judgments of some spoon is exercised in any judgment to the effect

effect that

every knife

54

MENTAL ACTS
is
. .

that some spoon

.,

and

is

in fact the ability to

frame judgments of this form.

respective concepts in

And

exercises of these

be Ideas of every judgments and of some spoon. knife My theory of judgment can now be stated. Suppose James judges that every knife is sharper than every spoon. This judgment comprises Ideas of every knife and of every spoon; let us call these two Ideas a and ft
will

respectively. theory is that James's act of judgment consists of his Idea a, of every knife, standing in

My

the relation ( sharper than) to his Idea ft, of every spoon. This analysis gets over the difficulties about 'sense'; the difference between James's judging that

every knife is sharper than every spoon and his judging that every spoon is sharper than every knife would be the difference between the Idea a's standing in the relation ( sharper than) to the Idea ft and /?'s standing in that relation to a. We have also overcome another that we did not know difficulty of Russell's theory how to extend it to (say) three-termed relations. A judgment to the effect that a three-termed relation R holds will consist in a man's having Ideas a, ft y such that we have (R)(a, ft y); for "(R)", like "R", will stand for a triadic relation, one therefore which can
relate a, ft

and

y.

In the dyadic case just discussed we pose that

may
ft

well sup-

James's Idea of every spoon, and a stands in the relation

is

"a

James's Idea of every knife, and

ft"

is

(sharper than) to

and: "a is James's Idea of every knife, and ft is James's Idea of every spoon, and ft stands in the relation
(blunter than) to a"

A REVISION OF RUSSELL'S THEORY

55

are equivalent, mutually inferable, statements. This supposition seems to be demanded in order that we may say (as we surely must) that "James judges that every knife is sharper than every spoon" and "James

judges that every spoon is blunter than every knife" are merely alternative descriptions of one and the same judgment. But can we consistently hold that in general (using the Principia Mathematica notation for
relation

"converse of") Cnv'(R) and (Cnv'R) are the same e.g. that the converse of (sharper than) is the relation (blunter than)? We certainly need to distinguish between "James judges that some knife is sharper than any spoon" and "James judges that any spoon is blunter than some knife"; one might indeed hold that in judging the first thing James eo ipso judges the second thing, but the converse is not true. Thus we must also say of the two following statements made in terms of Ideas and -relations:

"a is James's Idea of some knife, and /? is James's Idea of any spoon, and a stands in the relation
(sharper than) to
",
/?

"a

is

James's Idea of some knife, and

ft

is

James's

Idea of any spoon, and (blunter than) to a",

that the second
is

stands in the relation

implied by the first but not conversely. Accordingly, we must deny that the relations and (blunter than) are converse rela( sharper than) is not and in general when the relation tions; there will be a distinction between the symmetrical

relations

Cnv '(R) and (Cnv'R).

this time,
]8

A further example will show the power of my theory.

Let a be,
knife]

and

James's Idea of [some Idea of some James's Idea of [some idea of any spoon];

56

MENTAL ACTS
)

and suppose that a stands

than)
to
ft.

On my theory,

in the relation (( sharper these Ideas in this relation

judgment that some Idea of some stands in the relation (sharper than) to some knife Idea of any spoon. Now since the relative terms formed with the operator "( )" were introduced to signify relations holding between Ideas in a single person's mind (viz. within a single act of judgment), tve may
constitute James's
well take this to be part of their sense; so "some Idea of some knife stands in the relation (sharper than) to

some Idea of any spoon"

will

be tantamount to saying

"somebody's Idea of some knife stands in the relation to his own Idea of any spoon". But this ( sharper than) last is our analysis of "somebody judges that some knife is sharper than any spoon"; so a's standing in the relation (( sharper than) ) to ft will constitute a judgment on James's part that somebody judges that some knife is sharper than any spoon. This shows how our method of analysis can be extended to reports of judgments to the effect that some judgment has occurred reports involving a double oratio obliqua ."). ("James judges that somebody judges that to account for such double oratio obliqua Ability constructions is in any case a necessity for an adequate theory of judgments. For as soon as we understand the verb "to judge", the oratio obliqua construction, and the sense of the particular bit of oratio obliqua abbreviated to "p", we can understand "James judges that someone judges that p", "James judges that someone judges that James judges that p", and so on indefinitely (until sentences become too long for us to take in); and this surely involves that after the first stage
.

we must have no new

duced into the game.

logical piece (so to speak) intro-

My

iterated use of the operator

A REVISION OF RUSSELL'S THEORY

57

"( )" certainly satisfies this condition. (To make my account of judgments about judgments tenable, it would also be necessary to assume, as I tacitly did, that given the sense of the term "Idea" and the virtual
oratio obliqua construction, as we knife" in "Idea of some knife",

may call it,

we have
all

of

"some

that

we

need in order to understand the double virtual oratio obliqua in "Idea of [some Idea of some knife]". For if this were not so, we should get something 'logically new' introduced; which, as I have said, would disqualify our analysis.) It may be objected that the following two statements certainly cannot be treated as equivalent, even though, according to our theory, the sentence from which they are obtained by prefixing "James judges that ..." would be equivalent:
A. "James judges that somebody judges that some is sharper than any spoon"; B. "James judges that somebody's Idea some knife stands in the relation (sharper than) to that person's Idea any spoon".
knife

How could B be true unless,

also

not only were

A true, but

James had accepted Geach's

analysis of judg-

ments? B, then, cannot be equivalent to A; so our

analysis of double oratio obliqua in reports of judgments seem to break down.

But suppose instead of A I had written: A'. "James judges that somebody judges that some knife stands in the relation sharper than to any spoon".

A and A' equivalent? Does the truth of A' require, over and above the truth of A, that James shall have done some philosophizing about relations? Not at all; my
Are
writing

down A'

instead of

A would make no difference

58

MENTAL ACTS
it

to the statement I made,

would merely display

my

familiarity with a bit of philosophical jargon. The relation between and is, I think, similar; unlike A', is not just a pedantic circumthough B,

own

James expresses his judgment in the sentence "somebody judges that some knife is sharper than any spoon", eo ipso James is in fact relating the phrases "some knife" and "any spoon" to the realm of
locution for A. If

what / call Ideas, and taking "sharper than" to stand not for the relation it ordinarily stands for, but for the relation that .Z express by "( sharper than)"; he understands the words "some knife", "any spoon", "sharper than", in a special oratio obliqua sense in their ungerade Sinn, as Frege would say. But this understanding is a matter of 'knowing how' rather than 'knowing that'; it comes out in James's practical mastery of the oratio obliqua construction, not in his ability (if he has any) to give a philosophical account of that construction. When I, on the other hand, am concerned to bring out the logical force of James's judgment, it is open to me to use logical devices and terms of my own invention, regardless of whether James would understand them. In RUSSELL (3) there is an argument against the view that judgments comprise ideas; the vague word "idea" is here used by Russell in very much the same way as I use "Idea", and his arguments, if correct, would go against my present theory, so I shall take the liberty of writing "Idea" for his "idea" throughout. Russell holds that the view that a judgment is built up out of Ideas probably depends on a failure to reach a correct theory of descriptions. There is a popular use of "idea" in which "my idea of Julius Caesar" means much the same as "what I know about Julius Caesar",

A REVISION OF RUSSELL'S THEORY

"what

59

I take to hold true of Julius Caesar". In this Russell would allow some truth to the statement sense, that in order to make a judgment about Julius Caesar
I

must

first

have an idea of Julius Caesar;

for Russell

in fact actually contending, that 'in order to discover what is actually in mind when I judge
is

admits, he

my

about Julius Caesar, we must substitute for the proper

name a, description made up of some of the things that I know about him'. This certainly shows that Julius
Caesar himself does not enter bodily into my judgment as a constituent; but it does not show that in his stead there does enter some Idea. The description that expresses how I think of Julius Caesar could be so phrased as to contain only words for entities (universal and particular) that are objects of my awareness; and
these entities, Russell thinks, which are 'the actual objects concerning which we judge', do enter as constituents into the judgment-complex; no need to postulate any Ideas that shall go proxy for them.

Secondly, Russell objects that if in an act of judgment there are included Ideas of what we are judging about, then 'Ideas become a veil between us and outside things

the things only to the Ideas of these things'. Surely it is preferable to hold that there is a mental constituent which stands in the two-termed relation of awareness to the several things that are judged about, and also stands in the irreducibly many- termed relation of judgment to all of them together? Is not the appeal to Ideas just due to a lingering prejudice against relations, so that an Idea 'in' the mind seems preferable to a relation 'between'
the

we never really, in knowledge, attain to we are supposed to be knowing about, but

mind and other

talk

things?

Russell's

of

'complexes'

and

'constituents'

60

MENTAL ACTS
y

embodies, I think, a radical confusion. If a word for x is part of a description of y it does not follow that x is u a constituent of y\ "Denmark" is a part of the capital of Denmark", but Denmark is not a part or constituent of the capital of Denmark. So there is no reason why a thing that is judged about should be a constituent of my act of judgment, merely because the description of the act mentions the thing. Sense-data and universals may appear better candidates for the position of elements in my act of judgment than a man who died two thousand years ago; but there is no better reason in logic why they should be elements or constituents of it. The argument that if our judgment contains Ideas of things, not the things themselves, then what we judge about is the Ideas, not the things, is a sophism frequently used by idealists; Russell's use of it (of course to prove an anti-idealist conclusion) may well be a

hangover from

his idealist beginnings.

One might

as

well argue that all disputes are about words, because they are conducted in words. If Ideas are part of our

judging about things, it does not follow that Ideas are what we are really judging about; no more than words which are part of our talk are what we are really talking about. I do not think, then, that there is anything in

own recourse to Russell's objection to Ideas. And Ideas in the analysis of judgment certainly involves no

my

reluctance to bring relations into the account.

15

JUDGMENTS ABOUT SENSIBLE

PARTICULARS
of the judgments considered in JL the* last section was wholly universal or general; a verbal expression of such a judgment would contain no demonstratives and no proper names. This stands in marked contrast with Russell's examples: judgments that Desdemona loves Cassio, that Charles I died in his bed (or: on the scaffold), that the Sun is shining, that Bismarck was an astute diplomatist, that Scott was the author of Waverley; and again, judgments that this is yellow, that this is before that, that I am acquainted with A (where "this", "that", and "A" are given unique reference by someone's direct experience). (RUSSELL (1), (2), (3).) In point of fact, Russell only attempts to analyse the second sort of judgment, which relates, not merely to singular objects, but to objects of direct acquaintance. His treatment of the other sort of judgment, which is expressed by using ordinary proper names like "Bismarck" and "the

CONTENT

Sun", is entirely unsatisfactory. Sometimes we find such objects as human beings treated as constituents
of the judgments in whose expression their proper names occur a procedure quite inconsistent with Russell's general view. No doubt this is done in order to get simple, familiar, illustrations; Russell makes believe that the proper names stand for judgment-constituents, i.e. objects of acquaintance that are being judged about, although on his own showing this is

never true. Sometimes, on the other hand, in discussing

61

62

MENTAL ACTS

examples Russell is expressly concerned to bring out their concealed logical form: when I say "Bismarck is an astute diplomatist", my use of the word "Bismarck" takes the place of some description,

such

and I am using in my judgment what this description would mean. What I really judge is (say) that the first Chancellor of the German Empire was an astute diplomatist; and by Russell's theory of descriptions the italicized words would mean something like: "there was somebody who was a Chancellor of the German Empire, before whom nobody was a Chancellor of the German Empire, and who was an astute diplomatist". But now the trail is lost; for Russell makes no attempt (in RUSSELL (1), (2), (3) ) to deal with judgments whose verbal expression would even approach the complexity
of the sentence just quoted. Russell's theory of descriptions enables us to analyse a sentence containing

a definite description, only so long as that description does not occur in oratio obliqua. For although, if
pass from "James judges that shot himself" to "James judges he who shot Lincoln that somebody shot Lincoln and shot himself and that
Russell
is

right,

we may

no two

(different) people shot Lincoln", this latter

is

a finally satisfactory analysis; for it still presents the appearance of asserting a two- termed judging relation the contents of the "that"-clause are not split up, so as to reveal the real polyadicity of the special judging relation here involved, and we have no indication how to
not,
principles,

statement

on Russell's

do

this.

Thus Russell only deals with singular judgments, and only with those singular judgments which relate to
objects of direct acquaintance. I think this is a further reason for saying that an attempt to explain judgments

REFERENCE TO PARTICULARS
on
his lines is likely to fail.

63

There

is

much more hope

that an account designedly adequate for general judgments will turn out adaptable to singular judgments, than there is of the reverse adaptation, to which Russell would be committed; and we find no hint (in

RUSSELL
to
fit

(1), (2), (3)

how

his

theory could be adapted

general judgments. reject as spurious his examples of judgments expressible with proper names, Russell actually
If

this is before that.

analyses only such judgments as that this is red or that But are the expressions "this is red", "this is before that", genuine expressions of any actual

not a genuine expression of judglike "that flash was before this bang"? Russell might reply that this last statement is analysable as: "that is a flash, and this is a bang, and
Is

human judgment?

ment something more

that is before this". I propose,

however, to ignore this analysis and attack the problem another way. How, I shall ask, is a judgment expressible as "that flash was before this bang", which refers to a particular flash and a particular bang, to be distinguished from a judgment, with the same verbal expression, which refers to another flash and another bang? What constitutes this
reference to definite particulars? Or again, difference between "there are white cats",

what is the "some cats are white", on the one hand, and on the other hand "these cats are white"? How do the judgments they
express differ? What constitutes the reference to a particular set of cats? In all such cases, I should maintain, there is no difference to be found on the side of the judgment itself. What we may call the intelligible content of the judgment is the same in all judgments expressible as "that
flash

was before

this

bang", regardless of which flash

64

MENTAL ACTS

and bang are in question. (In terms of the analysis in 14, there would always in such cases be an Idea, flash standing in the relation (before) to an Idea bang. But what I now say does not stand or fall with this
analysis.) may see how the special reference to that flash and this bang comes in, if we ask ourselves the

We

question:

How could the utterance ''flash before bang"

be taken to refer to a particular flash and bang? The answer is obvious; the utterance can be, and probably will be, so understood in a sensory context in which the hearer notices a flash and a bang. Similarly, the utterance "some cats, white" could be taken to refer
to particular cats if its hearer was looking attentively in the right direction. The content of the judgment is

always intelligible and conceptual acquaintance with a particular sensible thing is no part of the judgment itself but an act of judgment performed in a particular sensory context may thereby be referred to particular sensible things. It
is

clear, indeed, that the act

of

judgment must bear a

closer relation

than mere

simultaneity to the context of sense-perception that gives it its special reference to these particular sensible
things; I am not prepared to characterize this special relation it must bear to its context. (So far as I can

quite useless to say the relevant senseperceptions must be being attended to; either this does
see,
it
is

not give a sufficient condition, or else "attended to" is a mere word for the very relation of judgment "James to sense-perception that requires analysis to such-and-such sense-perceptions" judges, attending meaning no more than that his judgment refers to what he thus perceives.) But I do not think this throws any doubt upon what I have said; although more remains to be said.

REFERENCE TO PARTICULARS
The problem

65

I have just been discussing how we about sensible particulars was much agitated judge in the Middle Ages; and in my solution of it I believe

am

relation of the 'intellectual' act of

following Aquinas. Aquinas's expression for the judgment to the

,

context of sense-perception that gives it a particular was "conversio ad phantasmata" "turning round towards the sense-appearances". This metaphorical term is obviously a mere label, with negligible explanatory value; but it does not pretend to be more than a label. Aquinas has, in my opinion, at least rightly located the problem; the problem is not how we advance from judgments like this is before that to more general judgments, but contrariwise how a judgment inherently general can be tied down to referring to particular things (la q. 86
reference
art. 1).

Quite similar considerations apply to judgments involving tense. The difference between judgments to the effect that a hydrogen bomb will be exploded and that a hydrogen bomb has been exploded is an intellia specifically different gible or conceptual difference
exercise of concepts is involved. But there is no conceptual difference between judgments formed in dif-

ferent years to the effect that a hydrogen

bomb has been exploded, although such a judgment formed in 1940 would have been false and one formed in 1956 would be true. The difference between the two judgments

by their standing in relation to contexts (by there being a different different sensory conversio ad phantasmata AQUINAS, la q. 85 art. 5 ad 2 um); but it is a great mistake to try to bring in these
is

constituted

contexts into any setting forth of that which is judged cf. the articles (a mistake very frequently committed

66

MENTAL ACTS
e.g.

on tensed statements that have appeared,

1 Analysis of recent years ).

in

16

JUDGMENTS INVOLVING
IDENTIFICATIONS
TUDGMENTS
-

expressed by using proper names cJ present a much more difficult problem than is raised by demonstratives and tenses. What I have to say is meant to rule out solutions that seem to me clearly wrong, rather than to offer a positive solution. There are two statements about proper names that have become widely accepted slogans: first, that

proper names are disguised (definite) descriptions; secondly, that proper names have no connotation at all. At any rate these slogans cannot both be true; my own view is that they are both entirely false. It is anyhow important to examine them, because on the truth of the matter there will depend what we ought
to say about judgments expressed

by means of proper

names.

The theory that proper names are disguised definite descriptions was first devised by Russell (cf. RUSSELL, (3) ) and has since been maintained in an even more intransigent form by Quine (in QUINE (1) and other
a psychological point of view it is I refer to a person by a proper palpably false; name, I need not either think of him explicitly in a form expressible by a definite description, or even be
works).

From

when

of these have been reprinted in the volume Philosophy and Analysis, ed. Margaret Macdonald (Blackwell, 1954).

Some

JUDGMENTS OF IDENTIFICATION

67

prepared to supply such a description on demand (not, that is, with any confidence that the description really is exclusive). Quine, however, has made a strenuous attempt to get over the difficulty. "We may insist", he says,
"that what are learned by ostension, or direct connames but solely predicates. This we may insist upon at the level strictly of logical begrammar, without prejudice to epistemology cause we grant the epistemologist any of the words which he traces to ostension; we merely parse them differently. Instead of treating the ostensively learned word as a name o/the shown object to begin with, we treat it to begin with as a predicate true exclusively of the shown object; then we construe the name, as
.

frontati&n, be never

" amounting to "(w)Fx" where "I* represents 1 Given any singular term that primitive predicate such, as
1

of ordinary language, say "Socrates" or "Cerberus" the proper choice of "F" for translation of the
.
. .

term into "(?#)1<V need

in practice never detain us";

"is-

Quine

is

prepared to allow such 'predicates' as

Socrates" and "is-Cerberus" as interpretations of the predicate-letter "F", "since any less lame version", he says, "would, if admissible as a translation at all, differ at most in expository value and not in meaning".

(QUINE

(1),

pp. 218-19.)

suppose somebody chooses to call his dog "Cerberus", and introduces him under that name:
"this
is

Now

Cerberus".

follower of Quine 'insists'

on

parsing "Cerberus" as a predicate. Later on, he says

1

"(ix)Fx"

is

to be read "the thing that

is

F", "JF" representing a

logical predicate.

68

MENTAL ACTS

of another dog: "here is another Cerberus". On Quine's view, this would be merely a factual mistake, which the owner could suitably correct by saying "No, there's

only one Cerberus". Would it not then be in place to ask the owner how he knows there is only one Cerberus? and is not this plainly a silly question, in fact? As the owner intended "Cerberus" to be used, such expressions as "here is another Cerberus" or "there are several Cerberuses" are not just false statements they are excluded altogether as moves in the languagegame played with "Cerberus". The owner did not, then, mean "Cerberus" as a predicate; and it is useless to 'insist that it be' one this 'insistence' is insistence on playing a different game with the word, and is irrelevant to the owner's use of it. Of course there are grammatically proper names that can be construed as disguised definite descriptions: e.g. "Homer" as "the man who composed both the Iliad and the

Odyssey". But these are altogether exceptional. The theory that proper names have no connotation historically forms part of a highly muddled logical doctrine about 'connotation and denotation'. A clearer way of stating it is to be found in Locke: that there is nothing essential to an individual. To a thing to which a general term is rightly applied, those attributes may be said to be essential which logically follow from the general term's being true of it; but no attributes logically follow from a thing's being given a proper name. (Essay, III. vi. 4.) Notice that I spoke of 'rightly' applying a general term, but not of 'rightly' giving a thing a proper name; for, on the theory I am considering,

or
to

give proper names by fiat and there is no right wrong about the procedure. A thing is introduced
its

we

me by

proper name, and thereafter I can refer to

JUDGMENTS OF IDENTIFICATION

69

the same thing by using that name; but the name tells one nothing about the attributes of the thing. The sense of a proper name certainly does not involve anything about the peculiarities of the individual so named, which distinguish it from other individuals of the kind; a baby, a youth, an adult, and an old man may be unrecognizably different, although the same name is borne throughout life. But it is meaningless to say without qualification that the baby, the youth, the adult, and the old man are 'the same', or 'the same

and that this is what justifies us in calling them by the same name; nor yet is it a matter for our free decision whether or not they are to be deemed 'the same'. "The same" is a fragmentary expression, and has no significance unless we say or mean "the same X", where "X" represents a general term (what Frege
thing',

a Begriffswort or Begriffsausdruck). What is implied by our use of the same name throughout a period of years is that the baby, the youth, the adult, and the old man are one and the same man. In general, if an individual is presented to me by a proper name, I cannot learn the use of the proper name without being able to apply some criterion of identity; and since the identity of a thing always consists in its being the same X, e.g. the same man, and there is no such thing as being just 'the same', my application of the proper
calls

justified only if (e.g.) its meaning includes its being applicable to a man and I keep on applying it to
is

name

one and the same man. On this account of proper names, there can be a right and wrong about the use of proper names. Suppose that an astronomer observes two different fixed
stars at

two

different times, erroneously believes that

one and the same planet has been observed on both

70
occasions,

MENTAL ACTS

and christens this supposed planet "Vulcan". His use of the word would be justified if he had succeeded in identifying an astronomical object as one
and
same planet on different occasions; and since in of fact his identification was a mistake, his use of point the term "Vulcan" was also a mistake; he has given
the
is

the term no reference, no use, whatsoever, and there nothing he can do except to drop it. A similar situa-

tion arose in a detective story I once read (The Wraith, by Philip Macdonald). The man suspected of murder

was one A. G. Host; but the great detective discovered that two men had on different occasions acted the part of Host; so that the identification of one and the same man as the person who had done various things, which was expressed by using the proper name "A. G. Host", was misconceived. After this discovery, people who knew the facts could clearly no longer use "A. G. Host" as a proper name.
It has often been argued that it cannot be part of the meaning of a proper name that its bearer should be a man, because we cannot tell this just by hearing the name, and because there is nothing to stop us from giving the same name to a dog or a mountain. You might as well argue that it cannot be part of the meaning of "beetle" that what it is applied to must be an

because we cannot learn this meaning just from the sound of the word, and because "beetle" is also used for a sort of mallet. In a given context, the sense of "beetle" does include: being an insect, and the sense of "Churchill" does include: being a man. Again, my view of proper names does not commit me to an Aristotelian doctrine about 'essences'. I hold, indeed, that the sense of "Churchill" does include: being one and the same man (homo) and the term
insect,
9

JUDGMENTS OF IDENTIFICATION
"man" ("homo")

71

is traditionally assigned to the cateof substance; but similarly I hold that the terms gory "Venus" and "Thames" have as part of their respective senses: being one and the same planet, and being one and the same river. Now a defender of substance

would

"river" as a

not, I think, wish to regard either "planet" or word for a kind of substance; a planet

would Le an aggregate of substances of many kinds, and the identity of a river is not identity of the substance (the water) that occupies the river-bed. My thesis is that for any proper name there is some interpretation of "X" such that we can truly say "the continued application of this proper name requires, as part of the sense of the name, that it be always applied to the same X". But the general terms represented by "X" need not stand for a kind of substance, even if we allow a distinct category of terms that do so. My thesis relates only to 'nominal essence', and has nothing to do

with

'real essences'.

from this account of proper names that from a statement in which a proper name is used "Smith committed seven burglaries, then Smith committed a murder, then Smith was hanged" we can "a man distil out a statement in purely general terms committed seven burglaries; then the same man committed a murder, and was then hanged". I should also hold that this is all the intelligible content that can be
It follows

got out of the statement, since I hold that the sense of a proper name never includes anything about the individual peculiarities of its bearer. (Sometimes I may expressly advert in my mind to other people's using the name "Smith" for the man who had all these adventures: "a man known to others as "Smith" committed seven burglaries (etc.)". But this is not F

72

MENTAL ACTS

necessary no more necessary than that I should think of a cow always as an animal that other people call "a cow".) Thus on my view there is no intrinsic difference between a judgment that Smith had a certain series of adventures, and a judgment that some one man successively had the adventures (or perhaps: that some one of the men called "Smith" did so); there is nothing within the act of judgment itself to point to the right

Smith. lf God had looked into our minds, he would not have been able to see there whom we were speaking of (WITTGENSTEIN, Part II, p. 217; my italics). So far, what I have said about judgments expressed by means of proper names is like what I said about those expressed with demonstrative pronouns; and in this case too the singular judgment would have the same content as one statable in purely general terms. (As before, judgments' having the same content would be analysable, on the view I developed in the last Section but one, as their consisting of similar sets of Ideas, related by the same -relations.) But we cannot in our present case appeal to conversio ad phantasmata as what makes a judgment relate to the right Smith. We could explain the particular reference of judgments expressed with demonstrative pronouns by relating them to an immediate sensory context; but of course we can talk about Smith and make judgments about him when he is not present to our senses. We might try relating the judgment to mental images of one sort or another (which would likewise be phantasmata for Aquinas): e.g. to memory-images of Smith's appearance or of his name. But my visual image of how Smith looked may be very inadequate, and may be known by me to be so; and very many people have the name "Smith". So in general the mind will not contain
c

JUDGMENTS OF IDENTIFICATION

73

anything uniquely relating to Smith an unmistakable photograph or relic of him, so to say. So far as I can see, a description of the judgment as relating to
the right Smith

ment

is possible only if it places the judgin a real-life context described as including that

Smith. The problem how you call Smith, the right Smith, to mind is like the problem how you call him (WITTGENSTEIN, Part I, 691). Although lots of people are called "Smith", the summons "Smith!" may be quite effective to fetch the Smith I want if he is the only man of that name within earshot; and similarly, a judgment that might in principle relate to many men may yet in a particular real-life context be relatable
to just one.

The problem how I can continue to refer to a man as the same man over a period of time is a matter of my own mental acts; my identification may be mistaken, but I should none the less be thinking as it were of the same man; and if the identification is mistaken, then we can give no account of it by describing a con-

text that includes the man I was all along judging about, since there will have been no such man. Such
identification
I hold,
is

essentially appealed to in

my account

of proper names; in a series of statements about Smith,

"Smith" could be replaced the first time by "a man", and in later occurrences by "the same man" or "the man" or "he". Now from a logical point of view the string of statements thus obtained, with the name "Smith" eliminated, are just one long existentially quantified statement; "for some #, x was a man, and x committed seven burglaries, and x committed a murder, and x was hanged". But this would clearly not do as an account of how judgments are actually formed especially if the one long statement would

74

MENTAL ACTS

have to be very long, as may easily happen when I keep on learning things about somebody. Let us consider a clear case of the problem, uncomplicated by proper names. I say first of all 'There's a man on the quarry-edge"; five minutes later I say "Now he's gone he must have fallen in! "The identification expressed by my use of "he" cannot be ex-

by bringing the man himself into th& story; perhaps 1 was wrong and there never was a man on the
plained

quarry-edge.
to

be sensibly "the man I am now looking at, or pointing to" but "the man I meant a little while ago"; what is required in my referring to him as "he" is not a present sense-perception but a recent thought. (To use Aquinas's language: the pointing, demonstration effected by the demonstrative word "he" is not ad sensum but ad intellectum.) Although my later judgment need not be held to incorporate my earlier one (which, I argued, it would be very unplausible to hold if there were a long series of judgments all relating as it were to the same person), nevertheless there must be an intelligible difference of
content between my actual pair of judgments five minutes apart, and a pair of successive judgments that a man is on the brink of the quarry and that a man (not necessarily the same nor necessarily a different one) has fallen in. There must be some special relation of the two judgments in the former case; but I am at present unable to say more than this.

My use of "he" in this case do with conversio ad phantasmata; "he" its meaning from anything that could indicated. Crudely: "he" does not mean

has nothing does not get

17

ANALOGY THEORIES OF
I very* different
remarked
"
in
4,

PSYCHOLOGICAL CONCEPTS NOW TURN to a theory of judgment prima facie

from the one sketched
in
14.

As

there are

many
"

languages in which, as

in English, the only difference

between the statements

said that p" and James judged that p" is that have different main verbs. The theory I am now they

James

going to discuss takes this as its clue to the nature of judgment; the concept judging is viewed as an analogical extension of the concept saying. I must begin, then, with some general remarks about the 'analogy' theory of psychological concepts. This theory is not to be supported by appealing to the etymology of words
for

obscure

Such etymologies are often very that of "credere", "to believe", in Latin) (e.g. and often have no apparent relevance when they are clear (what has standing under to do with understandmental
acts.

ing?).

Again we must distinguish between what may be called casual and systematic analogies. In casual
analogies, a particular epithet may be suggestive or even 'inevitable' in its metaphorical use, but this gives us no indication for the transferred use of another

epithet of the same family; an understanding of "yellow that's the only word for him!" does not give us a clue as to what would be meant if another colour-

word were substituted

for "yellow". The sort of that is important is that in which a whole analogy system of description is transferred to an analogical use; the oratio obliqua construction, for example, whose

75

76

MENTAL ACTS

primary use

is to report actual speech, is transferred to the content of judgments. describing What is the meaning of calling a description of a

mental act "analogical" or "metaphorical", except by contrast with a literal description? and if the literal description is available, how can the analogy be of any
importance for the concept of the mental act? This objection to the expression "analogy" or "metaphor" is easily got over. What makes the description to be analogical is not that there is a literal description of the mental act, to replace the metaphor, but that there is another application of the description, and of the whole system of description to which it belongs; an application that may be called literal, in contrast to the transferred application to mental acts. We may agree that if a metaphorical description of a mental act is replaceable by a non-metaphorical description of that very act, then the metaphor can have little importance for the concept of that sort of act. Someone may, however, argue thus: "Even
if

in our

actual language there is no way of replacing a metaphor used to describe a mental act (except perhaps by another metaphor), this can only be an accident of language. For, in order to see that the metaphorical

system of description is appropriate to the facts it is used to describe, we must already be able to discern those characteristics of mental acts which make the
description appropriate. Now the characteristics (for example) of a judgment that is described by a transferred use of the oratio obliqua construction will be characteristics not to be found in the pieces of language that this construction is primarily used to report; so a

concept of the mental characteristics must be an entirely new concept, which it is merely misleading to

ANALOGICAL CONCEPTS
call

77

an

in the

'analogical' extension of the concept exercised primary use of oratio obliqua. And if we can dis-

cern,

and form a concept of, these mental characteron their own account, then we could if we chose invent a special word, which would be used only to
istics

express our exercise of the new psychological concept. So the analogy of mental judgments to actually uttered statements cannot be an indispensable part of our concept judgment; and in general psychological

concepts cannot be irreducibly analogous in character."

concept consisted in extracting a feature from our experience and becoming repetitive able to recognize its recurrences, then indeed psychological concepts extracted from 'inner sense' woujd be

If forming a

no way dependent on the analogies afforded by other concepts extracted from sense-experience; only when we already possessed the psychological concepts could we go on to observe the appropriateness of such analogies. I have argued, however, that no concept whatsoever is to be identified with a recognitional
in

capacity. In all cases it is a matter of fitting my concept to my experience of exercising the appropriate concept not of picking out the feature I am inter-

ested in from

among

others simultaneously given in

exercise of a particular set of conexperience. cepts in a particular way (e.g. the exercise involved in

Now

describing something by means of a particular set of words in the oratio obliqua construction) may be appropriate both as applied to A and as applied to B,

although

A is not

'like'

B; and

we may

perfectly well
(in the

know

its

appropriateness in
is

both cases

'know-

to say, we may know when to ing-how' sense; that exercise our concepts thus, whether or not we can

78

MENTAL ACTS
so).

formulate to ourselves a rule as to when we ought to

do

our theory of concepts, there is no reason that can be offered in advance why psychological concepts, or some of them, should not be irreducibly analogous. It might very well be that we always get the knowhow of a particular set of concepts first of all in application to sensible things (e.g.

On

human

utterances),

and

then in psychological applications. Any alleged example of this would have to be investigated on its own account before we could decide about its genuineness. On the other hand, our theory destroys the old grounds for saying that psychological concepts must be irreducibly analogous. Somebody who held abstractionist views, and at the same time held that sense-experience offered all the material we have for performing abstraction upon a combination of views expressed in the old maxim "Nihil in intellectu nisi prius in sensu" would be obliged to maintain that we first extract concepts from sense-experience and then give them an

extended psychological use; his position would, in my opinion, both involve very unplausible accounts of certain psychological concepts, and be extremely hard to maintain on an abstractionist footing. Our rejection of abstractionism means that we are not antecedently committed either way as to the analogousness of a
given psychological concept. It is a mere matter of terminology whether we say that the psychological concept is a new concept derived by 'analogy' from a previous concept, or that we exercise an old concept in new, psychological applications 'analogous' to our former exercise of it. A concept is a capacity for certain exercises of the mind; and as regards capacities generally, if a man formerly

THE ORATIO RECTA ANALOGY

79

capable of certain performances in one domain comes to perform similarly in another domain, it is quite arbitrary whether we say that his former capacity now extends to a new domain of exercise or that he has acquired a new capacity closely related to his old one.
I shall say indifferently "analogical concepts" exercises of concepts".

and

"analogous

18

PSYCHOLOGICAL USES OF ORATIO RECTA

ONE

OF THE MOST convincing analogy-theories

is

of psychological concepts

the view that modes

of description primarily applied to actual bits of written and spoken language are transferable to the
role of describing the content of judgments that which is judged. As I have remarked, in many lan-

guages the same oratio obliqua construction is used to report a man's words and to report his thoughts. But I would rather not appeal to this fact in my discussion of the 'interior language' analogy. For one thing, the logic and grammar of oratio obliqua clauses involve a
lot of complications,
is

mostly trivial. What is worse, a plausible argument to show that it involves there a vicious circle to treat the use of oratio obliqua in
describing actual bits of language as being logically prior to psychological uses of the construction. Oratio
obliqua serves to give us, not the actual words that somebody said, but rather their gist or purport; but

does not this term "purport" smuggle in a psychological concept? Should we not say that oratio obliqua

80

MENTAL ACTS
tell

serves to

us what the speaker had in mind, or

his hearers'

what

he wanted to evoke in
said that p", so far

minds? If so, "James

had

it

in

from being logically prior to "James mind that p" and other such psychological

expressions, needs to be analysed in terms of them. I

think this objection could be answered, but it would take us too far round to do so directly; instead of this I shall turn from oratio obliqua to oratio recta, actual
quotation. The primary role of oratio recta is certainly not psychological; it serves to report what somebody actually said or wrote. But oratio recta can be used

metaphorically to report what somebody thought, 'said in his heart' (without, of course, implying that the thinker had the quoted words in his mind); such constructions are frequent in the Authorized Version of the Bible: "The fool hath said in his heart "There is no God" "; "They said in their heart "Let us destroy them together" ". 1 Clearly we could always describe

judgments by using
is

way; oratio logically superfluous. (Certain arguments obliqua that have been advanced against the replacement of oratio obliqua by oratio recta will be answered in 18.) And here we have a clear case of a linguistic device

oratio recta in this

whose psychological application

is

logically secondary

to its application to sensible things bits of actual written and spoken language. In order to understand how this device works, we shall have to discuss in some
detail the logic of oratio recta in its

primary or

literal

application.

In contemporary logic there

is

a pretty general

re-

cognition of the rule that when we write down an expression for the purpose of talking about that very
1

The Authorized Version of the Bible does not use quotation

I

marks;

have here adapted the texts to

my own

practice.

THE ORATIO RECTA ANALOGY

expression,

81

we miftt show what we are doing by enthat expression within quotes; if this rule is not closing strictly observed, actual fallacies are likely to arise.
Nothing
in spoken language corresponds to this use of quotation-marks; the beginnings of the sentences "man is an animal" and " "man" is a noun" are

phonetically quite alike. There is indeed a particular tone of Voice that is conventionally represented by using quotes, as in "He introduced me to his 'wife' "; but such quotes (which are sometimes called "scare-

quotes") are of course quite different from quotes used to show that we are talking about the expression they enclose. In this work I have tried to follow a strict rule of using single quotes as scare-quotes, and double quotes for when I am actually talking about the expressions quoted.
tion

I shall distinguish between a quotaand a quoted expression; the quoted expression is what stands between quotes, e.g. "man" in " "man"
is

" a noun", whereas the quotation, "man" " in this case, consists of the quoted expression and the
quotes.

The rationale of using quotes is, I think, widely misunderstood by contemporary logicians. Instead of saying that quotes signalize a special use of the quoted expression, many logicians use the terms "mention" and "use" as though they were mutually exclusive as though a 'mentioned' (i.e. quoted) expression were not at the same time being used. They would say that when an expression is quoted, not it but the corresponding
practical risk of confusing the two uses of find this precaution rather like the White Knight's armouring his horse's legs against possible sharkbites. But once bitten, twice shy I have actually been criticized in print for lack of 'rigour' because I used scare-quotes in a logical article without
1

There

is

very

little

quotes, so readers

may

warning

my readers

that I was doing so.

82

MENTAL ACTS

quotation is being used; and that although the quotation is physically built up out of the quoted expression and a pair of encircling quotes, the quoted expression is not logically or syntactically a part of the quotation, any more than the word "man" is logically a part of the word "emancipate", or the variable "#" of the word "six". Some logicians, it seems, would even " " deny that the quotation "man" is logically a part " "man is mortal" "; a quotation is of the quotation held to be, logically speaking, a single long word,

whose parts have no separate

p. 26.)

significance.

(QUINE

(2),

This last view

is

is

clearly wrong.

quoted

series of

always a scries of quoted expressions; expressions the quotes around a complex expression are to be read as applying to each syntactically distinct part of the expression; and if we fail so to read them, because we cannot discern the syntactical parts of the quoted expression, then we likewise fail to understand the quotation we do not know what it is a quotation of. The quotation " "man is mortal" " does not indeed " "man" " as a physical part, contain the quotation i.e. it does not contain "man" directly enclosed in a pair of quotes; but it does logically or syntactically contain " "man" " as a part, for the single pair of quotes in " "man is mortal" " is to be read as applied to the three quoted words "man", "is", and "mortal" severally. If we use an ampersand "&" to mean "followed by", then "man" & "is" & "mortal" is just the expression "man is mortal"; I should maintain that the " "man is mortal" " is rightly understood quotation " "man" & only if we read it as meaning the same as "is" & "mortal" ", i.e. read it as describing the quoted expression in terms of the expressions it contains and

THE OR AT10 RECTA ANALOGY

their order.
1

83

This Jesuit

expression must be taken

that quotation of a complex as describing the expression

in terms of its parts is vital for our purpose; if in said "let us destroy them together" " the "they

quotation were a single, logically unanalysable, word and not a complex description, there would be a breakdown, at the very outset, of the idea that oratio recta is a systfem of description applied in the first instance
to actual written or spoken language and secondarily to thought. In regard to the question whether a sign is to be

counted as occurring in a context in which it is quoted, Quine argues as follows. Let us suppose that "Cicero" " "Cicero" occurs, in a logically relevant way, within " has six letters", and "Tully" similarly within "Tully" has six letters". Then, Quine holds, the identity of Tully with Cicero would allow us to interchange these personal names, in the context of quotation marks as
in any other context (salva veritate, that is to say); " hence we could pass from the truth "Cicero" has six " letters" to "Tully" has six letters"; which is absurd.

(QUINE

above argument is contained words (which are Quine's own): "the identity of Tully with Cicero would allow us to interchange these personal names, in the context of quotation marks as in any other context". This is an application of a
in the

The

(2), p. 26.) crucial step of the

making this point is largely an accident of our might well have decided to symbolize quotation by underlining the syntactically simple parts of the quoted expression; " "man is in that case, "man is mortal" would correspond both to " and to" "man" & "is" & "mortal" ", and it would be as mortal"
1

The

difficulty in

notation.

We

easy to see that the quotation "man" is syntactically part of the quotation "man is mortal" as that the quoted expression "man" or man is part of the quoted expression "man is mortal" or man is
mortal.

84
logical law,

MENTAL ACTS

which we may fittingly call "Leibniz's that two names of the same thing are mutually Law", replaceable salva veritate. But why is Quine so sure that Leibniz's Law admits of no exceptions? cannot both hold this and treat 'occurrences' of expressions within quotes as genuine occurrences of them; but, so

We

far as I can see, we might simply regard Quine's argument as proving that Leibniz's Law does not

always hold. But further, Quine's own words, used to introduce Leibniz's Law into the argument, raise difficulties if his own view of quotations is accepted. What does he mean by "these personal names"? Plainly, the names "Tully" and "Cicero". But has he, on his own view, the right so to refer to them, when he has not mentioned them before (but only used them)? Let us waive this, and write out explicitly: "the identity of Tully with Cicero would allow us to interchange
"Tully" and "Cicero" salva veritate"; or, still more clearly: "if Tully is identical with Cicero, then "Tully" and "Cicero" are interchangeable salva veritate". In order to accept this as (a case of) a sound principle of we must inference, we must recognize its logical form be able to see that the same pair of names occur unquoted in the antecedent and quoted in the consequent (as I should say), or are used in the antecedent and mentioned in the consequent (as the current jargon has it). But according to Quine the logician's eye can no more see "Tully" and "Cicero" in the consequent than it can see "man" in "emancipated"; so the way
in
this application of Leibniz's Law is made is something that, on his own principles, Quine ought not

which

to understand.
indispensability of quotation is recognized in the practice of logicians, in the expository parts of

The

THE ORATIO RECTA ANALOGY

85

their works; I caa see no justification for their not attempting to construct formal systems with a device

corresponding to quotation. (For a pioneer investigation, see REACH.) It would of course be out of place to attempt such a formal construction in this work. Admittedly many 'quotations' are not logically part of the text at all, but serve rather as illustrations or diagrams. For example, in the discussion of the Existentialist in 4 I used quotes (as the reader may verify) to enclose bits of sheer nonsense. If quotation marks are always to be used as a logical sign, this will be incorrect; an added logical sign cannot turn sheer nonsense into sense, and so in a logically well-formed quotation the quoted expression must be a genuine bit of language. But it is quite easy to rewrite the discussion so as to

make

it

no longer open to

this objection,

as follows:

Nothing noths
Figure
1.

lie judged that

Nothing nothed
2.

Figure

Suppose that an Existentialist utters in a tone of conviction sounds transcribable as in Figure 1, then I
could not use Figure 2 as a report in written English of what he judged; for neither Figure 1 nor Figure 2 is a genuine bit of English. I ought rather to say: "He uttered sounds transcribable as in Figure 1, under the impression that they mean something that is true". I think one source of the accepted contrast between 'use' and 'mention' is a wrong assimilation of quotations generally to the untypical quotations of 4. Properly speaking, these are not quotations at all; we cannot quote sheer nonsense, we can only parrot it, or copy its visible pattern, or make believe that it is

86
(say) English

MENTAL ACTS

and that we are reading it aloud. Here 1 not including under "sheer nonsense" nonsensical combinations of genuine signs; for, as I have said, a complex quotation is to be read as a description, telling us that the quoted expression consists of such-andsuch signs in such an order; and this description, if the

am

signs are genuine, quotable, signs, is always significant, even if the signs in that order do not make up a wellformed expression. There is an important logical point to be made about equiform expressions (i.e. expressions of the

same

German "ja"
equivocation:

sensible form) in different languages; let us say "I"). I ( "yes") and Polish "ja" (

hold that the following syllogism involves a fallacy of

"The word "ja" is a personal pronoun in Polish; The word "ja" is a sign of affirmation in German; Ergo, some word that is a personal pronoun in Polish is a sign of affirmation in German."
For the Polish word "ja", which is a personal pronoun, is simply a different word from the German word "ja", which is not; and it is hard to imagine circumstances in which one and the same bit of written or spoken language would be an instance of both words simultaneously. (One might imagine a man, who had been asked one question by a Pole in Polish and another question by a German in German, answering them both at once with a monosyllabic utterance that meant "I" to the Pole and "yes" to the German.) Quotation of a word is, on my view, use of the word; and a quotation of the Polish word "ja" will not also be quotation of the German word "ja", any more than any other use of the one word is also use of the other.

THE ORATIO RECTA ANALOGY

In the
44

87

premi&e of our 'syllogism' the quotation contains the Polish word, and in the second "ja" " " contains the premise the equiform quotation "ja" German word; the phrases "in Polish" and "in Gerfirst

"

man"

logically attach not to the predicates of these

premises but to the subject they serve to show which word is being quoted, the German or the Polish word. An understanding of this point will be needed in order to refute certain fallacious arguments in the next
Section.

19

FALLACIOUS ARGUMENTS AGAINST PSYCHOLOGICAL USES OF ORATIO RECTA

r
I

CHOUGH the primary use of oratio recta is to reis

it can be used, often used, instead of oratio oblique. , in order to report the use of words tantamount to those quoted. Can we then dispense with oratio obliqua and stick to oratio rectal For example: Is "James said that man is

JL port a use of the quoted expression,

and

mortal" satisfactorily replaceable by "James said something tantamount to: "man is mortal" "? If not, then it would appear even more hopeless to try to report judgments by an analogical use of oratio recta. But, as my illustrations from the King James Bible show, judgments and thoughts can be reported this way; we know, then, in advance that any arguments against the satisfactoriness of oratio recta for the pur-

pose must be fallacious. One such argument is as follows. The statement

88

MENTAL ACTS
said

"James
mortal"

"

something tantamount
it

to:

"man

is

is

unsatisfactory as

stands; for the

quoted

expression might be an expression of some language other than English, and might in that language not mean that man is mortal. But if we insert "the English expression" after "tantamount to", then the resulting statement will not do as a substitution for "James said that man is mortal"; for this statement makes no reference to the English language, though it is made in
English. (Cf. CHURCH.) To this I reply that against the possible ambiguity here alleged no provision is practically necessary, and

none

is

work

theoretically sufficient. Theoretically, the frame" in which the quotation "man is mortal" "

occurs might likewise be equiform with an expression of some exotic language; and no provisos, sub-intents, and saving clauses inserted in the framework can remove this theoretical risk, for the exotic language might contain expressions equiform to them also. In practice, if I say to a normal English-speaking adult

tantamount to "man is morboth understand the framework of the quotation, as an English expression, and understand what expression I am intending to quote; there is no

"James
"

said something
will

tal"

he

practical risk of misunderstanding if I fail to insert "the English expression" after "tantamount to".

The argument

just considered gets its plausibility

from an assumption that one and the same expression "man is mortal" could occur both in English and in some exotic language, with different meanings; in that
case,

we could not mean anything definite by "tantamount to: "man is mortal" " if we failed to specify the
language in which "man is mortal" is to be taken to be. But, as I argued just now, an expression in an

THE ORATIO RECTA ANALOGY

"man
is

89

exotic language, fcquiform to the English expression

mortal" but different in meaning, would just be a different expression. Explicit mention of English would be needed only if one's hearer were likely not " to know that the quotation "man is mortal" " is a
quotation of the English expression; in practice not necessary.
it is

Another argument against replacing

oratio obliqua

by

oratio recta
is

is this.

If a sentence containing a quota-

translated, the quoted expression must be left untranslated; for the translation must be a statement

tion

about the same expression as the original sentence was

about, and therefore must contain quotation of that expression, not of some other. But if a sentence con-

must be translated; e.g. mortal" in oratio obliqua would become in Latin "hominem esse mortalem". So, since "James said something tantamount to "man is mor" is essentially different, in its behaviour under tal"
tains oratio obliqua, then this

the clause

"man

is

translation,

from "James said that

man

is

mortal",

it

cannot be a satisfactory Ersatz for it. (Once more, cf. CHURCH.) It is true that if we construct an oratio-recta Ersatz for "James said that man is mortal" we shall use the " quotation "man is mortal" ", whereas if we construct such an Ersatz in Latin for "Jacobus dixit hominem esse " mortalem" we shall have to use the quotation "homo est mortalis" "; and it may seem obvious that the one quotation is not a translation of the other. But the one quotation will occur in the context "tantamount to: "man is mortal" "; the other will occur in a Latin context translatable as: "tantamount to: "homo est mortalis" ".

Now these contexts are equivalent; for "tanta"homo

est

mount

to:

mortalis" ", the English version

90

MENTAL ACTS

of the Latin context, is substitutablfc in many English sentences, salva veritate> for "tantamount to: "man is
est

an expression is tantamount to "homo and only if it is tantamount to "man is mortal". Moreover, only a person who knows both English and Latin will understand "tantamount to "homo est mortalis" ", since otherwise he would not know what expression was being quoted; but tor such a person the description "tantamount to: "homo est " is of the same force as "tantamount to: mortalis"
mortal"
", since
if

mortalis"

".It follows that the Latin context of which "tantamount to: "homo est mortalis" " is a translation is likewise of the same force as "tantamount to

"man is mortal"

"man
lish

is

sentence containing the quotation "man is " is not fairly rendered by a Latin sentence " "homo est correspondingly containing the quotation mortalis" "; nevertheless the Latin oratio-recta Ersatz

mortal'" ". Thus, although in general "

an Eng-

mortal"

for "Jacobus dixit

hominem

esse mortalem",

which

latter quotation, is a reasonable equivaof the English Ersatz for "James said that man is lent

would use the

mortal", which would use the former quotation. Somebody may complain that it is not enough for the English Ersatz and the Latin Ersatz to be 'reasonably equivalent' to one another, since the sentences they replace are synonymous; if they are only 'reasonably equivalent', something important has been lost. I should demand of such a critic: "What is your criterion of

synonymy?" with very little hope of getting a coherent answer. (On the obscurity of the term
"synonymous", see QUINE
(3).) Two expressions are to the extent to which they are intersubstiequivalent tutable salva veritate; but we cannot define synonymy

as a supreme degree of equivalence, intersubstituta-

THE ORATIO RECTA ANALOGY

91

bility salva veritati in all contexts; for if occurrence in quotation is recognized, no two distinct (non-equiform)

expressions are intersubstitutable salva veritate at all occurrences. E.g. there is a high degree of equivalence between "furze" and "gorse", but "furze" is not sub" stitutable for "gorse" salva veritate in "gorse"

rhymes with "horse"

".

Nor could we say that syno-

nymy

intersubstitutability of expressions salva veritate at all non-quoted occurrences. For then, in order to
is

know whether "furze" and "gorse" are synonymous, we should have to know whether "furze" can be substituted salva veritate for the first occurrence of "gorse" in "Any statement made about gorse is tantamount to

the same statement made about gorse"; i.e. whether any statement made about furze is tantamount to the same statement made about gorse; and to know this we should have to know already whether "furze" and "gorse" are synonyms. So no objection stated in terms of "synonymy" is going to have much weight; an

attempt to define "synonymy" strictly is likely either to yield a relation that never holds between non-equiform expressions, or to involve a vicious circle when we try to apply it. For this reason I have deliberately used familiar words like "equivalent" and "tantamount to" rather than "synonymous"; there is less danger that way of making a show of precision where precision is not attainable. (Carnap's talk of "intensional isomorphism" is even worse; for the term is just as vague as "synonymy", but is even more calculated to give the impression that scientific rigour has been achieved.) "Correct translation" is of course just as imprecise a term as "synonymy". Very often, what we count as a
correct translation will include translation of quoted expressions; a translator of Quo Vadis would not feel

92

MENTAL ACTS

obliged to leave all the conversations in the original Polish, and we should count it as perversely wrong, not

pedantically correct, if he did so. Translating on this principle, we should get in the Latin version of "James " said something tantamount to "man is mortal" the " " " "homo est mortalis" rather than "man quotation

reasonably so, because nothing is being is mortal" that is not equally true of "homo est mortalis". But then the 'translation' argument against the oratio-recta Ersatz of "James said that man is a rational animal" breaks down. What about double oratio obliqua? The Ersatz for "Smith says that James says that man is mortal"
is

mortal"

":

asserted of

"man

would be: "Smith says something tantamount to: "James says something tantamount to: "Man is " ". What I have already said about 'reasonmortal" able' equivalence, replaceability salva veritate, and correct translation all carries over to this more complex case; I shall not here work the matter out in detail, but I do not think any new problem arises.

20

QUANTIFICATION INTO ORATIO OBLIQUA AND ORATIO RECTA

A PROBLEM we must
certainly tackle
is

that of

JLJL quantifying into (to use Quine's expression) oratio obliqua and oratio recta. So simple a psychological state-

ment as "More people came than he thought would come" involves 'quantifying into' oratio obliqua; one might try to show its quantification as follows:

QUANTIFIED ORAT1O RECTA

93

(A) "For some fcumber n, more than n people came, and he thought that at most n people would come", where the quantifier "for some number n" is meant to be picked up by (to 'bind') an occurrence of "n" in an oratio obliqua clause. But such a simple analysis will not do; in fact, there are plausible arguments to show that
it is

always nonsense to 'quantify

(3), chap. viii). assign a range to a quantification "for some #" that reaches into oratio obliqua; to answer the question

(see

QUINE

One

into' oratio obliqua of the difficulties is to

Quine's argument is certainly too in its conclusion, because it is certainly quite sweeping intelligible to say "More people came than he thought

"some what!"

would come".
clusion. If

We must try, then, to get round the conthis

we can do

by our method of using oratio

be a point in favour of the method. Let us begin with quantification into plain oratio recta. Suppose a letter I get from Smith contains the words: "James says one his colleagues "is out on bail after arrest on a serious charge"; I am not at liberty to give you his name". I assume Smith's meaning to be that James used the actual words "is out on bail after arrest on a serious charge". A first shot at making explicit the quantification in Smith's statement might
recta as Ersatz for oratio obliqua, this will

run

like this:

(A) "For some x, x is one of James's colleagues and James says "# is out on bail after arrest on a serious

charge" ". There are, however, two troubles about (A). First, if we interpret the quotation within (A) in the ordinary way, then the variable "#" would have to be the actual expression used by James to refer to his unlucky

94

MENTAL ACTS

all

colleague which is improbable. Secondly, though it is right to use a proper name first unquoted and then
in:
is

quoted, as

(B) "Johnson
serious charge" ",

James says "Johnson

one of James's colleagues, and is out on bail after arrest on a

there are both technical difficulties and difficulties of interpretation about replacing the two occurrences of

"Johnson" by occurrences of the variable "#", both bound by the same quantifier. We need not here discuss technical difficulties; what makes such a use of "#" hard to interpret at all is that we cannot consis." in (A) as tently interpret "For some a?, ... x .", nor yet as meaning "There is some person who "There is some word which ." The first meaning is ruled out by the quoted occurrence of interpretation
.
.

"a?"; for, even if we waive the objection that this ought to refer to an actual use of "#" by James, it certainly

ought to refer to something that James said; and a person cannot be part of what somebody else said. The second interpretation is equally ruled out by the unquoted occurrences of "#"; a word cannot be somebody's colleague. But there is an easy way out of our
troubles. If

catenation'
press
(C)

we use the ampersand, as in 17, for 'con(i.e. to mean "followed by"), we may exas follows:

what Smith presumably meant

"For some x, one of James's colleagues has the name #, and what James said was a? & "is out on bail after arrest on a serious charge" ".
Consider

"More would

quantification into or olio obliqua: people came to Smith's party than James said come". first shot at analysing this would be:

now a

QUANTIFIED ORATIO RECTA

95

(D) "For some* n, more than n people came to Smith's party, and James said that at most n people would come to Smith's party".

As before, there are difficulties over the interpretation of the variable; but they can be got over fairly easily if we resort to an oratio recta paraphrase:
^For some numeral n, "more than" & n & "people came to Smith's party" is a true statement, and James said something to this effect: "at most" & n &
4

(E)

"people
(E)
is

will

come

to Smith's party" ".

obtainable by existential quantification from a

like:

statement
"
(F)

"More than twenty people came to Smith's a true statement, and James said something party" to this effect: "at most twenty people will come to
is

Smith's party"

".

For, using our ampersand,

we may

rewrite (F) thus:

& "twenty" & "people came to a true statement, and James said something to this effect: "at most" & "twenty" & "people will come to Smith's party" ".
(F')

"

"More than"
is

Smith's party"

(E) appears reasonably close to the way one would intend "more people came to Smith's party than James

would come"; the quantification in (E) will be seen to relate to expressions, not to the intensional entities' sometimes brought into such analyses. If now we want
said
c

instead an analysis of

"more people came to Smith's party than James thought would come", all we have to do is to replace "James said" in (E) by "James said in
his heart".

21

THE PREDICATE
ROLE of the predicate "is a true statement"
__ in (E) of 20 should be noticed. In (F) ofJ20 the use of "is true" or "is a true statement" is superfluous; we could cancel out these words along with the pair of opening and closing quotes preceding them:

"More than twenty people came to Smith's and James said something to the effect: "at party, most twenty people will come to Smith's party" ".
(F*)

The expression
"

"is (a) true (statement)"

is

like the sign

Iq" in musical notation. After the sign "#", all notes written on the "C" line of the stave are to be read as

cancels the precedmeaning C sharp; but the sign and thereafter such notes are to be read as ing "#"> meaning C natural. Similarly, we write a statement in
t|

"

"

quotes in order to mention that statement;

write after
it

if

we now

"is (a) true (statement)", this cancels the

quotes, and it is as though we simply used the quoted statement. By the current false doctrine about "use and

mention' this simple point has been made quite unintelligible; since the logician is supposed to turn a blind eye to the occurrences of expressions within quotes (each quotation being regarded as one long word, logically unanalysable), he must make believe that he does not see the rationale of the inference (say) from "man is mor" "man is mortal" is true" and conversely. The tal" to genuine significance of "true" is thus missed, and then queer 'theories' and 'definitions' of truth are framed. If "is true" thus cancels out with the preceding pair
96

THE PREDICATE "TRUE"

97

of quotes, why not simply write down the statement that is being presented as true, without quotes and without any use of "true"? Is not the word "true"

we cannot always do this cancelwe cannot do it with "What the policeman ling-out; said was true", nor can we do it in (E) of 20. Again,
superfluous? No, for

a true statement" is a perfectly into a hypothetical statement, but "if telligibfe protasis every man" is not; here again the cancelling-out is un"if

"every

man"

is

workable. It

is

only

quoted

in full that

when a genuine statement is being we can use "is (a) true (statement)"

to cancel the enclosing quotes. must not think that the role of "the statement " is true" is that of a (Fregean) assertion sign; and ".

We
.

the policeman said is true" does not mean something like "I corroborate what the policeman said". For "is true" can stand quite happily in an "if" clause, which is not asserted; to say "if the stateagain,

"What

ment "the Earth is flat" is true, astronomy is bunkum" does not commit the speaker to asserting that the Earth is flat. Again, "if what the policeman said is true" is not at all the same thing as "if I corroborate what the policeman said". In main clauses, to be sure, "I corroborate" has a 'performatory' role its utterance is an act of corroborating and it loses this role in
an "if"
clause; but this very fact shows how differently "I corroborate" works from "is true", which can go into an "if" clause without any change of its ordinary role. When I say " "jam dies" is true in Latin if it is now day", "in Latin" logically attaches not to "is true" but " to the quotation "jam dies" "; we cannot argue that because "jam dies" is false in English, there is something true in Latin but false in English this is a crude

equivocation, already exposed.

The English "jam dies"

98

MENTAL ACTS
used (quoted) in " "jam dies" " and the
is

and the Latin "jam dies" are simpty

sions; the former
false in English",

different expresis

in Latin
self of

latter in "jam dies" is true when it is now day"; you may convince yourthis by reading the two sentences aloud. The

'semantic' theory that "in Latin" and "in English" could qualify the predicate "true" is thus a howler; whatever paradoxes like the Liar may arise, no*solution in terms of "true-in-L" and "true-in-M" is acceptable. I have been trying to make out that "true" is a predicate properly and primarily applicable to statements in language, rather than to thoughts or judgments or objective 'intensional entities'. This contention is a necessary part of my general thesis that language about thoughts is an analogical development of lan-

guage about language.

22

COMBINATION OF OUR TWO THEORIES OF JUDGMENT

HOW

account of judgment related to the account given in 14? That first account was

IS

THIS

both sketchy and schematic: sketchy, in that I discussed only a few sample judgments; schematic, in that "Idea" and "". The I used two uninterpreted terms "" latter expression was used quite formalistically; was an unspecified logical operator upon relative terms. "Idea" on the other hand was not wholly uninterpreted; Smith's Idea every man would be an exercise of his concept every man, which in turn, on my explanation, is a mental capacity exercised in judgments that

TWO THEORIES COMBINED

are expressible

99

of the words "every man". But be very variously interpreted, conformably to these rough explanations. Now can interpre"" be found within the limits tations of "Idea" and of of our oratio recta system for describing judgments? Let us provisionally say that Smith's Idea every man

by

tise

"Idea" could

still

consists in his saying-in-his-heart something to the same effect as "every man" (which, let me repeat, need not consist in his having mental images of these or

other words). I shall say "mental utterance of" as an abbreviation for "saying-in-his-heart something to the same effect as"; thus, Smith's Idea every man is Smith's mental utterance of "every man"; and "Smith's Idea of [an Idea of every man]" will be interpreted as "Smith's mental utterance of "a mental utterance of " ". This satisfies the condition, stated "every man" on p. 57, that once we understand "Idea" and the

"Idea of every man" we must already have all the logical devices we need in order to understand "Smith's Idea of [an Idea of every man]". For "Smith's mental utterance of "a mental " " is understandable so utterance of "every man" as we understand "mental utterance" and also long understand the logic of quotations. In order that the relation (sharper than) should hold between two Ideas, there must be a judgment that that is, in our present language, a is sharper than mental utterance of some statement using "(is) sharper
construction of "every
in
. .

man"

than". (Let us, to simplify matters, erase the word "blunter" from the language we are using to describe judgments, and replace it by "conversely sharper".)

What we need is some relation

%(sharper than), defin-

able in terms of "sharper than", satisfying this condition: When Smith's judgment consists of an Idea a in

100

MENTAL ACTS
(

sharper than) to an Idtea /?, then there are expressions A, B, C, such that a, /? are Smith's mental utterances of A, B, respectively, and Smith's judgment as a whole is a mental utterance of C, and any (physical) occurrence of C consists of an occurrence of A in the relation %(sharper than) to an occurrence of B. If we could actually define " %( sharper than)", we should have completed the interpretation of our former theory of judgments in terms of our oratio recta system for
the relation
describing them. Now we can at least see the main outlines of the defin" ition of %(sharper than)". Let us define "|(sharper

than)" as follows; x is in the relation f(sharper than) to y if and only if: x and y are utterances of the same and such that person, and there are expressions x is the utterance of X, and y of Y, in that person's utterance either of & "is" & "sharper than" & Y or

of

Y&

"is conversely"
is

&

"sharper than"

&

X. (The

as before the sign of 'concatenation'.) ampersand For example, if the person had said out loud "some knife is sharper than some spoon", or if he had said

conversely sharper than some knife", "some knife" would have in the relation f( sharper than) to his utterance of been "some spoon"; if, on the other hand, he had said "some spoon is sharper than some knife", then his utterance

"some spoon

is

in either case his utterance of

of "some spoon" would have been in the relation f( sharper than) to his utterance of "some knife". It is because we make relations like f(sharper than) hold between utterances of expressions, that we are able to talk about the relation sharper than effectively just

having the relative term "sharper than" would not suffice unless we used it to establish the relation ^(sharper than) between expressions. (This point is

DANGERS OF ANALOGY

101

made, rather obscurely, in Wittgenstein's Traciatus.) It is no objection to our using such a relation in explaining the relation (sharper than), that in the definiens of "t(sharper than)" there is a non-extensional (quoted) occurrence of "sharper than"; for I said already in 13 "" must be that the operator regarded as nonextensional. 1

We

c&nnot regard the relation %(sharper than) as

actually being the relation f(sharper than); for it is easily seen that Smith's utterance of "some knife"

would stand
knife
is

in the relation f(sharper than) to his utterance of "any spoon" alike in the utterance "some

sharper than any spoon" and in the utterance "any spoon is conversely sharper than some knife"; whereas the corresponding judgments are different, But this only means that the definition of " %( sharper than) would have to be more complicated than the above definition of "f(sharper than)".

23

DANGERS OF THE INNER LANGUAGE ANALOGY

c '

thing about analogical exten sions of a concept is that we should know (in prac tice at any rate) how far to carry the analogy. Som< people have certainly carried the analogy of thought t(
In an 'extensional' logic the Leibniz law of 18 would hold un and predicates or relative terms would be freely inter substitutable provided that they held good of the same object
1

IMPORTANT

restrictedly,
14).
is

(c/.

sions,

easily

Our language, since it contains quoted occurrences of expres shown not to be 'extensional' in either respect.

102

MENTAL ACTS

language too far. Thus, for William 6f Ockham, besides the spoken, conventional, languages, all men have a

common,
call it

natural, language; for convenience, I shall

"Mental". The grammar of Mental turns out to be remarkably like Latin grammar. There are nouns and verbs in Mental; nouns have cases and numbers, and verbs have voice, mood, tense, number, and person. On the other hand, there is nothing in Ment&l corresponding to the different Latin declensions and conjugations; nor are there any deponent verbs in Mental. Ockham's criterion for transferring Latin grammatical terms to Mental was very simple-minded. Nouns of different declensions, or verbs of different conjugations, may be synonyms, and then presumably correspond to the same Mental noun or verb; so there is no reason
to ascribe differences of declension or conjugation to

Mental words. But a change of case or number or voice may quite alter the sense of a Latin sentence; so Mental words must have case, number, and voice.

Without being able to say just how far the analogy of inner language can be carried, I think men of good sense would see immediately that Ockham carries it much too far. He merely transfers features of Latin

grammar to Mental, and then regards this as explaining

why such features occur in Latin they are needed there if what we say inwardly in Mental is to be outwardly
got across to others in Latin. But clearly nothing is explained at all. Presumably Ockham's reasons for thinking that the supposed grammar of Mental had explanatory force were that Mental is a natural and universal language, and that Mental words, unlike Latin words, are immaterial entities. But if all men had a natural and universal spoken language, that would not mean that its grammar was any more self-explana-

DANGERS OF ANALOGY

103

tory than Latin grammar. And what carries significance in a language is its structure, not its medium the structure that can be transferred from spoken to written language and to Morse code; but Ockham takes for granted the grammatical structure of Latin, and supposes that Mental, unlike Latin, is intrinsically intelligible, simply because its medium is not material but spiritual. In point of fact, any problems that arise as to the significance of a grammatical device will arise equally for the alleged Mental uses of this device; and Ockham's saying that the words of Mental are immaterial

would merely raise such footling problems as how something immaterial can be in the genitive case, without throwing any light on the use of the genitive. To do Ockham justice, he wastes little time on such futilities; most of his, often acute, enquiries into the logical
syntax of Latin are undisturbed by the reflection that Latin is really an imperfect reproduction of the Mental original, which on his view is the proper study of a
logician.

little

consideration will

show us various points at

which the analogy of inner language must break down. The grammatical properties ascribed by Ockham to Mental words may all be easily dismissed, except mood and tense. "The farmer has one donkey, which is white" and "the farmer's one donkey is white" plainly both
express the same judgment, although the cases of "farmer" and "donkey" are different in the two sentences; and so for most of the grammatical attributes. Tense, however, does enter into the content of our

thoughts; "the farmer has often been drunk" cannot be replaced by any sentence to the same effect in which the tense is future instead of past. And I am inclined to think that there are also modal differences between

104

MENTAL ACTS

thoughts though the moods of a natural language like Latin are a very inadequate indication of this, being cluttered with a lot of logically insignificant idiomatic
uses.
is more important is the difference between and thought as regards temporal duration. speech Spoken words last so long in physical time, and one word comes out after another; the time they tdke is, as Aquinas would say, the sort of time that is the measure of local motion one could sensibly say that the utterance of the words was simultaneous with the movement of a body, e.g. the hand of a clock, from one place to another. The same would go for the duration of mental images of words, or any other mental images; one could sensibly say "I rehearsed the words in my head as I watched the beetle crawl from one side of the

What

table to the other".

it is quite different. Even if we the view sketched in 14 that a judgment is a accepted complex of Ideas, we could hardly suppose that in a thought the Ideas occur successively, as the words do in a sentence; it seems reasonable to say that unless the whole complex content is grasped all together unless the Ideas, if Ideas there are, are all simultaneously present the thought or judgment just does not exist at all. All the same, when we read of the French politician who maintained that in French the word-order perfectly represented the order in thought (WITTGENSTEIN, Part I, 386), our first reaction may be e.g. to think of the French negative "ne pas", and protest that even a Frenchman could not negate the 'verb'

With a thought

part of his judgment in two bits, half an act of negation preceding the Verb' part and half an act following it; that is to say, we are inclined to think that the Ideas

DANGERS OF ANALOGY
do have a tempofttl

105

order, only not quite the same order as that of the French words expressing them. Only on further reflection do we come to the view that there can be no question of their occurring in any temporal order. On my analysis of judgment in terms of Ideas that which corresponded to the 'main verb' in the verbal expression of judgment would be, not an Idea, but the relation binding the Ideas together; and it would be plainly absurd to enquire which position of the main verb best corresponded to the temporal position of this relation as compared with the Ideas it relates. Of course that analysis was only tentative and provisional; at the same time, it was suggested to me by the rather stringent requirements of logic, which

became apparent in criticizing Russell's theory; it seems most unlikely that analysis of a judgment as a temporal succession of Ideas could meet these requirements, whatever temporal order we assume. If a judgment is a non-successive unity, it seems to follow that we cannot assign to judgment more than a loose connexion with physical time. First, could a judgment be regarded as occupying a mere moment of physical time? If so, then either in any finite stretch of time there would be an illimitable number of judgments (or other thoughts), or there would be flashes
of thought at discrete instants separated by gaps of thoughtlessness; the first alternative at least is plainly false, but the second might seem possible. Let us con-

however, the time-relation between an act of judgment and the words that express it. Once the words have been spoken, we suppose, the man goes on to think of other matters; so the act cannot be put later than the utterance of the end of the last word. If the man was not in a position to rehearse his words
sider,

106

MENTAL ACTS

mentally before he said them, the actfof judgment might be said not to come before the beginning of the utterance. Let us further suppose that the man's attention is not distracted while he speaks. Are we now to say that the judgment occurs in just one instant of the utterance, the rest of the time being taken up with uttering parrotwise? or, that it is repeated in a series of flashes, so as to keep the utterance on tTie right
track?

going on continuously during the utterance? That is worse nonsense than ever: one can say "at 12.10 I had a sudden stab of pain, and the pain went on for two minutes", but not: "at 12.10 I had a sudden thought, and the
to say that the judging
is

Or are we rather

thought lasted just long enough for me to utter it it went on all the while that I was uttering it, and then it stopped". All that we can say is that the judgment is loosely bound up with physical time, in that (e.g.) it did not occur before the beginning or after the end of the words in which it is uttered; if we try to assign it to a definite moment or moments, or to a definite stretch of time, we find ourselves in a bog of nonsense. What I have just said about the 'loose' relation of judgments to physical time is a logical point about applying time-specifications in our discourse about judgments; it does not imply e.g. that judgments are really performed in a super-physical realm. Excluding certain questions about the time of judgments as unreasonable can have no such momentous consequences; and for judgments to be tied loosely to physical time is still for them to be tied.

24

THE NOTION OF 'INNER SENSE

IN concepts,
INVESTIGATING
I shall

some other psychological approach them from the side of

psychological judgments in which they are exercised. Psychological judgments have very often been held to be based primarily on the deliverances of an 'inner sense' whereby we are cognizant every one of his own psychical states; just as judgments about physical realities are based on the evidence of our senses. The

supposed 'inner sense' is compared sometimes to looking ('introspection'), sometimes to feeling.

With the ordinary

senses there

is

associated the

power of forming mental images; I see in my mind's eye past scenes that are no longer before me, or things that I might see now in another position, or even quite
imaginary scenes.

No

mental images, however, are

commonly assigned to 'inner sense' in the same way. People have ridiculed the idea of McTaggart that I may
have mental images of my own past mental states which I have introspectively perceived, of other people's mental states which I happen to be unable to inspect (a removable limitation, McTaggart believed), and of purely imaginary mental states that nobody has actually been in. (MCTAGGART, pp. 106-8.) Of course McTaggart's idea is quite wrong, but why is it absurd?
If "looking into the

mind" has a genuine

logical

similarity to "looking into the box", then it ought to make sense to talk about introspective mental images as we do about visual images; lack of introspective

imagery would be an idiosyncrasy

107

like

being unable

108

MENTAL ACTS
and might be expected to disable a man

to visualize,
for
it

McTaggart's idea is absurd, then ought to make us suspect the comparison of introtasks. If

some

spection to real looking.

Hume seems to have supposed that there were menimages of past mental states less vivid copies of them, related to them as a perfume I 'smell' in memory is to something I really smell. But what Hume says is shot all through with his confusion between two senses of "idea" the mental image of something, and the exercise of a concept of it in judgment. This exercise is in no way dependent on the presence of a pale replica of the thing judged about; if it were, then, as St Augustine remarked, we should habitually feel a certain reluctance to use the words "grief", "fear", "pain", lest they should arouse in our minds faint reproductions of
tal

the unwelcome experiences so called (Confessions, book X, chap, xiv). 1 People suppose that I can give meaning to such

words as "seeing", "hearing", "thinking", "hoping", etc., only by observing in my own case sample occurrences of what these words refer to; failing the relevant experiences, or failing attention to them when I have had them, I must either lack a concept altogether or possess it in a very imperfect form, comparable to a colour-blind man's concept of colours. Let us fasten upon this comparison. The defect colour-blindness can be tested for, not only by looking for oddities in a man's colour-concepts, which come in his use of colourwords, but also by a non-linguistic investigation of what colours he is practically able to discriminate,
1 Augustine here uses the term "notiones" for the mental acts involved in an understanding of such words as "fear", and denies that notiones are imagines. I wonder if this is the origin of Berkeley's puzz1 ling talk about 'notions as opposed to "ideas'.

THE NOTION OF INNER SENSE

6

'

109

without using wofds.

can even investigate what colour-discriminations brutes are capable of, although of course they cannot tell us and have no colour- concepts at all. Now could there be, let us say, anger-fear
emotion-blindness, as there can be red-green colourblindness? Could a man's introspective 'sense' be unable to discriminate between his being angry and his

We

being afraid, so that his use of the words "anger" and "fear" depended precariously on other people's use of them, in the way that a colour-blind man's use of colour-words does? In regard to colours, we can distinguish between a colour-blind man with a sensory defect and a mentally defective man who is unable to form colour-concepts and learn the use of colourwords; could we make a similar distinction about emo-

we say of somebody: "He's very intellibut he keeps on using words for emotions gent, wrongly; the psychiatrist says he has a congenital defect of the 'inner sense' that discriminates emotions from each other"? I chose to set emotion- words and colour- words side by side, because there really is a considerable logical similarity. Both colours and emotions can occur in different intensities, and can wax and wane in intensity; there can be an emotion that is a blend of anger and fear, as there is a colour that is a blend of red and blue; you can ask of a feeling of fear, as you can of a coloured light, whether it came on suddenly or gradutions? Could
ally,

and how long

it

lasted, etc. If in spite of this

we

find a radical dissimilarity between colour-language and emotion-language, in that we could not apply a term"emotion-blindness" comparably to "colour-blindness", then the conclusion we ought to draw is surely that the idea of an introspective 'sense' is an illusion

110

MENTAL ACTS
for the

Somebody might
have no use

try saying that*the reason why we term 'emotion-blindness" is that

4

our 'inner sense' is not liable to such defects as our eyesight is; our inner sense represents our emotions just as they occur, even if we are unable to describe them correctly. Now a sense that was in fact not
affected
etc., is

by any

illusions,

any

failure to discriminate,

indeed conceivable; but plainly what our present objection is really after is a sense that not merely does not but cannot mislead us. But this "cannot" would be a logical "cannot"; and the inclination to use "cannot" here points to a logical difference between our knowledge of the outer world by our senses and our knowledge of the mind by 'inner sense'. Of bona fide sense-faculties, it is impossible to say that they cannot be defective or inaccurate. If anyone should think that in criticizing the idea of 'inner sense' I am flogging a dead horse, or knocking down a stuffed dummy of my own creation, I may reply by instancing Freud's use of the idea. "What role is there now left, in our representation of things, to the phenomena of consciousness, once so all-powerful and overshadowing all else? None other than that of a senseorgan for the perception of psychic qualities." (FREUD, p. 544; his italics.) "The unconscious is the true psychic reality; in its inner nature it is just as much unknown to us as the reality of the external world, and it is just as imperfectly communicated to us by the data of consciousness as is the external world by the reports of our senseorgans" (FREUD, p. 542; his italics.) Freud, as may be seen, transfers a naive 'representative' theory of perception from its usual application (to the bodily senses, that is), and holds it to be no less valid of 'inner sense'. Presumably on his view 'inner sense' would not be in-

SENSATION AND THE BODY

errant; but I fin<Mn him no clear account of error of 'inner sense' would be like.

111

what an

25
COU.LD SENSUOUS EXPERIENCES

OCCUR APART FROM AN ORGANISM?

C

TNNER

SENSE'

is

A the only thing that shows us,

supposed to show

us,

and to be
to see,

what

it is like

hear, be afraid, etc. With this there goes a view that the connexion between such 'sensuous' experiences and a bodily organism is only a well-established empirical generalization. Such experiences are indeed dependent

upon material things in the sense of being occupied with them; but they are not identifiable with any describable physiological processes in a living organism, and their connexion with such processes is only something empirically determined. There is no necessary, conceptual, connexion between the experience we call
and the processes that physiologists tell us " happen in eye and brain; the statement James can
'seeing"
4

although his optic centres are destroyed" is very unlikely on inductive grounds but perfectly inafter all, people used the word "see" long telligible before they had any idea of things happening in the optic centres of the brain. It therefore appears to be clearly conceivable that seeing and other 'sensuous' experiences might go on continuously even after the death of the organism with which they are now associated, and that the inductive reasons for doubting
still see,

112

MENTAL ACTS

whether

this ever happens might be H^ut weighed by the evidence of Psychical Research. I think it is an important conceptual enquiry to con-

sider

whether

really

disembodied seeing, hearing, pain,

are so clearly intelligible as is supposed in this common philosophical point of view. (I stress "really disembodied". Some people believe that there is a subtle body endowed with its own senseorgans, which persists after the dissolution of the body

hunger, emotion,

etc.,

commonly

so called. This view, so far as I can see,

is

philosophically speaking both unobjectionable and uninteresting. It is clear off-hand that the 'mind-body problem' is just the same whether the body is gross or
subtle.)
'to see' has its meaning for me because I have that experience!" Nonsense. As well suppose that I can come to know what a minus quantity is by setting out to lose weight. What shows a man to have the concept seeing is not merely that he sees, but that he can take an intelligent part in our everyday

"The verb
I

do see

use of the word "seeing". Our concept of sight has its life only in connexion with a whole set of other concepts, some of them relating to the physical characteristics of visible objects, others relating to the behaviour of people who see things. (I express exercise of this concept in such utterances as "I can't see, it's too far off now it's coming into view!" "He couldn't see me, he didn't look round", "I caught his eye", etc., etc.) It would be merely silly to be frightened off admitting this by the bogy of behaviourism; you can very well admit it without also thinking that "seeing" stands for a kind of behaviour. Our investigation is put on the wrong track by an
abstractionist prejudice.

For an

abstractionist, the

SENSATION AND THE BODY

113

possession of the Concept seeing must be taken to be a capacity for finding and recognizing some recurrent feature or at least to be that primarily; and I can find instances of seeing only in my own mind, by 'inner

nothing but characteristic pieces of behaviour, from which however it could justiIn fact, fiably be inferred (how?) that they also see. of course, I learn to use the word "see" of others and of myself simultaneously; and if we reject the doctrine of abstractionism, we need not distinguish between exercises of the concept seeing as primary ones, when I catch myself in the act of seeing something, and
sense'; in other people I find

on

secondary ones, when I (with great rashness, surely, this view) form the judgment that others likewise see. To have the concept seeing is not even primarily a matter of being able to spot instances of a characterrepeatedly given in my ('inner- sense ) experiences; no concept is primarily a recognitional capacity. And the exercise of one concept is intertwined with the exercise of others; as with a spider's web, some conistic
1

nexions

may be broken with impunity, but if you break

enough the whole web collapses

the concept becomes unusable. Just such a collapse happens, I believe, when we try to think of seeing, hearing, pain, emotion, etc., going on independently of a body. When I apply this sort of concept to a human being, I do so in connexion with a whole lot of other concepts that I apply to human beings and their natural environment. It is easy enough to extend the concepts of
'sensuous' experience to creatures fairly like human beings, such as cats, dogs, and horses; when we try to extend them to creatures extremely unlike human

beings in their style of life, we feel, if we are wise, great uncertainty not just uncertainty as to the facts,

114

MENTAL ACTS

or as to the possibility of finding thfcm out, but uncertainty as to the meaning of saying: "I now know how
is angry". One is of course to say: "That's easy; an earthworm is angry tempted if it is feeling the way I feel when I am angry!" But this

to

tell

when an earthworm

is

just like saying:

"Of course

know what

it is

for the

time on the Sun to be

the

five o'clock; it is five o'clock

on

Sun when

it is

the same time as

when
I,

it is five

o'clock here!" (WITTGENSTEIN, Part

350)

which

clearly gets us no for'arder. There is just the same difficulty in extending the concept the same time as in extending the concept five o'clock. So in the psychological case: I know how to apply the concept anger to myself and to James, and I know how to apply the concept feeling the same way as between myself and James, or James and Smith; I get into the same difficulties if I try applying the concept feeling the same way as between myself and an earthworm as I do over 1 applying the concept anger to it.

Even an earthworm, though,

cepts;

affords

some handholds

for the application of 'sensuous' psychological con-

we connect

its

writhings

when
it

injured with our

own

pain-reactions.

But when

comes to an auto-

1 On a point of interpretation, I think it is a mistake to read Wittgenstein as having intended to show that I cannot apply a concept like anger both to myself and to others, and that it is meaningless to speak of others' feeling the same way as I do (unless indeed I just mean that they behave as I do). The difficulty of transferring the concept anger from myself to others is a spurious one, arising from the abstractionism that Wittgenstein consistently rejected; and the solution that the term "anger" is an equivocal term, applied in my own case to a recurrent experience and in other cases to a recurrent pattern of behaviour, is plausible, I think, only on abstractionist presuppositions; I do not believe, as some people I have had discussion with apparently do, that Wittgenstein really held this view and only shrank from a brash statement of it.

SENSATION AND THE BODY

115

maton, or again if *ve are invited to apply the concepts to a supposed disembodied existence, then we may be sure that we are right in refusing to play; too many
threads are broken, and the conceptual

web has

col-

is not automaton, by lapsed. even alive; we know this, though we may be uncertain whether to call a virus (say) alive or not. (Doctors may not agree whether a patient is yet dead; but we know

An

all sorts

of criteria,

that Queen

Anne

is

dead.)

Between what

is

certainly

inanimate and ourselves there is far too little similarity for us to be able to pick out anything in its behaviour corresponding to the context in which we judge that

human beings are in pain, or hungry, or afraid; we know that any particular movement which might even
remotely suggest similarity is performed because the designer of the automaton intended such an imitation, and we ought to be no more inclined to ascribe feelings to the automaton than, after childhood, we think that a doll is in pain because it has been so constructed as As for disembodied sensato cry when it is smacked. more connexions are broken in tions and feelings, even this case; there is no handhold for applying 'sensuous* concepts to disembodied existence at all we just do not know what we are doing if we try. A good illustration, I think, of a concept's losing its applicability through connexions being broken is the following. Certain hysterics claimed to have magnetic sensations; it was discovered, however, that their claim to be having them at a given time did not go with the presence of a magnet in their neighbourhood but with their belief, true or false, that a magnet was there. It would now have been possible to say: "We must take
the patients' word for it that they have these peculiar sensations, which are quite different from ordinary

116

MENTAL ACTS

people's sense-modalities; it is merdy the term "magnetic sensations" that has turned out to be inappropri-

they had formed a wrong hypothesis about the physical cause of their sensations". But nobody even considered saying this; it was decided that the patients had just been indulging in the sort of pointless talk that hysterics often do indulge in. This decision just to
ate;

drop the idea of magnetic sensations and to ignore the patients' 'reports' of them was taken after a much smaller breakdown of the ordinary connexions than we are asked to tolerate when it is attempted to apply sensation-concepts to automata or to disembodied
existence.

Denying sense to the attempt

ism

to think of feelings,

sensations, emotions, etc., apart from a living organmay seem to be practically the same as denydisembodied mind altogether. Such a denial does ing
historically always been held to follow. Aquinas, for example, believed that there were

not follow, nor has

it

wholly disembodied intelligences, but that they were not liable to any such experiences as seeing and hearing and feeling afraid and having a pain: the evil spirits in hell are tormented not by aches but by the frustration of their wicked will. (la q. 54 art. 6, q. 59 art. 4, q. 64 art. 3.) Sensuous experiences are possible only in connexion with a living organism (la q. 77 art. 8). Only since Descartes has the main problem become: "How is cogitatio related to bodily processes?" ("cogitatio" covering, for him, everything 'in the mind', from a toothache to a metaphysical meditation); the old problem was rather: "How can a being that thinks and judges and decides also have sensuous experiences?" It was 'intellectual' acts like judgment, not just anything that would now be called 'conscious-

THE FALLACY OF "COGITO ERGO SUM"

ness',

117

which seemad to Aquinas to be wholly incommensurable with events in the physical world; for him, the 'unbridgeable gulf was at a different place. The
usefulness of historical knowledge in philosophy, here as elsewhere, is that the prejudices of our own period may lose their grip on us if we imaginatively enter into

another period, when people's prejudices were different.

26

THE FALLACY OF "COGITO ERGO SUM"

IS

WORTH WHILE

to

show what

is

with the Cartesian "cogito ergo mm". Many people find this part of Cartesianism attractive when they begin studying philosophy; and many unphilosophical people think that such uses of "I" as in "I am
feeling hungry,

IT wrong

really

and remembering what

was thinking

about yesterday" can enable us straightway to understand such a question as "shall I still be conscious

body?" The idea is that introspection can give the word "I" a special sense, which each of us can learn on his own account. "I" in this sense would not mean the man P.T.G., when P.T.G. used it; for nobody would wish to know whether the maix P.T.G. was still there after his body was
after the destruction of

my

destroyed. What is supposed is that P.T.G. can use "I" to express knowledge of something distinct from
P.T.G., which is directly discernible to one who gazes within himself. Let us begin by reminding ourselves how "I" is used

the

man

118

MENTAL ACTS

in ordinary life with psychological v^rbs. If P.T.G. says "I see a spider" or "I feel sick", people will ordinarily

who says this, P.T.G. sees a or feels sick. The word "I", spoken by P.T.G., spider serves to draw people's attention to P.T.G.; and if it
think that the speaker
,

is

not at once clear

who

is

speaking, there

is

a genuine

question sider Descartes brooding over his poele and saying: "I'm getting into an awful muddle but who then is

"Who

said that?" or

"Who

is

T?" Now con-

who is getting into a muddle?" When "I'm into a muddle" is a soliloquy, "I" certainly getting does not serve to direct Descartes's attention to
this

Descartes, or to

who

show that it is Descartes, none other, into a muddle. are not to argue, getting though, that since "I" does not refer to the man Ren6
is

We

Descartes it has some other, more intangible, thing to refer to. Rather, in this context the word "I" is idle,

used only because Descartes is habituated to the use of "I" (or rather, of "je" and "rnoi") in expressing his thoughts and feelings to other people. In soliloquy he could quite well have expressed himself without using the first-person pronoun at all; he could have said: "This is really a dreadful muddle!", where "This" would refer back to his presuperfluous;
it
is

vious meditations. (We have here an example of the puzzling demonstratio ad intellectum which I mentioned
in
15.)

Moreover, what is going to count as an allowable answer to the question "What is this 'I'?" or "Who then am I?" These questions might have a good clear sense in certain circumstances e.g. if Descartes had lost his memory and wanted to know who he was ("Who am I?" "You are Ren6 Descartes"), or if he knew that somebody had said "I'm in a muddle" but

THE FALLACY OF "COG/TO ERGO SUM"

119

not that it was hirilself ("Who is this who said he was in a muddle?" "You did"). The states of mind that

would give the questions sense are queer and uncommon, but they do occur. But no such rare circumstance was involved in Descartes's actual meditation;
in the actual conditions, it is simply that the questions is this T?" are deprived of I?" any

"Who am

"Who

ordinary use and no new use has yet been specified. When William James tried to pose Descartes's question to himself, and to answer it, he came out with the answer that what he meant by "I", his 'Self of selves', was a collection of feelings in his head and throat. Now of course "I am getting into an awful muddle", said as a soliloquy, did not mean to William James "These head-and-throat feelings are getting into an awful muddle". But how did he manage to make such a mistake? Unlike Descartes, William James was a skilled and trained professional psychologist; how did he manage to miss his own genuine 'Self of selves' when he took an inward look, and only happen upon headand-throat feelings? If even a professional psychologist can make such a gross mistake, who can be sure of avoiding such mistakes? Who can give directions to make it at least likely that my inward glance will hit

upon
I

my

real 'Self of selves'?

it?

have glimpsed

I to take your word for it? And why cerned yours? should I take your word? I may be sure you are not deliberately deceiving me; but may you not, like William James, be honestly mistaken? What I maintain is that if William James said to himself "I am very puzzled about this problem", his soliloquy would not mean either that the man William James was very puzzled or that certain head-and-throat

Am

How shall I know when How do I know that you have dis-

120
feelings

MENTAL ACTS
were very puzzled; "I" wduld not be serving show who was puzzled. The use of "I" in such

to

soliloquies is derivative from, parasitic upon, its use in talking to others; when there are no others, "I"

redundant and has no special reference; "I am very puzzled at this problem" really says no more than "This problem is puzzling" (demonstratio ad intellectum again). Similarly "I have (had) frightful pain" really says no more than "That pain is (was) frightful"; the question whose pain it is does not arise if the remark is a soliloquy. (This demonstratio will be discussed in the next Section.) If you insist that the soliloquistic "I" is not redundant in this fashion, and likewise does not stand for the human being who is speaking, then it is up to you to explain your use of "I"; not only do I not yet understand it, but I have positive grounds for suspecting that your own apparent understanding of it is a familiar philosophical illusion. People have had the oddest ideas about words having to stand for something; "nothing" has been quite often regarded as naming an entity, or rather nonentity (cf. the fourth of the Meditations, ad init.); and I even once read an author maintaining that "there" in the idiom "there "there is a snake in the bath" is" signifies Reality would mean "Reality has the characterization snakein-bath"! For all I know, a German philosopher has written something like this about the *es* referred to in "es gibt": "das Es steht vor Allem, auch vor Gott; denn das Es gibtja auch Gott, wenn es einen Gott gibt". When you look for something, other than the human being who speaks, to be the reference of the soliloquist's "I",
is

you are surely

just falling into this ancient trap. These are the sort of arguments I should deploy

DESCRIBING ONE'S

OWN SENSATIONS

121

against somebodjf who thought the Cartesian "Cogilo ergo sum" afforded a way for him to grasp some im-

material part of his make-up signified by "I". I have not said, nor do I think, that the only way of getting at a concept of soul or spirit is the "Cogt'fo"; I just wanted to show that the "Cogito" at least is a blind
alley

27

DESCRIBING ONE'S OWN SENSATIONS

psychological judgments. In the last Section I remarked that such judgments are expressible with a demonstrative instead of a first-person pronoun: "This pain is lessening", "this is terribly puzzling", instead of "I am having less pain", "I am terribly puzzled". The question is how we should account for these sorts of demonstratio. I shall consider first-person uses of psychological verbs for feelings and sensations. Part of the ground is covered by Wittgenstein's idea of the Aeussenmg, utterance, of a mental state. "How that hurts!" or "it hurts here" would be Aeusserungen of pain, serving not to express judgments in which the pain is described but as learned replacements of primitive (animal) pain-reactions. Similarly, "I am extremely angry with you!" is not a description of the degree of my anger but a replacement of primitive angry behaviour. (WITTGENSTEIN, Part I, 244-5.) All the

other account REJECTING of first-person

the "Cogito",

we must look for some

122

MENTAL ACTS
is

such a thing as deliberate description of feelings; and in particular "I had (or: that was) a nasty stab of toothache five minutes ago" can only be so taken, not as replacing a primitive painreaction, for there is no way of writhing and groaning in the past tense. (In WITTGENSTEIN, Part II, we find it expressly recognized that the concept of Aeufserung cannot be applied to all first-person psychological statements about feelings; see pp. 187-9.) Where there
same, there
one's

own

is
is

deliberate description, there

is

judgment; and that

problem. Let us recollect what we said about statements involving demonstratio ad sensum. In agreement here with Aquinas, we said that the judgment expressed in
eating the liver" did not differ as regards its own content from one expressible as "a cat eats liver"; the references to a particular cat and a particular bit
is

my

"The cat

of liver and a particular time

judgment

is

made

in

because the a particular sensory context, of

in

come

seeing a cat eat liver. The relation of the judgment to the sensory context was what Aquinas called "conversio ad phantasmata"; it cannot be mere simultaneity, but we were no better able than he to specify

what the relation must be. Now when I make this judgment about the cat and the liver, it is not the cat and the liver that are parts of my sensory experience, but my seeing the cat and the liver. 1 The concept of cat eating liver is exercised,
1 must here make a protest against the view that there is something wrong about counting sight (and perhaps hearing) as sensation that the term "sensation" properly applies only to feelings of eyestrain (say) and not to seeing. It so happens that modern English has no native, Germanic, word covering all the senses alike; "sensation", originally a philosopher's word, is all that we have. But this is a
1

trivial accident of linguistic history; in Latin, in

Greek, in German,

DESCRIBING ONE'S

OWN SENSATIONS

123

then, in a content of sensation; my judgment is a judgment about the cat and the liver because I see

them. If we believed in 'inner-sense', we might say in parallel fashion: "In order that I should say "this pain waxes and wanes regularly", it is not enough that I should think of pain regularly waxing and waning and the pain be there; the pain must fall before my mind's
eye,

m\ 'inner sense', as the cat and the liver fall before

my bodily eye." But this is just mythology. All I need in order to judge about the pain is that my thinking of pain-waxing-and-waning should stand in the conversio-ad-phantasmata relation (whatever that may be) to the pain itself not to some sensation of 'inner

whereby I am cognizant of the pain. Similarly, suppose I momentarily think I have lost my keys, ferret for them in my pocket, and then report: "How odd! A pang of panic came after my fingers had actually touched the keys!" The bits of my senseexperience to which there is now conrersio are the feeling of the keys and the feeling of panic; there is no 'inner' sensation in which I see or feel the panic and
sense',

discriminate

its

panicky quality
itself.

there

is

only the

feeling of panic

in Polish there is a current native word that applies naturally to the senses and is not a borrowing by the layman from the discourse of philosophers. Nowadays in vulgar use "sensation" has acquired overtones (as in "I've got a peculiar sensation") that make it sound wrong as applied to seeing and hearing; but this no more shows that philosophers are wrong to use it in the old way for all the senses than one could refute scholastic talk about essence by saying that 'properly', 'in ordinary language', "essence" means essence of peppermint. (This is meant only as a defence of calling sight a kind of

and
all

sensation; I hold
sense-data.)

no

brief for philosophical talk

about sensing and

28

SENSATIONS AND PICTURES

say, looking at the Moon): ''This looks a yellow disc one foot across; ergo, this is a* yellow disc one foot across." If we ask what "this" stands for both times, we are already beginning to get off the
like

of Epicurus, LET

US

NOW consider the utterance (of a follower

track.
is

Although

in

RUSSELL

(3)

we

are told that "this"

logically a proper name and ordinary 'proper names' are not, in reality "this" is not a name at all. (The

whole point of a name is that we can use it to talk about the thing named in its absence; contrast Swift's learned men, who each carried a pedlar's pack of the things they needed to talk about.) What gives the
the sensation that the speaker has his seeing (as it were) a yellow disc a foot across; but "this" is not the name of the sensation, nor is it the name of a 'private' object of which the speaker is cognizant in having the sensation. The speaker has the thought of a yellow disc a foot across; the concepts exercised in this thought are primarily applied to physical objects; and what gives his exercise of these concepts a particular application is a certain bit of his sense-experience of his visual sensations. This last sentence would serve as an account both of his judgment in "This looks like a yellow disc a foot across", and of his judgment in "This is a yellow disc a foot across". Where, then, does the difference between them lie? Both the thought and the conversio ad phantasmata seem to be the same. Many philosophers have maintained that "yellow
its

word "this"

use

is

124

SENSATIONS AND PICTURES

disc a foot acrossi'

125

is understandable primarily as a of a 'private' object, a 'sense-datum'; "a description foot across" must of course then be taken as standing for a 'visual magnitude', not for a relation to the

Standard Yard. The judgment expressed in "this looks like a yellow disc a foot across" would be the primary one, at least in theory; it would simply apply to the
'private? object, given in

my

sensation, concepts of

qualities that

it

ment expressed
would
effect that

possesses. The seemingly simpler judgin "this is a yellow disc a foot across"

really be

more complex;

it

would be to the

my

'sense-datum' stood in a relation

to

object; my Epicurean would moreover be rashly predicating of the physical object what he had originally predicated of his sense-datum. What relation the relation R might be would be a matter for philosophical enquiry (a matter on which many volumes have in fact been written).

some physical

view is the reverse of this: that "this looks a yellow disc a foot across" is not only a more complex expression, but also the expression of a more developed sort of judgment, than "this is a yellow disc a foot across". In the former case, the concepts being
like

My own

exercised,

which we acquire in their application to physical things around us, have a transferred or

analogical use, serving here to describe the sensation itself. There is no 'sense-datum' language for what I
see,

and no reason to believe that one could be con-

we can describe our visual sensations only by describing how it looks as though things were applying in this description certain of the concepts we use
structed;
in describing physical things, viz. concepts of shape, size, spatial relations, and colours. I believe that one

reason for assuming a distinct realm of sense-data

126

MENTAL ACTS

is as follows: "What I actually isee is completely describable in terms of shape, size, spatial relations, and colours; physical objects are not completely describ-

what I actually see is never a physical object." (This argument is pretty plainly to be seen in Berkeley's Three Dialogues.) But this 'syllogism' is a mere paralogism. The word "deable in those terms; ergo,
scribable" might have

warned us to look

fof-

buried

when the

oratio obliqua; and the fallacy ceases to be plausible oratio obliqua is unburied. From the restated

premises: "I can completely answer questions as to how the things I see look, in terms of shape, size,

and colours; I cannot completely answer questions as to what physical objects are like, in those terms": it does not even seem to follow that none
space-relations

of the things I see are physical objects, so that they must belong to some other realm. The description of

the look of things, in which I use words for shapes, colours, etc., is not a description of queer objects that really have such shapes and colours, but only of the look of things, of my sensations. Not that, when I look
at the

a sensation, a state of my mind, round, yellow, and a foot across; but: I am having sensations like seeing a (real) object which is round, yellow, and a foot across. And the only way to

Moon, there
is

is

which

describe these sensations

is to mention physical proof way; the analogy is irreducible perties in this sort there is nothing to the sensation except what it represents something as being.

transferred use of language in describing sensations may be compared to our way of describing might think that pictures by what they represent.

Our

We

"That

is

a man, and

picture, really

meant

that is his eye", when said of a something like "Somebody drew

SENSATIONS AND PICTURES

this

127

with the intention that that should represent a that his eye"; in fact, however, a young child learns the use of pictures and learns the transferred use of words whereby "man" and "eye" are applied to bits of pictures, long before it has any power of formulating such a complex concept as intention to represent. (It is one of the big differences between men and brutes that brutes take no notice of pictures as such.) The child, moreover, is not at all likely to get mixed up between "eye" meaning an eye and "eye" meaning a drawing of an eye. The use of shape and

man and

colour words to describe sensations is like their use to describe pictures. The difference is that we can describe the picture as a physical object, not as a picture, e.g. by saying it is a set of pencil lines with certain

geometrical attributes; whereas we cannot replace the mention of colour in describing a sensation, by mention of some inherent quality that represents the colour (as various black-and-white designs represent colours in printed books about heraldry). It seems as though we could say nothing about what visual sensations are like on their own account (formaliter, as Descartes would say); only about the attributes they represent things as having (attributes that Descartes would say were in the sensation obiective). A sensation is like a picture in no medium; and isn't this
impossible?

But we can

formaliter. physical time-scale (in a way that thoughts have not). It always makes sense to say of a sensation, or of any bit of sensory experience, like a feeling of fear, that it

A sensation,

after all say something about sensations for instance, has a place in the

lasted continuously all the while that the hands of a watch moved over a certain space; sensations thus
i*

128

MENTAL ACTS

have their being in physical time in the time of local motion of bodies. It is not merely that in our sensations there is a representation of physical events as going on in time and occurring one after the other; sensations themselves take up time and occur at different times. If I have sensations of (a flash of red followed by a
flash of blue), the sensation of the blue flash really,

formaliter, follows that of the red flash;

and hearing

a prolonged note is really, formaliter, a prolonged hearing of a note. Again, just as the pictures of parts of a thing are parts of the picture of the thing, so it might well be held that the seeing of a complex whole is itself really

complex and composed of parts. Suppose that part of what I am seeing is obscured by an opaque object that comes between, may we not reasonably say that part of my seeing goes on (the seeing of the still unobscured
parts)

when another part

ceases?

Or

again, consider

double vision. Sense-datum philosophers have argued that there must be two sense-data in this case; may we not at any rate say that there is something, my seeing, that splits into two when I push my eyeball, and afterwards fuses together again? It does seem to to be reasonable to say this sort of thing. (See also

MCTAGGART, pp. 96-7.) And if so, sensations would be formaliter, really, composed of parts, and really capable (as in double vision) of dividing and then fusing again,

But though time-determinations (and

if

division into

right) can really be ascribed to sensaparts, tion we are not just saying that it represents physical entities as having such attributes nevertheless we are not saying the same thing when ascribing

am

them

to sensation as

physical world;

when we apply them in even here we have an analogous

the
ex-

SENSATIONS AND PICTURES

tension of a concept.

129

As

I said before, the great thing

about such analogous extensions is to know how far the analogy can be carried to know when to stop. And, although there is much more to say on the problems raised in this work, I

am

going to stop here.

APPENDIX
HISTORICAL NOTE ON 11: AQUINAS AND ABSTRACTIONISM*
have been a good deal surmy using arguments adapted from to refute abstractionism. Aquinas is very Aquinas often regarded as an abstractionist, and many of his professed followers are abstractionists; and of course he does use the term "abstractio" for the process of forming concepts. All the same it can be decisively shown that in his maturest work, the Summa Theologica, his views are opposed to what I have called

SOME prised

PEOPLE may
at

abstractionism.

In accepting the comparison whereby the intellectus agens, the mind's concept-forming power, is likened to a light that enables the mind's eye to see the intelligible features of things, as the bodily eye sees colours, Aquinas is careful to add that this comparison goes on
if we suppose that colours are generated the light that the light is not just revealby kindling ing colours that already existed in the dark (la q. 79 art. 3 ad 2 um). Furthermore he says that when we frame a judgment expressed in words, our use of concepts is to be compared, not to seeing something, but rather to forming a visual image of something we are not now seeing, or even never have seen (la q. 85 art. 2 ad 8 um). So he expresses anti-abstractionist views both on the formation and on the exercise of

all

fours only

concepts.
ISO

HISTORICAL NOTE

131

Again, an abstractionist, as we say, cannot allow that we possess 'proper' concepts of the various kinds of material substance in our environment, and it is arguable that he ought to reject the term "substance"
as nonsensical.

But

it is

theory of knowledge that

a main thesis in Aquinas's what our understanding

grasps primarily and most readily is the specific nature (quod ^uid est) of material substances, in spite of his
also holding that the senses are in no way cognizant of this nature. In fact, he greatly exaggerates the ease

and certainty of this knowledge; as when he says that the term "stone" signifies the nature of stone as it is
in itself, since it expresses the definition of stone,

by

which we know what stone is (la q. 13 art. 8 ad 2 um). His soi-disant followers who adopt abstractionism have been obliged to reject his epistemology on this cardinal point; how they can think that this departure is not fundamental I have never understood.

BIBLIOGRAPHY
AQUINAS, St
T.,
4

CHURCH,

A.,

Summa Theologica On Carnap's Analysis

of Statements of

Assertion

and Belief, Analysis, Vol. 10, 1950. Reprinted in Philosophy and Analysis, ed. M.
S.,

Macdonald, Blackwell, 1954

FREUD,

Basic Writings, The Modern Library, York, 1938

G.,

New

HUMPHREY,
JAMES, W.,
1901

Thinking, Methuen, 1951

The Principles of Psychology, Maemillan,

J.

MCTAGGART,

E.,

The Nature of

Existence, Vol.

II,

Cambridge, 1927
PRICE, H. H., 1953

Thinking and Experience, Hutchinson,

(1)

QUINE,
(2)
(3)

W.

v. O.,

Methods of Logic, Routledge and

Kegan

Paul, 1951

Mathematical Logic, Harvard, 1951 From a Logical Point of View, Harvard, 1953
K., 'The Name Relation and the Logical Antinomies', Journal of Symbolic Logic, September

REACH,

1938

RUSSELL,

B.,

(1)

The Problems of Philosophy,

Home

University Library
(2) Philosophical Essays,
(3)

Longmans, 1910 'Knowledge by Acquaintance and Knowledge by

Description',

Aristotelian

Society,

Proceedings,

1910-11
Inquiry into Meaning and Truth, George Allen and Unwin, 1940 RYLE, G., The Concept of Mind, Hutchinson, 1949
(4)

An

WITTGENSTEIN,
well,

L.,

Philosophical Investigations, Black-

1953
132

INDEX
Abilities,^?ee Capacities

Abstract counting, 30f. numbers, 31 units, 27-30

Abstracting, abstraction, 18-44, 76-8, 112 from, 18, 28f., 32f., 37 Abstractionism, 4, 18-44, 78,
112f.

Bedeutung, 3 Begriff, 14, 28n.

Begriffsausdruck, -wort, 69 Behaviour, 4-6, 112, 114f., 121f. Behaviourism, 3, 112
Belief, 8f.

Acquaintance, 45, 47f., 61 f. Acts, mental, description of,

75f., 79f.,

Berkeley, G., 108n., 126 Blind (men born), 35f. Brain, 19, 111 Bridge, 13
Brutes, see Animals
Capacities, 12-15, 25f., 31, 36, 44, 53f., 78f.; see also Con-

1,

83, 99,

121f.,

125-7
existence of, 2f.

privacy

of, 3f.
9f., 79f.,

cepts
85,

reports of, 1-3,

Carnap, R., 91
Categoricals, see Hypotheticals Chess, 3, 13-15 Church, A., 88f. Cogito, Cartesian, 45n., 47, 117-

87
Aeusserung, 121f. Analogy, see Concepts irreducible, 78, 126

Animals, 17, 42, 109, 113f., 127 Aphasia, 13, 44 Apprehension, simple, 14 Aquinas, St T., 3, 40, 44n., 65,
72, 74, 104, 116f., 122, 130f.
Aristotle,

21 Colour, see Concepts

-blindness, 108f.

chromatic, 37-9 Concatenation of

82f., 94f.

expressions,

7O

Assertion sign, 97
Attention, 34f., 64 discriminative, 18, 21, 28, 32,
37, 44 to one's

Concepts, Ideas

see

Abstractionism,

own

experiences, 4,

analogical (extension of), 20f., 75-9, 101, 124-9 and acts of judgment, 7f., 11-

21
Attributes, 38f.

16

and

linguistic capacities, 12f.,

Augustine, St, 108

16f., 43f.

Automata,

114f.

arithmetical, 27-32
collapse of, 113-16

Awareness, see Acquaintance

133

134

INDEX
Descriptions, analogical transference of, 75-9 deliberate, of mental acts,
121f.

Concepts, colour-, 33-40, 108f. experiments on formation of, 17-19, 42-4

falsify reality?,

39-42

interrelation of, 43, 112f.

logical,

22-7

non-linguistic exercise of, 13, 17, 26, 42-4 number of, exercised in judg-

and metaphorical, 76 quotations as, 82f., 86 Russell's theory of (definite), 62, 66-8; see also Names, proper
literal

ment, 15-27
psychological,
) (

Disembodied
7f.,

existence,

1 1 1 f. ,

20-2, 75-9,

115f.,

117

107-17, 121-9
recognitional capacities, 17, 21, 25, 33, 40, 42, 44, 47, 77,

Dispositions, 5, 8, 14 Duration of judgments, 104-6

of sensuous experiences, 104,

127f.

113
relational, 32f.

sameness of, 14, 23, 36, 78 sensory, 124-9

alleged primacy of, 20, 41f. subjective nature of, 13f. Conditionals (counterfactual,

Ego, see Self Emotion-blindness, 109f. Emotions, 108f., 112, 114, 128 Equiform expressions, 86 9, 97f.
Essences, 70f., 123n., 131

subjunctive, unfulfilled),
15; see also

6f.,

Events, mental, sec Acts, mental

Hypothetical

E volution,

17

Connotation, 66, 68 Consciousness, 110, 116f. Constituents of judgment, 46,

59f.

Exponentiation, 30 Expression of thought in words, 4, 8f., 11-14, 103-6

Frcge, G.,
3, 14, 28n.,

Content, 1, 61, 63f., 71f., 104, 122 Conversio ad phantasmata, 05, 72, 74, 122f., 124

39, 58, 69,

97
Freud,
S.,

110

Counting, 30f.

Hume, I)., 108 Humphrey, G.,

Definition,

as exercise of con-

4-7; Hypotheticals, Conditionals

12n., 19, 42-4 also see

cept, 44 ostensive, 23, 34, 67, 96

Hysterics, 115f.
I,

Demonstratio ad intellectum, 74, 118, 120

see

Cogito,

Descartes,

Pro-

nouns, Self
Idealism, 60 Ideas, 53-60, 64, 98f., 104f.
Identification, 69f., 71, 73f.

ad sensum, 74, 121f. Demonstratives, 62, 63-5, 124 Denotation, 68

127

73f.,

Images,

Descartes, R., 45n., 46f., 116-21,

107f.,

mental, 130

72,

99,

104,

Inference-licence, -ticket, 7

INDEX
Intensional entities, 95,*98 isomorphism, 91 Introspection, see Sense, inner

135

Meinong, A., 48 Mention, see Use Mind-body problem, 11 If., 116f.

James, William, 17, 23, 119 Joyce, James, 5

Name, concept of, 43 Names, proper, 58f., 6 If., 66-73,

124 Quine on, 66-8
Russell on, 45f., 61-3, 66

Judgment

(act of), 7-10 about particulars, 45-7, 61-6 about relations, 32f., 50-8, 99101, ?05 and deliberate description, 122 of identity, 66-74

Negation, concept 104

-operator, 52f.

of, 12f.,

22-6,

tensed, 65f., 102f., 122 theories of:

analogical,
92f.,

75-7, 79f.,

87,

95

Non-extensional operators, 98-101 Nonsense name, 43 reports of, 10, 85

52f.,

combined, 98-101 Ockham's, 102f.

revised Russelliun, 52-60
Russell's,

Object of mental acts, private, 3, 124-5

Object-expression, If. Objective entities, 14,

If.

45-52

time-relations of, 9, 104-6

48f.,

98

Knowing how

) (

knowing

that,

Objectives, see Meinong Oratio obliqua, 9f., 11, 49, 62, 75-

16, 35n., 43, 77f., 101,

129

80
buried, 126

Language
-game,
3,

and concept -forma68

tion, 13, 43f.

quantification into, 92-5 replaceable by oratio recta*?,

87-93
Oratio recta, 80-95, 99f.

inner or mental, 102-6 structure ) ( medium of, 103 Leacock, S., 19

quantification into, 93-5 Other, concept of, 26f., 42

Learning, 12, 74, 127

Leibniz's

18f., 27, 30f., 34,

Law,

84, 101 n.

Pain, 3f., 108, 114f., 120, 121-3 Particular, see Judgment about
particulars

Liar paradox, 98 Locke, J., 20, 21,

36f.,

68

Pavlov, I. P., 24 Philosophy, history

116f., 120, 130f. jargon of, 57f.

of,

19-21,

Macdonald, P., 70 McTaggart, J. E., 107, 128 Meaning, see Sense

Plato, 46
Price,

and private experience,! 12,1 14 and reference, 8f. and use, 12 and verification, 8
Russell's view of, 45, 47f.

H. H.,

19, 22, 25, 82-5,

44

Pronouns, demonstrative, see Demonstratives first-person, 12, 45, 117-21 Psychical research, 112

136
Quine,

INDEX
W. van
) (

O., 66-8, 82-4,

90, 92f.

Quotation
81

quoted expression,
81-7

Quotes, use

of,

Sense ofrelations, 5 If., 54 Sense-data, 45f., 124-8 Sinn, 8, 58 private, 3 ungeradc, 58

Skills, see

cancelling of, 96f.

Capacities
L.,

Smoke, K.
Reach, K., 85 Recurrent characteristics, 33, 113 Reference = Frege's Bedeutung, 3, 49, 120
of judgment to things, 63-5, 72f., 123, 124

42-4

Soul, spirit, 121

Substance, 19f., 71, 131 Substitution salva verttate, 83f.,

9O-2

Synonymy,
82f.

90f.

Syntactical parts of expressions,

Reports of mental mental

acts, see Acts,

Thought,
sion,

see

Duration, Expres-

Russell, Bertrand, 14, 22, 24f.,

Ideas,

Judgment,

45-60, 61-3, 66, 124 Ryle, G., 3-8, 14f.

Scholastic maxims:

Language
Translation, 89-92

ab

esse

ad posse

valet conse-

Truth, 96-8 does not copy reality, 41 semantic theory of, 98

Universals, 45, 47f., 59f.

quentia, 15 eadem est scientia oppositorum, 25, 52 nihil dat quod non habet, 17
nihil in intellectu nisi prius in sensu, 19, 20, 78
agit in quantum est in actu, 5f. relativa sunt simul natura et
intellectu,

Usage,

12f.,

16
12, 16, 69, 76,

Use of expressions,

omne agens

33

Scholasticism, decadent, 17, 19f.,

130f.

117-20 ceremonial or ritual, 28, 84 ) ( mention, 85f., 96 performatory, 97 redundant, of "I", 120 of "true", 96 transferred, 126
Utterances, 64, 75f., 78, 99-101,
105f.

Seeing, 111-13, 122f., 124-8 Self, 12, 45n., 47, 117-21

Aeusserungen, 121 f.

Semi-hypotheticals, 4, 6 Sensation, 35-7, 41, 111-17, 125-8; see also Pain

Variables, 50, 93-5

Verb, psychological,
Verbalization, 43f.
Verification, see

1 , 9,

75, 1 12f.

and and
Sense

pictures, 126 sight, 35n., 122n.

Meaning

Frege's Sinn,

8f.,

58

inner

21-5, (introspective), 38, 77, 107-11, 113, 117,

119, 123

William of Ockham, 102f.

Wittgenstein, L.,
101, 104, 121f.
8f.,

18,

72f.,