symposium: intentionality and intensionality

Symposium: Intentionality and IntensionalityAuthor(s): William Kneale and A. N. PriorReviewed work(s):Source: Proceedings of the Aristotelian Society, Supplementary Volumes, Vol. 42 (1968), pp.73-106Published by: Wiley-Blackwell on behalf of The Aristotelian SocietyStable URL: http://www.jstor.org/stable/4106586 .Accessed: 06/10/2012 00:00

Your use of the JSTOR archive indicates your acceptance of the Terms & Conditions of Use, available at .http://www.jstor.org/page/info/about/policies/terms.jsp

.JSTOR is a not-for-profit service that helps scholars, researchers, and students discover, use, and build upon a wide range ofcontent in a trusted digital archive. We use information technology and tools to increase productivity and facilitate new formsof scholarship. For more information about JSTOR, please contact [email protected].

.

Wiley-Blackwell and The Aristotelian Society are collaborating with JSTOR to digitize, preserve and extendaccess to Proceedings of the Aristotelian Society, Supplementary Volumes.

http://www.jstor.org

http://www.jstor.org/action/showPublisher?publisherCode=black

http://www.jstor.org/action/showPublisher?publisherCode=aristotelian

http://www.jstor.org/stable/4106586?origin=JSTOR-pdf

http://www.jstor.org/page/info/about/policies/terms.jsp

INTENTIONALITY AND INTENSIONALITY

By WILLIAM KNEALE AND A. N. PRIOR

I-William Kneale

?1 IN this paper I shall argue that entertainment of propositions is essential to thinking in that wide sense of the word in which thinking includes all mental life except bare sentience. I hope this is not a difficult thesis to understand or accept; but in order to explain why it is interesting, and perhaps even important, I must plunge into the debates connected with the two technical terms which appear in my title.

It has been realised for some years that there is a close connexion between the notions signified by these terms, and it has even been supposed by some good philosophers that 'intensional' is no more than a modern corruption of 'intentional' , to be tolerated, if at all, only because the technical use of 'intentional' in philosophical psychology may possibly be confused with its common use in moral and legal contexts, where it is closely related to 'voluntary' and 'deliberate'. But in fact the phonetic resemblance of the words is mere accident. For they are quite independent in origin, and in order to make my point clearly I shall have to consider them separately.

?2 In mediaeval philosophy the word idea was reserved for the

sense of 'exemplar in the mind of God ', but the sense which idea and its derivatives came to have in seventeenth century philosophy was expressed by a technical use of intentio which occurs first in a Latin translation of Avicenna's account of Aristotle's theory of knowledge. St. Thomas, for example, says in his Summa Contra Gentiles, IV 11, that whenever the mind

74 WILLIAM KNEALE

thinks anything it conceives in itself a certain likeness of the thing thought. This is an intentio, but because it may be signified by an outer word, it is sometimes called an inner word. Clearly it is to be distinguished both from the thing of which it is the intentio and from the mind to which it belongs. For the study of the intentio is not the same as the study of the thing, and the being of the intentio, unlike the being of the mind, consists precisely in its being thought (esse intentionis intellectae in ipso intelligi consistit).

Because an intentio is supposed to be a certain sort of representation, the adjective intentionalis is used occasionally in the wide sense of 'appropriate to a representation', as when St. Thomas tells us in his Summa Theologiae, [ 56 (2), that the form of colour has natural being in a coloured wall but only intentio na

being (esse intentionale) in the transmitting medium. The context shows that his purpose in saying this is to explain by an

analogy how according to Aristotle's theory of knowledge a soul may have in itself the form of something else without being quite literally of the same form as the other thing. But intentional

being is usually equated with being as an object of though (esse objective), and this is the usage which interested Brentano when he revived the word 'intentional' in his Psychologie of 1874. So far as I know, it occurs only twice in that work (ed. Kraus 1924, I p.124, II p.8), and each time Brentano makes clear that to say something has intentional (or mental) inexistence is

simply to say that someone has it in his mind as an object of

thought, i.e. thinks of it. Since he believed firmly that it is

possible to think of what does not really exist in any proper sense of that word and that the explanation, or rather the

description, of this situation belongs to philosophic psychology, he was much distressed in his later years by Meinong's presentation of the theory of objects as an autonomous science, prior not only to psychology but also to metaphysics. It must be admitted, however, that there is a germ of paradox already in such phrases as esse intentionale and 'existence in thought '.

INTENTIONALITY AND INTENSIONALITY 75

By tolerating them we seem to commit ourselves to the odd- sounding thesis that things can exist somehow in thought which do not really exist at all, and many of the surprising features of Meinong's doctrine are no more than developments of this, as when for example, after telling us that the pure object is beyond both being and not-being, he goes on to say that it has beyond- being (Aussersein).

Because Brentano's use of the word 'intentional' occurred in his presentation of the doctrine that mental events of all kinds are directed on objects, it came to be thought that intentional

objects (i.e. objects in the old strict sense) were so called as being the objects of an intention or direction of the mind. And once that erroneous belief was established it was a small step to using the adjective 'intentional' of mental events with the sense of 'directed on an object'. This development is to be found in the writings of Husserl and many of his followers. I think it is due to a misunderstanding of Brentano's quotation from mediaeval

Latin; but it may nevertheless be useful, and I do not wish for a moment to suggest that it should be abandoned, though I shall not adopt it myself.

More recently some philosophers have tried to lay down rules for recognition of what they call intentional sentences or intentional verbs, by which apparently they mean any sentences or verbs whose meaning involves the notion of the direction of the mind upon an object. It is maintained, for example, that intentional verbs include not only 'believe', 'wish', and 'wonder', which have often been said to signify mental events, but also ' ridicule ', ' worship ', and ' hunt ', which signify pieces of overt behaviour guided by thought. Apart from the influence of a general fashion for philosophizing in linguistic style, there

are, I think, two reasons for this development. The first is a

very proper wish to avoid any phraseology that may lead into the paradox noticed above. And the second is a wish to discuss

intentionality of the Husserlian kind (i.e. what Brentano called

76 WILLIAM KNEALE

the directedness of mental events) in an idiom similar to that often used by formal logicians for intensionality considered as a peculiar feature of certain expressions. This may or may not be a good reason. It is good if it is connected with a plan for discussing the relation between the two notions whose names are accidentally similar. But it is bad if it is based on a confusion between them or on an unwillingness to try to work out a general account of the way in which thinking is concerned with objects. In practice, however, philosophers who have written in this way have commonly tried to characterize verbal phrases of an intentional kind by the peculiarities that may be found in their grammatical objects, and so we are brought back again to a place near our starting point.

A substantive phrase that is used to indicate an object of thought may also in fact refer to an ordinary thing in an ordinary way, as when for example someone pointing at the ruins of Bolton Priory says 'Wordsworth wrote a poem about this place '. But a phrase which successfully indicates an object of thought may nevertheless exhibit one or more of the following three features. (1) The phrase may fail to refer to any real thing, as when for example we say 'According to Greek mythology earthquakes are caused by Poseidon, the sea god'. Sometimes indeed there may be no possibility of the existence of anything answering to the phrase, as when we say ' It is useless to search for the largest prime number '. This pecularity of some talk about objects of thought was stressed by Brentano. Recognition of it is implicit in the mediaeval distinction between intentional being and natural being. (2) The phrase may be essentially indeterminate or incomplete, as in the statement 'I was dying for a cup of tea '. If anyone hearing this asked ' Which was the cup of tea you wanted so much?' or 'Was the cup of tea you wanted China, India, or Ceylon tea?', he would show that he did not understand how the phrase 'a cup of tea' was used in such a context. What was wanted was neither a certain cup of tea nor yet a cup of tea of a certain kind but just a cup of tea.


In recent times the possibility of indeterminate objects of thought has been maintained by Meinong, but the point was noticed already by the mediaeval logicians Burleigh and Buridan in a sophism about a man who after promising to deliver a horse argues that there is no particular horse he has promised to deliver. (3) The phrase may occur in such a way that its replacement by another name or description of the same thing would result in a sentence expressing a proposition of different truth-value. Thus in the sentence ' Oedipus wished to kill the haughty stranger who barred his way' the grammatical object cannot be replaced by ' his father ', although the latter phrase is supposed to apply to the same man. The interest which philosophers have shown in such complications during recent times is due to Frege's discussion of indirect reference in his paper Ober Sinn und Bedeutung, but it is possible that Eubulides wished to illustrate the point by his paradox which is sometimes called the Electra and sometimes the Hooded Man.

Since each of the three peculiarities may be lacking in some types of sentences about the direction of human thought on an object, none of them can be a necessary condition for intentionality in the sense to be defined. But it is important to notice also that none of them can be, as some philosophers have supposed, a sufficient condition, since each may occur equally well in a sentence which is not concerned with the direction of human thought. This is shown by the following examples in which I have used once more the interesting phrases of my earlier examples: (1) 'If any number n were the largest prime number, n! +-1, being a greater number, would have to be divisible without remainder by n or some number less than n; but that would be impossible'; (2) 'What you need is a cup of tea'; (3) 'It is a contingent truth that the haughty stranger who barred the way to Oedipus was his father '. The moral of all this is that propositions about the direction of thought on objects must share with modal propositions (i.e. propositions about necessity or possibility) a feature in virtue F

WILLIAM KNEALE

of which their expressions may exhibit the peculiarities we have just noticed. That feature is intensionality. But the explanation and justification of the moral requires a new start.

?3 In his paper Uber Sinn und Bedeutung Frege maintained that

a phrase occurring as a nominal clause (i.e. a clause with the syntactical r81e of a noun, e.g. 'the earth is flat' in 'Some people believe the earth is flat') names or designates the sense which it would express if it occurred by itself as a declarative sentence. This doctrine is presented as part of general theory in which declarative sentences are treated as designations of truth-values. For according to Frege 'The earth is round' and '2 + 2 = 4' both designate one and the same object, namely the True, though the thoughts they express are, of course, different; and the relation between them is to be conceived as similar to that between 'the husband of Xanthippe' and 'the teacher of Plato ', which both designate the one man Socrates, though in different ways, i.e. by expressing different senses. I think we should refuse to allow that propositional signs occurring as declarative sentences designate anything at all, but should nevertheless retain Frege's distinction between expression and

designation. For I want to say, as he does, that a propositional sign occurring as a declarative sentence expresses a proposition, or complete thought capable of being true or false, and that if

anyone has occasion to talk about that same proposition it may be designated for this purpose by a nominal clause corresponding to the original sentence. In order to provide an example that seems to fit Frege's explanation exactly I have taken advantage of the English idiom whereby indirect speech may be employed after an appropriate verb without use of' that ' as a nominalizing prefix. But when such a device as 'that' is available, we may say very simply that its application to an expression of a proposition produces a designation of the same proposition. This, I believe, is the best pattern to follow in any attempt to develop a


precise symbolism for discussion of the topics Frege had in mind when he wrote about nominal clauses. I recognise, however, that many philosophers from Antisthenes onward have been profoundly suspicious of the linguistic move which

grammarians now call nominalization. Often they show their fears by saying that it is unnecessary reification and dangerous metaphysics to admit abstract entities such as propositions and propositional functions. But since they are quite content to use phrases which express propositions or propositional functions, it seems that the abstraction which they dislike must be designation as opposed to expression. In short their nominalism consists precisely in a refusal to nominalize.

In natural languages we have various idioms by which we can avoid nominalization in certain contexts. Instead of 'It is necessary that 2 + 2 = 4 ' we can say, if we wish, ' Necessarily 2 + 2 = 4 '. And instead of ' Socrates taught that the soul is immortal' we can say 'According to Socrates the soul is immortal '. With these possibilities in mind Mr Prior once suggested (P.A.S. Sup. Vol. XXXVII, 1963, p.116) that instead of thinking of 'that' as a nominalizing prefix we might treat it as a terminal sign of certain functional expressions like ' Socrates taught that . . .', i.e. expressions which can be applied to propositional expressions in order to produce more complicated propositional expressions. It seems to me that this analysis is permissible in many contexts and that it can make for simplicity of philosophical formulations, as I shall show later. In these days of ecumenical conciliation I am therefore prepared to say that, if I were Pope of the antiqui, I should try to get the good will of the moderni by offering them the privilege of a special Priorese rite in which they would indeed be required to say 'that' on solemn feasts but would be allowed to make a longer pause after the word than before it. I should do so, however, in the hope that acceptance of this concession would in the end frustrate their schismatic tendencies. For I should reassure myself about the abiding strength of the old doctrine by

80 WILLIAM KNEALE

reflecting that a nominal clause formed by means of the prefix 'that' is replaceable as a whole by other names or descriptions of the proposition it designates. Instead of 'Socrates taught that the soul is immortal' we may say, for example, 'Socrates

taught the doctrine which Epicurus most wished to reject'. And instead of' It is true that the square on the hypotenuse of a

right-angled triangle equals the sum of the squares on the other two sides' we may say 'Pythagoras' theorem is true'. The second example is especially interesting because anyone who insists on Prior's analysis in this case will have to treat the phrase 'it is true that' as a unitary sign with the sense of an identity operator and will thereby commit himself to the so-called

redundancy theory of truth, missing the important fact that 'is true ' works as inverse to ' that '.

In his Grundgesetze Frege made no effort to develop his theory of nominal clauses any further, because he thought the only complexity to be found in mathematical propositions was truth-functional. In the first edition of Principia Mathematica a similar attitude was taken by Whitehead and Russell, though their interest in semantic paradoxes led them to say something there about indirect speech. Whatever view we may take of modal statements and reports of mental events, it is clear that, if there are any propositions at all, there must be some which are not about other propositions in that strict sense for the rendering of which nominal clauses seem especially appropriate; and since these form a very important class, it will be useful to give them a name. They might perhaps be called first-order propositions, but that phrase has been pre-empted by Russell for another

purpose, and I shall therefore try to suggest what I have in mind by calling them basic. It will be seen that according to my usage a basic proposition may well be a truth-function of other basic propositions. In fact a basic proposition may be defined as a proposition such that, if it is a function of any other proposition, it is a truth-function of that other. Following Prior's plan for avoidance (or concealment) of nominalization, we may express this definition in logical symbols as follows:


D 1. Basic (P) = (_)(q)(r) {(P// b(q)) D [(qE r) : (b(q) =b(r))]}. Here 'Basic (P)' is to be understood as short for 'The proposition that-P is basic ', '. .. // - ' as meaning 'The proposition that ... is the same as the proposition that ', and '0b(q)' as exhibiting a pattern exemplified by any proposition which is in any logical sense a function of another proposition.

By analogy a basic function may be said to be one satisfying the definition

D2. Basic (Flc) H (b)(g)(h)(x) {(Fx// Jb(gx)) D [(gx= hx) D

( f(g x) - (hx)]} But it should be noticed that according to this definition a basic function need not have basic propositions for its values. Thus the proposition expressed by 'Socrates said that the soul is immortal ' is not basic according to D1, but the function expressed by 'Socrates said that p' is basic according to D2, because (unlike that expressed by 'Plato said that Socrates said that p ') it cannot be rendered in the form' '1b(gp)' where '0' represents a function other than a truth-function and 'gp' something distinguishable from 'p'. This is important because it allows us to formulate a definition of identity applicable not only to individuals and classes but also to propositions and propositional functions, however they may be designated, namely D3. (X = Y) = (f) [Basic(fz) :

(fX- fY)].

In Dl and D2, which prepare the way for D3, I have used the sign '//' with an explanation which may perhaps arouse suspicion. If it is to be understood in the way I have explained, is there not a circle in my attempt to define identity ? In order to dispose of this objection it is important to make clear that in well-formed formulae the sign '=' can occur only between designations (including of course, propositional designations that, like 'Pythagoras' theorem', are not produced by nominalization of

expressions) for the purpose of indicating identity of reference, whereas our Priorese convention provides for the occurrence of

82 WILLIAM KNEALE

'//' between expressions as a sign for identity of meaning. Since in my view reference is always incidental to meaning and therefore less fundamental in the theory of language, I do not find it surprising that the notion of identity of meaning should be

required for a satisfactory account of identity of reference, though I think that the possibility of distinguishing them as I have done is an uncovenanted bonus from the use of Mr Prior's technique. But there is another difficulty. Sameness of meaning is notoriously an ill-defined concept. Even if we are not inclined to the radical

scepticism of Quine, we must admit with Bernays ('Zur Rolle der

Sprache in erkenntnistheoretischer Hinsicht' in Logic and

Language, Studies dedicated to Professor Rudolf Carnap on the occasion of his Seventieth Birthday) that whether or not we say two expressions mean the same may depend on our interests and in particular on the level of abstraction at which we are working. In physics, for example, two expressions may be described as different versions of the same hypothesis if each can be obtained from the other by means of mathematical transformations, though mathematicians who are professionally interested in the transformations maintain that there is a difference of sense.

Similarly in pure mathematics two expressions which are related

by simple logical transformations may be described as different formulations of the same theorem, though logicians insist that even the difference between 'P' and '-'~ P' is enough to justify them in talking of two propositions. How, then, are we to understand '//' in Dl and D2 ?

At first I thought it might be sufficient to require the sense of Lewis's strict equivalence, which (in spite of its name) is, so to say, the lower limit of propositional identity. For with this interpretation of the sign Dl is already narrow enough to exclude reports of

speech or thought from the class of basic propositions. But Mr Prior has convinced me that strict equivalence is certainly not strict enough for my purpose since with this interpretation of '//' Dl and D2 do not discriminate adequately between basic and modal. On the other hand, for reasons that will become clear


presently, it is important not to put the requirement for propositional identity so high that no sentences of different surface structure can be said to express the same proposition. I confess that I am unable to suggest a simple formulation which seems to me evidently the right one for this context, but I am inclined to think that what is needed is two-way derivability in accordance with principles of entailment which are not peculiar to modal logic.

In his Tractatus Logico-philosophicus, 5.54, Wittgenstein declared that one Satz can occur in another only as a basis of a 'truth-operation' such as negation or conjunction. This is obviously false if a Satz is supposed to be, as Wittgenstein says in 3.12, 'a propositional sign considered in its projective relation to the world'. But in view of the ambiguity of the German word the assertion may perhaps be taken to mean that truth-functional dependence is the only kind of functional relation that can hold between one proposition and another. If, however, all propositions are basic, as this interpretation implies, the right-hand side of Dl may be asserted with any propositional sign we please in place of 'P'. Then, since substitution of 'b(q)' turns the conditional clause into a tautology, we may write quite simply

T1. (q E r) :) ( 0(q) _=-0 (r)),

using free variables to express universality. Similarly from D2 we may obtain

T2. (gx =_ hx) D ( (gx) E- (hx)),

which signifies that all functions are basic. And from this together with D3 we can get

T3. (x = y) ( fx f y),

which is sometimes called the principle of the indiscernibility of identicals. The three formulae are equivalent, but Tl is the simplest and T3 the one most often cited.

In his preface to the second edition of Principia Mathematica (p. xiv) Russell accepted Wittgenstein's contention, with the

84 WILLIAM KNEALE

interpretation suggested above, and went on to assert as a consequence (p. xxxix)

T2'. (x) (gx - hx) D (0(gx) d (h)), which he paraphrased by a statement that all functions of functions are extensional, i.e. indistinguishable from functions of classes. Since that time Wittgenstein's doctrine has commonly been called the Thesis of Extensionality; and by a further development of the terminology which Russell borrowed from Hamilton, those who reject Wittgenstein's doctrine have often been said to plead for intensional logic. In Hamilton's usage 'intension' had nothing at all to do with 'intention', being merely a rather unfortunate substitute for the word 'comprehension', which the Port Royal logicians used for the range of attributes signified by a general term in distinction from the extension or range of examples to which it might be applied truly. Hamilton's new word and his associated doctrine of the inverse variation of intension and extension were both suggested to him by a passing remark of Cajetan that there is an extensive sense in which a genus may be said to have greater collective power than a species, because it contains more members, and an intensive sense in which a species may be said to have a greater collective power than a genus, because it unites its members more closely (Hamilton, Lectures on Logic, vol. I, p.141).

?4 Having worked this bit of history out of my system, I can

now return to my main theme. On the assumption that thinking always involves entertainment of a proposition it is easy to explain the three peculiarities that philosophers have discovered in substantival phrases which are said to stand for objects of thought. If my thesis is correct, any account of a mental act or attitude or of a piece of behaviour which is essentially guided by thought must be concerned with the thinking of a proposition and so be capable of exhibiting some of the features of indirect


speech. Let us apply this consideration to each of the three peculiarities in turn.

(1) When we say 'It is useless to search for the largest prime number ', our remark is to be understood as a short way of saying what might be put more fully by saying 'If anyone were to go through mathematical procedures in the hope that he might thereby find the largest prime number, he would be wasting his time'. But in the lengthier version the phrase ' the largest prime number' occurs inside a passage about a person's hoping that something might happen, and so is not to be understood as though it conveyed an assumption by the speaker that there is a prime number larger than all others. If a witness in a court of law says 'The prisoner claimed to have bought the jewellery as a gift for his wife ', it is not necessarily to be assumed that he believes the prisoner to be married. In some cases, of course, we assume that a witness himself believes the existential proposition pre- supposed by a definite description which occurs in his report of another man's speech, either because we all accept it ourselves without question or because he has said something else which reveals his belief. But it is a well-established rule that a definite description may be used without such commitment in indirect speech and discussion of other men's thoughts. I shall consider later the bearing of this fact on a much debated question about quantification into oblique contexts.

(2) In colloquial English the remark 'I was dying for a cup of tea ' is taken to be equivalent to ' I very much wished there were a cup of tea I was drinking '. In the longer version the existential quantifier can be seen to belong to a nominal clause which designates what Meinong would call the objective of wishing, and that seems clearly to be the reason why it is foolish to suppose that there must be answers to all conceivable questions about the cup of tea the speaker wanted. When a man wishes that there were something of a certain kind, it is not necessary that there should be any real thing of that kind nor even that he should have conceived such a thing in more detail than would be implied by a nominal clause such as we have considered. The situation is very

86 WILLIAM KNEALE

different, of course, if anyone says truly ' There was a cup of tea I very much wished I were drinking', since here the existential

quantifier has for its scope all the rest of his report. And it is

precisely because there can be reports like this that we have to allow not only for non-basic propositions but also for non-basic functions such as that expressed by the phrase 'x was a cup of tea and I very much wished I were drinking x'.

(3) In the sentence 'Oedipus wished to kill the haughty stranger' we already have two verbs, and it is therefore easy to see why replacement of 'the haughty stranger' by 'his father' may involve misrepresentation. The phrase 'to kill the haughty stranger' is a nominal clause used to designate the objective of

Oedipus' wish, and the fact (if it is a fact) that 'the haughty stranger' applies to the father of Oedipus and to him alone is no sufficient reason for attributing unfilial thoughts to Oedipus. But to admit this is to reject the thesis of extensionality in its most widely received form, T3, and so to accept the existence of non-basic propositions.

Each of the criteria by which philosophers have tried to help us to recognize intentional sentences has been shown to lead back to the fact that the sentences under consideration express propositions about propositions. But it may perhaps be argued that more is needed to prove the thesis that thinking (in the wide sense which includes wanting, hoping, fearing etc.) always involves entertainment of a proposition. I do not propose to try to produce a proof by summative induction over all varieties of thinking. For although I believe that by taking account of dispositions as well as activities it would be relatively easy to deal with any apparent counter-examples that might be suggested, I do not think it would be possible to make sure by systematic enumeration that all varieties had been examined. A better course is to consider why philosophers have sometimes thought it plausible to assume that there is a kind of thinking which does not entail entertainment of propositions.


In the first chapter of his De Interpretatione (faithfully translated by Professor Ackrill) Aristotle says: 'Just as some thoughts in the soul are neither true nor false while some are necessarily one or the other, so also with spoken sounds. For falsity and truth have to do with combination and separation. Thus names and verbs by themselves-for instance " man " or " white " when nothing further is added-are like the thoughts that are without combination and separation; for so far they are neither true nor false. A sign of this is that even " goat-stag " signifies something but not, as yet, anything true or false- unless " is " or " is not " is added (either simply or with reference to time).' Here the argument goes from the organization of thought, with which we are supposed to be familiar, to the organization of speech, which we are supposed to learn by noting how its elements correspond to elements of thought. Until

recently almost all philosophers have professed to follow Aristotle's lead. But it now seems clear that their accounts of thinking were determined by their customary analysis of talking, and that

they were misled by a false analogy when they assumed that there must be simple apprehensions corresponding to terms, or even

perhaps to words, as judgments correspond to statements and inferences to arguments. Even before the discoveries of modern

philologists about languages very different from those of the

Indo-European family there was already reason for suspecting a mistake when a mediaeval philosopher said that thought took

place by means of natural signs corresponding to all the various

parts of speech and that the natural nouns of thought were distributed among all the various grammatical cases. If thought has its grammer no less than speech and writing, why should its accidence be like that of Latin rather than like that of Greek, which has no separate ablative case ? And what about word order? Is Greek, Latin, German, or English nearer in this

respect to the internal discourse of the soul? No doubt we must

speak words in order to speak sentences, and we can, if we wish, pronounce a single word all by itself, recognizing various possibilities of continuation; but that is no sufficient reason for supposing

88 WILLIAM KNEALE

that we can think a complete thought which, as Aristotle says, is without either truth or falsity. On the contrary, since words are only material for the building of sentences and there need be no word-for-word correspondence between synonymous sentences of different languages, it is rash to put much faith in any doctrines about thought that depend upon the old story of internal discourse constructed from natural words. It is true that we occasionally use the phrase 'think of' to mean ' choose silently '. But if in the course of a game someone tells me to think of a number in this way, my saying the name of a number to myself in English or some other language will not count as the carrying out of his instruction unless I think that what I name is a number.

Another source for the doctrine of simple apprehension as a mode of thinking is a confusion between the internal objects of sensation and the intentional objects of thought. This is to be found at all stages in the tradition from Aristotle to Brentano, but it was perhaps at its most dangerous in the use of' idea' by seventeenth century philosophers to cover pains and after-images as well as what Frege would call concepts. I think it is a sound linguistic instinct which makes us unhappy when we read in Descartes or Locke that having a pain is a kind of thinking, and it is this instinct I am following when I say, without begging the question at issue by arbitrary stipulation, that in any reasonable use thinking involves entertainment of propositions.

?5 Once this point is conceded it is not difficult to find a solution

for the old puzzle about the existence of objects of thought. Let us consider the true statement ' Newton sought the philosopher's stone '. Since we do not believe that there is any such thing as the philosopher's stone, we do not ordinarily wish to accept as a consequence that there really is something which Newton sought in his study of alchemy. On the other hand when we reflect that it was not in Newton's character to work aimlessly, we feel inclined to say that he certainly sought something. How can we


escape from the apparent contradiction? The situation becomes clearer if we substitute for the original statement a paraphrase which reveals the non-basic character of the proposition under consideration, e.g. 'Newton did chemical experiments in the hope that he might find the philosopher's stone', Here 'the philosopher's stone' occurs in such a way as to show that its existential presupposition is not necessarily shared by the speaker. And it is obvious that we can if we like derive from this the consequence 'For some 0, Newton did chemical experiments in the hope that he might find the b thing'. This is, indeed, exactly what we wish to say when we reflect that Newton did not work aimlessly, and it is acceptable because it commits us to nothing which can possibly be in dispute among persons who have found the original statement intelligible. If anyone believes in the existence of the philosopher's stone, he will think himself entitled to go further and say 'For some x and some 0, Newton did chemical experiments in the hope that he might find the b thing, and x alone is 0b'. This is not a consequence of the original statement according to the strictest interpretation, but it is entailed by the original statement if the speaker has given it to be understood that he himself accepts the existential presupposition of 'the philosopher's stone'.

The difference between the two existential statements is important. In the first we have quantification into an oblique context, but only over attributes. The combination of quantification with use of the definite description notation is peculiar but unobjectionable; the effect is the same as that of quantification within Lesniewski's system called ontology. In the second we have also quantification over individuals, but not into the oblique context. The logical complexity of the conclusion is obscured, however, in this and similar cases by the phraseology we use to express it in ordinary speech. If we believe that there is someone who can properly be called Plato's philosophical teacher, we go straight from 'Plato honoured his philosophical teacher' to 'There is someone whom Plato honoured', without regard for

90 WILLIAM KNEALE

the peculiarity of 'honoured' as an intentional verb. That, I think, is one reason why some logicians have wrongly supposed that it is possible to quantify over individuals into oblique contexts. But when everything is made explicit by means of the blessed nominalizer it can be seen that oblique contexts involve what might perhaps be called quotation of thoughts and that the presuppositions of the referential phrases occurring in such contexts are not to be imputed to the speaker without special warrant. Such warrant is given automatically by demonstratives and other token reflexives (except when they occur, as some of them may, in a purely resumptive rOle), but there is nothing else in the grammar of ordinary language to show when the presuppositions of names and definite descriptions in oblique contexts are accepted by the speaker.

II-A. N. PRIOR

1. I share Professor Kneale's view that it is best to regard "intensionality ", or if you like " non-extensionality ", as a rather general phenomenon of which " intentionality " provides some of the most interesting examples. But I want to make a sharper distinction than Kneale seems to do between the question as to whether there are intensional functions and that as to whether there are intensional objects. I shall argue, or at all events assert, that there are indeed intensional functions as well as extensional ones, but that there are no such objects as "intensions ", and no such objocts as " extensions " either.

2. To say that a function F of propositions, properties, relations, etc. is an extensional one, is to say that it satisfies the appropriate member of the series of formulae which begins as follows :

(i) ) ( (q) : p =q. :3. F(p) - F(q) (ii) ( 0) (0) :) x) (x O- x). z. F(y) = F(0)

(iii) ( 0) ( ): (x) (y) (yxy = Oxy). 3. F(y) - F(0) (iv) (yp) (0) : (x) (y) (z) (ypxyz - Oxyz). z. F(c) - F( ). It is obvious, I take it, how to continue this series. To take

case (i), a function of propositions F is an extensional one, if whenever p and q have the same truth-value (p - q), F(p) and F(q) have the same truth-value. Negation is an obvious example. And to take case (ii), a function of properties F is an extensional one if whenever cp and 0 characterise the same objects, F(cp) has the same truth-value as F(0).

To say that a function F is a non-extensional or intensional one, is to say that it does not satisfy the appropriate member of this series.

An extensionalist may be defined as a person who holds that

92 A. N. PRIOR

every function F of propositions, properties, relations, etc. is extensional in this sense, and a non-extensionalist or intensionalist may be defined as a person who holds that some functions of propositions etc. are not extensional in this sense.

In this sense, both Kneale and I are intensionalists. And we are both, I think, partly led to our intensionalism by the consideration of " intentional " functions, i.e. ones involving psychological attitudes. For example, the intentional function of propositions " Jones knows that- " is also an intensional one; for it is clear that the fact that " grass is green " and " log10 100 = 2" have the same truth-value does not guarantee that "Jones knows that

grass is green " will have the same truth-value as " Jones knows that log10100 = 2".

It seems to me worth adding, however, that this is by no means the only sort of example that makes extensionalism

unplausible, not to say incredible. To take a quite different sort of example, one that is not " intentional " at all: Everything that is red occupies one or another of certain regions of space, and everything that occupies one or another of those regions of

space is red, but who wants to infer from this that since red is a colour, the occupation of those regions is a colour? It is not, on the contrary, perfectly plain that the occupation of those

regions is not a colour but a location ? With cases of this sort in mind, I am simply flabbergasted when extensionalists assert that their principles are perfectly compatible with the outlook of Science, though we may be led to question them if we descend to inferior forms of discourse such as Philosophy, Literature and Introspective Psychology.

3. The question as to whether or not there are any such entities as intensional objects or intensions is an altogether different

question, and is part of the wider question as to whether or not there are abstract objects, i.e. the question which is sometimes described as that of Platonism versus nominalism. If we take the so-called Platonist view that there are indeed abstract objects, we may divide these into intensions, which include propositions,


properties and relations as ordinarily understood, and extensions, which include truth-values, classes and what are called " relations in extension ". I shall have something to say about extensions later; in the meantime, let us look at the question as to whether there are any such entities as propositions and properties.

In referring to my own views on this subject, Professor Kneale expresses himself in a very generous and accommodating way, though I think it remains clear that he and I are a little further apart on this question than we are on that of intensionalism versus extensionalism. But his representation of my version of nominalism is entirely accurate and just. I do still wish to say that in " Socrates taught that the soul is immortal " we ought not to regard the word "that " as forming, from the sentence " the soul is immortal ", a name of the object of Socrates' belief, but we ought rather to regard the word " that " as attaching to the verb " taught " to form a sort of half-connective linking a name, not to another name, but to a sentence. And I am committed, as Kneale says I ought in consistency to be, to the " redundancy theory of truth ", that is, the theory that " The proposition that snow is white is true " is merely a blowing up of the plain " Snow is white ". Talk which is ostensibly about properties and relations I would deal with similarly. I would advocate, for example, what might be called redundancy theories of characterisation and of holding-between; to say that the property of cp-ing characterises the object x is in my view to say no more and no less than that x y's, and to say that the relation of p-ing holds between x and y, in that order, is to say no more and no less than that x p's y.

Of course we do all talk as if there were such objects as propositions, properties and relations. For example, in the last section I myself said, " A function of properties Fis an extensional one if whenever p and 0 characterise the same objects, F(Q) has the same truth-value as F(0) ". I would, however, regard such ways of talking as philosophically justified only if they can be given a good sense, i.e. a nominalistic one. For example, the sense of my own bit of Platonistic talk just quoted, or rather the G

94 A. N. PRIOR

sense of the latter part of it, is given by the symbolic formula (ii) of which it is an English version; in that formula there are no variables that stand for names of properties but only (apart from variables that stand for names of individuals) variables that stand for verbs. (These do, indeed, occur bound by quantifiers, and there are those whose eyebrows lift at this; but that is another topic.) Again, the proposition that "there are intensional functions of properties " amounts to " For some F, it is not the case that (ii) ".

4. What is the logical relation between these two issues which I have been so insistent on distinguishing? There is a rather obvious relation between them which belongs not so much to logic as to the sociology or politics of learning. In the language of William James, extensionalism and nominalism both count as "tough-minded" philosophies, and tend to be propagated by the same philosophical gang, while intensionalism and Platonism both count as "tender-minded ", and again have a tendency, though in this case not such a strong tendency, to be held by the same people. From this sociological point of view, I am probably a rather deviant type in advocating a combination of intensionalism and nominalism. But I think this particular deviation can be supported by reasons.

It seems to me that the sort of philosophising which aims at being tough-minded at all costs is better called escapist philosophising; the world just isn't simple and tidy enough for all the tough-minded theories to be true. Nominalism, in particular, seems to me just hopelessly unplausible unless you admit non- extensional functions. It is clear, at all events, that if you get rid of the suggestion that the noun-clause " that grass is green " names an abstract object, by saying that in " James believes that grass is green " the expression "James believes that -" forms a sentence not from a name but from a sentence-if you eliminate Platonistic suggestions this way, you are immediately landed with connectives that are not truth-functional. Since it seems to me that there obviously are non-extensional functions anyway -even being a colour and being a location, and in general falling


under this or that determinable, seem to be obviously non- extensional functions of properties-this consequence of nominalism, or of the way of carrying out nominalistic reductions which I have sketched, seems to me to have nothing awkward about it whatever.

Whether, conversely, being a Platonist would make it easier to escape intensionalism, I do not know; I cannot see how it could; in any case I do not myself have this motive for being a Platonist. Nor, of course, has Kneale, so it is not very clear to me why he is not a nominalist too, though I shall suggest a reason for it at the beginning of Section 6.

5. About some sorts of abstract object, Kneale is quite as nominalistic as I am. I do not entirely understand the principles by which he draws the line between those abstract objects which he regards as genuinely named by abstract nouns and noun phrases and clauses, and those abstract objects of which he prefers to give a nominalist account. It is clear, however, that what are called extensions fall, for him, into the latter class. I am, of course, entirely in agreement with Kneale about this, but it might nevertheless be useful for me to discuss the subject of extensions in my own way.

Take, to begin with, classes. To say that x is a member of the class of 9-ers is, in my view, just an inflated way of saying that x p's. But I have already indicated that I would give exactly the same brief paraphrase for the assertion that x is characterised by the property of p-ing. The difference between classes and properties does not emerge at this point but at another. This difference is sometimes represented as one between the ways in which we count these various entities or quasi-entities. There are, for instance, innumerable unexemplified properties but only one null class, although to say that the property of p-ing is

unexemplified is just the same as to say that the class of p-ers is empty, both being the same as saying that nothing p's. This difference in ways of counting arises from different ways of identification. The class of mermaids is the same class as the

96 A. N. PRIOR

class of centaurs, though to be a mermaid is one thing and to be a centaur quite another thing.

There is a difference, then, between the following two identity- statements:-

(a) The property of y-ing is the very same property as the property of cp-ing.

(b) The class of 9-ers is the very same class as the class of p-ers.

In neither of the statements, I should think, is the word " same " being used in exactly the same way as it is in

(c) x is the very same individual as y, though some of the theses which Professor Kneale lumps together as varieties of extensionalism suggest that he doen't see much difference between (b) and (c). I would say that (b) differs from both (a) and (c) in that the identity which (b) ostensibly asserts is only a fictitious or figurative identity; for what (b) means can be equally well conveyed without suggesting that anything whatever is identical with anything. For (b) is just a way of saying

(b)' Whatever cp's O's, and whatever O's y's.

For instance, to say that the class of mermaids is the very same class as the class of centaurs, is just to say that all the mermaids there are are centaurs and all the centaurs there are are mermaids, and this is trivially true because there are no mermaids and no centaurs either.

I do not think that any such account can be given of the identity expressed by (a). We can, however, informally elucidate (a) a little by observing that it is in one way like (c) and in another like (b). It is like (c) in that the identity involved is indefinable- it is what it is and not another thing-but all the same, the identity in (a) is not the same as the identity in (c). That it is the same could, indeed, be asserted by a Platonist, who could say that (a) ascribes the kind of identity that is appropriate to individuals to an individual of a slightly odd sort, namely, a property (that property which is at once the property of cp-ing


and the property of 0t-ing). My own view, however, would be that " the property of 0-ing " no more forms a genuine name in (a) than it does in " The property of 0-ing characterises x "; it is, rather, an inseparable part of the whole expression

" The property of ( )-ing is the very same property as the property of [ ]-ing ",

which forms a sentence not from a pair of names but from a pair of verbs, just as does the expression

" Whatever ( )s [ ]s and whatever [ ]s ( )s" in our paraphrase of (b). More or less idomatic paraphrases of (a) would include

(a)' To yp is to h

(a)" Whatever p's ipso facto O's, and whatever i's ipso facto p's. Similar remarks are to be made about the two forms (d) The proposition that p is the very same proposition as the

proposition that q (e) the Truth-value determined by the proposition that p

is the very same truth-value as that determined by the

proposition that q. Here (e) is simply an inflation of

(e)'p if and only if q, where the " if" is, as we say, merely " material ", i.e. where the whole is tantamount to

(e)" Either p and q, or it is not the case that p and not the case that q.

(d) expresses an identity as irreducible as that in (c) or (a), but one which falls into the syntactical category of a connective; that is, (d) is not to be read as predicating ordinary identity of an extraordinary individual, the proposition that is at once the

proposition that p and the proposition that q; rather, this apparent name-former "the proposition that-" is here an inseparable part of the expression

"The proposition that () is the very same proposition as the proposition that [ ] ",

which is not (as " ( ) is the very same individual as [ ] " is) a

98 A. N. PRIOR

two-place predicate forming a sentence from a pair of names, but is rather a two-place sentential connective forming a sentence from a pair of sentences, just as "( ) if and only if [ ]" does. We could roughly paraphrase (d) as

(d)' For it to be the case that p, is the same as for it to be the case that q,

or as (d)" If p then ipso facto q, and if q then ipso facto p.

In brief, we have invented locutions which represent material and formal equivalences as identity-statements, and extensions are those quasi-objects which are said in these statements to be identical, but in fact the whole force of the apparent identity statements is given by the statements of material or formal equivalence, in which these quasi-objects are not mentioned.

To say what I have just said is to take sides, and to take a slightly unpopular side, in an important issue in the Philosophy of Mathematics; for it is to adopt a form of type theory rather than a form of set theory. It would be chicken-hearted either to attempt to disguise this fact, or to be much abashed by it; I mention it for candour's sake, and pass on.

6. An intensionalist who is a little put off by the foregoing hierarchy of indefinable identity functions might well wish to combine his intensionalism with a comparatively moderate sort of Platonism in which extensions are paraphrased away but intensions regarded as genuine, though abstract, individuals. For the paraphrases in which extensions cease to be even ostensibly mentioned only employ perfectly ordinary functions of sentences and of predicates, namely the truth-functional "if and only if" and its universal quantification. If we used these to scrap extensions but took intensions seriously as individual objects, the latter could then be regarded as terms of the ordinary relation of individual identity, and my hierarchy of what one might call " ipso facto functions " dispensed with.

Professor Kneale's own form of moderate Platonism seems to be quite close to this position. If I have understood correctly


his account of " basic propositions ", he would say that of those propositions which appear to be about other propositions, the ones which really are about other propositions, and so are in his sense " non-basic ", are precisely all those which are not truth- functions of their constituents. His account of" basic functions " is, however, more complicated, and so is his account of identity. These complications appear to arise from a desire to escape from Leibniz's principle of the "indiscernibility of identicals ", i.e. the law which may be symbolised as

(x) (y) (F): x = y. ) . F(x) - F(y), with analogues employing variables of different logical types. Kneale seems to regard the acceptance of this principle as a mark of extensionalism, but this seems to me a mistake.

7. Let us consider Leibniz's law firstly as a principle about the identity of individuals. It asserts that if x is the very same individual as y then whatever is true of x is true of y. Possibly this law is not true; I have my own doubts about it, which are connected with the possibility of one individual actually becoming two individuals and vice versa. (Cf. my " Time, Existence and Identity ", Proc. Arist. Soc. 1965-6.) But the point on which I wish now to insist is that a belief in non-extensional functions of propositions, properties, etc. does not in itself provide any reason whatever for having misgivings about Leibniz's law. To take one of the stock examples, if Tully is the very same individual as Cicero, it really does follow that if Tom believes that Cicero denounced Catiline then he believes that Tully did so; at least this follows if we so understand " Tom believes that Cicero denounced Catiline " as to make it a genuine case of F(x), i.e. as really telling us something that is " true of Cicero ", as ascribing to the man Cicero the property of being believed by Tom to have denounced Catiline. If we only understand by it that Tom believed that there was someone who was called " Cicero ", who was famous for this and that, and who denounced Catiline, then indeed it does not follow that he also believed that there was someone who was called " Tully ", etc.; but in this case the premiss does not tell us something that is " true of Cicero "- not because " Tom believes

that--" expresses a non-extensional

100 A. N. PRIOR

function of the proposition that someone was called " Cicero ", etc., but because this " intentional " act of Tom's is not directed towards Cicero or indeed towards any individual in particular. (ly)F(y) is not a case of the form F(x), nor is Ba(1y)F(y), where B = " believes that ".

This is not just an ac hoc solution to this particular problem. There are independent reasons for not regarding " Something F's ", or of course " Everything F's ", as propositions about particular things that F. For one thing, as Geach has pointed out (Reference and Generality, p.6) their meaning is unaltered if in fact nothing at all F's (though of course this would make them false). And formally we should be led into grave para- logisms if we treated them as cases of F(x). For example, no believer in Leibniz's law would regard it as legitimate to argue as follows:-

(1) If x is one and the same individual as y, then for any F, if F(x) then F(y)

Therefore (2) If x is one and the same individual as y then if for all x,

x = x, then for all x, x = y. But (3) For all x, x = x. Therefore

(4) If x is one and the same individual as y then for all x, x = y,

i.e. if x is the same individual as y then everything is the same individual as y. The fallacy here would be universally agreed to lie in the step from (1) to (2), i.e. in treating " For all x, x = x" as substitutable for F(x).

We may note, however, that although Tom's believing that someone was called " Cicero ", etc., is not a truth about the man Cicero, the verification of Tom's belief by the man Cicero's being called " Cicero ", denouncing Catiline, etc., is a truth about the man Cicero, and incidentally does imply (given the identity of Cicero and Tully) the verification of Tom's belief by Tully's being called " Cicero ", denouncing Catiline, etc.

If we turn now to the analogous principle about functions of

propositions, i.e.


(p) (q) (F):'p - q. 0. F(p) - F(q), this too seems to have nothing specially extensionalist about it. If the proposition that all bachelors are unmarried really is the very same proposition as the proposition that all unmarried men are unmarried, then whoever believes or doubts either ipso facto believes or doubts the other, though a man who does not know what " bachelor" means might doubt whether whatever it is that is expressed by " All bachelors are unmarried " is true without doubting whether whatever it is that is expressed by " All unmarried men are unmarried" is true. Though, once again, if we describe a doubt as " verified " when what is doubted is true, it will follow even in the case of this semantically ignorant man that if his doubt is " verified " by the fact that all unmarried men are unmarried it is ipso facto verified by the fact that all bachelors (i.e. unmarried men) are.

What does entail extensionalism is the identification of the propositional identity expressed by "p= q" in the above formula, with material equivalence, and the identification of the

property-identity expressed by " 9 - $ " in the formula

(yp) (0) (F): 9p - ?.:D .F(o) - F(0)

with formal equivalence, i.e. characterisation of the same objects. It is also difficult to avoid extensionalism if extensions are treated as genuine individual objects to which Leibniz's law applies. But Kneale no more wishes to make either of these moves than I do.

8. Intentional functions of propositions, properties, etc. are a sub-class of intensional ones. There is also a terminology in which people speak of " intentional objects "; would these too be a sub-class of intensional ones ?

The position here is rather complicated. Intentional objects are, in brief, those objects towards which our thinking, knowing, wanting, hoping, etc. are directed. These certainly include " intensional " objects in the sense in which these latter include

propositions and properties. If James believes that grass is

green, we may say that the object of his belief is the proposition

102 A. N. PRIOR

that grass is green; though I would repeat that to say this is just to say over again tha the believes that grass is green. Propositions as objects of thought, we now know what to do with.

What about properties ? I suppose we could say that when Tom ascribes to Cicero the property of denouncing Catiline, this property is at least one of the objects of this rather complicated "intentional act" of Tom's. But I would wish, naturally, to give this way of talking a harmless sense by providing a paraphrase in which this apparent name of a property is replaced by the corresponding verb. And what can this paraphrase be but: Tom asserts, or believes (depending on what sort of " ascription " we have in mind), that Cicero denounced Catiline? Surely for x to ascribe the property of 9-ing to y, is simply for x to assert or believe that y p's.

But we must be careful here. To ascribe the property of 9-ing to someone could be rather more than merely to assert or believe that someone p's. For to say that Tom ascribes the property of denouncing Catiline to someone could be to say that there is someone to whom he ascribes this property, i.e. that there is a y of whom he asserts or believes that y has it. Whereas to say that Tom asserts or believes that someone denounced Catiline could be merely to say that Tom asserts or believes that there has been at least one Catiline-denouncer, without his ascribing such denunciation to anyone in particular. However, in both senses it is easy to get rid of the " ascription of a property " location--in the one case there is a y such that Tom believes that y denounced Catiline, and in the other Tom believes that there is a y such that y denounced Catiline.

To get back now to the question as to whether all intentional objects are intensional ones: in the case in which there is a y such that Tom believes of that y that it denounced Catiline, it would seem that while this " intentional act " quasi-relates Tom to the intensional quasi-object denunciation-of-Catiline, it also genuinely relates him to a certain y which is not an intensional


quasi-object at all but a perfectly ordinary human being (well, one now dead; but let that pass). In general, it would seem that the objects of our beliefs etc., in the sense of the objects that these beliefs etc. are about-the subjects of these beliefs, etc., as we might more readily say-may be perfectly ordinary individuals. As I insisted in the last section, it is only when the objects of our believing etc. are such individuals that we can expect Leibniz's law for the identity of individuals to apply to them; but it does seem that they sometimes can be such.

9. This seeming, however, may be deceptive. Miss Anscombe opens one of the discussions in her well-known paper on " The Intentionality of Sensation " by saying, quite in line with what has just been suggested, that "if I am thinking of Winston Churchill then he is the object of my thought ". [Analytical Philosophy, Second Series (ed. R. J. Butler, Blackwells, 1965), pp.160-1.] The italics are mine, but Miss Anscombe herself seems to intend some emphasis here. For when she goes on to consider cases in which the object of thought is non-existent, she dismisses the suggestion that in this case we are really thinking of the " idea " of the object by saying that

" if the idea is to be brought in when the object doesn't exist, then equally it should be brought in when the object does exist. Yet one is thinking, surely of Winston Churchill, not of the idea of him, and just that fact started us off. "

I am not at all sure what the people Miss Anscombe is attacking mean by " ideas ", but if Miss Anscombe is right in insisting that thoughts of Mr. Pickwick and thoughts of Mr. Churchill (or, let us say, of Professor Kneale) must be given similar analyses, then surely Churchill and Kneale cannot be the objects of my thought in the direct way in which they might be, e.g., the objects of my bumping if I bump into them. At most, certain thoughts of a fundamentally general sort which I have-thoughts to the effect that someone is p-ing, maybe that someone is uniquely cp-ing--can be verified by the performances of Kneale and Churchill. For certainly my thoughts of Mr. Pickwick can only, in the end, be general ones, and cannot constitute relations

104 A. N. PRIOR

between myself and the actual man Pickwick; for there is no actual man Pickwick.

The alternative is to say that the existence of the object does make a difference, and that only what exists can be thought to (, though one can think that something that y's O's, or that

something that uniquely 9's O's, whether or not anything in fact

9's (or uniquely 0's) at all. On this view we would also have to say that only what exists can be even thought by A to be thought by B to p, though A might think that there is some p-er which B thinks to 0, or that B thinks there is some p-er that O's, whether or not there are any p-ers at all, and in the last case whether or not A thinks there are.

10. Perhaps, then, there are intentional objects which are not also intensional objects, but are individuals of a philosophically ordinary sort, like Churchill; and perhaps there are not. There are, at all events, many cases in which intentional objects- objects of thought, of discussion, of wishing, etc.-are also intensional objects, i.e. not ordinary objects but propositions, properties and the like. About these, Professor Kneale has a

concluding thesis which I should very much like to think he has succeeded in establishing. As Kneale himself expresses it, this thesis is that "intentional sentences" always "express propositions about propositions ". This is Platonistic language, but a nominalistic paraphrase of it is not difficult, and a successful nominalistic programme seems to me almost to depend on its being true.

We speak sometimes of thinking of such and such an object, sometimes of thinking that such and such is the case; similarly we speak sometimes of wishing for or wanting such and such an object, sometimes of wishing that such and such was the case; and similarly with other intentional verbs. I take it that what Kneale's final thesis amounts to is that every genuine case of " thinking of", " wishing for ", etc., is also a case of " thinking that ", " wishing that ", etc. He gives the example "I was

dying for a cup of tea ", for which his paraphrase is " I very


much wished (that) there were a cup of tea I was drinking "; and his contention appears to be that this sort of thing can be done in all cases. To this, I can only say that I hope it is true. For the objects that we may ostensibly think of, wish for, etc. include some very strange ones-abstract objects, fictitious propositions, the philosopher's stone-and the strangest of all are perhaps those which might be described as concrete but indefinite, like " a cup of tea ", or " a saucer of mud ", when there isn't any particular cup of tea or saucer of mud that I think of or want, but I just think of or want a cup of tea or a saucer of mud. There is, clearly, no such object in the real world as a cup of tea which is just a cup of tea without being any cup of tea in particular; and yet it does seem perfectly possible to think of or want a cup of tea without thinking of or wanting any cup of tea in particular. These individua vaga do disappear when we replace " wishing for (thinking of) a (p-er " by forms like " wishing that (imagining that) something p's ", or by more complicated forms of this general type; and moreover this is the only way to make them disappear that I can think of (apart from violently ad hoc solutions like treating "( ) is thinking of a [ ]-er " as an expression, forming a sentence from a name and a verb, which is just not analysable.) That is why I hope that Kneale's final thesis is true.

The thesis does, all the same, have certain difficulties about it which I'm not quite confident can be resolved, though I hope they can. One (put to me by P. T. Geach) is this: Suppose there is an F.B.I. agent who would like to catch a Communist, though there is no particular Communist he is after (he possibly doesn't even know who the Communists are). It would clearly be wrong to say that in this case there is an x such that x is a Communist and the F.B.I. agent would like it to be the case that he has caught x; for ex hypothesi there is no such x. But is it then the case that the F.B.I. agent would like it to be the case that for some x, x is a Communist and has been caught by him? This would seem to imply that the agent would like it to be the case that there are Communists, but surely our original proposition needn't imply this. The best solution I can think of is that the

106 A. N. PRIOR

suggested analysis doesn't imply it either, i.e. that "a wishes that for some x, both yx and Ox " doesn't imply "a wishes that for some x, cpx "; but I cannot pretend to be wholly happy about this, and if Kneale has a better solution I should welcome it.

symposium: intentionality and intensionality

Documents