The difficulties which as we have seen, face Chisholm's program in "Sentences about Believing" seem to rule out the possibility that this program can be made to do the tasks which Chisholm set for it. Nonetheless, it is worth examining certain features of Chisholm's proposal, as well as asking what, if anything, can be salvaged from Chisholm's program. A part of the present chapter will be devoted to these tasks. In particular, the first section will deal with Chisholm's claims in connection with a number of sorts of putative counterexamples to these claims, while a second section will deal with the relationship between Chisholm's marks of intentionality of sentences and two suitable marks of non-extensionality of sentences.

Since the publication of "Sentences about Believing," Chisholm has rejected the claims which he had made there, in favor of certain new claims involving both a new thesis of intentionality and several new definitions of intentionality. Thus following the first two sections of this chapter, two further sections will be devoted to discussing two of these more recent proposals. A fifth section will be devoted to some concluding remarks concerning Chisholm's several proposals.

I. For the purposes of the present section, we shall assume, contrary to the results of the preceding chapter, that Chisholm's claims as presented in "Sentences about Believing" do not need modification in the various ways discussed above. Making this assumption will permit us more easily to focus on one particular aspect of Chisholm's presentation, namely his proposals for dealing with sentences which either exhibit the characteristic but fail to be psychological, or fail to exhibit the characteristic but nonetheless appear to be psychological. Let us begin by considering the following three sentences:

(1) George is talking about Atlantis.

(2) It is obligatory (or imperative, or permissible) that George keep his promise.

(3) It is logically necessary that 9 is greater than 7.

These three sentences each exhibit the characteristic, for the three exhibit, respectively, the first, second and third of the logical marks discussed by Chisholm.{1} None of them, however, appear to have psychological subject matter.

Chisholm's proposal for dealing with such sentences, as we have seen, is to show either that the sentences are expendable in favor of non-intentional sentences which also contain no technical term, or that their subject matter is implicitly psychological. The second of these moves seems especially appropriate in the present cases. Thus, it might be argued that sentence (1) amounts to an assertion that George is in the process of uttering one or more sentences in some language, intending thereby to say something about Atlantis. For it seems reasonable to distinguish cases in which George merely utters sentences the grammatical subject of which is 'Atlantis' from those in which he is talking about Atlantis. For the former might occur without the latter, as in cases where George is simply translating certain sentences, or simply utters certain sentences without knowledge of what they mean; and similarly the latter can occur without the former, as in cases where George talks about Atlantis by means of using some other expression than 'Atlantis'. Thus for George to be talking about Atlantis, it seems required that we qualify the fact of his simply making certain utterances by specifying what his state of mind is, or what he has in mind in making such utterances. Thus we might paraphrase (1) as

(1') George makes certain utterances which he intends to be taken as about Atlantis.

Similarly with sentences about other speech acts, such as asserting, denying, saying, and so forth, we may argue that they are suitably paraphrased so as to exhibit explicitly their covert psychological reference.

It might, of course, be objected that such sentences do not mean more than that the person in question is uttering certain sentences which are describable in certain formal terms. Thus it might be argued that (1) amounts to the assertion that George uttered certain sentences the grammatical subject of which is 'Atlantis'. The adequacy of this approach has been criticized by Church, who has argued that no paraphrase of this sort is adequate to capture what is intended by the original sentence. Church 97-99] Church's criticism seems to be rephrasable, in the terms used in the foregoing paragraph, as the claim that no such paraphrase allows us to retain a distinction between someone merely making certain utterances, and someone actually saying something about something. It was the desire to retain this distinction which led us to suggest paraphrasing (1) as (1'). The merits of arguments like those of Church need not, however, be considered here. For if we were to paraphrase (1) not as (1') but as

(1'') George uttered at least one sentence the grammatical subject of which was 'Atlantis'.

then while we have failed to exhibit sentence (1) as making covert reference to psychological subject matter, we seem also, on this suggestion, to have shown that (1) is expendable in favor of a sentence which neither is intentional, is psychological, nor contains a technical term. For involved in the objection which we are considering is the assumption that (1'') asserts no more about George's utterance than we can say by making reference to no properties of the utterance other than its formal properties.

Let us turn to our third sentence, which involves the use of the logical modalities. One standard sort of account of such sentences argues that they are to be understood, in effect, as asserting that a certain sentence is analytic.{2} Thus on this view, (3) would be paraphrased as

(3') ''9 is greater than 7.' is analytic.

understood as asserting that the denial of '9 is greater than 7.' is self-contradictory. Since it is clear that this sentence is not self-contradictory on the basis of the laws of elementary logic, it seems reasonable to claim that it is self-contradictory on the basis of the meanings of the component terms.

If we adopt this understanding of sentence (3), then it appears that the sentence is to be understood as one which is, in effect, about the uses and meanings of words. As we have seen, Chisholm accepts the idea that such sentences are ordinarily intentional, but claims that they are also elliptical for sentences which are about psychological phenomena. [Chisholm (17) 516] Thus sentence (3) could be regarded as about how people intend their use of terms like '9', 'is greater than' and '7' to be understood, and so regarded, (3) would be not only intentional but also covertly psychological.

As with sentences describing speech acts, sentences which assert a logical modality could be argued to be paraphrasable in a way which fails to make any psychological reference. Thus we might regard (3) as correctly paraphrased by (3'), but argue that (3') is true in virtue of certain semantical rules governing the language in question, or certain regularities to be found pertaining to the use of expressions in the language.{3} So viewed, (3') fails to be about any psychological phenomenon. It is a consequence of such a reading, however, that such sentences are also no longer intentional. For since such sentences, so regarded, are explicitly about certain words, the sentences fail to fall under any of the logical marks which Chisholm advances.

It seems clear that sentences like (3') fail to exhibit either the first or the second of Chisholm's marks. That it also fails to satisfy the third may be seen as follows. Since (3') makes use of single quotes, the expression ' '9 is greater than 7.' ' as it occurs (3') is to be viewed as simply the name of a certain expression, namely, '9 is greater than 7.'. Thus the syntactical form of (3'), so understood, prevents us from applying the third logical mark to it. Thus just as in the case of sentences like (1), sentences like (3) may be regarded either (a) as nonpsychological, non-intentional sentences containing no technical terms, or (b) as sentences which are both psychological and intentional.{4} Just as there appear to be a variety of ways in which to understand sentences like (1) and (3), so we face a variety of competing accounts of sentences like (2), which seem to involve assertions of ethical obligation. To show that such sentences are not counter-examples to Chisholm's claims, therefore, would involve either considering an exhaustive set of paraphrases of such sentences, or advancing one particular ethical system as defensible to the exclusion of others. Since the aim of the present section is simply to show how such putative counter-examples might be dealt with, it will be useful to consider only three ways of understanding sentences like (2): those suggested by some form of emotivism, some form of act utilitarianism, and some form of rule utilitarianism, respectively.{5}

A strictly emotivist view of ethical discourse, such as that defended by Ayer, avoids the whole question with which we are concerned. For since on this view sentences containing normative expressions fail to express any statement whatever, "even a statement about . . . [one's] own state of mind," such sentences can hardly be psychological. [Ayer (1) 107] By the same token, such sentences have no truth values, and therefore have no entailment relations with any other sentences whatever. [Ayer (1) 111] For this reason, such sentences do not seem to be susceptible of the sorts of logical conditions which Chisholm's characteristic involves. Thus on this view, sentences such as (2) are neither psychological nor intentional.{6}

An act utilitarian position such as that advanced by Smart (cf. [Smart (1)]) might result in our paraphrasing sentence (2) as

(2') If George keeps his promise, then certain events will occur such that the expected utility of George keeping his promise is at least as high as that belonging to any alternative course of action.

Such a sentence is not intentional, for neither component of the sentence exhibits any of the logical marks in question. On the other hand, the sentence does not seem to be psychological. Thus on the reading suggested by an act utilitarian view, sentence (2) fails to provide a counter-example to Chisholm's claims.{7}

A rule utilitarian account does not seem to lend itself to a paraphrase of the sort suggested above. For the purposes at hand, it will be useful to examine the sort of position defended by Rawls. Rawls argues that act utilitarianism fails to capture the force of normative statements because such statements, in at least certain cases, make implicit reference to a rule-defined practice in terms of which certain ethical obligations are to be justified. On this account, "a particular action which would be taken as falling under . . . [a given] rule given that there is the practice would not be described as that sort of action unless there was the practice." Rawls 25. Emphasis original] Thus a given action would be described as promise keeping or promise making only given the existence and general recognition of the practice. The existence and general recognition of such a practice is, however, in part a psychological phenomenon.

On this account, then, while sentences like (2) are admittedly intentional, they are also psychological, for they involve implicit reference to the fact that a certain practice is generally recognized as such.{8}

Let us consider briefly a fourth sort of sentence which Chisholm himself offers as a possible counter-example to his own claims, namely, "certain sentences describing relations of comparison." [Chisholm (17) 520, fn. 5] An instance of such a sentence, Chisholm suggests, is

(4) Some lizards look like dragons.

Two ways of dealing with such sentences seem plausible for the purposes of the present discussion. The first involves reconstructing such sentences along the lines of

(4') Some lizards look like dragon-pictures.

where 'dragon-picture' is to be understood independently of any reference to dragons. (This sort of account is suggested in [Goodman 59-60].) On this account (4), although non-psychological, is expendable in favor of a sentence which is non-intentional and contains no technical terms. A second sort of move would involve arguing that sentences like (4) involve an implicit reference to psychological phenomena, on the basis of the claim that they are to be understood as asserting that there are lizards which look the way we imagine dragons might look, or that there are lizards which have certain perceptible properties which we generally ascribe to dragons, and so forth. On both sorts of account, Chisholm's case fails to provide a counter-example to his claims.

The foregoing arguments have been advanced not as conclusive, but merely as indications of how someone who wished to defend Chisholm's claims against these sorts of putative counter-examples might proceed. For in each case, the arguments we have presented rest on certain alternative ways of understanding or construing a sentence, and these matters are highly controversial. Thus the purpose of these suggested reconstructions of various types of sentences has not been to argue that any one of them is what is involved in a defense of Chisholm's claims against these cases, but rather that some sort of reconstruction along such lines would seem to be involved in giving such a defense. If no such reconstruction were itself defensible, then it would follow that Chisholm's claims could not be defended against such cases.

Whereas the foregoing sorts of argument depend on reconstructions of sentences which would, if no suitable reconstruction were possible, be counter-examples to Chisholm's claims, there is another sort of move which, as we have seen, Chisholm invokes. For it is sufficient to show that a psychological, non-intentional sentence is not a counter-example to show that it contains a technical term.{9} It will be useful, then, to consider a case in which such a move might plausibly be made.

It has been argued by Clark that sentences such as

(5) George knows that Alaska is the largest state.

are non-intentional, for, according to Clark, such sentences fail to exhibit the third of Chisholm's logical marks, and it seems clear that (5) exhibits neither of the other two. [Clark] Criticizing Kenny for claiming that "if John knows that Cicero was murdered it does not follow that he knows that Tully was murdered," Clark argues that "the Criterion of Indirect Reference [that is, the third of Chisholm's logical marks] is not a sufficient condition for" a sentence to be psychological{10} in cases of this sort. [Clark 126] If Clark is correct in his contention, then (5) is a case of a sentence which although psychological, is nonetheless non-intentional.

The immediate sort of response which Clark's argument suggests is that even if it is sound, (5) is nonetheless not a counter-example inasmuch as it contains the term 'knows', which is a technical term. Following the suggestions advanced in the foregoing chapter, it is clear that (5) entails the simple, non-analytic, intentional sentence

(6) George believes that Alaska is the largest state.

Since it seems plausible that any sentence containing 'knows' entails some sentence like (6), the term 'knows' seems a plausible candidate for a technical term.{11}

Since, however, the third of Chisholm's logical marks is invoked specifically to deal with cases like (5), it is worth seeing whether an alternative way of dealing with Clark's argument is possible. For in the case of sentences, such as (5), which both describe a propositional attitude held by some person and entail that its propositional content is true, the third of Chisholm's three marks is required.{12} It will be useful, then, to consider Clark's argument in somewhat greater detail.

Clark contends, in effect, that there are cases in which the name or description under which something is known is irrelevant to determining the truth-value of a sentence asserting such knowledge. While allowing that there is "some sort of implication" by such sentences that a person knows what he knows under a particular description, he denies that this implication is "an entailment: it is merely a consequence of the normal conventions or requirements of simple conversation." [Clark 126] In support of this view, Clark offers the following case:

Suppose a witness is being questioned about the activities of a criminal, Jones, hitherto known to him under the alias of 'Brown', and suppose that the witness has just been told his real name. If he now says, 'I knew that Jones intended to rob the bank', he does not imply that he knew that criminal under the name 'Jones' [Clark 126]

Clark is arguing that there are cases in which substitution of identicals does, in fact, preserve the truth value of sentences like (5). That this is so, however, is not a sufficient condition for a sentence to fail to fall under the third of Chisholm's logical marks. For as Chisholm states the logical mark in question, a sentence falls under it provided that such substitution may result in a sentence with a different truth-value. Thus although Chisholm's move involving technical terms is sufficient to deal effectively with Clark's argument, it does not seem that this argument is valid.

Clark's argument relies upon the idea that if a sentence like (5) is to satisfy the third of Chisholm's logical marks, then it must entail that the person who knows something, knows it under the description specified in the sentence. Clark argues that no such entailment obtains, and that the belief, in Clark's view mistaken, that such an entailment does hold is the result of noticing that "some sort of implication" does sometimes hold, and of confusing this with an entailment which is then believed to hold in all cases. Thus Clark's point is clearly not restricted to the case which he has presented: if Clark is correct, then the "implication" in question might never hold, and the entailment in fact never does. Thus it might be argued that Clark's point about the case which he has considered can be made entirely general. In this case, Clark's conclusion would presumably be that in no case of cognitive sentences such as (5) does substitution of identicals change truth-values. Thus, it might be urged, cognitive sentences such as (5) fail to satisfy Chisholm's third mark because in no case of such a sentence does substitution of identicals result in change of truth-value. Thus, it might be argued, it is empty to claim that even though such substitution never does alter truth-values, nonetheless it might.

Clark's argument, as well as the generalized argument sketched in the foregoing paragraph, seem to rest on the assumption that if a cognitive sentence does not entail that what is asserted to be known is known under the specified description, then the sentence fails to satisfy the third mark. It does not seem, however, that this assumption is an acceptable one. For there do seem to be cases in which two cognitive sentences which differ only in that the two contain two different substantive expressions which refer to the same thing do differ in truth-value. Thus even if Clark is correct in urging that sentences such as (5) do not entail that what is asserted to be known is known under the specified description, nonetheless it seems that certain sentences do satisfy Chisholm's third mark. (Indeed, Clark concedes just this point in agreeing with Kenny that 'John knows that Caesar's assassin killed Caesar.' and 'Caesar's assassin is identical to Brutus.' fail to entail 'John knows that Brutus killed Caesar.'. [Clark 126] In making this concession, moreover, Clark rejects the generalized argument presented in the foregoing paragraph.) In the absence of an independent argument in support of the assumption which, we have seen, Clark's argument hinges on, it does not seem that we are required to accept his conclusion.{13}

A somewhat different way of putting this point may be offered which depends on a consideration of the particular case which Clark has presented. Clark's argument is intended to establish that the witness in question--let us call him 'George'--may say that he knew that Jones intended to rob the bank without thereby committing himself to accepting as true that he knew anything about Jones under the name 'Jones'. What is at issue, however, seems to be a somewhat different matter. For unless a person (George, or anybody else) who is committed to the truth both of

George knew that Jones intended to rob the bank.

and of

Jones is identical to Brown.

must also, if he is to avoid subjecting himself to a warranted charge of inconsistency, be prepared to accept the truth of

George knew that Brown intended to rob the bank.

it seems that substitution of identicals may result in altering truth-values, in the relevant contexts. It is not clear, however, given what Clark believes that he has established, why that conclusion would lead us to accept that a person who is committed to the truth of the first of these sentences, but not to the truth of

George knows that Jones is identical to Brown.

need to be committed to the third sentence above.{14}

The difficulties involved with the move of invoking technical terms discussed in the last chapter aside, it would seem that this move is open to us in all cases of sentences which although psychological, are claimed to be non-intentional. Thus it might be argued, for example, that every psychological sentence entails some sentence about beliefs, and since this latter sort of sentence would be intentional in virtue of exhibiting the second of Chisholm's marks, we could argue that every psychological sentence contains some term such that whenever that term occurs in a non-self-contradictory simple sentence, it entails some non-analytic belief sentence. This sort of suggestion will be useful in dealing with a proposal of a different logical mark of intentionality which has been advanced by Chisholm, in section IV of the present chapter.

II. Chisholm's thesis of intentionality has been discussed, by Cornman, in connection with a certain version of the thesis of extensionality. Cornman (1)] Cornman's formulation of this latter thesis "states that a universal language of science may be extensional, or, as Carnap says, 'for every given [non-extensional] language S1 an extensional language S2 may be constructed such that S1 may be translated into S2.' " [Cornman (1) 46] By an extensional language, Cornman has in mind a language every sentence of which is extensional, and to make this latter notion clear, Cornman advanced two conditions which, he claims, are individually necessary and jointly sufficient for a sentence to be extensional:

1. The truth-value of a sentence which results from the replacement of any expression contained in the original sentence by an extensionally equivalent expression will not differ from that of the original under any conditions.

2. The truth-value of the sentence, if it is compound or complex, i.e. if it contains coordinate main clauses or at least one subordinate clause, is a function of the truth-values of the simple sentential elements which make up the compound or complex sentence. [Cornman (1) 46]

As Cornman points, out, these two conditions of extensionality bear a certain relationship to two of the three logical marks discussed by Chisholm. [Cornman (1) 47-48] Thus if a sentence satisfies the first of Cornman's conditions, then the third of Chisholm's marks will not apply to it, and conversely. Similarly, if a sentence satisfied the second of Cornman's conditions, then Chisholm's second mark will fail to apply to it, and conversely. On this basis, Cornman claims that aside from the first of Chisholm's three marks, Chisholm's characteristic serves to pick out non-extensional sentences, and for this (and related) reasons, Cornman argues that only the first of Chisholm's marks is distinctively a mark of the psychological.{15}

The importance of Cornman's comparison of Chisholm's logical characteristic with his own two conditions of extensionality seems to lie in the idea that if all intentional sentences were intensional, then it would seem that Chisholm's thesis would, if true, rule out the truth of Cornman's thesis of extensionality. For given that the language of science is to contain psychological sentences, it seems that if Chisholm's thesis is true and all intentional sentences are non-extensional, then the language of science would have to contain some non-extensional sentences.

This argument seems to rely on ignoring the move, explicitly mentioned by Chisholm, involving technical terms. For it seems that Chisholm clearly allows for the possibility that all psychological sentences could be translated into sentences which are not intentional (or, given the truth of Cornman's claim, into sentences which are not intensional) given that we are prepared to coin the requisite sort of technical terms. (Cf. [Chisholm (17) 513].) Thus even if all intentional sentences were non-extensional, Chisholm's thesis would clearly fail to be incompatible with Cornman's thesis of extensionality.

Independently of this consideration, however, Cornman is concerned to show both that not all intentional sentences are non-extensional, and that not all non-extensional sentences are intentional. That the latter is the case seems to follow from a consideration which Chisholm points out, namely, that since a compound sentence is intentional if, and only if, at least one of its simple sentence components is in its declarative form intentional, a subjunctive conditional is not intentional unless it contains such a simple sentence component. (Vide [Chisholm (17) 512], where Chisholm makes this point in connection with his program of arguing that certain non-psychological sentences which exhibit the characteristic are expendable in favor of certain conditionals (in fact, subjunctive conditionals) which fail to exhibit the characteristic.) Thus although subjunctive conditionals with no simple sentence components which are intentional,when transformed into declarative form are nonetheless non-extensional in virtue of failing to be truth-functional, they are not intentional as well. (Cf. [Cornman (1) 47].)

That not all intentional sentences are also non-extensional can be seen, Cornman argues, from the fact that "unlike the second and third criteria [that is, marks invoked by Chisholm], there seems to be no necessary condition of extensionality comparable to" the first of Chisholm's marks. [Cornman (1) 49] This suggests that there are sentences which are extensional, and intentional by the first mark but by none of the others. As an example of such a sentence, Cornman suggests

(6) John is thinking of Alaska.

For while it is clear that (6) is intentional, Cornman argues that since it "is a simple sentence in which any expression may be replaced by an extensionally equivalent one under any conditions without change of truth-value," it is therefore extensional. [Cornman (1) 49]{16} Thus, given that Alaska is identical to the largest state, Cornman would maintain that under no circumstances would (6) differ in truth-value from

(7) John is thinking of the largest state.

Let us suppose that a person, whom we may call 'John', fails to know that Alaska has become a state, and on this basis believes that Texas is still the largest state. Let us suppose further that at a certain time he is thinking about Alaska, and that he assents to the truth of (6) but explicitly denies (7). The question which John's verbal behavior raises for us is whether such behavior counts as a case in which substitution of identicals fails to preserve truth-value in the case of sentences like (6). For on the face of it, it seems that John's reports concerning what he is thinking of count as true, unless there is some overriding evidence to the effect, say, that he is lying, mistaken about his own thoughts, or mistaken about his use of language. It will be useful, therefore, to consider how one might try to deal with this sort of evidence that sentences like (6) are non-extensional by the first of Cornman's conditions of extensionality.

The most straightforward way in which we might envisage Cornman dealing with this sort of objection is as follows. One might argue that since John has certain mistaken beliefs, namely, that Texas is the largest state and that Alaska is not a state at all, John therefore is mistaken about what he is thinking about. Thus one might imagine correcting such a person, as a result of which correction he would come to concede that he had, in fact, been thinking about the largest state; he just had not known that he was. The trouble with this sort of rejoinder, however, is that it seems to apply to too much. For by this sort of argument we should find ourselves unable to argue that psychological verbs ever govern contexts to which the law of substitutivity for identicals fails to hold. Thus on this view,

(8) George believes that Cicero denounced Cataline.

together with the identity of Cicero and Tully would entail

(9) George believes that Tully denounced Cataline.

For were George to affirm the first and deny the second, we should simply discount this fact, arguing that we could correct George's mistaken belief and thereby get him to assign the same truth-value to both (8) and (9). This result seems,however, clearly counter-intuitive. For even if we should find ourselves able to alter a person's verbal behavior so as to eliminate such discrepancies, this does not discount the fact that at a certain time, the person held the discrepant beliefs in question.{17} Unless, therefore, we are able to find another reply for Cornman to make to this objection, it seems that we must reject his claim that (6) is extensional.{18}

Aside from the argument presented in the foregoing paragraph, there is another sort of move which might be made by someone who wishes to contend that sentences which exhibit the first of Chisholm's marks are non-extensional. Like the foregoing argument, this second sort of move also presupposes that sentences like (8), which are intentional because they exhibit {not only the first but also) the second of Chisholm's logical marks, are non-extensional. Thus instead of arguing, as above, that any way of showing that sentences like (6) are non-extensional would also show that (8) is non-extensional, one might take for granted the non-extensionality of sentences like (8) and argue rather that sentences like (6) are to be analyzed along the lines of

(10) There is some proposition, p, such that John is thinking that-p, and p is about Alaska.

Since (10) fails to satisfy the second of Cornman's necessary conditions of extensionality, it follows that (10) is non-extensional, The trouble with this argument, however, is that it does not seem to allow us to infer that (6) is therefore non-extensional, for in order for the second of Cornman's conditions to apply to a sentence, the sentence has to be either compound or complex, and (6) is neither. This second sort of argument must be rejected, therefore, unless we are to count a sentence as compound or complex if it is analyzable into one which is; and this last condition seems unacceptable because it is not clear that applying it would not result in every sentence being compound. In any event, it seems that the first of our arguments is more satisfactory in that it does not, like the second, rest on a claim about the proper analysis of sentences like (6).

In the absence of any arguments other than that advanced by Cornman, then, we may accept the idea that even though there are sentences which are non-extensional and nonetheless non-intentional, if any given sentence is intentional, given Chisholm's logical characteristic, it is non-extensional. Since, moreover, our examples of non-extensional, non-intentional sentences are all compound sentences such as subjunctive conditions, it seems that we are warranted, in the absence of any countervailing reasons, in claiming that every non-extensional sentence which is simple is also intentional.{19} Thus it seems that Chisholm's criteria for the intentionality of simple sentences amounts to a criterion for the non-extensionality of simple sentences. Since according to Chisholm's claims every intentional sentence which is not expendable in favor of a non-intentional sentence containing no technical term is psychological, it seems that we may conclude that if Chisholm's claims were correct, then every sentence at least one of whose simple sentence components is, in its declarative form, non-extensional, and which is not expendable in favor of a sentence each of whose simple sentence components is extensional and which contains no technical term, is psychological.{20} Thus the arguments of the first section of this chapter may be seen as arguments against the claim that this last conclusion leads to the counterintuitive results that sentences (1), (2), (3) and (4) are psychological. For according to these arguments, these sentences if viewed as non-extensional are also to be seen as covertly psychological; while if we deny that they are psychological, we are led to the conclusion that they are extensional.

III. Chisholm has recently proposed several different definitions of intentionality, and, accompanying them, an altered version of the thesis of intentionality. It will be useful to take note of several of these, both in order to compare them with Chisholm's earlier proposal in "Sentences about Believing," and in order to evaluate their success. One of these recent proposals is the following. Chisholm begins by considering a class of linguistic expressions each of which exhibits the following characteristics: (1) the result of prefixing it to a sentence is another sentence: (2) it contains no proper part which is logically equivalent to a sentence or a sentence-function; and (3) the result of prefixing it to any sentence whatever is always a logically contingent sentence. Any expression which satisfied these three conditions is, Chisholm proposes, an intentional prefix,and any sentence which entails a sentence which is the result of prefixing an intentional prefix to some sentence is an intentional sentence. (Thus all sentences formed by prefixing such an expression to some sentence are intentional in virtue of the fact that they entail themselves.){21}

In connection with this proposal for a new definition of intentionality, Chisholm proposes the following thesis of intentionality: if a sentence is intentional, then it is psychological.{22} The first thing to be noticed in comparing these claims with those advanced in "Sentences about Believing," is that on this proposal Chisholm does not rule out the possibility that there may be psychological sentences which are not characterized by his claims. For according to this proposal, the thesis of intentionality amounts simply to the claim that there are no intentional sentences which are non-psychological. Thus the correctness of this thesis would not rule out the possibility of saying whatever we wished to say about psychological phenomena while never making use either of intentional language or of some special sort of term. It is difficult to see how such a weakened thesis, if established, would yield any significant insights into the nature of psychological phenomena. For at best, it might be argued, the truth of such a thesis would be an accident of language; for there could be a language, no weaker in descriptive power than English, which contained no intentional sentences, and this language would behave in a way compatible with the truth of this weakened thesis of intentionality. At best, it seems, the truth of such a thesis might be used to come to understand better not the nature of expressions which may be used adequately to describe psychological phenomena, but rather the nature of expressions which are intentional.{23}

There is a second contrast which is important to notice between this proposal and Chisholm's earlier one. On the basis of the logical characteristics involved in Chisholm's earlier definition of intentionality, sentences like

(3) It is logically necessary that 9 is greater than 7.


(2) It is obligatory that George keep his promises.

are intentional. As Chisholm points out, however, according to the present proposal prefixes such as 'it is logically necessary that' and 'It is obligatory that'{24} are not intentional prefixes.{25} If sentences like (2) and (3) are to be shown to be intentional, it must be shown that they entail certain sentences which are themselves intentional, say, sentences about beliefs. For since instances of neither sort of sentence contain an intentional prefix, such sentences will be intentional, on the current definition of intentionality, only if they entail sentences which are intentional. And sentences about beliefs seem the most plausible candidates for the entailed intentional sentences. The same thing, however, is required to show that such sentences are psychological. Thus with regard to these sorts of sentences, it seems a trivial matter that they are, if intentional, also psychological. This is not paralleled in the case of Chisholm's earlier proposal, however; for in that case, the arguments used in support of the intentionality of such sentences are, as we have seen in section I, not the same as the arguments used to show that such sentences are psychological.

It will be useful to consider, in spite of the weakened form of this thesis of intentionality, certain possible objections to its truth, and certain possible ways of dealing with these objections.{26} Let us begin by considering a case which it seems clear that Chisholm's claims apply to. We may take 'George believes that' as an intentional prefix, for this expression seems clearly to satisfy each of the three conditions mentioned above: no proper part of it is logically equivalent to a sentence or to a sentence-function, and the result of prefixing it to any sentence whatever is always a logically contingent sentence. For it is never logically true or logically false that a given person believes what he does.{27}

It might be thought, however, that if 'George believes that' is prefixed to the sentence

(11) Socrates is mortal, or either it is raining or it is not raining.

the result, namely,

(12) George believes that Socrates is mortal, or either it is raining or it is not raining.

would be logically necessary. For the comma seems to indicate that the scope of 'George believes that' stops at that point, and the logically necessary disjunct following the comma would make the whole sentence logically necessary. This result can be handled, however, by stipulating that whenever we form a sentence by prefixing an expression to another sentence in order to test the expression for intentionality, we consider the scope of the prefix to extend throughout the sentence to which it is prefixed. On this basis, {12) would be regarded as a sentence asserting a compound disjunctive belief of George's, and it would properly be regarded as logically contingent.{28} Thus (12) would not constitute a counter-example to Chisholm's revised thesis.

A second possible sort of objection would involve the |condition, stated by Chisholm, that an intentional prefix contain no proper part that is logically equivalent to a sentence or to a sentence-function. To see the motivation for this condition, we may consider the expression 'Either it is raining or'. It might be argued that this is an intentional prefix, for whenever it is prefixed to a sentence, the result is logically contingent. The second of the conditions mentioned by Chisholm would, however, rule out this possibility, for a proper part of the expression is logically equivalent to a sentence. One might, however, introduce the term 'glub' as meaning the same as 'Either it is raining or', and since no proper part of 'glub' is meaningful, no proper part can be logically equivalent to a sentence or to a sentence-function.{29} Its seems, however, that this difficulty can be handled more easily. For it is clear that when 'Either it is raining or' is prefixed to sentence

(14) It is not raining.

the result is not a logically contingent sentence. Thus our prefix fails to be intentional not because it fails to satisfy the second of Chisholm's conditions, but because it fails to satisfy the third, that whatever sentence the expression is prefixed to, the result is a logically contingent sentence.

There is, however, one revision which is required in formulating this definition of intentionality. Since any sentence which entails the result of prefixing an intentional prefix to some sentence is an intentional sentence, and since every self-contradictory sentence entails any sentence whatever, it would seem that as the definition stands, every self-contradictory sentence is intentional. This result can be avoided by stipulating that the entailing sentence must be self-consistent. Thus the definition would provide that every self-consistent sentence which entails the result of prefixing an intentional prefix to some sentence is intentional.

As we have seen, Chisholm does not formulate a thesis which asserts that every psychological sentence is intentional, but rather only that every intentional sentence is psychological. This last restriction on the definition of intentionality makes clear that Chisholm cannot claim that every psychological sentence is intentional. For it seems clear that self-contradictory psychological sentences will not qualify as intentional if the above restriction on the definition of intentionality is adopted. Similarly, and independently of the above restriction on the definition, it seems clear that no analytic psychological sentence will be intentional. For since no analytic sentence entails any contingent one, and since by hypothesis the result of prefixing a sentence by an intentional prefix is a contingent sentence, no analytic sentence can qualify as intentional. It does not seem impossible, however, that there should be sentences which are both psychological and analytic. Such sentences would entail no contingent sentences, and no sentence can, on the current definition of intentionality, be intentional unless it entails some contingent sentence. If there are any analytic, psychological sentences, therefore, they will entail no intentional sentences and will therefore fail to be intentional themselves. Thus we are not warranted, in connection with the present definition of intentionality, in claiming that all and only psychological sentences are intentional.

It does not seem, however, that we will be able to generate any examples of psychological, non-intentional sentences, on this definition of intentionality, which are logically contingent. For it seems that every contingent psychological sentence will entail some contingent sentence about beliefs or thoughts which can be reconstructed as a sentence involving an intentional prefix in the required fashion.{30} Thus we might conclude that we are warranted in asserting the following as the thesis of intentionality to accompany the present definition: Every intentional sentence is psychological, and every logically contingent psychological sentence is intentional. Nonetheless the restriction is important in comparing the present proposal with that advanced in "Sentences about Believing," for on Chisholm's earlier view, the device of invoking technical terms, if it can be made to work at all, can be made to handle non-contingent psychological sentences. The fact that we are able to defend even this strong a thesis for the current definition rests, moreover, on the idea that all logically contingent psychological sentences entail sentences about beliefs or thoughts. Thus the truth of such a thesis does not, on this definition of intentionality, allow us to claim also that we have characterized psychological language independently of the fact that it has the sort of subject matter it does. For if all contingent psychological sentences do entail sentences about beliefs and thoughts, this would seem to be a fact about the subject matter of such sentences.

IV. Chisholm's most recent definition of intentionality involves the idea that expressions which are psychological in subject matter behave differently from other expressions when they are embedded in quantificational contexts. For the purposes of formulating this definition, Chisholm considers the following two sentence schemata:

(15) (Ex) (Ey) (y = a & xRa)

(16) (Ex) (Ey) (y = a & xRy)

where 'a' stands for some singular term, and 'R' for some two-place predicate. Any such two-place predicate, 'R', is intentional, Chisholm claims, if, and only if, it satisfies the following three conditions: (a) sentence (15) does not entail (16); (b) sentence (16) does not entail (15); and (c) no proper well-formed sentence part of (15) is logically non-contingent.{31} [Chisholm (4) 22-23] (Cf. also [Aune 197, fn.].) A well formed sentence is intentional if either (I) it contains an intentional expression and other than that only individual terms and/or quantifiers and bound variables; or (II) it is consistent and entails an intentional sentence. [Chisholm (4) 22-23]

Chisholm provides us with an example which is intended to clarify how this definition of intentionality is to apply. We may let 'R' be an abbreviation for 'believes that the Mayor of New York is', and 'a' an abbreviation for 'the next president'. We may suppose, further, that Robert Kennedy will, in fact, be the next president. Now we may consider the following two independent states of affairs: (A) There is a person who believes that Mayor Lindsay will be the next president; and (B) There is a person who believes that Senator Kennedy is the Mayor of New York. If the first of these states of affairs, (A), is the case, but not the second, then, Chisholm argues, (15) is true but (16) is false: whereas if (B) is the case but not (A), then (16) is true but (15) false. Since it seems clear that states of affairs (A) and (8) are such that either could obtain without the other, it would follow that for the case we are considering (15) and (16) fail to entail each other, and that 'believes that the Mayor of New York is' is therefore an intentional expression. [Chisholm (4) 22-23]

One might be tempted, in reaction to this proposed definition of intentionality, to advance the following objection. If state of affairs (A) obtains, then not only is (15) true, but (16) must be as well. For by hypothesis, there is a single person denoted by both the expressions 'Robert Kennedy' and 'the next president'. Thus if there is someone, let us call him George, who believes that Lindsay will be the next president, then there is some person, namely Kennedy, who is identical to the next president and who, referred to as the next president, is believed to be the same as Mayor Lindsay. But all that (16) asserts is that there is someone who is such that he is identical to the next president and someone believes that Mayor Lindsay is that person. Thus, it might be argued, if state of affairs (A) obtains, then both (15) and (16) are true, and thus Chisholm has not shown us that (15) and (16) do not, in the particular case envisaged, entail one another.{32}

It is not required that we evaluate the merits of objections such as these, for it seems that there is a way of reconstructing Chisholm's proposal which avoids any objections of this sort. Instead of using sentences (15) and (16) in a test for the intentionality of a given two-place predicate, we might instead use the following:

(15') (Ex) (Ey) (x is a singular term & y is a singular term & y is not identical to 'a' & when y is substituted for '. . .' in '. . . = a' and x for '---' in '---Ra', the resulting sentences are both true)

(16') (Ex)(Ey)(x is a singular term & y is a singular term & y is not identical to 'a' & when y is substituted for '. . .' in both '. . . = a' and '---R . . .' and x for '---' in '--- R . . .' the resulting sentences are both true)

These revised versions of (15) and (16) avoid the problems suggested above. For in this case, we are no longer quantifying into contexts governed by psychological verbs such as 'believes', and so it can be shown that given Chisholm's example, (15') and (16') may differ in truth-value. This reconstruction has the additional advantage that it allows us to make clear sense of the third condition which must be satisfied if 'R' is to be counted as an intentional expression, namely, that (15) contain no proper part which is a well-formed sentence and non-contingent. For given our reconstruction, we can readily reformulate this condition to provide that if 'R' is to be counted as intentional, then neither sentence resulting from the substitutions specified by (15') is non-contingent.

To see how these reconstructed sentences provide us with a test for intentional expressions, let us return to the case presented by Chisholm. For the purpose of constructing appropriate substitutions, let us imagine that the person who entertains the beliefs in question is called 'George'. Thus the conjunction of the two sentences resulting from the substitutions specified by (15') would be

(17) Robert Kennedy = the next president & George believes that the Mayor of New York is the next president.

while the conjunction of those resulting from (16') would be

(18) Robert Kennedy = the next president & George believes that the Mayor of New York is Robert Kennedy.

Since both (15') and (16') stipulate that whatever is to be substituted for '. . .' cannot be the same term as 'a', and since Chisholm's example provides that 'a' is an abbreviation of 'the next president', these two sentences, (17) and (18), could not turn out to be the same.{33} Since (17) and (18) seem clearly to be logically independent, and since (15') asserts the truth of (17) but nothing about (18) and (16') asserts the truth of (18) but nothing about (17), it seems to follow that neither (15') nor (16') entails the other. Since, finally, no substitution instance which is generated by (15') or (16') is logically non-contingent,{34} 'believes that the Mayor of New York is' is intentional on Chisholm's most recent definition of intentionality, reconstructed as above.{35}

In connection with this most recent definition of intentionality, Chisholm proposes the following thesis: "All intentional sentences pertain to what is psychological." [Chisholm (4) 23] It seems therefore that much of what was said in connection with the proposal of Chisholm's discussed in section III applies as well to the current proposal. For like the former, the latter proposal involves claiming not that all psychological sentences are intentional, but simply the converse of that claim. Furthermore, like the former definition of intentionality, the latter definition rests in part on the idea that any consistent sentence which entails an intentional sentence is itself intentional. Thus if a suitably wide variety of sentences about beliefs and thoughts can be ruled intentional by virtue of their containing intentional two-place predicates such as 'believes that the Mayor of New York is', then any consistent sentence which entails one of these will also be intentional. But since the fact that sentences about hopes, fears, desires, meanings and so forth entail sentences about beliefs or thoughts seems to be a fact about their having psychological subject matter, their being ruled intentional in this way seems not to be a matter which is different from the fact of their having the sort of subject matter they do.

One final observation may be made about this most recent of Chisholm's proposals. If a given two-place predicate, 'R', is intentional, then (15') and (16'), read as containing the name of an expression containing such a predicate, fail to entail one another. If this is true, then there are two substantive expressions, 'b' and 'c', such that the sentences

(19) b=a & cRa


(20) b=a & cRb

fail to entail one another. For if (15') and (16') fail to entail one another, then one may be true and the other false; and if this were so, then one of the two sentences (19) and (20) could be true while the other was false. Since, moreover, (19) and (20) differ only in regard to their second conjuncts, it follows that if 'R' is intentional, then there are three different substantive expressions, 'a', 'b' and 'c', such that

(21) cRa


(22) cRb

fail to entail each other. If this is so, however, then both sentences (21) and (22) satisfy the third of the logical marks advanced by Chisholm in "Sentences about Believing." For a sentence satisfies that third logical mark if it contains a substantive expression such that if that substantive expression is replaced by another referring to the same thing,the resulting sentence neither entails, nor is entailed by, the original sentence. [Chisholm (17) 511] If, therefore,a two-place predicate, 'R', is intentional by this most recent definition, then a simple, contingent sentence containing 'R' would seem to be intentional by the definition advanced in"Sentences about Believing." Except in so far as the later definition invokes the idea that a sentence is intentional if it is consistent and entails an intentional sentence, it is not clear how this more recent definition is intended to be more useful than the earlier proposal.{36} And as we have seen, there are certain disadvantages to making use of this new sort of move, for if we adopt the move, then what it takes to show a sentence intentional on this basis is, it seems,the same as what is involved in one way of showing the sentence to be psychological.{37}

V. Given that contingent psychological sentences generally entail sentences about beliefs and thoughts, we have seen that it is reasonable to suppose that on each of the definitions of intentionality examined in the foregoing two sections, logically contingent psychological sentences are generally intentional. That this is so, it has been argued,is a result of the fact that such sentences have psychological subject matter. It is important to see, however, that it is compatible with Chisholm's claims as examined in these last two sections that (a) we should be able to describe all the psychological phenomena we presently can, and (b) there should be no intentional sentences whatever. For there are,on each of the two definitions of intentionality, two ways in which a sentence may be ruled intentional. One is by determining that it is consistent and entails an intentional sentence. But if there were no other way of showing a sentence to be intentional, then this test would of course be inapplicable. Let us call the other way of showing a sentence to be intentional which is invoked in each definition, the primary definition of intentionality. As a matter of fact, there are sentences in, say, ordinary English, which are intentional by the primary definition on each of the proposals examined above. Examples of such sentences are, as we have seen, belief-sentences. Thus contingent sentences, in whatever language, and no matter what other logical properties they may exhibit, are intentional provided only that they entail English sentences which are intentional by the primary definition. But there is nothing in any of Chisholm's claims which rules out the possibility that there might be no sentences, in any language whatever, which are intentional by the primary definition. Thus nothing in these claims is incompatible with the possibility that there should be no intentional sentences whatever in any language. It might be argued, therefore, that our being able to show certain sentences intentional by the primary definition is a result of certain facts about certain languages, among them English, and thus that the fact that we are able to show any sentences intentional at all is simply a fact about certain languages.

Given that there are sentences which are intentional by the primary definition, however, the force of Chisholm's two recent proposals regarding intentionality seems to be the following. Since one can generally analyze a psychological sentence into terms involving reference to beliefs or thoughts, contingent psychological sentences can generally be shown to entail sentences involving reference to beliefs or thoughts. Thus such sentences will be intentional.

In the case of Chisholm's original proposal in "Sentences about Believing," a similar observation can be made. For it seems that nothing in the claims Chisholm makes in that earlier article is compatible with the possibility that there should be no sentences, in any language whatever,which are intentional. For it is compatible with these claims, as we have seen, that all psychological sentences should be non-intentional and contain some technical term instead. There is, however, an important difference between the two sets of claims. For if a sentence contains a technical term, then, as we have seen, it uses language which we need not make use of in describing non-psychological phenomena. In particular, it would seem that we need to use terms which need never be used in formulating the language of the physical sciences.

It does not appear, however, that a parallel claim is made in Chisholm's two more recent proposals. For all that is involved in these latter claims is that contingent psychological sentences either satisfy the primary definition of intentionality or entail sentences which do. These proposals do not, however, involve the claim that such sentences contain expressions which need not be used when we are describing non-psychological phenomena, or when we are formulating the language of the physical sciences. For this reason, it seems that these two recent proposals are considerably weaker than Chisholm's original one as advanced in "Sentences about Believing."{38}

As it was suggested at the close of Chapter I, the most important problem which Chisholm's original proposal faces is that of finding a suitable mark of technical terms For it seems that on any plausible construction of 'technical term', any intentional sentence will contain such a term Thus the notion of an intentional sentence, relative to this proposal seems to be superfluous In the absence of a suitable test for technical terms, however, it is not clear how this proposal can be salvaged Thus in spite of the fact that we seem to be justified in construing Chisholm's thesis as the claim that it is not possible to define the peculiar vocabulary used in psychological sentences in a way which breaks out of that vocabulary,{39} it does not seem possible to defend such a thesis Nonetheless, Chisholm's earlier proposal seems to have two advantages as compared to his more recent ones. For it does not rely on the idea that a contingent sentence is intentional if it q entails an intentional sentence, and it is in a crucial respect stronger in what it claims than the more recent ones. In what follows, therefore, it will be useful to use the following claim as expressing Chisholm's thesis of intentionality: There is a class of terms such that each member of this class satisfies the following three conditions: (a) no psychological phenomenon can be adequately described with out using at least one of these terms; (b) no one of these terms is required for describing any non-psychological phenomenon, or for formulating the language of the physical sciences; and (c) none of these terms can be defined without making reference to another of them.{40}

[Table of Contents] [Go to Chapter 3]


{1} For neither sentence (1) nor its denial entails either that Atlantis exists or that it does not; neither sentence (2) nor its denial entails either that George does or that he does not keep his promise; and sentence three conjoined with 'The number of planets is identical with 9.' does not entail that it is logically necessary that the number of planets is greater than 7. [Back]

{2} Cf. [Quine (2) 143]. Similarly, Carnap argues that such sentences are to be understood as asserting that certain sentences are, in his term, L-true. [Carnap (1) 174] The arithmetic example, which is Quine's, might be viewed as controversial if it were thought that such truths are, although necessary, nonetheless not analytic. If this were to be argued, however, it would seem that a plausible alternative account of necessity would have to be forthcoming before such a case could be evaluated as a possible counter-example to Chisholm's claims. [Back]

{3} The first of these approaches seems to be that adopted in Carnap (1)], while the second is that defended by Ziff. Cf. [Ziff]. [Back]

{4} Although the use of 'necessary' to mean 'epistemically necessary' results in sentences which are intentional by the third of Chisholm's marks, the resulting sentences also have clear psychological reference since they are about what is known or not known. Similarly, the notion of physical necessity, understood perhaps as involving deducibility from certain natural laws, appears to result in sentences which are non-intentional, non-psychological and contain no technical term. This latter notion will not be discussed in the present section, since it involves issues beyond the scope of our treatment. [Back]

{5} It seems worth noting that Chisholm's own view concerning sentences like (2) seems to require that such sentences be accepted as intentional in virtue of exhibiting the second of his logical marks. (Vide [Chisholm (5) 35].) Thus it would seem that Chisholm is committed to showing either that such sentences are implicitly psychological, or that such sentences are expendable in favor of sentences which are not intentional and contain no technical term. The acceptance of sentences such as (2) as intentional does not, however, follow from Chisholm's views as expressed in "Sentences about Believing. [Back]

{6} An alternative understanding of Ayer's position might run as follows. Since moral argument is possible given that some system of values is presupposed [Ayer (1) 111], we might suppose that the presupposing of such a system of values allows us to talk about the entailment relations of ethical sentences. On this basis, such sentences would seem to be intentional in virtue of exhibiting the second of Chisholm's logical marks. Ayer also concedes that there is a certain way of talking about the meanings of ethical terms: "We may define the meaning of the various ethical words both in terms of the different feelings they are ordinarily taken to express, and also the various responses which they are calculated to provoke." [Ayer (1) 108] Given this understanding of ethical sentences, it seems clearly arguable that such sentences are about psychological phenomena. [Back]

{7} Since the notion of expected utility, understood as the product of the numerical values assigned to the probability of an event and the utility of its occurring, involves the notion of probability, the consequent of conditionals such as (2') might be viewed as exhibiting the characteristic. Chisholm raises such possibilities in his paper, however, and makes clear how sentences involving probability may be construed either as both psychological and intentional or as neither psychological nor intentional. [Chisholm (17) 512] [Back]

{8} A different version of rule utilitarianism might, of course, be advanced according to which normative ethical sentences do not involve such reference. On such a reading, (2) might be paraphrased as 'If George does such-and-such, then his action is a case of promise-keeping, and the rules governing such cases state that so-and-so.'. The question whether such a sentence is to be regarded as intentional or as psychological would thus seem to hinge on the question whether we are to regard the component of this latter compound sentence which asserts that the rules state that so-and-so as intentional or as psychological. As we have seen above, such a component will arguably be either both intentional and psychological or neither. [Back]

{9} As we have seen, it would have also to be shown that the sentence is not expendable in favor of a sentence which is psychological, non-intentional and contains no technical term. This complication will be ignored, however, in what follows, for the purpose of the present discussion is simply to show how someone defending Chisholm's thesis might go about doing so. [Back]

{10} Clark actually writes that this mark "is not a sufficient condition for intentionality in these cases." [Clark 126] I have reported Clark's contention using the word 'psychological' instead of the word 'intentional' because it seems clear that there is a terminological divergence between the usage adopted by Clark and that which, following Chisholm. has been adopted above. For Clark's concern is primarily with what he calls intentional objects, and if a phenomenon involves an intentional object, Clark asserts, then it is a psychological phenomenon. [Clark 123] Thus Clark appears to be talking of sufficient conditions of a sentence's being psychological, when he uses the expression 'sufficient condition of intentionality'. [Back]

{11} Cf. Chapter 1, page 13, fn. 1, in which this sort of move is adopted to deal with sentences asserting that someone knows someone or something, in the sense of being acquainted with someone or something. [Back]

{12} Cornman has argued that such cognitive sentences should not be regarded as describing a psychological activity but rather an achievement of a certain sort, and that for this reason Chisholm might profitably drop the third mark, which is useful only in picking out such sentences, given the adoption of the first and second marks. There does not seem to be any basis, however, for supposing that Chisholm was concerned only with psychological activities as opposed to psychological "achievements" such as seeing, knowing, and so forth. Thus the force of Cornman's argument is not clear. [Cornman (1) 51-52] [Back]

{13} Clark's formulation of what has been discussed as Chisholm's third mark actually follows a proposal of Kenny's. [Kenny 171] On this account, a sentence satisfies the mark if (i) it contains a substantive expression, and (ii) the conjunction of it and an identity statement containing a substantive expression contained in the sentence in question does not entail the result of replacing that substantive expression in the original sentence by the other term of the identity statement. This formulation. which seems to be equivalent to that offered by Chisholm, has the merit of making it clear precisely what is involved in a sentence satisfying or failing to satisfy the third mark. For even if every substitution of identicals allowed by the language results in a new sentence having the same truth-value, this does not allow US to infer that the original sentence fails to satisfy the third mark. Like Chisholm's formulations of the first two marks, this alternative formulation invokes the notion of entailment. [Back]

{14} In Clark's example, the witness is imagined as knowing, at the time of his utterance, the truth of the relevant identity statement. [Back]

{15}Cf. second fn. on page 57 above. [Back]

{16} In a footnote to "Sentences about Believing," Chisholm claims that "by adopting Frege's theory of meaning--or his terminology--we could make . . . [the third mark] do the work of the first two." [Chisholm (17) 520, fn. 3] If Cornman is correct in the foregoing argument, then, as he points out, Chisholm's claim would seem to be refuted. For sentence (6) would be intentional by the first mark, but not by the third, and so "at least the first criterion of intentionality is not replaceable by the third." [Cornman (1) 49] [Back]

{17} A defender of Cornman's position might argue, in response to this line of attack, that the logical behavior of sentences like (6) and (7), containing the expression 'thinking of', and sentences like (8) and (9), containing the expression 'believes that', are different in ways which undercut the assimilation of the two sorts of sentence presupposed above. For such an argument to be persuasive, however it would be required that a defender of Cornman's position specify the relevant sorts of differences. It is not clear, however, how this might be done. [Back]

{18} This contention of Cornman's has been attacked, on rather different grounds, by Brown. [Brown (2)] Brown argues that there are two senses of 'thinking of', such that if we read (6) in one of the two possible ways, it is both non-extensional and intentional, whereas if we read it the other ways, it is both extensional and non-intensional. As Cornman points out in his reply to Brown [Cornman (3)], Brown's argument in favor of this claim is somewhat obscure, and for this reason, it will not be discussed here. [Back]

{19} As we have seen in Chapter I, since every self-contradictory sentence entails any sentence whatever, it seems that the claims in this paragraph must be restricted to self-consistent sentences. [Back]

{20} This claim is, of course, subject to the various sorts of modifications and complications introduced in the course of Chapter I. [Back]

{21} This proposal is advanced at the close of a rejoinder to Luce and Sleigh in connection with a different, and earlier, proposal [Chisholm (3) 269], and in a rejoinder to Sleigh concerning the same earlier proposal [Chisholm (16) 54]. [Back]

{22} As Chisholm formulates this thesis, it reads as follows: "The psychological, unlike the non-psychological, can be adequately described only by using sentences that are intentional." [Chisholm (3) 269] This formulation suggests that Chisholm wishes to assert that if a sentence is psychological (if, that is, it can be used in adequately describing psychological phenomena), then it is intentional. Despite this misleading formulation, Chisholm makes clear, in a letter dated September 25, 1965, that he intends to advance the converse claim, as formulated in the text, above. [Back]

{23} It does not seem that it is open for us to understand the thesis as asserting that any intentional sentence can be used to describe only psychological phenomena, for just as in the case of Chisholm's proposal in "Sentences about Believing" it seems that there are intentional sentences which can be used to describe non-psychological phenomena. Thus one of Chisholm's examples in "Sentences about Believing" is a sentence describing the behavior of a broken computing machine by asserting that it believes falsely that 7 plus 5 is 11. [Chisholm (17) 512-513] Thus it would appear that the sentence 'The machine believes that 7 plus 5 is 11.', while intentional on the above definition, can be used nonetheless to describe a non-psychological phenomenon--namely, an aspect of the behavior of a machine. Unlike Chisholm's proposal in "Sentences about Believing," this more recent proposal does not involve suggesting a means for dealing with such sentences by specifying a certain condition which any such sentence must satisfy--say, being expendable in favor of a non-intentional sentence containing no technical term. [Back]

{24} Chisholm's example of a prefix expression which has evaluative import is not 'it is obligatory that' but rather 'it is wrong that'. Chisholm argues that 'it is wrong that' is not an intentional prefix because there is a sentence such that when 'it is wrong that' is prefixed to it, the resulting sentence is self-contradictory, and therefore logically non-contingent. The sentence which Chisholm offers is 'There is not anything of which it can be truly said that it is wrong.'. [Chisholm (3) 269] It is not clear, however, that if 'wrong' has the same meaning in both occurrences in the resultant sentence, that it is not correctly to be understood in roughly the sense of 'it is not the case that', and not in the sense of the normative expression 'it is ethically wrong that'. So far as the prefix expression 'it is obligatory that' is concerned, it seems that it can be prefixed to a sentence such that the resulting sentence is logically necessary, say, to the sentence 'One should do whatever is obligatory.'. If this is correct, then on the current definition of intentionality, 'it is obligatory that' would not be an intentional prefix. [Back]

{25} This conclusion presupposes, Of course, that sentences like (3) are themselves logically necessary. This presupposition might be defended on the basis of reconstructing sentences like (3) along the lines of ''9 is greater than 7.' is true in virtue of the meanings of the component terms alone.', and then arguing that this latter sentence is itself true in virtue solely of the meanings of its component terms, and that it is therefore logically necessary. Both this argument and the conclusion it is designed to support, however, are controversial matters. [Back]

{26} Both of the objections to be considered were mentioned by Chisholm in a letter dated September 25, 1965, in the course of discussing the present proposal. [Back]

{27} Thus even 'An omniscient person believes that it is raining or it is not raining.' is contingent, given that the existence of an omniscient person is a contingent matter. [Back]

{28} Such considerations could likewise be used to handle the case of prefixing 'George believes that' to the sentence (13) It Is raining, and it is not the case that George believes that it is raining. For the resultant sentence would be logically contingent given that the scope of the initial 'George believes that' extends to the end of sentence (13), unless it were argued that it is impossible that a person should believe that p, and not believe that he does. This latter argument, however, seems entirely unwarranted. [Back]

{29} In his September 25, 1965 letter, Chisholm attributes this objection to Robert Sleigh. [Back]

{30} Thus a sentence like 'George is afraid of lions.' might be viewed as entailing, for example, 'George believes that lions are fearful beasts.', and so forth. [Back]

{31} Expressed more rigorously, this third condition seems to amount to the requirement that no substitution instance of (15) contain such a proper part which is non-contingent. For (15) cannot contain any proper part which is a sentence. [Back]

{32} This objection is the sort of point raised by Quine [Quine (2) 141-144], in order to exhibit the difficulties involved in quantifying into contexts governed by psychological verbs such as 'believes'. In another location, Quine proposes a way of reconstructing sentences which involve such quantifying. [Quine (1)] A discussion of these difficulties, and the merits of Quine's proposed reconstruction would take us, however, beyond the scope of the present discussion. [Back]

{33} Thus on Chisholm's example, while 'the husband of Ethel' could have been substituted for '...' instead of 'Robert Kennedy', 'the next president' could not have been, even though all three of these expressions, by hypothesis, denote the same object. [Back]

{34} For the first conjunct of each cannot be non-contingent, given that whatever is substituted for '...' is not the same expression as 'a', and the second conjunct, being a sentence about beliefs, also cannot be non-contingent. [Back]

{35} This reconstruction of sentences (15) and (16) follows certain suggestions by Ruth Barcan Marcus. [Marcus 253-259] Mrs. Marcus' proposal concerns a way of understanding all sentences involving quantification, and is somewhat controversial. As applied simply to Chisholm's two sentences, however, the suggestion seems unobjectionable. [Back]

{36} One clear difference seems to rest with the third condition which must be satisfied if 'R' is to be ruled intentional, namely, that (given the reconstructed version presented above) neither sentence resulting from the substitutions specified by (15') be non-contingent. The point of this restriction seems, on the basis of the strategy employed in the proposal discussed in section III, to exclude predicates like 'is necessarily the same as' from being ruled intentional. For it seems reasonable to suppose that Chisholm would argue that

c is necessarily the same as b.

is non-contingent. This difference between the application of this logical mark And the third of Chisholm's original marks could, of course, be readily eliminated by providing that a sentence satisfies the third of the original marks only if it is contingent. [Back]

{37} In addition to these last two proposed definitions of intentionality, Chisholm has constructed a third, which like the last involves quantificational contexts, and which like the earlier, invokes the notion of an intentional prefix. (Cf. [Chisholm (9)], [Chisholm {10)] and [Chisholm (11)].) On this proposal, taking 'M' to be an expression which when prefixed to a sentence results in another well-formed sentence, 'M' is held to be intentional if, and only if, the entailment relations between '(x)MFX', 'M(x)Fx', '(ax)MFx' and 'M($x)Fx' conform to a certain specified pattern. In addition to facing possible difficulties involved in quantifying into psychological contexts, this proposal has the disadvantage that it is difficult to evaluate whether or not Chisholm is correct in holding that for some given expressions 'M' and 'F', certain of the entailments hold while certain others do not. For this reason, this alternative proposal will not be discussed here. Vide, however, the following comments on this proposal: [Luce], [Sleigh (1)] and [Sleigh (2)]. Cf. also Chisholm's replies: [Chisholm (3)] and [Chisholm (16)]. [Back]

{38} Cf. [Aune 197, fn. ], where the belief is expressed that Chisholm's most recent proposal is less illuminating than that advanced in "Sentences about Believing." The difference in the strength of the claims made in the two proposal- might justify this sort of conviction. [Back]

{39} Cf. Quine's rephrasal of Chisholm's thesis, that "there is no breaking out of the [psychological] vocabulary by explaining its members in other terms," [Quine (3) 220] and Putnam's contention that "psychological states are characterizable only in terms of their relations to each other, and not as dispositions which can be 'unpacked' without coming back to the very psychological predicates that are in question." [Putnam (5) 70] Both of these contentions seem to be similar to the formulation of Chisholm's thesis adopted (though not advanced as defensible) in this paragraph Quine, while he accepts such a thesis, rejects the usefulness or indispensability of such expressions [Quine (3) 221] Putnam's more involved view, on the other hand, seems to involve commitment to a functionalism of the sort advanced by Fodor (Cf. [Fodor passim].) [Back]

{40} It should be made clear that this thesis is not advanced, here, as defensible; rather this formulation is introduced simply as an attempt to capture what Chisholm appears to wish to claim, on the basis of his discussion in "Sentences about Believing." [Back]

[Table of Contents] [Go to Chapter 3]