The Rosenthal-Sellars Correspondence on Intentionality *

DAVID M. ROSENTHAL AND WILFRID SELLARS


Professor Wilfrid Sellars
Department of Philosophy
University of Pittsburgh
Pittsburgh, Pennsylvania

July 5, 1965

Dear Professor Sellars:

In response to your kind offer to read through portions of the typescript of my thesis pertaining to your views on intentionality, I am sending you a copy of an introductory section to such a chapter.{1} The enclosed typescript represents a first draft, for which I apologize, but I thought it might be useful to get any comments you might have in at the ground floor, so to speak.

The section I am sending leaves off, as you will see, just at the point at which I intend to take up the question of the reducibility (or not) of semantical discourse to psychological discourse. For this reason, I have not yet discussed the problems involved with the 'means'-rubric, nor those examined in "Some Reflections on Language Games."{2} Aside from these omissions, however, I wrote this section by way of attempting to present a general account of your position respecting mental discourse (leaving out, of course, language which exhibits what you call, in "Being and Being Known,"{3} pseudo-intentionality).

A point at which I was conscious of not understanding your views as well as I would like to involves the nature of the so-called reporting role which uses of sentences in the language of Jones' theory come to have. I have gone through, in particular, the letter of yours to Castañeda (April 3, 1961, #10) {4} in which you discuss this, and the passages you refer to, but I find myself unclear on what, for the language of thoughts, is meant by your statement that "the conditioning is itself caught up in a conceptual framework" (#13). I would be immensely grateful for any help you could give me with this.

Sincerely yours,
David Rosenthal

[Back to Contents]


Mr. David Rosenthal
82 Graduate College
Princeton, New Jersey

September 3, 1965

Dear Rosenthal,

I must apologize for the delay in responding to your letter and chapter. I moved from the Princeton frying pan to the Pittsburgh fire. The Spring (!) term is now over and I am caught up and ready to move ahead. I read the Ms through when it first arrived, and have just gone through it again. It strikes me as remarkably lucid and to the point. I think you give an excellent account of what I was up to in "Empiricism and the Philosophy of Mind."{5} There is nothing which sets my teeth on edge. I shall limit myself to a few comments which relate what you have said to the larger setting of the problem.

My argument requires that the term 'psychological' is ambiguous in a way which, if not taken into account, can muddy the waters.

(a) Pre-Jonesean psychological statements employ Rylean resources enriched by semantical devices and categories. They neither coincide with behavior-plus-dispositions-to-behave statements nor reach beyond them to inner episodes proper. Pre-Jonesean psychological statements are, of course, entailed by statements characterizing someone as speaking nonparrotingly. But that is simply because to say of an utterance that it is non-parroting is to say that it is for example, a tokening of a certain proposition--e.g., the proposition that two plus two equals four. If the utterance is a .two plus two equals four. or a .I shall raise my hand. or a .Lo! this is green. then it is, respectively, a thinking-out-loud that two plus two equals four, an intending-out-loud that he himself will raise his hand, or a taking-out-loud of something to be green. The relation of

Utterance U1 is a .two plus two equals four.

to

Utterance U1, being a 'zwei und zwei gleicht vier', means two plus two equals four

and

'zwei und zwei gleicht vier's are .two plus two equals four.s

was explored in "Abstract Entities" and "Notes on Intentionality."{6}

(b) Inner episode psychological statements. When I have denied that

'. . .' means -----

is to be analyzed in terms of psychological statements, I have had primarily in mind the claim that it is to be analyzed in terms of inner episode psychological statements. Notice, however, that even in the case of the pre-Jonesean use of 'means' the above rubric is not to be analyzed in (pre-Jonesean) psychological terms. It is, rather, constitutive of the latter. For, roughly, to be a pre-Jonesean psychological event is to mean something; where to mean something is to be a .two plus two equals four., etc. [where I am taking the liberty of using pre-Jonesean English in employing the dot-quoting device].

The above considerations are, I believe, relevant to your discussion on pp. 25-26.{7} It also indicates my strategy with respect to Chisholm's claim which you summarize on p.6.

As for the question which you ask in your letter: "What . . . is meant by . . . 'the conditioning is itself caught up in a conceptual framework'?"--the point I had in mind was intended to be the exact counterpart of the additional condition I lay down for "Lo! this is green" to count as observational knowledge, and which you correctly describe on p. 20.

One small point. On p. 21 you suggest that my view requires that pre-Joneseans do not think 'in the full sense.' I do not see this. I would rather say that since they do not have the concept of thought in the full sense, the scope of their thinking is substantially restricted.

Finally, it is 'I think . . .' rather than 'I am thinking . . .' which I count as a dispositional cousin of 'I believe . . .'. 'I am thinking . . .' is a variegated locution which implies a sequence of episodes (a process?), and can also be used to refer to processes-cum-dispositions ("I am thinking much, these days, about Vietnam").

I enjoyed reading the pages you sent me, and promise to be more prompt in responding to any further material you might wish to send.

Cordially,

Wilfrid Sellars

[Back to Contents]


October 2, 1965

Dear Professor Sellars:

[1] Thank you very much for your letter of September 3. I found your comments very helpful and encouraging, and was delighted to learn that you felt I had given an adequate account of your views. I had intended, by this time, to have written the sequel to that paper, dealing with the points of controversy between you and Chisholm which I had mentioned. The comments you made in your letter, however, led me to formulate some new ideas on that subject. Because I am not confident that I have not misunderstood you, I thought I might first write you to ask whether you think that the following rough line of argument does justice to what you had in mind.

[2] You write that your "argument requires that the term 'psychological' is ambiguous" in a way which you then trace out. This fact indicates, you write, your "strategy with respect to Chisholm's claim" about the analysis of 'means'-statements in terms of psychological statements. If I am correct at the end of my paper, then, this would indicate your strategy in dealing with the problem of showing that sentences like ' "Spectre" means ghost' and 'John's utterance was meaningful (in the sense of nonparrotingly uttered)' do not mean the same as certain psychological sentences.

[3] As I understand your views, your construction of psychological discourse from certain linguistic resources revolves around the claim that thoughts (and other psychological items) may be viewed as exhibiting two aspects, viz., their intentionality and their determinate factual character. It is in terms of these two aspects which, as I understand the broad line of your argument, you hope to account for the various properties which we believe psychological items to have, and, as you say in your March 11, 1962, letter to Castañeda, "save the appearances."

[4] In your letter, you contrasted pre-Jonesean psychological statements with what you called inner episode psychological statements. It occurred to me that it might be possible to understand this contrast in terms, roughly, of the contrast between the intentionality and the determinate factual character of psychological items. Let me try to sketch what I have in mind in a bit more detail.

[5] You write that the 'means'-rubric is constitutive of pre-Jonesean psychological terms, and for this reason it is fruitless to attempt to analyze 'means'-statements in terms of pre-Jonesean psychological statements. You also write that "statements characterizing someone as speaking nonparrotingly entail pre- Jonesean psychological statements," and that this amounts to saying no more than that a nonparrotingly uttered utterance tokens a certain proposition (or something of this sort). This suggests to me that you may wish to assert that 'means'-statements are no more nor less than pre-Jonesean psychological statements.

[6] This assertion seems on the face of it, however, to be mistaken. For while 'means'-statements are about more than simply the "characteristics which marks and noises can have as marks and noises" (Chisholm-Sellars Correspondence, Aug. 31, 1956), {8} they are also about certain marks and noises as marks and noises. Pre-Jonesean psychological statements, I would assume, are not.

[7] As we may with psychological items, we may assert of semantic items that they exhibit two aspects, their intentionality and their (linguistic) determinate factual character. We might then distinguish two senses of '"means"-statement' as follows: (1) 'statement which is about intentional items (no matter what their determinate factual character is)' and (2) 'statement which is about items which exhibit intentionality and a certain (linguistic) determinate factual character'. Would it be possible, then, to re-express what you have in mind by saying that pre-Jonesean psychological statements are no more nor less than 'means'-statements (in sense (1) )?

[8] This sort of account suggested to me that we might be able to describe the ambiguity of 'psychological' which you discuss as follows. Pre-Jonesean psychological statements are simply statements about intentional items; we may call them psychological because psychological items are intentional. Post-Jonesean (presentday) psychological statements, on the other hand, are psychological because they are about items which exhibit both intentionality and the sort of determinate factual character (whatever it may be) that psychological items exhibit.

[9] One might summarize these points as follows.'Psychological statement' can mean the same as (1) 'statement about intentional items (no matter what their determinate factual character)' or the same as (2) 'statement about items which exhibit both intentionality and a certain (psychological) determinate factual character'. It is clear that psychological statements (in sense (2)) "reach beyond," to use your words, "behavior-plus-disposition-to-behave statements . . . to inner episodes proper." We might then, if I understand your views correctly, go on to claim the following. It is a sufficient condition for a statement to be about items with a certain (psychological) determinate factual character that it be about inner episodes proper. (Would it, as I believe, be a necessary condition as well?) So we can characterize what you call inner episode psychological statements as being psychological statements (in sense (2)).

[10 (At this point I am unclear concerning whether you may, by 'inner episode psychological statement', have meant 'statement about inner episodes (no matter whether intentional or not)' rather than 'psychological statement (in sense (2))'. Although I shall assume that you meant the former, I believe that what follows can be suitably altered if, in fact, you did mean the latter.)

[11] Given this account, it is clear that 'means'-statements (in sense (1)) are psychological statements (in sense (1)). But it also seems clear that neither 'means'-statements (in sense (1)) nor 'means'-statements (in sense (2)) either are, or are to be analyzed in terms of, psychological statements (in sense (2)). If this is correct so far, then I am, I believe, in a position to compare this account with that advanced by Chisholm.

[12] Let us take Chisholm's claim as being that 'means'-statements may be (or are to be) analyzed in terms of, or entail, psychological statements. But Chisholm does not draw (nor, it seems, does he recognize) a distinction between our two senses of 'psychological statement' nor between our two senses of ' "means"-statement'. So it is necessary to see if Chisholm's claim can be made more precise. (It may be easier if, in what follows, I use 'P1 statement' as an abbreviation for 'psychological statement (in sense (I))', 'M1 statement' for ' "means"-statement (in sense (1) ), etc.)

[13] It is clear that by ' "means"-statement' Chisholm means the same as either 'M1 statement' or 'M2 statement.' For the alternative is that he is talking, about marks and noises as marks and noises. But it is just this that he appears to wish to deny by saying that "linguistic entities (sentences, etc.) are also intentional" (CSC, Oct.3, 1956, sentence (C-2)). Similarly, by 'psychological statement' Chisholm must mean the same as either 'P1 statement' or 'P2 statement'. For he denies that he is talking about merely inner episodes (no matter whether intentional or not) when he says that "thoughts (i.e., beliefs, desires, etc.) are intentional . . . " (CSC, Oct. 3, 1956, sentence (C-1)). Since further, we may assume that he is talking about psychological statements and 'means'-statements in the ordinary, everyday (and, therefore, post-Jonesean) sense, it is reasonable to suppose that he is claiming that M2 statements entail P2 statements.

[14] Both M2 statements and P2 statements are about items which exhibit a certain determinate factual character. M2 statements, however, are about marks and noises which exhibit certain characteristic patterns, while P2 statements are about inner episodes of a certain sort. From these considerations, it appears that neither M2 statements entail P2 statements (as we supposed that Chisholm was claiming) nor do P2 statements entail M2 statements.

[15] We might, however, wish to assert that P2 statements do entail P1 statements (conjoined with certain statements about inner episodes), just as we might assert that M2 statements entail M1 statements (conjoined with certain statements about marks and noises as marks and noises). We would then be in a position to claim that M2 statements entail P1 statements (for P1 statements are M1 statements). But since P1 statements do not, in turn, entail P2 statements, we are still without warrant in supposing that M2 statements entail P2 statements.

[16] If we were to understand Chisholm to be claiming that M2 statements entail P1 statements, then the account sketched above would lend support to his claim. The reason that we had for believing that Chisholm was talking about P2 statements was our supposition that he was talking about ordinary, presentday (and, therefore, post-Jonesean) psychological statements. Such statements could not, on our account, be P1 statements. We might, therefore, diagnose Chisholm's error as being the result of his belief that (at least in certain respects) presentday psychological statements behave as if they were (pre-Jonesean) P1 statements.

[17] This diagnosis seems to gain some plausibility from the following considerations. In CSC (Oct. 3, 1956), Chisholm characterizes thoughts as being intentional items and a "source of intentionality." He does not, so far as I can discover, speak of thoughts as the sort of item which has a certain factual character; nor does he speak of thoughts as inner episodes. It appears that Chisholm may be using 'psychological statement' to mean, in effect, the same as 'statement about intentional items without any determinate factual character'.

[18] I have suggested, roughly, that Chisholm is so set on stressing the intentionality of thoughts that he ignores their determinate factual character as inner episodes. Can Chisholm, in responding to this suggestion, reasonably claim that ordinary psychological statements are P1 statements? An argument that he cannot might run as follows. Our ordinary ideas about thoughts involve that they are inner episodes which are (although we may not know how) causally connected to other episodes and to dispositions to behave. (They are, for example, causally connected to overt, nonverbal behavior, to certain forms of sensory stimulation, to dispositions to utter 'I have a thought that-p', etc.) Therefore if a statement is about thoughts as thoughts, then it is about thoughts as inner episodes.

[19] How might Chisholm respond to this line of argument? If we accept that 'John thinks that-p' is a psychological statement, then it seems that Chisholm would have to deny that it is about an inner episode as inner episode (and that it entails a statement which is). This denial might open the door, then, for a view which looked upon the statement as expressing a relation between John and an intentional item (namely, the thought that-p), or as expressing an intentional relation between John and whatever the proposition that-p is about. (Such a view might be suggested by Chisholm's formulation of the problem of intentionality in the abstract of his paper for the symposium at West Virginia last May.) {9}

[20] In any event, it seems that Chisholm would be committed to looking on the statement 'John has a thought that-p' as if it were about "a thought in Frege's sense--an 'abstract entity' rather than a mental episode" ("Notes on Intentionality,"). And if I understand your distinction in that paper, looking upon such a statement as being about a thought in Frege's sense alone would be no more nor less than looking upon it as being a P1 statement. (Similarly, an M1 statement might be '.Spectre.'s are the same as .ghost.s', or would this be a mistake?)

[21] I know that this account, as it stands, would at the very least need to be both tidied up and filled out in point of detail. It represents, however, an account which I thought you might have in mind on the basis of reading your remarks on the ambiguity of the term 'psychological'. I would be immensely grateful to learn if you think that I am on the right track, and for any suggestions and comments you might have.

[22] There are two other points in your letter on which I am not clear and about which I thought I might write you. One is the question that I asked in my earlier letter about what you meant by "the conditioning is itself caught up in a conceptual framework." If I understand your account, it runs as follows. If (a) 'I have a thought that-p' expresses direct self-knowledge, then both (b) the uttering was the result of a conditioned response and (c) "the conditioning is itself caught up in a conceptual framework" (Castañeda Correspondence, April 3, 1961, #13). But if (c), then (d) the person who utters 'I have a thought that p' must recognize the speech act as a report (EPM, Sect. 8). Finally if (d), then (e) the speaker recognizes that such reports are reliable symptoms of the person who reports having the thought that-p (EPM, Sect. 8). If this is correct so far, then my question is whether (e) is a sufficient condition for (c), and whether (c) and (b) are jointly sufficient for (a). Put differently, is any more involved in what you call meta-thinking (Castañeda Correspondence, Dec. 8, 1961) than conditions (b) and (e)?

[23] It was on the basis of this same page in your correspondence with Castañeda that I took your view to require "that pre-Joneseans do not think 'in the full sense'." What I believed that you had in mind might be suggested by saying that 'John thinks that-p, but he is never able to be aware of what he is thinking' is incoherent. You write that pre-Joneseans "do not have the concept of thought in the full sense," and that for this reason "the scope of their thinking is substantially restricted." Is what you meant by this (and by the passage at the top of the page 17 [Dec. 8, 1961] of your correspondence with Castañeda) no more than that pre-Joneseans do not think about their own thoughts (or, for that matter, about others' thoughts), and for this reason they "do not have the concept of thought in the full sense"? If so, might one then say that the respect in which they lack the concept of thought in the full sense is that whatever they think of as thoughts, they do not think of them as inner episodes? (That is, that their concept of thought is no more nor less than a concept of intentional items.)

[24] Let me once again thank you for having read my earlier typescript, and for your very helpful comments on it. Let me also apologize for the undue length of this letter; I hope that the ideas I have sketched reflect, to some extent, what you had in mind in your letter, and I shall be immensely grateful for any help you can give me concerning respects in which they do not.

Sincerely yours,
David Rosenthal

[Back to Contents]


November 8, 1965

Dear Mr. Rosenthal,

Thank you very much for your clear and searching letter. You are, indeed, on the right track, though I have some reservations, as you will see, about the exact way the points are to be made.

1. To begin with, all pre-Jonesean statements are (by definition) Rylean in character--at least as far as their descriptive content is concerned. The same will be true a fortiori of pre-Jonesean statements. Thus, though we may allow pre-Joneseans to conceive of intentional items in a way which makes abstraction from 'determinate factual character', we may not allow them to conceive of intentional items, as we can, in a way which makes abstraction from their Rylean character. Thus, their concept of an explanation of behavior in terms of intentional items coincides with their concept of an explanation in terms of what we would call Rylean intentional items, i.e., in terms (roughly) of what people say or are disposed to say. Thus, even though these items are specified by them in a way which, as in indirect discourse, abstracts from the determinate linguistic materials (including gestures, etc.) in which the relevant linguistic roles are embodied, they are not specified by them in a way which makes abstraction from their Rylean character.

2. Thus the contrast between explanation in terms of determinate Rylean role-players (thus, by quoting) and explanation in terms of Rylean items which are specified in terms of the role which they play, but not in terms of determinate linguistic materials, is as close as they could come to our contrast between explanation in terms of Rylean role-players, whether or not their determinate factual character is specified, and explanation in terms of Ockhamite role players (inner episodes).

3. Thus, pre-Jonesean psychological explanations are explanations in terms of what I have called 'thoughts-out-loud' and long- or short-term dispositions to have thoughts-out-loud. Of course, if we were to follow ordinary usage and restrict the phrase 'psychological explanation' to explanation in terms of thoughts (inner episodes) and dispositions to have them, then we would not speak of pre-Jonesean explanations of behavior as 'psychological', but rather as 'linguistic' or 'symbol-behavioral'. But if the argument of EPM is sound, there is every reason to extend the term 'psychological' to cover pre-Jonesean explanations--provided the necessary distinctions are drawn.

4. Thus, when you suggest (Oct. 2, 1965, par. 6) that pre-Jonesean psychological statements are not (in addition to being about items qua intentional) about marks and noises as marks and noises, I am unhappy. For though they do not require their subject matter to be marks or noises{10} (they may be gestures), let alone specific kinds of marks or noises, their subject matter must, in order for them to qualify as pre-Jonesean psychological statements, be conceived in Rylean terms.

5. Now since we can distinguish between explanation in terms of Rylean intentional items and explanations in terms of Ockhamite intentional items, it is possible for us to draw a distinction between at least three senses of 'psychological explanation'.

(1) Explanation in terms of Rylean intentional items: PR
(2) Explanation in terms of Ockhamite intentional items: PO
(3) Explanation in terms of intentional items, Rylean or Ockhamite: PRO

Of these three the third (PRO) corresponds roughly to your P1, the second (PO) to your P2. Thus (at par. 9) you characterize a P1 statement as a "statement about intentional items (no matter what their determinate factual character)" and a P2 statement as a "statement about items which exhibit both intentionality and a certain (psychological) determinate factual character," by which latter expression you mean their inner-episode-ness or Ockhamicity.

6. Thus, when we speak of a Jonesean statement as psychological, we can mean either that it is PR or that it is PRO. As for 'means'-statements, philosophers may stretch this expression by using the 'means'-rubric not only to give the sense of Rylean linguistic expressions (its normal use) but also the sense of Ockhamite linguistic expressions (inner speech). If we do so stretch it, then we have three senses of 'means'-statement.

(1) 'Means'-statement about Rylean items: MR
(2) 'Means'-statement about Ockhamite items: MO
(3) 'Means'-statement about either Rylean or Ockhamite items: MRO

Thc first of these (MR) is roughly your M2. It is not quite so easy to pick out a counterpart to your M1.

7. Now MR statements are the core of pre-Jonesean psychological statements in the sense that pre-Joneseans explain what people do in terms of the rubrics,

x thought-out-loud that-p

x was disposed to think-out-loud that-p

where

x thought-out-loud that-p

has the sense of

x (nonparrotingly) uttered a .p.

which, in turn, has the sense of

x (nonparrotingly) uttered something which means p

given that

'___' means ------

has the sense of

'____' are ._____.s

for to classify an utterance as a ._____. is to classify it in terms of its role in a language-life game, i.e., in terms of its place in a system of language entry transitions, language departure transitions and intralinguistic moves.

8. Thus, there is a perfectly legitimate sense in which MR statements are psychological statements. They are PR statements.

9. Again, MO statements are PO statements. To say of a thought that it means that-p is to classify it as an inner episode which is doing a job analogous in important respects to that which thinkings-out-loud that-p do in behavior built around spontaneous Rylean intentional episodes. Such inner episodes, furthermore, are conceived to be links in the explanation of, among other things, the occurrence of thinkings-out-loud that-p and of short term propensities to think-out-loud that-p.

10. Now it is clear from the above that an MR statement is not a PO statement, nor is it to be analyzed in terms of PO statements. Thus when you write (Oct. 2, 1965, par. 11) "neither 'means'-statements (in sense 1) nor 'means'-statements (in sense 2) either are, or are to be analyzed in terms of, psychological statements in sense 2," I agree with this in so far as it concerns 'means'-statements (in sense 2). As for 'means'-statements in sense 1 the situation is less clear. For the class of 'means'-statements about either Rylean or Ockhamite items includes the class of 'means'-statements about Ockhamite items and these are psychological statements in your sense 2. They are "about items which exhibit both intentionality and the sort of determinate factual character (whatever it may be) that psychological items exhibit." Here, I take it, by the expression "the sort of determinate factual character . . ." you have in mind the determinate inner-episode-ness, whatever it may be, by virtue of which it does its job.

11. Thus, I would prefer to say that from the fact that a statement is a 'means'-statement in your sense 1 it does not follow that it either is or is to be analyzed in terms of a psychological statement in your sense 2. For the class of 'means'-statements about either Rylean or Ockhamite items also includes the class of 'means'-statements about Rylean items, and these, as we saw, neither are nor are to be analyzed in terms of Ockhamite psychological statements--my PO statements, your P2 statements.

12. You now go on to explore what Chisholm's claim might be, against the background of the sort of distinctions we have been drawing. You give it an initial formulation as

'means'-statements may be (or are to be) analyzed in terms of, or entail, psychological statements.

You continue by pointing out that Chisholm does not draw the sort of distinctions we have drawn, and you raise the question whether Chisholm's claim can be made more precise by using these distinctions.

13. You then write:

It is clear that by ' "means"-statement' Chisholm means the same as either 'M1 statement' or 'M2 statement'. For the alternative is that he is talking about marks and noises as marks and noises. But it is just this that he appears to wish to deny by saying that "linguistic entities (sentences, etc.) are also intentional" (CSC, Oct. 3, 1956, (C-2)).

Clearly, for Chisholm a 'means'-statement does not simply characterize marks and noises in terms of mark or noise characteristics (e.g., pitch, shape, color, intensity, etc.). On the other hand he surely thinks that the subject of a 'means'-statement is always a linguistic item in the literal sense (as contrasted with the analogical sense in which Ockhamite items are 'linguistic'). Is not, for Chisholm, a subject of a 'means'-statement always a mark or a noise or a gesture (or class of such?) and, hence, Rylean in character? I conclude, provisionally, that a 'means'-statement, for Chisholm, is always about a Rylean item.

15. If so, then according to my account of the 'means'-rubric, the role attributed to a Rylean item by a 'means'-statement must be a Rylean role, thus

'Himmel's (in German) are .sky.s

uses a Rylean 'sky' in our language to characterize the job Rylean 'Himmel's do in German.

16. Now the Rylean 'means'-statement

'Himmel' (in German) means sky

has as its 'relational' counterpart (see "Notes on Intentionality")

'Himmel' (in German) stands for the concept sky.

If this is put as

'Himmel' (in German) expresses the concept sky

there is danger of confusing the sense of 'express' in this context with the relation of manifesting, and 'concept of . . .' in this context with 'ability to have thought-episodes about . . .' The result would be to confuse

'Himmel' (in German) stands for the concept sky

with

'Himmel' (in German) manifests an ability to think about the sky.

This confusion would infect the original 'means'-statement, generating the idea that

'Himmel' (in German) means sky

is to be analyzed as something like

German sentences containing 'Himmel' give expression to thought episodes about the sky.

17. You write that

Similarly, by 'psychological statement' Chisholm must mean the same as either 'P1 statement' or 'P2 statement'. For he denies that he is talking about merely inner episodes (no matter whether intentional or not) when he says that "thoughts (i.e., beliefs, desires, etc.) are intentional" (CSC, Oct. 3, 1956, sentence (C-1)).

My guess is that Chisholm so uses 'psychological' that it includes items which I would deny to be intentional (e.g., sense impressions, tickles, pains, etc.)--cf. my discussion of the pseudo-intentionality of sense impressions in EPM. Unless I am very much mistaken, he thinks that they have intentionality, or confuses a broader sense of 'intentionality' in which non-conceptual items can have intentionality, with a narrow sense in which they cannot. But this is an aside. In any event, I agree with you that "it is reasonable to suppose that he is claiming that M2 statements entail P2 statements."

18. In other words, he is claiming that 'means'-statements about Rylean items entail psychological statements about Ockhamite inner episodes. In my terminology, he is claiming that MR statements entail PO statements.

19. The only sense in which I am prepared to grant that MR statements entail PO statements is that in which statements about bricks entail statements about molecules.

20. You proceed (Oct. 2, 1965, par. 16) to raise the possibility that Chisholm is to be understood as "claiming that M2 statements entail P1 statements" (i.e., that statements about linguistic intentional items entail statements about intentional items, which do not specify that they are linguistic). If this were his thesis it would, of course, be correct. You find that

this diagnosis seems to gain some plausibility from the following considerations. . . . He does not, so far as I can discover, speak of thoughts as the sort of item which has a certain determinate factual character, nor does he speak of thoughts as inner episodes. It appears that Chisholm may be using 'psychological statement' to mean, in effect, the same as 'statement about intentional items without any determinate factual character.'

I take it that by "without any determinate factual character" you mean without specifying their determinate factual character. If so, then Chisholm would be holding that a statement which made no reference to inner episodes, or dispositions pertaining to such, might properly be called a psychological statement. You suggest that he might have, at the back of his mind, the Fregean sense of 'thought' (Gedanke) in which this term refers not to inner episodes, but to certain abstract entities, so that to say that a piece of verbal behavior expresses a thought would not be to say that it is related to inner episodes, but rather to say that it stands in a direct relation to a nonmental, public, abstract entity, the sort of thing that Platonists call a proposition.

21. But the above can be interpreted either as a tough- or as a tender-minded thesis. The tough thesis would be that there are no thoughts in the sense of non-Rylean inner episodes. This would amount to a philosophical behaviorism which differs from the usual form only by postulating an irreducible relation of 'expressing' to obtain between meaningful verbal behavior and Fregean abstract entities. The tender-minded version would be that both inner episodes and verbal behavior can 'express' Fregean abstract entities.

22. The tough-minded view would hold that 'means'-statements imply psychological statements because statements about what marks and sounds mean imply statements about meaningful verbal behavior, which latter are psychological not because they relate verbal behavior to thoughts as inner episodes, but because they relate verbal behavior to Fregean abstract entities. Clearly, this tough-minded view is simply the pre-Jonesean outlook transformed into a self-conscious philosophy which misconstrues meaning as a relation.

23. Needless to say, I find it difficult to believe that Chisholm is a Fregean pre-Jonesean. But it would be interesting to press him on this point. As for the tender-minded version, his thesis could approach mine if it amounted to the claim that 'means'-statements about marks and sounds are to be analyzed in terms of psychological statements in the sense of Ryle-Frege statements about verbal behavior and dispositions to behave; and also imply (by virtue of the connection of verbal behavior and dispositions to behave with thoughts in the sense of inner episodes), psychological statements in the more ordinary sense.

24. My own conviction is that if confronted with these distinctions, Chisholm would claim that verbal behavior gets its relation to Fregean entities indirectly by virtue of being the manifestation of inner episodes which alone are directly related to these entities, and that he would correspondingly claim that verbal behavior and dispositions to behave are psychological only by virtue of the fact that they manifest such inner episodes. Thus, as I see it, he is claiming that what I have called PR statements are to be analyzed in terms of PO statements, and hence that MR statements are to be analyzed in terms of PO statements. In other words, he is claiming that M2 statements are to be analyzed in terms of P2 statements, which is the thesis we have already rejected.

25. It seems to me to be clear that the distinctions we are developing enable a spelling out of the issues. I look forward eagerly to your reactions.

26. As for your concluding questions, you are again on the right track with respect: to my views on direct self-knowledge. The only point I would want to add is that the avowal 'I have a thought that-p' (1) asserts the occurrence of a thought that-p, (2) is a reliable symptom of the occurrence of such a thought, and (3) gives overt expression to a metathought (inner episode) which is an .I have a thought that-p., exactly as the candid assertion 'it is raining' gives overt expression to a non-metathought which is an .it is raining.. In other words, we post-Joneseans explain the connection between thoughts that is raining and reports that one has a thought that it is raining, in terms of a connection between thoughts that it is raining and apperceptive metathoughts that one has a thought that it is raining. Thus I cannot agree that nothing more "is . . . involved in what [I] call meta-thinking . . . than conditions (b) and (e)."

27. As for your next to last paragraph, I would not say that "their concept of thought is no more nor less than a concept of intentional item" not because it is false, but because it is misleading. It must be stressed that their concept of thought is a concept of thought-out-loud. In other words we must emphasize that their concept includes only Rylean intentional items--though, of course, they do :not have the contrast Rylean versus non-Rylean.

Let me say, once again, how admirably lucid I found your letter. I hope you will find my comments helpful.

Sincerely yours,

Wilfrid Sellars

[Back to Contents]


December 17, 1965

Dear Professor Sellars:

Thank you very much for your detailed letter of November 8. I was delighted to learn your reactions to my October 2 letter, and have found your remarks a very great help in clarifying the points of discussion. I was also grateful to learn of several ways in which my distinctions did not accurately capture what you had in mind. Because of this initial misunderstanding on my part, I think it may be worthwhile to say something more about these distinctions.

1. I must first apologize for an unfortunate phrase which I used throughout my letter. On Oct. 2, 1965, paragraph 7, I describe 'means'-statements (in sense (1)) as statements which are "about intentional items (no matter what their determinate factual character is)," and at paragraph 9, I describe psychological statements (in sense (1)) in the same way. This phrase does not express what I intended. Again, at paragraph 17, I considered statements "about intentional items without any determinate factual character." In paragraph 20, you suggest that "by 'without any determinate factual character' . . . [I] mean without specifying their determinate factual character." This is correct, and also correctly expresses what I had in mind at paragraphs 7 and 9, but to make the point more clearly, I might say that such statements fail to specify anything at all about the determinate factual character of the items in question.

2. As I envisaged the distinction, therefore, the statement 'George said that-p' is not an M1 statement. For although the statement does not specify the determinate factual character of the intentional item (that is, what words were uttered), it does specify that the determinate factual character of the item is Rylean.

3. Although I am no longer altogether happy with the notion of a P1 (or M1) statement, I believe that instances of such statements might be statements about Fregean intentional items, that is, Fregean Gedanken. For Fregean intentional items are abstract entities, and thus have no determinate factual character. Statements about them, therefore, will not specify anything about such factual character. (I no longer believe, as I went on to suggest at paragraph 20, that '.Spectre.s are the same as .ghost.s' is an M1 statement, for it clearly specifies that the items under discussion have a Rylean character, whatever it may be.) An instance of a statement about (only) Fregean intentional items might be 'The Fregean Gedanke that-p contains (or involves, or entails) the Fregean Gedanke that-q'.

4. The fact that pre-Jonesean psychological statements are about certain events is one reason why such statements cannot, as you point out in paragraph 4, be P1 statements.

5. In paragraph 1, however, you write that "we may not allow . . . [pre-Joneseans] to conceive of intentional items, as we can, in a way which makes abstraction from their Rylean character." (I assume that this amounts to roughly the same thing as your writing, in paragraph 4, that the "subject matter [of statements] must, in order for them to qualify as pre-Jonesean psychological statements, be conceived in Rylean terms" [emphasis mine].) Am I correct in supposing that you mean to rule out the possibility that pre-Joneseans might use their enriched Rylean language to talk about abstract entities, and, a fortiori, about Fregean abstract entities?

6. Since a statement which specifies anything about the determinate factual character of the items involved cannot be a P1 or M1 statement, the classes of PRO and MRO statements cannot correspond, respectively, to the classes of P1 and M1 statements. For all PR and PO statements are PRO statements, and all MR and MO statements are MRO statements. But PR, PO, MR, and MO statements all specify at least the sort of determinate factual character involved. (Although your PR/PO/PRO distinctions are among various sorts of explanations in terms of, I assume they may be recast as distinctions among various sorts of statements about (and in terms of.){11}

7. Thus I think that your reservations about my statement which you quote in paragraph 10 may have resulted from a confusion which was due to my unclear account of P1 and M1 statements. For although M1 statements neither are, nor are to be analyzed in terms of, P2 statements, I agree, as you suggest in paragraph 11, that "from the fact that a statement is . . . [an MRO statement] it does not follow that it either is or is to be analyzed in terms of a psychological statement in . . . [my] sense 2."

8. In any event, I think that the distinctions developed in your letter are far preferable for the purpose of giving an account of the issues to the ones suggested in my October 8 letter. In what follows, therefore, I shall use your terminology, as I understand it. (I agree that P2 and PO statements are the same, just as M2 and MR statements are.) It will be useful for what I would like to say, however, to add a fourth class of psychological statements for consideration:

(4) Statements about Fregean intentional items: PF

(Here I do not wish to exclude statements which, although about Fregean abstract entities, are also about items conceived as having a certain (sort of) determinate factual character.)

9. I would like, then, to try again to characterize Chisholm's position as I understand it. The clearest shortcoming of my earlier characterization was that it failed to account for Chisholm's conviction that thoughts are uniquely a source of intentionality. Thus I entirely agree when you write that

if confronted with these distinctions, Chisholm would claim that verbal behavior gets its relation to Fregean entities indirectly by virtue of being the manifestation of inner episodes which alone are directly related to these entities, and that he would correspondingly claim that verbal behavior and dispositions to behave are psychological only by virtue of the fact that they manifest such inner episodes (paragraph 24).

10. You continue by writing that "he is claiming that . . . PR statements are to be analyzed in terms of PO statements, and hence that MR statements are to be analyzed in terms of PO statements." Although Chisholm may indeed be claiming this, it does not seem to me that such a claim is embodied in the claims attributed to him above. For it seems to me possible to interpret that latter claims as being about no more than matters of fact, and not about the appropriateness of analyzing certain statements in terms of others. So the task remains of trying to give an account of those of Chisholm's claims described in the previous paragraph.

11. From the fact that George nonparrotingly utters (or writes) 'p', together with Jones' theory, we can infer, I believe, that George had some thoughts. (If he speaks or writes sincerely, for example, then he had the thought that-p.) Let us suppose, now, that there simply are no thoughts. From this fact together with the truth of Jones' theory, we can infer that if there are any utterances or inscriptions, then they are not meaningful (in the sense of nonparrotingly produced). We cannot, however, turn this inference around. From the fact that there are no nonparrotingly produced utterances or inscriptions together with the truth of Jones' theory, we cannot infer that there are no thoughts.

12. It is reasonable to suppose, I think, that if an utterance or an inscription is not meaningful (in the sense of non-parrotingly produced), then it is not intentional. For such an utterance would not be produced as tokening a properly formed expression in a language, nor would it properly be understood as tokening such an expression (although it might, mistakenly, be so understood). For this reason, it would be properly described as not being about anything, and hence as not intentional. (It does not seem to me that the converse holds, for an utterance may fail to be about something by failing, in fact, to token a properly formed expression, even though it was nonparrotingly uttered.)

13. If tllese remarks are correct, then we can infer the following. From the fact that there are no thoughts together with Jones' theory it follows that if there are any utterances or inscriptions, they are not intentional. On the other hand, from the fact that there are no intentional utterances or inscriptions, together with Jones' theory, we cannot infer that thoughts are not intentional.

14. On Oct. 3, 1965 of CSC, Chisholm explicates his claim that "thoughts are a 'source of intentionality' " as follows: "Thoughts would be intentional even if there were no linguistic entities. . . . But if there were no thoughts, linguistic entities would not be intentional" ( (C-4), (C-5) ).

15. Chisholm goes on to caution that he is not talking here about the proper analysis of one sort of statement into another (CSC, Oct. 3, 1965 (C-6)). The alternative seems to be that he is talking instead about causal relationships. Thus I would like to suggest that one way of spelling out what he is claiming is provided by my remarks in paragraphs 11 through 13. That is, Chisholm may be viewed as claiming the following. If we take certain statements about causal relations as axioms, then 'There are no thoughts' entails 'If there are any utterances or inscriptions, they are not intentional', but 'There are no linguistic entities' does not entail 'If there are thoughts, they are not intentional'. One suitable choice of causal axioms would, of course, be Jones' theory. So Chisholm might be viewed as in effect taking for granted the truth of Jones' theory, even though in fact he does not state the causal axioms which he is presupposing. At the least, it seems that Chisholm's claims are certainly, thus far, compatible with the truth of Jones' theory.

16. These remarks may possibly provide an account of the claims Chisholm makes in the seven labeled sentences [Oct. 3, 1956 of CSC]. It remains, however, to consider his (independent) claim that 'means'-statements are to be analyzed in terms of psychological statements, and not conversely (CSC, Aug. 24, 1956).

17. In my October 2 letter I suggested that we might view Chisholm's position as being partially the result of an implicit belief "that (at least in certain respects) presentday psychological statements behave as if they were . . . P1 statements". We would thus understand his claim that 'means'-statements are to be analyzed in terms of thoughts as amounting to the claim that MR statements entail statements about Fregean Gedanken, that is, that they entail PF statements.

18. You suggest in paragraph 21 that such a claim is compatible with each of two sorts of positions--a tough-minded thesis and a tender-minded thesis. You describe the former as involving the assertion that "there are no thoughts in the sense of non-Rylean inner episodes." Thus whenever Chisholm talks about thoughts, he would, on this reading, be talking either about Rylean items or about Fregean abstract entities.

19. In spite of my having written that Chisholm "does not, so far as I can discover, speak of thoughts as the sort of item which has a certain determinate factual character, nor does he speak of thoughts as inner episodes" (vide your paragraph 20), he does speak of people having thoughts, and of the thoughts of people. (For example, in the last sentence of CSC.) This seems to rule out the possibility that he is talking about Fregean abstract entities tout court, for it is difficult to imagine that he supposes that people have abstract entities of any sort whatever. Thus since it seems unlikely that he is using 'thought' to talk about Rylean items, I share your skepticism with regard to Chisholm's position being the tough-minded thesis.

20. What you describe as the tender-minded thesis seems to hold more promise. This thesis involves, if I understand you correctly, that there exist both Ockhamite and Rylean intentional items, and that, although MR statements do not entail PO statements, both MR and PO statements do entail PF statements. Thus, f'or example, 'He had a thought that-p' would be held to entail 'He had an inner episode which stands in relation to the Fregean Gedanke that-p', just as ' " . . . " means - - -' would be held to entail ' " . . . " stands in relation to the Fregean Gedanke---'.

21. These claims would indeed be independent of the question of what sort of items are related directly, and what sort are related indirectly, to Fregean abstract entities. For this question would be settled by appealing to considerations of the sort described in paragraph 15. Thus, a certain sort of intentional item would be said to stand in direct, immediate relation to Fregean items if, and only if, from the nonexistence of such items, together with the truth of Jones' theory, it follows that other sorts of items (which are in fact intentional) would not, if they existed at all, be intentional. Intentional items would then be said to be indirectly, and in a mediated way, related to Fregean items if, and only if, they do not satisfy the above condition. And we could express this by saying that the former (but not the latter) sort of items are a source of intentionality.

22. In paragraph 23 you write that

As for the tender-minded version, his thesis could approach . . . [yours] if it amounted to the claim that 'means'-statements about marks and sounds are to be analyzed in terms of psychological statements in the sense of Ryle-Frege statements about verbal behavior and dispositions to behave; and also imply (by virtue of the connection of verbal behavior and dispositions to behave with thoughts in the sense of inner episodes), psychological statements in the more ordinary sense.

I am unclear concerning the second part of this claim, because I am not sure how to understand the force of 'imply', but it seems to me that you may have in mind the following. The connection between Rylean and Ockhamite items which you mention is, I assume, a causal connection. If so, then perhaps what you are saying might be expressed as follows: that MR statements together with the truth of Jones' theory, logically entail certain PO statements. I shall in any case return to this idea in paragraphs 28-30.

23. The chief feature of this thesis in which it departs from yours, as I understand it, is that it (like the tough-minded thesis) "misconstrues meaning as a relation" (paragraph 22). For whereas you enrich Rylean discourse so as to include semantical categories by means (essentially) of the semantical device of dot-quotes, the tender-minded thesis is able to deal with intentionality rather because it posits Fregean abstract entities. Thus on the tender-minded view MR statements entail PF statements. And if I am correct in my suggestion at the end of paragraph 5, pre-Jonesean simply do not have the linguistic resources required to talk about abstract entities at all. (Since your account in "Notes on Intentionality" seems to involve that MR statements entail PF statements, it may be more accurate to say that the tender-minded thesis, unlike yours, must involve claiming that such an entailment holds if it is to be able to account for intentionality.)

24. These differences raise the question as to what advantages the use of dot-quotes has as against the introduction of Fregean abstract entities, and the related (and more basic) question as to whether or not meaning is, after all, a relation. If the tender-minded thesis does accurately represent Chisholm's position, then his qualms at this point might be expressed by the question whether it is possible to explicate the term 'role' used (in "Notes on Intentionality,") in explicating the function of dot-quotes. In the context of the present discussion, I would be more interested in learning why, aside from reasons of parsimony, pre-Joneseans must be disallowed from having the concept of an abstract entity.

25. These questions aside, however, attributing the tender-minded thesis to Chisholm has the following disadvantage. Since Chisholm uses 'thought' to cover both Fregean and Ockhamite items, either he is using the term ambiguously, or he is simply confusing two senses.

26. Your remarks in paragraph 16, if I understand them correctly, are intended to make plausible the idea that Chisholm might have in fact simply confounded two senses of 'thought'. In terms of your distinctions in "Notes on Intentionality," Chisholm would have confused the logical and causal senses of 'express', and the sense of 'thought' as thinking with its sense as proposition. (The way in which I tried to make this confusion plausible was slightly different, I think, for it rested on the idea that "Chisholm is so set on stressing the intentionality of thoughts that he ignores their determinate factual character as inner episodes". Thus Chisholm would be viewed as having allowed himself, at certain times, to treat inner-episode thoughts as if they were intentional items which altogether lack a determinate factual character, that is, as if they were Fregean abstract entities. ) In any event, if Chisholm held the tender-minded view, then his claim as indicated in paragraph 16 would, properly understood, be about inner episodes and not about Fregean abstract entities.

27. If we must choose between supposing that Chisholm has used 'thought' ambiguously (but without even so much as indicating that he has) and supposing that he has just confounded two senses of the term, the latter does seem more likely. Having backed off somewhat, however, in paragraph 19, I think it may be worthwhile looking for an interpretation which does not so indict him.

28. If my suggestions in paragraph 15 are correct, then it follows that 'There exist Rylean intentional items' (logically) entails 'Either there exist Ockhamite intentional items or Jones' theory is incorrect'. But if we take Jones' theory as axiomatic, then it follows that 'There exist Rylean intentional items' entails 'There exist Ockhamite intentional items' tout court. But any MR statement entails that there are somc Rylean intentional items, and 'There exist Ockhamite intentional items might be taken. as itself a PO statement. So if we take Jones' theory as axiomatic, then any MR statement entails at least one PO statement.

29. But we can go farther than this, along the lines of my suggestion in paragraph 22. If Jones' theory is true, then if '. . .' means --- then whenever '. . .' is nonparrotingly and sincerely uttered it (causally) expresses a thought episode that ---. Thus if we take Jones' theory as axiomatic, then ' " . . . " means ---' entails 'Whenever '. . .' is nonparrotingly and sincerely uttered, it (causally) expresses a thought episode that ---'. But this last statement is about Ockhamite (as well as Rylean) intentional items. It is thus a PO statement. In general, then, if we take Jones' theory as axiomatic, then MR statements entail PO statements.

30. I assume that this account does not depart very far from your own, for you write that "the only sense in which . . . [you are] prepared to grant that MR statements entail PO statements is that in which statements about bricks entail statements about molecules. Thus one might claim that if we adopt '(x) (x is a brick only if x is a set of molecules)' as an axiom, then statements about bricks will entail statements about molecules (vide your paragraph 19).

31. Moreover, this account is compatible with one which does not admit Fregean abstract entities. (On the other hand, of course, Chisholm's position, if so understood, would not involve any account of the intentionality of mental or linguistic items; it would be purely descriptive.)

32. Against this background, we might diagnose Chisholm's error as follows. Being a post-Jonesean, Chisholm uses a language which presupposes Jones' theoretical innovations. But he so takes for granted the truth of Jones' theory, that he treats the claims made by Jones as if they were logical axioms (or rules of inference). Thus he makes assertions about entailment (or logical derivability) relations which if true, presuppose the use of Jones' claims as axioms (or rules of inference). His error would thus consist in his presupposing certain causal truths without making them explicit, and his treatment of them as if they were logical truths. (The strength of his conviction that such axioms may be taken for granted might be indicated by his insistence that pre-Joneseans, if they "gave just a little bit of thought to the semantical statements they make," would see that such statements entail statements about thought-episodes [CSC, Nov. 19, 1956, emphasis mine].)

33. I am not by any means completely persuaded by the account I have sketched in the last five paragraphs, but I think it may capture what Chisholm is claiming more accurately than an account which commits him to having talked about Fregean Gedanken. (He writes, for example, in the third footnote of "Sentences about Believing" that he wishes to prevent confusion of "what . . . [he wants] to say with what Frege had to say about meaning." This may make it less plausible to suppose that he confuses thought-episodes with the abstract entities in terms of which Frege discusses meaning. But this is, of course, not at all conclusive.) In any case, I would be very grateful to learn your reactions to these suggestions, as well as to my earlier ideas.

34. I have one final remark, which is not directly relevant to the foregoing discussion. In paragraph 17 you write that you believe that "Chisholm so uses 'psychological' that it includes items which . . . [you] would deny to be intentional (e.g., sense impressions, tickles, pains, etc.) . . . . " Although I agree with this, I do not think that Chisholm wants to claim that all psychological items are intentional, that is, about something. For he is careful, in formulating his thesis of intentionality, to restrict it to "certain psychological phenomena--in particular . . . thinking, believing, perceiving, seeing, knowing, wanting, hoping, and the like . . . " ("Sentences About Believing",{12} [emphasis mine]). And although he does include peceiving and seeing, I think that he is clear that he is talking about episodes which involve not only sense impressions but a judgmental component as well. Thus he claims, for example, that seeing a cat involves taking something to be a cat (ibid, p. 39). And in general, it seems to me that he avoids, for the most part, talking at all about the sort of items to which, in "Being and Being Known," you ascribe pseudo-intentionality. But I may well have overlooked features of his position that are relevant here.

I want again to express my appreciation for your very helpful and detailed letter. I have very much enjoyed studying it, and it has clarified a great number of points for me, as well as helping me to see more clearly, I think, the general setting of your position. I shall be immensely grateful to learn any reactions you may have to my suggestions.

Sincerely yours,

David Rosenthal

[Back to Contents]


January 4, 1966

Dear Rosenthal,

I was delighted to receive your admirably lucid letter of December 17. From my point of view, at least, we are making exceptional progress in clarifying the ambiguities and misunderstandings which prevented the Chisholm correspondence from achieving more decisive results.

Now for some specific reactions.

1. Your paragraph 3 introduces a number of themes which run throughout your letter. The most important of these is the concept of a 'Fregean intentional item'. (I assume that our aim is not Frege exegesis, but to spell out and evaluate the implications of the distinctions drawn in "Notes on Intentionality" for the questions with which this correspondence began.) I am not clear whether by 'Fregean intentional items' you are referring to particulars (e.g., episodes) or to nonparticulars. Your characterization of them as abstract entities strongly suggests the latter; but your use of the phrase 'determinate factual character' which we have been using in connection with psychological episodes and dispositions suggests the former. Since your paradigm seems to be a Fregean Gedanke, I suppose you are saying (a) that Fregean Gedanken do not have 'determinate factual character'--which is reasonable enough, though I shall comment further below; (b) That a particular (e.g., episode) which gains its intentionality through its relation to a Fregean Gedanke is not thereby pinned down to one determinate factual character rather than another--or do you mean that it need have no determinate factual character at all? Again, by 'determinate factual character' you might mean such characters as Ryleanism or Ockhamicity in general, or you might have in mind more appropriately determinate forms of these (e.g., on the Rylean side, specific patterns of meaningful English behavior).

2. My view, of course, is that every particular which has intentionality has, and necessarily has, a determinate factual character. A Rylean intentional item must have a determinate factual character by virtue of which it is a rule to play a certain Rylean role. An Ockhamite intentional item must have a determinate factual character (though we may not know what it is, save in very generic terms) by virtue of which it is able to play a certain Ockhamite role (conceived by analogy with a certain Rylean role). What, then, in my view, would Fregean intentional items be? The unpolished answer is that they are items which are either Rylean intentional items or Ockhamite intentional items.

3. Let me introduce three different quotation devices: dot quotes for Rylean items, diamond quotes for Ockhamite items and dot-diamond quotes for Fregean items. Consider the following statements

(a) a is a .not.
(b) a is a <>not<>)
(c) a is a .<>not.<>

The first, (a), tells us that a is an item playing the Rylean role played in our language by 'not's; (b) tells us that a is an item playing in inner speech an Ockhamite role analogous to that played in our language by 'not's (and in any language by .not.s); (c) tells us that a is either a .not. or a <>not<>. In each case a must have a determinate factual character in order to be a .not., a <>not<>), or a .<>not.<>.

4. Notice that each of the three statements (a), (b) and (c), has the form . . . is a K', so that what corresponds to 'K' is a predicative expression rather than an abstract singular term of the form 'K-kind'. Now just as

Fido is a dog

has as its 'platonistic' counterpart

Fido is an instance of dog-kind.

so (a), (b), and (c) have as platonislic counterparts

(a') a is an instance of .not.-kind
(b') a is an instance of <>not<>-kind
(c') a is an instance of .<>not.<>-kind

which correspond, on my analysis, to

(a'') a stands for Rylean negation
(b'') a stands for Ockhamite negation
(c'') a stands for Fregean negation

Thus, just as, on my analysis, the move down from the platonistic

Fido is an instance of dog-kind

to

Fido is a dog

is to be reconstructed, in first approximation, as

Fido is an instance of dog-kind
The .is a dog. is true of .Fido.
The Fido is a dog is true
Fido is a dog

so the move down from the platonistic

a stands for Rylean negation [ = a is an instance of Rylean negation]

to

a is a .not. [ = (Rylese) a means not]

is to be reconstructed as

a stands for Rylean negation
The .is a .not.. is true of .a.
The .a is a .not.. is true
a is a .not..

5. Notice that, as the above examples bring out, not all abstract entities (in our examples, kinds) are instantiated by intentional items. Negation is, dog-kind is not, at least as they occur in our examples. Thus the distinction between Rylean, Ockhamite, and Fregean abstract entities with which we are concerned applies only to a subset of abstract entities--those which have as their instances psychological items, that is, either Rylean or Ockhamite particulars. If, therefore, we assume that our (post-Jonesean) use of '-kind' (also '-ity','-hood','-ness', 'that-', etc.) can be reconstructed as dot-diamond (Fregean) quoting, both

dog-kind

and

.<>not.<>-kind

would be Fregean in the sense that they both use Fregean quotes, thus

The .<>dog.<>
The .<>(.<>not.<>).< >

(I throw in the parentheses to make the iterated quotes more perspicuous), whereas the latter but not the former is a Fregean quoting of a Fregean quote. Thus the latter, but not the former, can be contrasted with

<>not<>-kind

which, given the above interpretation of '-kind' translates into

The .<>(<>not<>).<>

which is a Fregean quoting of an Ockhamite quote.

6. Notice that the above assumption about the role of '-ity', '-hood', '-ness', etc., is appropriate only to the post-Jonesean era. Obviously neither <>-quoting nor .<>-quoting can occur in pre- Jonesean times, for they are conceptually bound up with the Jonesean revolution. Pre-Jonesean philosophers, confronted with the problem of abstract entities, might hit upon the device of reconstructing singular-term-making prefixes and suffixes as a use of quotes to make sortal words, together with that use of 'the' which turns sortal words into distributive singular terms, but the quotes would have to be, from our point of view, Rylean, for the sortal words they formed would appIy to overt linguistic behavior. Thus a pre-Jonesean Sellars would construe

Dog-kind

as

The .dog.

and would have no use for

The <>dog<>

or

The .<>dog.<>.

7. Against this background, let us take another look at the statement I quoted from your paragraph 3: "Fregean intentional items are abstract entities, and thus can have no determinate factual character" (I now underline the 'thus'). It seems likely that Frege did not think that his abstract entities had 'determinate factual character', if 'factual' is to have anything like the sense of 'empirical'. On the other hand, he would, I take it have granted that, though they could not be, e.g., red or noisy, they nevertheless could have structural characteristics, indeed that they must have structural characteristics. This enables me to bring out one of the strengths of my view. Consider

(d) Triangularity is noisy.

There is clearly something odd about this statement. Now on my view (d) is equivalent to

(d') The .<>triangular.<> is noisy

and this, in turn, to

(d'') .<>triangular.<>s are, as such, noisy.

In other words, the oddity of this predication can be explained as a clash between the implication of dot-diamond quotes and the predicate '(as such) noisy'. The dot-diamond quotes permit events of any kind to be .<>triangle.<>s in 'outer' or 'inner' speech provided that they can participate in the same structurally characterized moves and transitions as our word 'triangle'. The predicate '(as such) noisy' clashes with this permission. In this respect

The .<>triangular.<> is noisy

resembles

The pawn is made of ivory.

8. The above remarks on 'intentional items' and 'abstract entities' are designed to show that a reference to 'abstract entities' does not bypass the conceptual progression Rylean-Ockhamite-Fregean, where 'Fregean' stands for the common character of Rylean and Ockhamite roles. Bluntly put, the abstract entities which intentional items 'stand for' (i.e., instantiate) divide into Rylean (R), Ockhamite (O) and Fregean (RO) exactly as do the intentional items which 'stand for' (i.e., instantiate) them. Indeed, on the above analysis, this claim is a truism.

9. Thus, when, in paragraph 5, you ask "Am I correct in supposing that you mean to rule out the possibility that pre-Joneseans might use their enriched Rylean language to talk about abstract entities, and, a fortiori, about Fregean abstract entities?" The answer is a ringing "No"! I take it for granted that pre-Joneseans can speak of abstract entities (use abstract singular terms). These abstract entities are, however, from our point of view, Rylean abstract entities. Since, ex hypothesi, they cannot speak of Ockhamite abstract entities, it follows--and here is where 'a fortiori' belongs--that they cannot speak of Fregean abstract entities, if the concept of a Fregean abstract entity is the concept of what is common to Rylean intentional items and Ockhamite counterparts. If, therefore, my "Notes on Intentionality" is on the right track, one cannot minimize the revolutionary character of the Jonesean development by arguing that independently of this development pre-Joneseans could come to conceive of intentional items which are freed from Rylean limitation, by conceiving of them as standing for Fregean abstract entities.

10. Let me expand a bit on pre-Jonesean discourse about abstract entities. I pointed out above that of those abstract entities which can be said to have instances some are instanced by intentional items (as when an intentional item stands for [instantiates] the Gedanke that snow is white), while some are instantiated either by particulars which lack intentionality (e.g., triangularity) or by intentional items in some respect other than their intentionality (e.g., temporal precedence). Let me call the former intentional abstract entities and the latter nonintentional abstract entities. The pre-Jonesean conception of the nonintentional abstract entity triangularity, would be the conception of the .triangular..

11. It might be thought that since .triangular.s are Rylean intentional items, the account I am giving is incoherent, for how could triangularity be a nonintentional abstract entity, if triangularity is the .triangular. and .triangular.s are intentional. The answer is that an intentional abstract entity was defined as one which is instantiated by intentional items qua intentional, and the statements

(e) Triangularity is exemplified by pyramids, which are nonintentional items

and

(f) Triangularity = the .triangular.

are perfectly compatible with

(g) The .triangular. is an intentional item (i.e., .triangular.s are intentional items) .

For, according to my analysis, (e) has, roughly, the sense of

(e') The .triangular. is true of pyramids, and pyramids are nonintentional items

and this, in turn, the sense of

(e'') The .pyramids are triangular. is true, and pyramids are nonintentional items.

Thus, whereas in (g) the dichotomy intentional/nonintentional is considered with respect to that which is expressed by the construction which uses the word 'triangular', in (e) it is considered with respect to that which is expressed by the use of the word 'pyramid'.

12. If I am right, then pre-Joneseans can, by the use of their enriched Rylean language, make such statements as

(h) 'Dreieckig' (in G) means triangular
( i) 'Dreieckig's (in G) are .triangular.s
(j) The .pyramids are triangular. is true
(k) The .triangular. is true of pyramids.

Of these (k) is the reconstruction of

(k') Triangularity is exemplified by pyramids.

They can also say

(l) 'Dreieckig's (in G) are included in .triangular.-kind

which is the 'platonistic', or abstract, counterpart of (i), and thus belongs at the level of double dot-quoting, thus

(l') The .are .triangular.s. is true of .'dreieckig's (in G).

which is equivalent to

(l'') The'dreieckig's (in G) are .triangular.s. is true

which reduces, by the truth-move, to (i). Since (e) is the proposed reconstruction of

(m) 'Dreieckig's (in G) stand for triangularity

it follows that, according to the account I am giving, the word 'triangularity' has different senses in the two contexts

. . . stands for triangularity

and

. . . exemplifies triangularity.

In the former, it is at the level of double dot-quoting, in the latter of single dot-quoting. The fact that words appropriate to the gap in the second context refer, in their primary use, to nonintentional items is, as I see it, the root of the mistaken idea that abstract entities are independent of intentionality. Another root is the classical equation of the psychological with the I>individual-psychological, which led people to suppose that to equate the domain of the intentional with the domain of the psychological is to make unitelligible the intersubjectivity of the intentional. One must recognize the 'community' character of the psychological in order to appreciate the fundamental truth of conceptualism.

13. As for intentional abstract entities, our pre-Joneseans can, of course, conceive only of the Rylean variety. To see this we need only pull together some points made in paragraph 11. Thus, if we specify that an abstract entity is instantiated by intentional items qua intentional, we are specifying, in effect, that the expression for that abstract entity is at the double quote level of the available type of quote (R, O, or RO). For, on the above analysis,

(intentional items) exemplify, qua intentional, (abstract entity)

has the sense of

(intentional items) stand for (abstract entity)

and is the abstract counterpart of a statement having the form

(intentional items) are .---.s

which is the schema for the reconstruction of the Rylean 'means'-statement

(intentional items) means -------

which is available, ex hypothesi, to our pre-Joneseans. Thus, pre-Jonesean concepts of abstract entities generally are Rylean concepts, and pre-Jonesean concepts of intentional abstract entities are Rylean concepts of Rylean abstract entities. To reiterate a point made in paragraph 9, the ability of pre-Joneseans to conceive of abstract entities does not provide them with an independent route (independent, that is, of the Jonesean revolution) for arriving at a notion of an intentional item which is freed from Rylean restriction.

14. In paragraph 8 you express the belief that it is "useful for what [you] would like to say . . . to add a fourth class of psychological statements for consideration:

(4) statements about Fregean intentional items: PF."

You are quite right, for this addition enables a sharpening of the issues. Thus, if to conceive of PF items were not the same thing as to conceive of PRO items, then the possibility would remain that although pre-Joneseans cannot conceive of PO items, and hence, a fortiori of PRO items, they can conceive of non-Rylean psychological items, for they can conceive of items which are PF without being overt verbal behavior or dispositions to behave. (The question would remain, of course, as to just what these items are conceived by pre-Joneseans to be, if they are not construed as Ockhamite inner episodes.) The peculiar relevance of the problem of abstract entities to problems pertaining to intentionality and the mental should not be prejudged by a terminology which requires that all intentional items be either Rylean or Ockhamite or Rylean-Ockhamite.

15. In paragraph 11 you write "let us suppose, now, that there simply are no thoughts. From this fact, together with the truth of Jones' theory. . . ." I do not understand this. From the truth of the Jonesean theory, it follows that where there is nonparroting speech, there are thoughts. Notice that pre- Joneseans have a concept of nonparroting speech (thinking-out-loud), and that Jones' theory adds to this concept a theoretical framework which, by providing an explanation of nonparroting speech, generates a richer concept of nonparroting speech.

16. From the truth of Jones' theory one can, however draw the weaker conclusion that if on a particular occasion there are no thoughts, then on that occasion there is no meaningful (nonparroting) speech. And you are correct in saying that the reverse implication does not hold. Jones' theory allows, of course, for unexpressed thoughts.

17 In paragraph 13 you make what seems to be an illegitimate step from 'we cannot infer that there are no thoughts' to 'we cannot infer that thoughts are not intentional'. You overlook, perhaps, that it is an analytic feature of the Jonesean concept of a thought that thoughts are intentional.

18. Your point in paragraphs 14ff. begins to probe to the heart of the issue. You will remember that in commenting on Chisholm's (C-1) through (C-7) (CSC, p. 239) I write "it isn't so much that I disagree with your seven sentences, for I can use each of them separately, with varying degrees of discomfort, to say something which needs to be said. . . . It is rather that I am unhappy about the force they acquire in the overall framework in which you put them" (CSC, Oct. 19, 1956). The key sentence of your paragraph 15 is "So Chisholm might be viewed as in effect taking for granted the truth of Jones' theory. . . ." Exactly. Taken together with your paragraph 16, the chips are down.

19. Your definition of mediate and immediate relation to a Fregean item contains the passage "from the nonexistence of such items, together with the truth of of Jones' theory, it follows that other sorts of items (which are in fact intentional) would not, if they existed at all, be intentional." My comments in paragraphs 15-17 apply here as well.

20. Your interpretation in paragraph 22 is essentially correct.

21. You write, in paragraph 23, "on the tenderminded view MR statements entail PF tatements." Only in conjunction with Jones' theory is this true. You also repeat your suggestion of paragraph 5 that pre-Joneseans do not have the linguistic resources required to talk about abstract entities at all. This, of course, I have rejected.

22. Your paragraphs 24-25 raise the whole question of absract entities and their connection with intentionality. I agree that the issue is central, and have tried to throw some Iight on it in the earlier parts of this letter.

23. It is certainly true, as you remark in paragraph 28, that "if we take Jones' theory as axiomatic, then any MR statement entails at least one PO statement." But this doesn't help the pre-Joneseans.

24. Your remarks in paragraphs 30-32 are pretty much on target.

25. I agree, on the whole, with what you say in paragraph 34 about Chisholm's use of the term 'psychological'. I quite agree that he explicitly attributes intentionality only to "episodes which involve not only sense impressions but a judgmental component as well." 0n the other hand his equation in Perceiving {13} of sensing with being appeared to, and his failure explicitly to distinguish between a sense of 'appear' in which being appeared to has a conceptual character from a putative sense in which it does not, lead me to suspect that the intentionality of conceptual episodes and the pseudo-intentionality of sensory states have not adequately been distinguished in Chisholm's philosophical psychology.

On rereading this letter, I am afraid that it contains whole sections of a rather opaque character. But, then, I could scarcely have omitted the paragraphs on abstract entities; while if I had expanded them in the hope of reducing this opacity, this letter would have turned into "Abstract Entities." Every philosophical issue, seriously pursued, takes one everywhere. I can only hope that our other philosophical experiences have had enough in common to make this correspondence a genuine and fruitful dialogue. That, at least, is how it appears to me.

Sincerely yours,

Wilfrid Sellars


[Main Page] [Back to Contents]
* This correspondence was published in Intentionality, Mind, and Language, edited by Ausonio Marras (Urbana: University of Illinois Press, 1972): 461-503. It is placed on the Internet with the permission of Professor David Rosenthal.

Editor's Note: Sellars' dot-quotes are rendered by bold periods, and where Sellars uses the Greek character alpha, I have used an italicized 'a'. I have numbered (in brackets) Rosenthal's Oct. 2 letter, and have changed references to this letter by paragraph numbers.[Back]


Notes:

{1} Since these letters were written my ideas have crystallized, and I have expanded my earlier views on the disagreement between Sellars and Chisholm. See my Ph.D. dissertation, "Intentionality: A Study of the Views of Chisholm and Sellars" (Princeton University, 1968), ch. IV [D.M.R.] [Back]

{2} Philosophy of Science, 21 (1954), 204-228. Reprinted with revisions in W. Sellars, Science, Perception, and Reality (London: Routledge and Kegan Paul, 1964). [Back]

{3} Proceedings of The American Catholic Philosophical Association, 34 (1960), 28-49. Reprinted in Science, Perception, and Reality. [Back]

{4} This correspondence between W. Sellars and H-N. Castañeda is available at this site. [Text][ed.] [Back]

{5} (Hereafter abbreviated as EPM). In H. Feigl and M. Scriven, eds., Minnesota Studies in the Philosophy of Science, vol. I (Minneapolis: Univ. of Minnesota Press, 1956). Reprinted in Science, Perception, and Reality. This essay is available at this site. [Text] Parenthetical references will be to sections of the essay. [Back]

{6} "Abstract Entities," Review of Metaphysics, 16 (1963), 627-671; "Notes on Intentionality," The Journal of Philosophy 61 (1964): 655-65. [Back]

{7} This and the following three page references are to the draft of Mr. Rosenthal's paper.[Back]

{8} Abbreviates as CSC. [Back]

{9} Abstract of "The Problem of Intentionality," presented at a symposium entitled "The Problem of Intentionality" with papers by R. Chisholm and W. Sellars, held at West Virginia University, May 13, 1965. [Back]

{10} I would rather say mark-makings and utterances, for marks and noises are psychological only as actually or potentially related to the linguistic activity of persons. [Back]

{11} I assume that you talk initially about explanation in terms of certain intentional items in order to make plausible the extension of the term 'psychological' discussed in paragraph 3. [Back]

{12} Appeared first in Proceedings of the Aristotelian Society 56 (1955-56): 125-148. Reprinted in Minnesota Studies in the Philosophy of Science, vol II, eds. H. Feigl, M. Scriven, and G. Maxwell (University of Minnesota Press, 1958): 510-20. Reprinted again in Intentionality, Mind, and Language, ed. Ausonio Marras (University of Illinois Press, 1972): 31-51. [Back]

{13} R. M. Chisholm, Perceiving (Ithaca, N.Y.: Cornell Univ. Press, 1957), ch. 8. [Back]


[Return to Main Page]