blockCJP

Differences Between Interpersonal and Intrapersonal Belief Ascription:

A Problem with Block’s Argument for Holism

Ron Mallon Ron.Mallon@mail.hum.utah.edu

Ned Block's "An Argument for Holism" isn't quite that. Block stops short of arguing for holism, and instead he argues for a conditional: "if there is such a thing as narrow content, it is holistic," where holism is taken to be "the doctrine that any substantial difference in W-beliefs, whether between two people or between one person at two times, requires a difference in the meaning or content of W" (153, 152). Such holism has come under harsh criticism recently because of its apparently extravagant consequences.[1]

For one thing, barring an analytic-synthetic distinction, belief holism is taken to not have any principled limits. In the absence of an analytic-synthetic distinction, there seems to be no viable way to distinguish a "substantial" difference in W-beliefs from any change of W-beliefs. So if any substantial difference in W-beliefs can change the meaning of W, then it seems any difference in W-beliefs will result in a change of meaning of W. This has the apparent consequence that two people can never share a belief, and one person can never change his mind.

Well-known examples from Saul Kripke, Hilary Putnam and Tyler Burge (hereafter known as the Twin cases) suggest that the meaning of a thought is, at least in part, constituted by factors external to the person (for example, the environment or linguistic community of a speaker). Contrast these with the Frege cases which suggest that meaning must, in part, be constituted by factors internal to a person -- so-called "senses" or "modes of presentation."[2]

The intuition behind narrow content is that my molecular twin and I are in some sense psychologically identical, if there is such a thing as psychology at all. After all, all sorts of psychological generalizations subsume us both. If I am thirsty, or my pants are on fire, or my plant is wilted, or my hands are dirty, or I would like to make some tea, I might utter, "bring me some water," as would my molecular twin in similar circumstances. In addition, the proponent of narrow content contends, "Hesperus" and "Phosphorus" thoughts are psychologically different though coreferential, and psychology should explain that too.

What is needed is a way to specify the relevant narrow psychological meaning in these two types of cases. Narrow content, in this context, is whatever it is that is the same between me and my twin, and whatever it is that is different in "Hesperus" and "Phosphorus" thoughts. Block himself stops short of endorsing any particular theory of narrow content, but we will have cause to wonder whether anything can do everything that narrow content is supposed to do.

In this paper, I will contend that Block's argument for holism fails precisely because it rests on a conflation of the narrow content needed in each of these two types of cases. The similarity between Twins and the distinctness of thoughts in a Frege case do not, prima facie anyway, have the same explanation. In Section I, I will set out Block's argument. In Section II, I will consider what Block calls the "small distance objection" and a refinement Block makes to cleanly avoid it. In Section III, I will argue that Block's argument understood in light of the refinement is invalid, and I will suggest that this is not surprising since narrow content is being called upon to perform incompatible roles.

Section I. Block's argument:

The Principles:

Block’s argument begins with a set of plausible assumptions:

NARROWNESS:

Narrow content supervenes on non-relational physical features of the body. Or in slogan form, narrow content is narrow. This is just a definition.

DIFFERENCE:

If at one time, a person has substantially different beliefs associated with term t1 and t2, then t1 differs in narrow content from t2 for that person at that time. So for any normal person, words like 'cat' and 'dog' and 'panda' have different narrow contents.

EXPLANATION:

Narrow content's main purpose is its role in psychological explanation.

INTER/INTRA: FIELD's PREMISE[3]:

The relation of same narrow content that holds between people is the same relation of same narrow content that holds within a single person. (154-55)

NARROWNESS and EXPLANATION fall right out of the demands on narrow content discussed above. NARROWNESS stems from the idea that whatever psychology ought to study, it is the stuff in the head, not exotic and idiosyncratic historical relations. EXPLANATION says that narrow content is to be central in psychology. In particular, then, EXPLANATION combined with NARROWNESS ought to explain why molecular Twins in different environments behave similarly.

DIFFERENCE is less straightforward, but is supported by the notion of EXPLANATION. Block explains:

One of the roles of EXPLANATION is to bolster DIFFERENCE. If a theorist holds that a single person's 'dog' and 'cat' have the same narrow content, one should wonder what this theorist thinks narrow contents are for. (155)

This makes a lot of sense, since if narrow content is to be (in part) the stuff that psychological generalizations are made of, it had better distinguish between mental states that common sense distinguishes. Pay close attention, though, as the relationship between EXPLANATION and DIFFERENCE will become more important shortly.

The story from Ruritania:

There are two parts of Ruritania, B and W. Bruce lives in B, his molecular twin Walter, in W. Their languages are identical except that in Bruce's dialect, "grug" refers to beer, while in Walter's dialect, "grug" refers to whiskey. Ruritanian is English except that it contains the word "grug."

At age 10, Bruce and Walter are molecular duplicates. Both believe exactly the same things about "grug," e.g. "grug" is a brownish liquid, "grug" makes you tipsy, and "grug" is purchased in liquor stores.

By age 12, both are more worldly. Bruce believes that "grug" translates into English as "beer" while Walter believes it translates as "whiskey." Bruce believes that "grug" is cheap while Walter believes "grug" is expensive. Moreover, Bruce believes that "grug" in the other Ruritanian dialect translates to English as "whiskey" and is expensive. Walter believes that "grug" in Bruce's Ruritanian dialect translates to English as "beer" and is cheap. Hereafter, we will refer to these two uses as "grugb" for Bruce's home "grug" (for beer) and "grugw" for Walter's home "grug" (and whiskey). So, using "=" to indicate sameness of narrow content,

I. At Age 10: "grugb"="grugw"

Since narrow contents are narrow (by NARROWNESS), and ex hypothesei Bruce and Walter are molecular dopplegängers, this conclusion quickly follows.

By age twelve, the two have come to have different beliefs about their home "grug"s, e.g. Bruce believes "grug" is cheap while Walter believes "grug" is expensive. Also Bruce believes it takes a lot of "grug" to make one drunk, while Walter believes it takes only a little "grug" to knock one out. Moreover, as we noted, both boys recognize a foreign use of "grug" different than their own (namely each other's use of "grug"), and each has all the same beliefs about the foreign "grug" as the other has about his home grug.

II. At age 12: Bruce's home "grugb"≠ Bruce's foreign "grugw"

(By DIFFERENCE)

Also, III. One boy's foreign "grug" is the same as the other's home "grug." So Bruce's "grugw" = Walter's "grugw".

Block believes this falls right out of the principles. Bruce and Walter are still dopplegängers at age 12, "except for indexical 'grug' beliefs having to do with whose 'grug' is in question" (158). So each knows a word "grug" used to mean whiskey, and another word "grug" used to mean beer. By NARROWNESS and EXPLANATION, Block argues that the indexical difference doesn't make a difference to narrow contents.

IV. Since Bruce's "grugb"≠ "grugw" and Bruce's "grugw" = Walter's "grugw", then Bruce's "grugb" ≠ Walter's "grugw".

(Assuming that a representation R that's not narrowly synonymous with one representation S, can't be narrowly synonymous with another representation T, when S=T.)

V. Now, at age 10, Bruce's "grugb" = Walter's "grugw".

By age 12, Bruce's "grugb" ≠ Walter's "grugw".

So one of their "grug"s must have changed narrow content between 10 and 12. Since they are symmetrical dopplegängers, they both changed. So Bruce-at-10's "grugb" ≠ Bruce-at-12's "grugb", and

Walter-at-10's "grugw" ≠ Walter-at-12's "grugw".

Thus Block concludes:

So 'grug' in Bruce's dialect changed narrow meaning between age 10 and age 12. And the same for Walter. So a substantial change in Bruce's 'grug' beliefs results in a change in narrow content of his 'grug'. (160)

And this is just about where we wanted to get since Block is arguing that, "any substantial difference in W-beliefs, whether between two people or between one person at two times, requires a difference a difference in the [narrow] meaning or content of W" (153, 152). The differences between Walter's age 10 "grug" and age 12 "grug" surely aren't that substantial; they consist only of a few added beliefs about his home "grug" and some additional beliefs associated with a new concept, the foreign "grug". So the conclusion Block draws is very far reaching. It suggests that even slight changes in the beliefs associated with a concept amount to changing the meaning of the concept.

Section II. The Refinements

The small distance objection:

One objection Block foresees is that the conclusion follows too quickly from the stated premises. In fact, it seems we can almost get the conclusion just from DIFFERENCE and the INTER/INTRA PREMISE. If DIFFERENCE could be applied to an individual at different times and if INTER/INTRA also implies that the relation of different narrow content that holds between people is the same relation of different narrow content that holds within a single person, then the conclusion trivially follows that:

T: If a person J has substantially different beliefs associated with a term t1 than another person K has associated with a term t2, then J's t1 and K's t2 have different narrow content.

Now we can make Bruce at age 10 person J and Bruce at age 12 person K, and we can appeal directly to the difference of beliefs to justify a difference of narrow content.

While INTER/INTRA and DIFFERENCE do not, as stated, strictly imply T, the question is what principled reason is there for stating INTER/INTRA and DIFFERENCE in such a way as to avoid T. Without such a reason, the Ruritania story seems to have been merely the long way to the conclusion we could have deduced quickly from the premises, had they been a little less ad hoc. If this is the case, then Block's argument can be dispensed of in favor of the shorter version, but the shorter version itself is uninteresting in virtue of the small distance between premises and conclusion. What Block needs to show is that his way of stating the premises so that they don't quickly imply T is principled and not merely some deft wordplay. Remember, the premises are themselves supposed to be intuitively plausible, so if it turns out they are ad hoc constructions to avoid the small distance objection, Block's argument is uninteresting.

Block's solution:

Block avoids the looming triviality by revealing the way in which DIFFERENCE is independently motivated, and this motivation reveals why it, combined with INTER/INTRA does not (and should not) imply T. On Block's refined account, DIFFERENCE is really a version of what Schiffer (1978) has called Frege's Principle. This is the principle that leads us to posit something psychologically distinct (e.g. senses or modes of presentation) within a person's mind rather than attribute to the person incoherent or irrational thoughts. In this case, Block suggests that we adopt the metaphor of a "file."

A file concept Z is a holder for all the beliefs that a person has in which Z figures.[4] Here we are to construe file concepts as mental constituents which are causally efficacious and maintain identity over time, despite altered "contents." The "contents" of a file are taken to be beliefs in which the file-concept figures. So if one has a "panda" file, and one also has the belief that "Pandas are cuddly," then "Pandas are cuddly" counts as being in the "panda" file (assuming, for simplicity, proper functioning and only one "panda" file in the head).

Notice that files amount to a distinct notion of narrow content from the one we began with in DIFFERENCE. Among other things, Block argues,

The file explanation doesn't depend on there being different beliefs in the two files. Suppose I have two different files headed 'panda'. Both contain 'Furry' and 'Found in Asia' and 'Not identical to the other animal called "panda"'. Both files have the same beliefs. In fact, this does describe my epistemic situation some years ago. (Now I know more – that one is the great panda and the other is the lesser panda.) all that we need for different narrow contents in a single person are different files – the contents can be the same. . . . .Differences in beliefs are only relevant because it would be hard for one to have 'X' beliefs that are different from one's 'Y' beliefs without thinking that Xs are distinct from Ys. (165-66)

Difference in beliefs (and thus DIFFERENCE as originally stated) are only a diagnostic tool in discovering an underlying difference of files. Our real interest is in difference of files, at least for the purpose of the DIFFERENCE principle revealing different narrow contents.

If Bruce has systematically incompatible "grug" beliefs, then, according to DIFFERENCE thus understood, we should ascribe to Bruce two difference "grug" files, in which the distinct beliefs can figure. In this way we avoid attributing to Bruce irrational beliefs about a single referent of "grug." Notice that the requirement to preserve rationality is itself a synchronic, intrapersonal constraint. A person can change her mind about a proposition P as frequently as she wants without irrationality. It is only when she, at a single time, has inconsistent beliefs about P that she exhibits irrationality. Similarly, while rationality is a constraint on what an individual can coherently believe at a single time, it is no constraint at all on the differences that may arise between individuals. DIFFERENCE, then, is a well motivated principle of belief attribution, but one that is rightfully limited to intrapersonal beliefs at a single time.

Interpreting DIFFERENCE as an epistemic principle that indicates distinct files avoids the small distance objection. Since DIFFERENCE is only a synchronic principle, it is not possible to apply it at different times. And since DIFFERENCE is only an intrapersonal principle, it blocks any move from INTER/INTRA to

(D) Difference of interpersonal content is the same relation as difference of intrapersonal content.

Principle (D), it appears, is false. This is because DIFFERENCE indicates a kind of difference of narrow content – difference of file concept – that arises only in the intrapersonal case. It simply has no parallel in the interpersonal case. So it is false to say that the relationship of difference of narrow content that obtains between two people is identical to that which arises within one person. Since (D) is false and DIFFERENCE is synchronic, the inference from DIFFERENCE and (D) to T is both invalid and unsound. So Block's thought experiment is necessary for his demonstration to work.

Section III. Files and Explanation:

While the introduction of files may show why the small distance objection is incorrect, the degree of freedom they provide shows exactly why Block's argument is wrong. Whereas before we had the DIFFERENCE principle, now we have what I will call F-DIFFERENCE:

F-DIFFERENCE: It is sufficient for different narrow contents in a single person at one time to have different files .

One crucial difference between DIFFERENCE and F-DIFFERENCE is that whereas before EXPLANATION "bolstered" DIFFERENCE, that is not the case with F-DIFFERENCE. F-DIFFERENCE no longer requires a difference of beliefs attached to two terms, but only a difference of files. But files can be different while being explanatorily the same.

Take Block's "panda" concepts. He says he has two panda files, each with identical associated beliefs, but with different narrow contents. If you believe in the psychological reality of files, this seems possible. But notice that the two panda concepts are explanatorily the same. Any behavior that you could explain by attributing a belief involving one, you could also explain by attributing a belief involving the other since the contents of both files are exactly the same.[5]

To see how this affects Block's argument, let's consider the argument again. Steps I-III still work just the same, but look at step IV:

IV. Since Bruce's "grugb"≠ Bruce's "grugw" and Bruce's "grugw" = Walter's "grugw", then Bruce's "grugb" ≠ Walter's "grugw".

(Assuming that a representation R that's not narrowly synonymous with one representation S, can't be narrowly synonymous with another representation T, when S=T)

When we interpret this step in light of F-DIFFERENCE, the parenthetical assumption does not hold. Consider Block, who has two panda concepts, which are F-DIFFERENT, but have the same associated beliefs. Give Block a Molecular Twin, Joe, also with two panda concepts. Notice that

i) Block's "panda1" ≠ Block's "panda2" (by F-DIFFERENCE). Since there are distinct files, there is distinct narrow content.

ii) Block's "panda2" = Joe's "panda1" (by EXPLANATION and NARROWNESS). Since the two are molecular twins, and differences of extension are irrelevant, NARROWNESS indicates this. Moreover, since the two concepts are explanatorily identical, EXPLANATION also supports this. (The reasoning here is just like that in Block's Step III.)

iii) Block's "panda1"= Joe's "panda1" (by EXPLANATION and NARROWNESS).[6] The argument here is identical to that for (ii).

Now we have a representation R (Block's "panda1") that's not narrowly synonymous with another S (Block's "panda2"), but nonetheless, R=T and S=T. Block's assumption about the transitivity of nonsynonymy is wrong, in this case, and importantly so. The problem is that a difference in files is only explanatorily relevant in the Frege cases, not in ordinary cases of psychological content attribution like that between Block and Joe. Two files that are in the same head, will, according to F-DIFFERENCE be considered different, even if the same files distributed across two people (even Twins) would be considered identical. I think that what we must realize is that there is more than one notion of (and more than one role for) narrow content here.

It does seem intuitively that IV should hold. But intuitions are what we used to stipulate our premises -- we cannot also use them to stipulate our conclusions. And without step IV, we cannot get the crucial fifth step which is the goal of Block's paper. Once you analyze concepts in terms of files, it is not really surprising that V does not hold. One can gain many new beliefs about the same old things, without ever getting a new file. Given this, it's hard to see why there would be a change in file for Bruce or for Walter between the ages of 10 and 12. (Although each has had to add a file, for the foreign "grug".)[7] The only way we know we can get a difference in narrow content is through applying F-DIFFERENCE.

Block believes there is some notion of difference of narrow content other than difference of files that arises for Bruce and Walter between the ages of 10 and 12. (It is the aim of his argument to show this). EXPLANATION suggests such a difference. After all, Bruce at age 10 is psychologically different in many ways from Bruce at age 12. One way of putting this difference is that what's in Bruce's "grug" file changes – Bruce's "grug" beliefs change. The question for Block is, what makes this a change of narrow content, and what justifies the claim that the meaning of "grug" changes just because the associated "grug"-beliefs change?

There are at least two roles for narrow content to play – one of which is psychologically relevant in the Frege cases, the other of which comes into play in comparing either two people, or one person over time. Block's mistake was to try to "justify an inter-personal claim by appeal to an intra-personal claim," (157). But the intrapersonal claim is only brought about by a difference in files. It is, in the relevant respects, a Frege case. Interpersonal claims, claims about narrow content identity between people, are Twin cases. They are claims of sameness or difference based on EXPLANATION and NARROWNESS.

Block's argument, I think, is assuming that the same kind of narrow content could solve both the Frege cases and the Twin cases, but it is not clear that this is possible. Putnam cases show that (at least for many terms) sameness of referent is necessary for sameness of belief. Frege cases show that sameness of referent is not sufficient for sameness of belief. Block does offer us the apparatus to provide a solution to each kind of problem, but he doesn't allow that each kind of case requires different apparatus.

Difference in files provides a solution to the Frege cases by providing syntactically type-distinct mental tokens to serve as modes of presentation (see Fodor 1994 for a similar account). But it is the beliefs in the files that seem to be relevant to complex interpersonal explanation more generally. It these beliefs which change over time, leading to explanatory difference between file concepts having the same reference, and explanatory sameness of file concepts that have different references (as is the case between Twins). This makes a lot of sense -- it is hard to know how to predict your "grug" behavior without knowing what your "grug"-beliefs are.

The thing is, though, you need both sorts of apparatus – files and their contents – to do all the things narrow content is supposed to do.[8] And neither files nor their associated beliefs have obvious holistic consequences. Block's argument conflates the apparatus needed for the Frege cases with that needed for the Twin cases and psychological explanation more generally. As a result, Block derives a radical conclusion. If what I have said is correct, his conclusion is unwarranted.

Bibliography

Thanks to Peter Lupu and especially Jonathan Weinberg for their comments on earlier drafts.

Block, N. 1995: An Argument for Holism. Proceedings of the Aristotelian Society. February 27, 151-169.

Block, N. 1994: Advertisement for a Semantics for Psychology. In Stich and Warfield (eds), Mental Representations. Cambridge: Blackwell, 81-141.

Block, N. 1993: Holism, Hyper-compositionality and Hyper-analyticity. Mind and Language, 8:1, 1-26.

Field, H. (unpublished).

Fodor, J. 1987: Psychosemantics. Cambridge, MA.: MIT Press.

Fodor, J. and Lepore, E. “Why Meaning (Probably) Isn’t Conceptual Role.” in Stich and Warfield (eds), Mental Representations. Cambridge: Blackwell, 142-156.

Fodor, J. and Lepore, E. 1992: Holism: A Shoppers Guide. Cambridge, MA: Blackwell, 1-35, 163-186.

Fodor, J. and Lepore, E. 1993: Reply to Block and Boghossian. Mind and Language, 8:1, 41-48

Putnam, H. 1975: The meaning of 'meaning'. Mind, Language and Reality, Cambridge University Press. 215-271.

Rey, G. 1984: Concepts and Stereotypes. Cognition. 15, 237-262.

Rey, G. 1985: Concepts and conceptions: A reply to Smith, Medin and Rips. Cognition . 19, 297-303.

Schiffer, S. 1978: The basis of reference. Erkenntnis, 13, 180.

Endnotes: