functional role and intentionality
Post on 29-Sep-2016
Embed Size (px)
Functional role and intentionality
AMIR HOROWITZ Tel-Aviv University
CONCEPTUAL or functional role theories (hereafter CRT) are theo- ries that try to specify what endows mental states with their con- tent. My purpose in this paper is to present versions of CRT, and to show that they are basically inadequate as theories of content.
The roots of CRT are, presumably, Sellarsiam2 Sellars (1963) de- scribes language as having language-entry rules, language- language rules, and language-exit rules. In that spirit, the content of a (type of a) mental state is characterized in terms of its causal connections to perceptual inputs, behavioral outputs, and other mental states. In Blocks words (Block 1986, p. 628):
... [the conceptual role] is a matter of the causal role of the expression in reason- ing and deliberation and, in general, in the way the expression combines and in- teracts with other expressions so as to mediate between sensory inputs and behavioral output^.^
It looks as though CRT is very close in nature to psychophysical functionalism. Fodor argues that the latter does not entail the former. According to him (1986, p. 14), functionalists claim that what makes something a belief is its functional role, but they do not claim that functional role makes a belief the belief that P. Fodor asks us to imagine a semantic theory according to which what makes something a water-belief is its being caused by water. Should
McGinn (1982) calls it cognitive role, and we can find in the literature various other names like causal role, inferential role, and computational role. Some philosophers distinguish between some of these roles (see, for example, Cummins, 1989, and Richard, 1990). I shall deal with some of the distinctions. * But since, as Loar (1982) points out, conceptual role theory can be viewed as a kind of a use theory of meaning, we can step back at least to Wittgenstein.
Among the advocates of CRT we can also find Field (1977 and 1978), Loar (1981 and 1982), Harman (1973, 1982, 1987), Lycan (1985), and Schiffer (1981).
198 AMIR HOROWITZ we then accept that the growth of strawberries is a water-beleif? We should not, according to functionalism, since it holds that for some- thing to be a belief it must have a certain functional role in think- ing. The growth of strawberries does not have whatever functional properties beliefs have to have (if functionalism is true); a fortiori it is not a belief; a fortiori it is not a water-belief. That is, there is a clear distinction between functionalism (a theory which accounts for what makes something a belief, a desire, etc.) and CRT (a theory which accounts for what grants a belief, say, its specific con- tent), and the former does not entail the latter.
I do not question the validity of this reasoning, but I think that it deals with a rather weak and uninteresting version of functional- ism. The strong and interesting claim of functionalism is that be- liefs (and other mental states) are functional states when the are is the are of identity (as opposed to the are of predication). That is, this theory individuates types of mental states in terms of their functional role.4 Functionalism so construed does entail CRT. Since I am going to show that CRT is false, I cannot accept func- tionalism in this stronger sense, and my criticism of CRT is also a criticism of it.
The term conceptual in CRT can be misleading by creating the impression that it characterizes content in terms of its abstract conceptual connections rather than in terms of its concrete psy- chological connections. This, of course, cannot be the case: first, because what we want is a characterization of the content of men- tal states, that is, of real psychological entities; second, because the idea of accounting for content in terms of its abstract conceptual connections is senseless, since if we look for an account of content we cannot appeal to conceptual connections which are semantic connections. There is an abstract conceptual connection between
That functionalism claims this seems plausible in the light of the functionalist attack against reductive materialism which individuates types of mental states in terms of types of brain states. But I dont want to argue about what someone meant, all the more so about meanings of words.
Block (1986, p. 631) raises (without answering) the question whether conceptual role should be understood in ideal or normative terms, or should it be tied up to what people actually do. My opinion on the matter will be revealed shortly.
FUNCTIONAL ROLE AND INTENTIONALITY 199 father and parent only once father and parent (already) have their meanings.
I think that the rationale for characterizing content in terms of conceptual role comes from the fact that mental states with identi- cal contents (which may be physically and/or phenomenologically different!) have identical causal powers, and mental states with different contents have different causal powers.6 Loar (1982, p. 275) says in a similar context that the conceptual role theory of content quite naturally imposes itself. What can be seen as a compelling reason in favor of CRT is this: there is quite a good correspondence between many psychological causal relations and semantic relations. For example, we tend to reason according to various logical patterns, either valid or not. Now how can these phenomena be explained? Only thus, it seems: semantic relations just are causal relations, and this is so since to be semantic in some specific way is to have typical causal antecedents and effects. Giving up the idea that causal relations constitute content will make the correspondence between causal and semantic relations miraculous. In the light of this reasoning it seems just natural to characterize content in terms of conceptual role.
Yet the reasoning is not valid. The mentioned correspondence indeed constrains theories of content in an important way, but, as will be shown at the end of this paper, it does not impose CRT.
It is common to speak of CRT as a theory of narrow content. The motivation for erecting the notion of narrow content emerges from acceptance of the thesis that meanings (or, for that matter, contents), as what determine reference, are not in the head. The most well-known argument for that thesis is Putnams Twin earth argument (Putnam 1975a. Putnam asks us to suppose that there is a planet named Twin Earth (hereafter TE), that is exactly like Earth except for the following difference: the liquid that is called water on TE, and which is perceptually indistinguashable from water (that is, from our water), is not H,O but XYZ. We are asked to suppose, further, that one Earthling, Oscar,, has a TE dupli- cate, Oscar,, who shares all his mental states (non-intentionally
This is true only regarding narrow contents-see later.
200 AMIR HOROWITZ described). According to Putnam, the extension of water when uttered by Oscar, is H,O, while the extension of water when ut- tered by Oscar, is XYZ. That is, despite the identity between the Oscars mental states nonintentionally described these mental states differ in reference. Whats in the head, then, does not determine reference, so meanings are not in the head. Similarly, the contents of the Oscars water-thoughts differ, so contents are not in the head.
Putnams Oscars are said to share narrow contents, which are functions from context to (broad) content. That is, they determine (broad) contents relative to a context. As opposed to (broad) content, narrow content is taken to be internal and solipsistic (a property of whats in the head), and it is said to be that aspect of content that is causally responsible for behavior. So the moti- vation to characterize narrow content in terms of causal connec- tions is quite understandable. Block (ibid., pp. 667-8) explains why narrow content is relevant to the explanation of behavior: To have an internal representation with a certain narrow meaning is to have a representation with certain likely inferential antecedents and effects ....
Narrow content characterized as a conceptual role retains a fea- ture that it must retain, namely, that it can be shared by mental states with different intentional objects, This is clear considering that Putnams Oscars share all their conceptual roles. Since the Oscars do not share (broad) contents, it is impossible to character- ize (broad) content in terms of conceptual role.*
A similar advantage of the theory is that it is fine enough to treat differently, as it should, beliefs such as The Morning-star is a distant planet and The Evening-star is a distant planet, beliefs which represent the same state of affairs, but which would not do so in all possible worlds. These beliefs differ in content according to CRT since they have different inferential and functional roles. Similarly, CRT retains the difference even between beliefs which
See Fodor 1987, chapter 2. Putnam himself (see Putnam 1988) opposes the very notion of narrow content. * But see the discussion of Harmans theory later.
FUNCTIONAL ROLE AND INTENTIONALITY 20 1 do represent the same state of affairs in all possible worlds, like be- liefs about equilateral triangles and equiangular triangle^.^ Such a distinction-ability is a good thing for CRT since one can surely have beliefs about the latter without also having beliefs about the former (and, of course, vice versa).
CRT and holism
An important feature of CRT is that it accounts for contents of thoughts in terms of their total conceptual role (Block 1986, p. 628). What exactly does total mean here? Certainly, not all causal rela- tions of a mental state are relevant for its content. As Block says (ibid.), the fact that you take longer than I do in reas