Page 91 - Encyclopedia of Philosophy of Language

(Japanese, Urdu, English, etc.) in that they have syn- tactic and semantic properties. That is, they are com- posed of constituents in particular structural configurations and their semantics is a function of the semantics of their basic elements and their syntactic structure. The grammar of mentalese may, however, differ quite radically from the grammar of any natural language. The basic elements (concepts, perhaps) denote entities and properties in the world. The full formulas are truth-conditional, so have truth-values as determined by the way the world is, and they bear logical relations to each other, such as entailment.
The following discussion is confined to descriptive thoughts (thoughts about states of affairs), but it should be noted that there are also what are called interpretive (or metarepresentational) thoughts (see Sperber and Wilson 1995). These are thoughts which represent other representations (such as thoughts or utterances); their relation with that which they rep- resent is not one of truth/falsity but of prepositional resemblance (which is a matter of degree). The lan- guage of thought hypothesis is just as relevant to them as it is to truth-based descriptive thoughts, but they introduce considerable additional complexity.
1. The Representational Theory of Mind
The relevant notion of a thought here is that of an 'intentional' state of mind, where intentional mental states are those that have the property of being rep- resentational, that is, of being about the world. Beliefs, desires, intentions, hopes, and fears are different types of intentional mental states. They are sometimes called propositional attitudes since they involve the having of an attitude to a content or proposition, for example, having the belief attitude to the 'Mrs Thatcher has resigned' content. These intentional mental states play a central role in cognitive psy- chology in the explanation of human intentional behavior. For example, it is because Jane wants to drink some cola and she believes that there is some cola in the refrigerator that she goes to the refrigerator and reaches inside it. On the LOT view, having a belief or a desire, etc., with a certain content entails being in a relation to an internally represented sentence with that content, so the explanation of Jane's refrigerator- oriented behavior will include a specification of the interaction of the sentences which represent the con- tent of her relevant beliefs and desires.
The LOT hypothesis arises then in the context of the current computational model of the mind, whereby mental processes, such as reasoning, are sequences of mental states and the transitions between states are effected computationally. Conceiving of these com- putations as formal/syntactic operations defined over mental representations gives a mechanical expla- nation for mental processes. That is, they operate on symbols in virtue of the form of the symbol, not in virtue of any semantic property of the symbol, just
like the operations performed by a computer or the transitions from line to line in a logic proof. This approach to the causal explanation of mental pro- cesses is known as 'methodological solipsism' (see Fodor 1981; Lycan 1990). It follows that as far as our cognitive life is concerned two beliefs or desires are distinct if and only if the representations of their con- tents are formally distinct. For example, the desire to meet the husband of Janet Fodor and the desire to meet the staunchest advocate of the language of thought hypothesis are identical in their truth- conditional content (given that the definite description in each case picks out the same individual in the world, namely Jerry Fodor). However, so far as cognitive activity is concerned these are quite distinct types of desire as they may be the effects of different sequences of thought and each may cause further different thoughts. Furthermore, they may issue in quite dis- tinct behaviors: in the first case one might telephone Janet Fodor to ask her and her husband for dinner, in the second one might seek out conferences on the philosophy of mind. The crucial point here is that thoughts have their causal roles as a function of their formal properties. Semantic properties are respected only insofar as they are mimicked by formal proper- ties, which of course they are to at least some extent since deductive reasoning, which preserves truth, plays a major role in human thought.
2. Why Should Thoughts Have Syntactic Structure?
One could be an 'intentional realist,' that is, one could accept (a) that beliefs and desires really exist, (b) that they are physically instantiated in the brain, and (c) that they play a causal role in sequences of thought and in overt behavior, without positing a 'language' of thought in which the objects of attitudes are couched. What is crucial about language is constituent struc- ture, that is, that a sentence is made up of parts and these same parts can occur in a range of different sentences. So what distinguishes the LOT view from other intentionally realist views is that it entails that belief/desire states are structured states. Fodor (1975) claimed that the language of thought was implicit in the computational approach to psychological explan- ation since computation presupposes a medium in which to compute. However, the emergence of an alternative computational approach, 'new con- nectionism' (see Sterelny 1990 for an introduction), indicates that more in the way of arguments for struc- tured thought is required, since according to con- nectionism the mental causes of intelligent behavior can be modeled by patterns of activation across net- works of nodes and connections, involving no level of symbolic representation. One of Fodor's arguments for syntactic thought (Fodor 1987a) involves an appeal to the 'productivity' and 'systematicity' of thought. The set of thoughts is potentially infinite and the ability to think any particular thought is intrin-
Language of Thought
69