This paper is about how to accommodate Functionalism and type-type Identity theories of mind, and why you ought to. It thus runs counter to the common view that one message of Functionalism is that type-type versions of Identity theories should be abandoned in favour of token-token ones.
We assume a basically functionalist viewpoint and if (like at least one author) you have doubts about Functionalism and Materialism, you should view this paper as addressed to the conditional question, "lf I am to be a functionalist and identity theorist, what sort of identity theorist should I be?" Our answer is -a type-type one 1.
Functionalism, as we will understand it, is the view that what is defmitive of, say, pain, is its functional r61e in an organism; its functional r61e is roughly its relations, principally causal, to stimuli, behavioural responses and other mental states. Accordingly, what pain is, is simply what has that functional r61e, and (for beings broadly like us at least) it is plausible that what has that functional r61e, what occupies it or realises it, is a state of the brain.
The natural way to read this conclusion is the type-type way. Pain, and mental states in general, are states which can be multiply instanced. Pain is like heat in that it is a state many things may be in. In any case we will use terms like 'pain' and 'desire' in this way throughout what follows, as naming states which are types in that they may be common to different organisms (and the same organism at different times) and not as naming tokens which make their appearance in this world at most exactly once. Likewise 'state' will be used to name a type, something many things can, at least in principle, share. (Nevertheless we will sometimes allow ourselves to write 'kind of state' to aid comprehension.) D. M. Armstrong and David Lewis read the conclusion in this natural way and are thus type-type identity theorists 2 . The view that they ought not be derives from the (in itself perfectly correct) point that the way a functional