Mental States, Natural Kinds and Psychophysical Laws

Abstract

I. Suppose (i) that there are causal relations between mental and physical events, (ii) that where there is causality a covering law exists which subsumes the events under some of their descriptions, and (iii) that there are no psychophysical laws, known or unknown. Then, Davidson argues, physicalism with respect to particular mental events may be deduced.' The third premiss has proved to be the stumbling-block, and Davidson admits that his defence of it is less than decisive. However, I think the case for (iii) can be substantially strengthened, and my main aim in this paper is to show why the claim should be accepted. Davidson's own reasons for denying the possibility of psychophysical laws, or something close to them, will emerge as corollaries of the considerations I adduce, but my arguments will make use of doctrines and principles foreign to his own defence of (iii). The issue is important beyond its rl1e in establishing an identity theory: for a demonstration of the in principle unavailability of psychophysical laws would have repercussions for such matters as the unity of science, the irreducibility and autonomy of psychological explanation, determinism with respect to human action, the prospects for psychophysiology as a science like others, and quite generally what the relation is between mental states of a person and physical states of his body. At first I thought that the absence of psychophysical laws would have interesting consequences for the question whether mental states should be conceived of as natural kinds, whose real essence is specifiable in physical terms of the brain. For, if there were no such laws, then mental states could not be so conceived, since the necessary co-satisfaction of mental and physical predicates required by that conception would be incompatible with the demonstrated lack of nomological tie between the two types of predicate. But I now think it is the other way about: it is precisely because mental predicates can be shown not to denote natural kinds, but rather to express concepts of a fundamentally different character, that authentic