Physicalism: From Supervenience to Elimination

Abstract

The varieties of physicalism and of anti-physicalism are customarily regarded as doctrines pertaining narrowly to the philosophy of mind. But they are, I believe, best seen as different and competing responses to a quite general problem, which can be posed in the following terms. In describing and explaining the world, we make use of many, apparently distinct, vocabularies; and these vocabularies derive from the sciences, both physical and special. (Let common sense count as an honorary special science.) Moreover, we usually take it that the entities and properties spoken of in these vocabularies are real; these entities and properties may be called the domains of the vocabularies. The problem, then, is to give some plausible general account of the relations between these vocabularies and also, less linguistically and more metaphysically, between their respective domains. An account of these relations will qualify as a version of physicalism to the extent that it assigns some descriptive and ontological priority or primacy to the physical vocabulary and its domain. Token physicalism is a physicalist account which unifies the different domains by insisting/that an ontology of events is constant throughout the sciences, so that all the sciences describe and explain the very same events, albeit at different levels; but it nonetheless accords physics a certain primacy, since physics alone is said to be completely general, that is, competent to describe and explain every event. But token physicalism is in my view unsatisfactory. In recent years, however, several philosophers have advanced a new version of physicalism which is claimed to enjoy the advantages of token physicalism without at the same time suffering its drawbacks. This version we may call supervenience physicalism (see Lewis 1983; Horgan 1981 and 1982; Haugeland 1982); and in what follows I shall criticise it. Like token physicalism, it constitutes an account of the relations between the different vocabularies and their domains; but unlike token physicalism, it does not affirm the token identity