Metaphysical Semantics Meets Multiple Realizability

Abstract

Sider�s Writing the Book of the World is an outstanding work. In an earlier review (Schaffer forthcoming) I called it �masterful� and �required reading�, while worrying about Sider�s notion(s) of structure. Here I offer the further praise of further discussion, centred on a separate worry concerning whether Sider�s metaphysical semantics can handle multiple realizability. Metaphysical semantics is supposed to connect the non-fundamental to the fundamental in a distinctively linguistic way, explaining how non-fundamental truths can be grounded in fundamental facts, and so inducing a radically eliminative vision of the non-fundamental as mere talk. I wonder how the story goes when a single non-fundamental truth can be grounded in many different fundamental facts. For instance, the truth that Moore has hands can presumably be grounded in many different distributions of fields, arrangements of particles, vibrations of strings, or whatnot. I am asking how metaphysical semantics can accommodate this sort of multiple realizability. Multiple realizability came to prominence in philosophy of mind in the 1970s, raising problems for reductive versions of physicalism. I think that it raises analogous problems for Sider�s eliminative view of the non-fundamental. Just as one cannot identify the mental state type pain with a specific brain state, so one cannot equate the non-fundamental truth that Moore has hands with a specific fundamental fact. And just as multiple realizability ultimately favours non-reductive treatments of mental state types, so I think it ultimately favours non-eliminative treatments of the non-fundamental truths. Overview: In Section 1 I describe metaphysical semantics, and identify three conditions of adequacy. In Section 2 I propose three ways for metaphysical semantics to treat multiple realizability, and in Sections 3�5 I argue that none of these three ways can satisfy more than one adequacy condition. I conclude in Section 6 that the underlying problem lies with Sider�s �