Abstract

Those who naturalize semantics concentrate on avoiding difficulties in getting the right sort of cause for the biological item which is to possess semantic properties (to be ‘true of or to be ‘about’ some physical item). Using an analogy with sense‐data, I argue that the real difficulties will be trying to get any proposed neural representation to be the right sort of effect of natural processes. The idea of a biological item which can be a semantic ‘primitive’ is as bankrupt as the idea of a mental item which could be a perceptual ‘primitive’. Eventually I propose a technique for finding naturalized representations without these problems.