The Conceivability Argument and the Referential Opacity of Natural Kind Terms

The conceivability argument against physicalism holds that we can conceive of a physical entity, such as the neural correlate of pain, in the absence of pain. If this were indeed possible, it would entail the falsity of physicalism, given the necessity of identity. The physicalist’s challenge is explaining why such conceivability is not a guide to possibility or a modal illusion. Previous accounts of this illusoriness have focused on phenomenal terms like “pain”. By contrast, we argue that the illusion stems from the referential opacity of natural kind terms, that is our ignorance of what they refer to. If a term like “neural correlate of pain” is a natural kind term, its reference is fixed to the essence of the actual samples denoted by that term, even if that essence were unknown to us. Thus, it is possible that the reference of “neural correlate of pain” is fixed to pain. When we conceive of pain not being identical to its neural correlate, we conceive of something that appears to be the neural correlate of pain but is not truly the same natural kind as the term denotes in the actual world, that is pain. We show that this argument can be developed within Kripke’s and Chalmers’ own semantic frameworks: given that the reference of a natural kind term is a posteriori , it is impossible to tell a priori what is metaphysically possible for the referent. This argument can be made on purely semantic or epistemological grounds, without invoking any metaphysical distinctions, as in Russellian monism. We conclude that a priori arguments cannot undermine physicalism since what is possible for natural kinds is an a posteriori question.