Materialism Is Not (or Shouldn't Be) a Metaphysical Thesis
Materialism is the view that the world is entirely material (or physical). There are no immaterial souls or properties. There is no ghost in the machine.
David Chalmers argues against materialism as follows:
(1.) I can conceive of a world in which everything material is as it is in the actual world, yet in which there is no consciousness. (This would be a zombie world, which has a counterpart Eric Schwitzgebel who says and writes and does exactly the same things as I do, but who has no light of consciousness inside and is in the relevant sense an empty machine.)
(2.) Although such a world may not be naturally possible -- that is, although such a world may violate the laws of nature that hold at our world -- the fact that it is conceivable shows that it is metaphysically possible. (Compare: It is conceivable, and metaphysically possible, that my coffee cup rise of its own accord into the air and circle around my head in violation of the laws of gravity and inertia. A three-sided square, in contrast, is neither naturally nor metaphysically possible.)
(3.) Since in that world my counterpart does not have the property of being conscious though he shares all material properties with me, the property of being conscious must not be a material property.
(4.) So the world is not entirely material.
(Obviously, this argument is condensed. See Chalmers's 1996 book for the full details!)
What has always struck me as strange about this argument is how it derives a conclusion about the fundamental structure of reality from facts about what we can conceive. How could that possibly work? How could doing thought experiments in my armchair reveal whether the world is purely material or not?
Most materialist responses to Chalmers either deny that we can really conceive of such a world or deny that conceivability is an adequate test of metaphysical possibility. However, I find Chalmers convincing in his responses to both lines of attack. My thinking, instead, is that we should conceive metaphysical possibility as conceptual possibility but deny that materialism is (or should be) a thesis about what is conceptually possible.
To make this work, I need to play around with the concept of a "property". For a simple, concrete example, let's say that that in all naturally possible worlds I'm in brain state #1117A if and only if I'm having the conscious experience of pain. My zombie counterpart without consciousness (in a conceptually possible but naturally impossible world) has #1117A but not conscious pain. We might define thinly-sliced properties as properties individuated such that if they diverge even only in conceptually possible worlds, they are different properties. Thickly-sliced properties, in contrast, might be individuated such that if two diverge only in conceptually possible worlds but never in naturally possible worlds, they really are only one property. Conscious pain and #1117A would thus be different thinly-sliced properties but the same thickly-sliced property.
Now the question is, should we think of materialism as a claim about properties thinly sliced or thickly sliced? Let me suggest that the proper spirit of materialism, as a scientific hypothesis, confines it to being a claim about what is naturally possible, not a claim about what is conceptually possible. So zombie worlds and thinly-sliced properties are irrelevant to its truth. Materialists can give Chalmers "property dualism" if "property" means thinly-sliced property. In some sense, I do have non-material properties, but that's just a function of the fact that such thinly-sliced "properties" are individuated in accord with the concepts of the person attributing them and the human concept of consciousness pulls apart from the human concept of the material.