Not all zombies are the same

David Chalmers argues that, since for any given function, that function could be performed somehow without phenomenal properties (that is, since we can, supposedly, conceive of some system that does this), then metaphysical dualism is true. But this conclusion doesn't follow from the counterfactual. The counterfactual (the zombie thought experiment) does not show that phenomenal properties are epiphenomenal (causally inert). It only shows that for any function (identifying color surfaces, say) that a human performs using phenomenal experience, we can imagine (supposedly) another being performing using some other means (like a robot that measured light amplitude without being conscious at all). This no more shows that phenomenal experience is epiphenomenal than it shows that light-amplitude measuring is epiphenomenal, in the other direction. At most this shows that human psychology is a subset of "pure" functionalist psychology, involving contingencies that "pure" psychology does not. It in no way shows that phenomenal experience cannot be naturalized.