The doctrine of physicalism can be roughly spelled out simply as the claim that the physical state of the world determines the total state of the world. However, since there are many forms of determination, a somewhat more precise characterization is needed. One obvious problem with the simple formulation is that the traditional doctrine of epiphenomenalism holds that the mental is determined by the physical (and epiphenomenalists need not assert that there are any properties except mental and physical ones, so (...) one can freely add to epiphenomenalism the claim that everything is determined by the physical state of the world). However, the orthodox view, which seems obviously correct, is that physicalists would and should balk at the claim that epiphenomenalism is a form of physicalism. The philosophical zombie thought experiment vividly reveals exactly why epiphenomenalism is not a version of physicalism. According to traditional epiphenomenalism the determination relation in question is causation, and causal relations do not hold with full necessity. They hold with at most nomological necessity. Thus there is a possible world, w, in which the causal laws (or, if one prefers a more Humean approach, where the cosmic regularities) are different in such a way that the physical states which in the actual world cause mental states either cause no mental states at all in w (the zombie option) or cause aberrant mental states (the inverted spectrum option). Why should the physicalist care about this ‘mere possibility’? Because it shows that the mental can vary independently of the physical and there can be no better demonstration of ontological distinctness 1 than independent variation. Therefore, physicalism requires that the sort of determination at issue must exhibit maximum modal force; it must be absolutely impossible for the mental to vary without attendant, determining physical variation (let us label this relation ‘logical determination’). I think the best way to state physicalism which meets this constraint is in terms of what are called minimal physical duplicates (MPDs) of possible worlds (an approach pioneered by David Lewis, see Lewis (1983b); see also Jackson (1998) and Chalmers and Jackson (2001)).. (shrink)