Supervenience physicalism attempts to combine non-reductionism
about properties with a physical determination thesis in such a way
as to ensure physicalism. I argue that this attempt is unsuccessful:
the specific supervenience relation in question is either strong enough
to ensure reductionism, as in the case of strong supervenience, or too
weak to yield physical determination, as in the case of global supervenience. The argument develops in three stages. First, I propose
a distinction between two types of reductionism, definitional and scientific, a distinction thanks to which we can reply to a standard objection against the ontological reductionism of strong supervenience.
Second, I claim that because of "the problem of random distribution,"
global supervenience needs strengthening to be an adequate relation
to capture our physicalistic intuitions; and I show, in accordance with
Stalnaker's relevant proof, why a natural strengthening of global supervenience renders it equivalent to strong supervenience. Finally,
I argue against Stalnaker about the possibility of a non-reductionist
global supervenience. The upshot is that despite appearances, supervenience physicalism is a form of reductive physicalism.