properties out of individuals. On this view, the set of properties does not correspond
to an ontological category since properties can be generated from individuals via the
constructive operation of set-formation.
One cannot use Westerhoff’s account of ontological categories toexplainfailures
of recombination since the account presupposes these failures. Westerhoff’s notion
of intersubstitutability is simply that which preserves the possible obtainment of a
state of affairs. And whether it is possible to substitute a constituent of a state of
affairs with another entity and still have a possible state of affairs is simply a brute
fact for Westerhoff. The brute limitations on how entities can be recombined are
what determine which entities go together in which form sets, which in turn
determines which sets correspond to the ontological categories.
In a sense, Westerhoff’s view identifies ontological categories with the behavior
they engender. I think this is a mistake analogous to the mistake made by the
psychological behaviorist who identifies mental states with manifestations of behav-
ior. Just as a mental state is not to be identified with its typical manifestation, but
rather should be thought of as the proximate cause of its manifestation, so too we
should not analyze the notion of an ontological category directly in terms of failures
of recombination or intersubstitutability of states of affairs, but should rather think of
ontological categories as those which are metaphysically responsible for such failures.
What then can Westerhoff’s account explain? Perhaps not much, but this might
not bother him. By his lights, the notion of an ontological category does little, if any,
real work for metaphysics. We employ the notion of an ontological category as a tool
for systematizing fundamental features of the world, but that is its sole function.^5 By
Westerhoff’s lights, the project of providing an account of ontological categories is
largely descriptive.
Westerhoff discusses several interesting consequences of his account; I will men-
tion those that I take to be costs, although Westerhoff does not evaluate them this
way. One consequence of Westerhoff’s (2005: 123–5) view islocal relativism: at some
worlds, there might be more than one set of form sets suitable to serve as base sets,
and hence there is no absolute fact of the matter concerning which form sets are the
ontological categories at those worlds. A second consequence isglobal relativism:
something might be an ontological category in one world but not in another, since
whether something is an ontological category in worldwdepends on what things are
in worldw.^6 No ontological category is essentially an ontological category.
Westerhoff takes global relativism to imply that things might belong to their
ontological categories only accidentally. Since there are plausible ontological schemes
in which things do not belong to their categories necessarily, perhaps this conse-
quence shouldn’t trouble us. (We will acquaint ourselves with one such scheme
soon.) But regardless of whether global relativism or local relativism have unintuitive
(^5) See Westerhoff (2005: 3). (^6) See Westerhoff (2005: 132–6).