The notion of kind inclusion induces a partial ordering on kinds; ontological
categories are those kinds that appear in the higher reaches of this ordering.^18
Intersubstitutability accounts of ontological categories appeal to the notion of
intersubstitutability of one expression for another. Roughly,xandybelong to the
same ontological category just in case for all (or perhaps only for some) meaningful
sentences in which an expression standing forxappears, an expression standing
forycan be substituted without converting the original meaningful sentence into a
meaningless one.^19
Generality accounts identify ontological categories with“the most general kinds of
entities,”but do not appeal to the idea that ontological categories come with
associated conditions of identity. This seems to be the account of Paul (2013: 90),
who writes that“The fundamental ontological categories are the most basic kinds or
natures of the world.”^20
In what follows, I will focus on two versions of the generality account, since by so
doing I highlight certain difficulties facing many accounts of ontological categories.
I won’t discuss in much detail the intersubstitutability accounts or the identity
accounts discussed by Westerhoff (2005), since the arguments he presents against
them are persuasive.
What is generality? In an earlier article, Westerhoff (2002: 288) suggested the
following definition along with an account of ontological categories that employ it:
Say that a classSis more general than the classTiffTis contained inSand, necessarily, if
Sis empty, then so isT. Ontological categories are classes such that none are more general
than them.
Note that here Westerhoff (2002) talks ofclassesrather thankinds. Moreover,
Westerhoff (2005: 18–19) suggests that many ontological systems take categories to
be sets or classes. Probably this is not so. Many think that sets (and classes) have their
members essentially. Supposematerial objectis a category. Then, on the assumption
that categories are sets, the categorymaterial objectexists only in worlds in which
exactly the same material objects exist. This is troubling. Perhaps this is why
Westerhoff (2005: 27) talks about the ontological dependence of onepropertyon
another, a notion which he defines as follows: F depends on G = df. Necessarily, if
there are no Gs, then there are no Fs. A similar definition of“kind K1 ontologically
(^18) This sort of account is favored by Lowe (2006). Rosenkrantz (2012: 90) holds that things essentially
belong to the categories they fall under. Gracia and Novotný (2012: 30–3) attribute a similar view to Suárez,
although the notion of essential predication they appeal to might be better understood as what chapter 9
will call 19 “strict essence”thande remodal predication.
Although I am not sympathetic tometa-ontological accounts that appeal to the notion of intersub-
stitutability, the positiveontologythat I describe in section 4.5 appeals to something like the notion of
intersubstitutability in order to justify a particular formulation of the principle of recombination. 20
Similar remarks are made by Norton (1977: 18–20) and Rosenkrantz (2012: 83). Van Inwagen (2014:
194 – 7) also identifies ontological categories with highly general kinds of things that“objectively go
together,”provided that these kinds are also sufficiently“modally robust.”