structural relations to each other.^3 These specifications are such that, among them,
one is the central specification and the others are to be understood via how they are
related to that central specification.^4 With respect to this kind of analogy, one
specification of an analogous property enjoys more naturalness or fundamentality
than the other specifications, since the central specification is more intrinsic than the
remaining specifications in the following sense: the remaining specifications are both
extrinsic features (while the central specificationmightbe an intrinsic feature, but it
needn’t be) and they are instantiated because of a relation that their exemplifiers
bear to an exemplifier of the central specification. For this reason, I say that the
central specification ismoreintrinsic than the remaining ones, and hence is more
natural or fundamental than the remainders.
A third kind of analogy is an attenuated version of the second metaphysical kind,
or at least a more general version that encompasses the second kind. With respect to
this kind of analogy, the specifications are both different from each other and alike,
but there might not be a central or focal feature to unify the specifications. Let’s try to
get a bit clearer about this kind of analogy.
Let us say that a possibly instantiated feature istopic-neutralif it can apply to
objects from any ontological category. For now, I’ll remain neutral on what it is to be
an ontological category and simply help myself to the idea that things belong to
different ontological categories. In chapter 4, we will examine more critically what it
is to be an ontological category.
A plausible candidate of a topic-neutral feature is self-identity: propositions are
self-identical, properties are self-identical, concrete objects are self-identical, and so
forth. Let us say that a possibly instantiated feature istopic-specificjust in case there is
some ontological category such that the feature cannot apply to entities in that
category. Being spatially located is a topic-specific feature, since no proposition is
or could be spatially located.
It will be helpful to have the following notion of relative topic-neutrality. Let F1
and F2 be possibly instantiated features and say that F1 ismoretopic-neutral than F2
if the class of kinds of things that F2 can apply to is a proper subclass of the class of
kinds of things that F1 can apply to. (A feature Fappliesto a class of things K just in
case a possible member of K is F.) This account here is not fully general, since it does
not allow us to compare the relative topic-neutrality of properties whose possible
extensions are completely disjoint. Moreover, this account of topic-neutrality pre-
supposes some background assumptions about what the ontological categories are
and how they relate to each other in order to yield the right results in some cases.
(^3) Similarly, Shields (1999: 59) distinguishes betweenfocal meaning, which is a semantic phenomenon,
and 4 focal connection, which is an object-level phenomenon.
As Shields (1999: 59) notes, it is very tricky to determine which relations are relevant to determining
focal connections. Shields (1999: 110–24) adopts the view that what determines focal connections are the
various Aristotelian forms of causation. Ward (2008: 80–1) criticizes Shields’s account.