artifactual and natural objects. Within the fi eld of cognitive psychology this capacity
is often explained by reference to the Design Stance (Dennett 1987), which presupposes
a metaintentional capacity on the part of users of technical artifacts, namely the capacity
to form intentional attitudes about the intentional attitudes of the designer or maker
of the artifact. He criticizes this approach and as an alternative he proposes the hypothesis
that the human cognitive apparatus comprises a Functional Stance, which enables
people to deal with functional knowledge, that is, with knowledge about what objects
are for. He argues that the concept of “for-ness” goes beyond the concept of “causality”
and that attributing for-ness to an object does not presuppose a metaintentional capacity.
The Functional Stance on its own does not lead to a distinction between natural and
artifactual objects, since empirical research indicates that in the early stages of human
development, functions (for-ness) are attributed indiscriminately to natural and artifi cial
objects. According to Romano, the distinction between natural and artifactual objects is
to be based on inferential reasoning on the part of the user about the intentional origins
of an object that is for something. Thus the Functional Stance would be a constitutive
component of a more general human attitude of categorizing objects into artifacts and
natural entities.
Let us assume that the distinction between artifactual and natural objects can be given
a fi rm basis. The objects in the classes of natural as well as artifi cial objects may be further
divided into many subclasses. What is the ontological status of these subclasses, in par-
ticular of artifactual classes? Do specifi c classes of artifactual objects correspond to real
kinds, that is, are they part of the structure of the world? That is the problem addressed
by Soavi. She defends a realist approach to artifact kinds provided that artifact kinds are
functional kinds. This is rather surprising because generally one of the main reasons
adopted for rejecting real artifact kinds is that artifact kinds are functional kinds. A real
kind of physical objects is a kind whose items must share a set of common physical fea-
tures used in explaining their behavior. A functional kind does not grant the existence of
such a set because of the multiple realizability of functions. Hence if artifact kinds are
functional kinds, they may bring together objects with completely different physical struc-
tures. So artifact kinds can refer only to nominal kinds. According to Soavi, the reality of
artifact kinds can be defended if the notion of “function” is defi ned in an appropriately
narrow way. From the main theories on functions she extracts three types of criteria for
the classifi cation of artifacts into functional kinds: the selectionist criterion, the intentional-
use criterion, and the intentional-production criterion. For each of these three types, she
sketches the ontological consequences of their adoption for artifact classifi cation into
kinds. Thereafter she suggests a strategy for a defense of real kinds for artifacts by indi-
viduating narrow functional kinds on the basis of a characterization of a function in terms
of the triple (1) input-output–relations, (2) system of interaction and (3) structure. She
points out that everyday functional classifi cation terms do not correspond to such narrow
functions and therefore do not individuate real kinds.
164 Part IV