Realism and Artifact Kinds 187
Devitt, those objects that are said to be tables do not share a common nature: being a table
is not a real property. We just have the word table to name those objects that are used in
a certain manner or play a certain role or are built according to certain intentions. This
can be labeled, as Losonsky labels it, the “Aristotelian position on artefacts” (Losonsky
1990: 81–82). Many other authors adopting strong realism defend the idea that artifact
kinds are not real kinds with different arguments. But artifacts are objects that we do
individuate and classify as easily as natural beings; different artifacts seem to have a dif-
ferent nature just like different animals or any other natural being of an acknowledged
kind. Thus once we admit the existence of kinds for natural objects, we need very good
reasons to deny the existence of kinds for artifacts. Moreover, since there are well-known
arguments according to which the individuation of an artifact never coincides with the
individuation of physical kinds, such as the amount of matter composing it, it is not even
easy to accept a reductionistic^3 solution as proposed by Devitt and others.
Among the arguments that have been put forward to deny the existence of real kinds
for artifacts, there are some that are based on the idea that artifact kinds cannot be real
kinds because they are functional kinds. In the present study I examine the relation between
artifact functions and artifact kinds in an attempt to establish reconciliation between the
functional characterization of artifact kinds and realism.
11.2 The Argument from Multiple Realizability
Why can functional kinds not be real kinds? Real kinds are supposed to collect objects
that have the same nature. According to some famous examples presented by Kripke
(1980) and Putnam (1970), the nature can be identifi ed with the molecular or atomic
structure for chemical substances, and with the genetic content for biological entities. That
is, objects of the same real kind have the same inner structure that causes them to react
with the environment and behave in similar ways, which is why real kinds strongly support
induction. For functional kinds, the following principle holds:
(FK) o is an item of a functional kind K iff o has the function F.
Generally speaking, “o has the function F” means roughly that o is used for or is produced
for F. This interpretation plus the widely accepted principle that functions are multirealiz-
able leads to the consequence that objects of the same functional kind may have very dif-
ferent structures and be composed of different materials. Identity of function does not
therefore guarantee any identity of nature, according to the Kripke-Putnam notion. Artifact
kinds, such that watch, chair, and pen are kinds of this type that collect objects with no
common inner structure, for this reason cannot be considered real kinds.
First, we need a more critical account of what “o has the function F” means. I do not
want to analyze the general epistemological problem of what the right criteria for function