200 Marzia Soavi
11.7 Final Remarks
Some antirealists argue that artifact terms behave like abbreviations for functional descrip-
tions, and that for this reason they cannot refer to real kinds but only to nominal kinds. I
propose the two following theses:
I. Real artifact kinds are those specifi ed by the author’s design—that is, by the specifi c
connection between a function and a structure;
II. The vernacular artifact kind terms correspond mainly to the Input-Output
characterization of the object—that is, to a generic functional description.
Common artifact terms, then, simply name clusters of real artifact kinds sharing the same
general functional characterization (input-output or merely output), and for the most part
we lack adequate terms for the real artifact kinds. The present proposal can thus, on the
one hand, provide an explanation for the fact that common artifact terms are normally
considered just as names of nominal kinds, while, on the other hand, it can render this fact
compatible with the existence of real artifact kinds, explaining the connection between
such nominal kinds and real kinds. The general terms in everyday language may or may
not refer to a general kind that can be a real kind. The problem is similar to that concern-
ing the biological taxa that are of a higher order than species. The present proposal is
neither meant to defend the reality of artifact kinds as ordinarily intended nor to reject
it.
Finally, it is important to note that the artifact kinds described here are not to be confused
with structural types. The fact that an object falls under a certain artifact kind has to do
with its history of production and not merely its actual structure.
Notes
- The distinction between strong and weak realism is drawn by Devitt (1997: 17–18).
- These are the traditional examples of the nonnatural kinds to be found in philosophical literature. Indeed it
is possible that kinds like bachelor or widow turn out to be theoretically relevant for a certain scientifi c discipline,
for example, for certain sociological theory. In the literature, the burden of proof is normally laid on those who
want to demonstrate that kinds of nonnatural science are natural kinds as well. This is a biased position I am not
endorsing here. - It is possible to roughly distinguish between a reductionist approach and an eliminativist approach. Claiming
that the kind tables, for example, is not a real kind and hence that tables are not real objects does not lead to the
idea that there are no objects that can be truly said to be tables. According to reductionists, such as Devitt (1997)
and Wiggins (2001), there are real physical objects classifi ed as tables. Being a table is a property like being the
fi rst child born in 2006 or being a Christmas tree. That is, being a table is a real property that is accidental with
respect to the nature of the objects having that property. Eliminativists, such as van Inwagen (1990) and Merriks
(2001), also deny the existence of those material objects identifi ed as tables, chairs, etc., hence according to these
authors there is nothing that is strictly speaking a table—there are only particles and properties of particles and
relations holding them together. - In the philosophical literature on function Millikan and Preston try to apply a selective criterion to function
attribution even to artifacts. But the main input to such an approach comes either from sociological literature or