188 Marzia Soavi
attribution to artifacts are, as pluralism seems to be a perfectly feasible solution. Instead
my concern is to try to understand which criteria for function attribution are involved in
our classifi cation of artifacts as kinds.
Traditionally since the work of Wright, two distinct types of criteria for function attribu-
tion have been proposed: they are the selective criterion for biological entities and the
intentional criterion for artifacts—either related to use or to production. More recently this
distinction has been challenged by those claiming that it is possible to adopt selective cri-
teria for function attribution to artifacts.^4 That makes it possible to individuate three main
criteria for attribution of function to artifacts: a selective criterion analogous to the criterion
defi ned for biological entities, and the two intentional criteria of use and production
criteria.
Then accepting the common functional classifi cation of artifacts mentioned in the fi rst
section,
(FA) o is an artifact of a kind K iff o has the function F^5 ,
it is possible to analyze “has the function F” according to three different interpretations
and to formulate the following three options:
- o is an object of an artifact kind K iff o has been selected for F.
- o is an object of an artifact kind K iff o is used for F.
- o is an object of an artifact kind K iff o has been produced for F.
In (FA) and (1), (2), and (3), K and F must be suitably chosen; that is, we must grant that
there is some connection between F and K. According to a pluralistic approach, all three
solutions can provide a good criterion for some attribution of functions to artifacts. My
interest is not in eliminating any one of them as improper in an account of the attribution
of functions to artifacts but simply in exploring which notion of function can provide the
basis for a defi nition of functional kinds that will meet the challenges of antirealists for
artifacts.
11.3 First Option
The fi rst statement corresponds to the attempt to apply the function attribution worked out
for biological entities to artifacts.
- does not require that o is directly selected for the function F. What it requires is that
there is an appropriate history of selection for o, according to which it is true that o has
been selected for F.
Undoubtedly natural selection necessarily involves inheritance, that is, some process
granting the transmission of characters. Any general theory of selection that is thus aimed