Realism and Artifact Kinds 191
process for copying characters grants that P—the character that gives some advantages to
the individual—is passed to the following generation. The increasing proportion of indi-
viduals with P in relation to individuals without P is then the consequence of the selective
action of E, the environment, and the inheritance process. For artifacts, there is no auto-
matic process of inheritance; it seems that the only copying process that can play the same
role of inheritance is the intentional copying of designers. The fact that some features are
copied more frequently than others is partly because such features are more effective in
the performing of a certain function F and partly because designers intentionally copy
those features that are more effective with respect to a certain function. Artifacts with
roughly the same function F, such as glasses, are produced with those features that design-
ers regard as apt for the performing of F. These features may or may not be copied from
previous types of glasses.
Of course the introduction of new features providing new solutions to a functional
problem is always possible, either intentionally or by accident, and they can play the same
role that the occurrence of variations plays in the natural process of selection. If we adopt
this perspective, we may have to admit that the reproductive process of artifacts is more
prone to variations than the biological process, but this seems a reasonable price to pay.
Nonetheless, it seems clear that according to a selective account of this kind, the function
F for which a certain object (or features of a certain object) has been selected is simply
the function for which it has been intentionally designed. It is the effi ciency in performing
F of an object with P, plus the intention of the designers in copying those objects that
perform F better, that can explain why P is copied over and over again. But if that is the
case, why not simply appeal to an intentional criterion for function attribution, which
would allow us to attribute function not only to copied features like P but also to those
that have been newly introduced?
Indeed it seems plausible to accept that the selection process for artifacts is different
from the selection process for biological entities and that it is possible to subsume both
the artifact and biological copy relations under the same type allowing for the existence
of a weak copy relation. That is, we do not need any counterfactual dependence, and we
can allow for a weaker causal role. Designs are almost never the result of a completely
new discovery; designers always take inspiration from previous designs. It seems reason-
able, then, to grant that there is a noncasual similarity of artifacts produced according to
the same design.
Nonetheless, even if we accept a weak copy relation and we assume that some kind
of suitable copying process exists, fi nding a similarity between the biological copy
relations and the artifactual one turns out to be very hard. It is then very diffi cult to give
sense to the hypothesis that in both biological and artifact cases we attribute function on
the basis of a unique and common selection theory. The main differences are the
following: