Realism and Artifact Kinds 189
at covering both a theory of selection for biological entities and artifacts must defi ne a
general relation of inheritance that is viable for the application to both.
In line with Millikan, we can introduce a general notion of “copy relation” that is
intended to cover biological inheritance and the corresponding phenomenon for artifacts.
To increase our conceptual tools we can distinguish between a strong and a weak notion
of “copy.”
- According to the strong notion the relation of copy implies a counterfactual dependence
of the features of the copy on the features of the original. - The weak notion of copy does not require any strict counterfactual dependence of this
type; it simply requires that there is a noncasual similarity between copies.
According to Millikan, any copy relation involved in a selective process has to be of the
strong type. It must grant that for every determinable property such as skin color or iris
color there are local laws governing the transmission or copying of the determinate char-
acters in such a way that counterfactuals can be warranted.^6 To defend a strong selectionist
approach to the attribution of function to artifacts, we need to individuate a mechanism
that can grant the holding of a strong copy relation for artifacts of the same kind. Indeed
artifacts do not reproduce themselves as do biological entities, but the notion of “copy”
allows us to discard the material aspect of the process. Alternatively if we do not succeed
with the strong notion of “copy,” we can still try to create a selectionist criterion for artifact
function that adopts a weak relation of copy rather than a strong one. It seems reasonable
in fact to allow some discrepancy between the processes of selection for natural beings,
and for artifacts. We may agree, for example, on the identity of type between the copy
process for biological entities and for artifacts while accepting that there is a difference
in the degree of predictability—the copying process for artifacts is simply less predictable
where the transmission of characters is concerned and leaves more space for incidence of
variation.
There is an important role that the relation of copy plays in the theory of selection; to
illustrate it I consider the general analysis of the theory of selection provided by Lindley
Darden and Joseph Cain (1989). They develop an analysis in fi ve steps: preconditions,
interactions, effects, longer-range effects, and even-longer-range effects. Preconditions are
a population of entities coexisting in the same environment E and differentiated by the
fact that some of them have a certain property P while others lack P. Interaction is due to
the fact that E has some critical factors C and to the fact that members of the population
that have P interact with C in a different way with respect to the other members. Effects
are due to the fact that, for example, interaction with C brings benefi ts only (or more) to
those members that have P. There are longer-range effects when the interaction and fi rst
effects are followed by an increased reproduction of members bearing P. Even-longer-
range effects are, for example, the accumulation over generations of benefi cial properties
like P and the subsequent production of lineages of individuals.