54 Françoise Longy
A selected effect is something real and objectively ascertainable if the mechanism of
selection is itself objective in that it operates on real effects. Now, both natural selection
and sociocultural selection operates on real effects. (The case of “mental selection” that
Wright wrongly put on a par with natural selection (1973: 163) is here left aside.) More
precisely, a selected effect is a type of real effect since it is supposed that some of these
effects (some token effects) have already occurred and have subsequently acted as a
cause. For example, circulating blood is the selected effect of hearts because many
hearts have circulated blood, and the fact that they did so caused hearts to be preserved
by natural selection. So a selected function supposes a connection between existing
items, let us say the Xs, and existing effects. There is no doubt that some, if not all, of the
Xs have the capacity to produce the functional effect F since some of them have
already produced it.
That is not the case with intended functions. The conviction that at least some Xs should
have the capacity to do F in the right circumstances is not suffi cient to ensure that it is
effectively so, however rational and justifi ed such a conviction might be. Rationality does
not preclude errors. So, by defi nition, an intended function does not necessarily refer to
real effects of the type of item envisaged—it refers only to rationally predictable ones.
Now the difference between an actual effect and a rationally predictable effect is by no
means superfi cial. A criterion that puts real effects on a par with rationally predictable
ones, or ontological conditions on a par with epistemological conditions, can determine
only a very heterogeneous class from an ontological point of view. Such would be the
case with the following disjunctive defi nition: X has function F if F is a selected effect of
the Xs or if F is an effect one can rationally expect some if not all of the Xs to have. So
the class of artifact functions demonstrates no substantial unity—but rather quite the con-
trary if it is made up, as is currently presumed, of both intended and socioculturally
selected functions.^8
What conclusions should we draw from the discrepancy between, on the one hand, the
current image of functions in which they separate easily into two homogeneous categories,
the biological and the artifactual functions, and, on the other hand, the image we obtain
when we deepen the analysis and avail ourselves of the current etiological theories? Should
we imagine a dividing line passing elsewhere than in between artifact functions and bio-
logical functions? For instance, between selected effect functions and purely intentional
ones? Should we try to discover another criterion in order to obtain the desired distinction?
Or should we simply abandon the very idea that there are different categories of teleofunc-
tions? I argue that at least as far as material artifacts and biological entities are concerned,
we should renounce the idea of putting their functions into different ontological categories.
This therefore implies that we must abandon the idea that we should or could have differ-
ent accounts of functions of material entities depending either on the nature of the entities
(natural or artifi cial, inert or living), or on the origin of the functions (selection or
intention).^9