52 Françoise Longy
or the effect that, pushed by natural selection, explains the diffusion or the conservation
of the trait in the population (Millikan 1984; Neander 1991). So when we begin to inves-
tigate the nature of functions, two different sorts seem to emerge depending on whether
we are considering artifacts or natural entities. On the one hand, there are those that result
from human intentions, on the other hand, there are those that are due to natural mecha-
nisms such as natural selection.
Upon closer analysis, however, the artifact case proves to be more complex. Objects
may now have functions that were not the functions for which they were originally
made, as is the case with old cart wheels used as decorative pieces on restaurants’ walls.
Moreover, some artifact functions may result from a long history without anybody
having apparently ever done anything with the explicit intention of obtaining the desired
effect. For example, some ergonomic forms for tools have probably been gradually
selected and copied without anyone planning them explicitly with the particular aim of
ergonomic correctness. These cases where parallels with natural evolution can be seen
have prompted many authors to suggest that SEL could be applied to artifact functions
with sociocultural selection replacing natural selection.^2 As a matter of fact, in long-stand-
ing categories of artifacts like hammers, clocks, and cars, an evolution has taken place
that can be attributed to innovations gradually selected by buyers and users. Regardless
of whether or not the positive effect justifying the diffusion of an innovation has been
foreseen by whomever introduced it, that will be the function with which the new feature
will be associated. Moreover, such a sociocultural mechanism is able to explain changes
of function. In fact the effect for which people buy and use a type of entity may vary
historically.
However, not every artifact function, whether that concerns the whole artifact or just
one of its features, can be identifi ed with the effect for which such an artifact has been
bought in the recent past (a coffee machine for making coffee) or preferred to others (a
coffee machine with a drop stop for stopping drops). As Houkes and Vermaas emphasize,
there are also the functions attributed to a fi rst generation of artifacts (2003: 264–65). Such
functions cannot result from any sort of sociocultural selection. Thus the only explanation
seems to be that they refl ect the designer’s or the producer’s intentions. The traditional
intentionalist conception of artifact functions is apparently an unassailable spot as far as
new artifacts are concerned.
To summarize, artifact functions are not as easy to analyze as one might have fi rst
thought. They may have two sorts of origin, either they may result from the intentions of
some inventor or from sociocultural selection. Nevertheless, such a difference in origins
might not be of great signifi cance, since intentions are involved in both cases. In the fi rst
case, there are the inventors’ intentions, and in the second case there are the buyers’ and
users’ intentions. So, even if the situation is more complex than it fi rst seemed, the current
assumption that all artifact functions belong to one and the same type because they all
depend on intentions might still hold. It requires a deeper analysis to judge whether the