opposed toXies-or-shot-gun-pellets? Fodor argues not, since in the en-
vironments in which the frog’s perceptual system has evolved and been
maintained, something is aXy if and only if it is a small black thing, and if
and only if it is aXy-or-shot-gun-pellet. And from the point of view of
evolution, it does not matter which way you tell the story. You can either
say that the function (and hence the content) of the frog’s state isto
representXies, since that state has evolved because of its eVect in causing
snappings which have led (given the nature of the frog’s digestive system)
to enhanced survival; or you can say that its function is to representsmall
black things, since that state has evolved because of its eVect in causing
snappings which have led (given the fact that all the small black things are
Xies) to enhanced survival; or you can even say that its function is to
represent Xies-or-shot-gun-pellets, since it evolved to cause snappings
which led (given that all theXies-or-shot-gun-pellets have beenXies) to
enhanced survival.
Fodor goes on to claim that the notion offunctioncan only be used to
discriminate between these cases if we move to a version of it framed in
terms of counterfactuals, rather than in terms of actual selectional history.
That is, we can ask whether the state in questionwould have beenselected
for had the frog lived in an environment where all the moving small black
things were shot-gun-pellets. Plainly, the answer is ‘No’. Then if the
function of a state is those of its eVects which ensure its existence in actual
and counterfactualcircumstances, we can say that the function of the frog’s
movement-detector is to representXiesandnoteithersmall black thingsor
Xies-or-shot-gun-pellets. (Actually, pursuit of this sort of approach to
functional identity would lead us to the conclusion that the frog’s percept
representsambient food, since it would still have evolved as it did in
environments where all the small moving things were bees, or wasps, or
bits of beef thrown by human experimenters. This seems right.)
We shall return to the idea that semantics is best approached by em-
ploying an a-historical notion of function below. ButWrst it is worth noting
an alternative response to the form of disjunction problem raised here,
suggested by our earlier discussion ofmouse. A teleo-semanticist might
concede the indeterminacy in the content of the frog’s representation
which causes it to snap, because the mechanism in question is such an
immediate and simple one. But in more complex cases, where the represen-
tation feeds into a distinctive pattern of inferences, these further eVects can
be used to resolve the indeterminacy. Thus, what shows thatmystate
representsXyand notXy-or-shotgun-pellet, on this account, is that I am
prepared unequivocally to infer from itliving thingandhas wings.If
these are amongst the eVects which the state is supposed to have, then they
can help to triangulate the correct function. On this approach, then, a
172 Content naturalised