Functions 535as he does) cannot possibly have been competent in the use of the SE
concept of function.
None of these problems is fatal, although they are a little awkward. As far as
scenarios A and B go, the SE theorist can simply bite the bullet. Function claims
are already historical explanations in disguise, which is why it is not quite right
to say that eyes exist because their function is to assist in vision. And when we
remember that we have to deny only that thepurposeof the heart of an instant
lion is to pump blood, not that instant lion hearts are able to pump blood, it is
not so clear how worrying the SE account’s pronouncements are. The instant lion
is an unlikely animal indeed, and he has all the capacities and causal powers of a
normal lion. In such cases the rule should be ‘spoils to the victor’, as David Lewis
sometimes argues. The odd awkward pronouncement is a cheap price to pay for
an otherwise strong theory. The cost is not so cheap if we want to use the SE
theory for work in teleosemantics, for there we end up denying not merely that
the mental states of instant persons have the biological purpose of representing
states of affairs, we deny also that they have representational capacities at all.
Scenario C is also less damaging than many suppose. The SE theorist simply
needs to make clear the scope of her analysis. She is trying to make sense of
function claims in modern biology. It is perfectly legitimate to give an account
that only applies to modern biology, although an additional requirement of the
theory is that it should be able to explain why there is such continuity in how
functions have been described in spite of a radical change in the processes thought
to underpin such function talk. The suggestion that selection is at least a design-
like process — a suggestion that seems to account for much of the appeal of the SE
account in grounding heavy function talk — promises just such an escape route
for the SE theorist who does not wish to belittle what Harvey did for us.
A more interesting objection, although again not fatal, is the claim that the SE
account does not suffice for the individuation of trait types (important articles on
this topic include [Matthen, 1998], [Neander, 2002] and [Griffiths, manuscript]).
Sometimes it seems that SE theorists think that their account gives us the resources
we need to say what makes two token traits instances of the same type. On this
view, what it takes to be a heart is to have a certain function. Yet that cannot
be the whole story, and the SE theorist needs only to retain her modesty here. A
good way to see why there must be some way of saying when two traits are the
same that does not rely on functions is to look again at the concept of malfunction.
A damaged eye cannot help for seeing, and neither can an ear. Yet an ear is not
malfunctioning in this respect, but a damaged eye is. The SE account must make
sense of this by saying that helping to see is not the function of ears. That, in turn,
is to say that helping to see does not explain why ears increased their frequency,
while it does explain why eyes increased their frequency. But if two traits are
pleiotropically linked, then the fitness contributions that explain the increase in
frequency of one will also explain the increase in frequency of another. Suppose,
then, that earlobes are present in some population of organisms because they are
pleiotropically linked to an essential part of the eye — the lens, say. If assisting