and not to shrews-or-mice. Thus frommousethe system infersnot shrew,
and also perhapscan be an indoor pest. And it will lead me to answer
‘Yes’ if asked whether there is a mouse and not a shrew nearby, and so on.
Somouserepresentations have a variety of further eVects, on inference and
action, whose success requires the presence of a mouse, and not of a shrew.
So ifmouseis to have the eVects which it issupposedto have, it needs to
carry the informationmouse, and notshrew-or-mouse. And that, accor-
dingly, can be said to be its truth-condition.
It is important to note that most of those who opt for some version of
teleo-semantics, such as Millikan (1984, 1989), Dretske (1988), and Pa-
pineau (1987, 1993), are explicit in endorsing anevolutionary,selectionist,
notion offunction. On this account, the functions of any property F are
those eVects of F which explain why the system in questionhasthat
property – that is, in terms of which we can explain how the property was
selected for and/or has been sustained in systems of that type. So functions,
on this account, are essentially historical. To know the function of a thing
or property, it is not enough to observe what itpresentlydoes. Rather, you
must discover which eVect, from amongst the things which it presently
does (perhaps only rarely), explains why that thing or property exists.
Thus, to know the function of the peacock’s tail, you have to ask which of
the eVects of such tails in ancestral peacocks (presumably, in this case,
attractiveness to female peacocks) explains why they were selected for
and/or preserved.
The contrasting notion of function,notadopted by defenders of teleo-
semantics, is an a-historical one. On this account, the functions of any
property F are those eVects of F which are beneWcial to some wider system
or process of which F forms a part, or which play a role in sustaining the
capacities of that wider system. So on this account, whether or not a
property has a function is entirely independent of the question of how that
property came to be possessed in theWrst place. To ask, in this sense, about
the function of the peacock’s tail is to ask what the taildoes forthe peacock
(what beneWt it confers), in a way which just brackets oVas irrelevant the
question of how the peacock came to have such a tail. This notion of
function is generally rejected on the grounds that it is not scientiWcally
respectable, and that its application might be vague and observer-relative
(but see Cummins, 1975, for replies). For:
(1) How are we to determine the boundaries of the system of which the
target property forms a part? Unless some principled way can be found
of picking out systems, then just about any eVect of a thing can count
as its function. (Consider the sound made by the heart when it beats:
this, surely, isnotits function. But now take the ‘system’ to be the
170 Content naturalised