doctor–patient relationship. Since it is the beat of the heart which
enables the doctor to diagnose heart disease, that beat will confer
beneWts on that ‘system’, and so come to be picked out as a function of
the heart, after all!)
(2) How are we to determine, objectively, what counts asbeneWtto the
system so deWned?
The objection is that there is no principled way of answering these two
questions (we return to answer this objection in section 3.5). In contrast,
the evolutionary notion of function is held to be objective in its ap-
plication, and just as well deWned as any other notion in scientiWc biology.
For whether or not a property F has some evolved function/comes down
to the question whether it was the fact that instances of F caused/in the
past which has caused it to be the case that F is now instanced. And this is a
matter of objective causal fact. (In fact, given poly-functionality, this is not
so clear. Almost any morphological feature or behaviour will have a
complex cost-beneWt proWle. So is there then any objective way of selecting
an element from that proWle asthefunction? But if not, then the problem of
indeterminacy of content may return.)
3.3 The disjunction problem strikes again?Plainly, teleological semantics can makesomeprogress with the dis-
junction problem. If, in the half-light of evening, I mistake a tiger-in-the-
distance for a nearby tabby cat, then, one presumes, we can say that my
perceptual statemisrepresents the tiger as a tabby, on the grounds that it is
the making of just this sort of discrimination which my perceptual mechan-
isms were selected for. A tabby-cat-representation issupposedto lead to
actions of stroking and feeding, or at least ignoring the presence of, its
object. These actions are (needless to say) not appropriate in respect of the
tiger. All the same, Fodor argues that teleological semantics must still be
fatally infected with a version of the disjunction problem (1990, ch.3). For,
since teleology cannot distinguish between properties which arereliably
co-instantiatedin the organism’s environment, it will have to be left inde-
terminate which of these properties is represented.
Consider the snapping-reXex of the frog. This is normally made in
response to aXyXying by, but it can in fact be induced by any small black
thing moving across the frog’sWeld of vision, such as a shot-gun pellet
thrown by an experimenter. We are inclined to say that the frog’s percep-
tual statemisrepresentsthe shot-gun pellet as aXy. But can we vindicate
this by an appeal to function? Can we demonstrate that the function of the
frog’s percept is to representXiesas opposed tosmall black things,oras
Teleo-semantics 171