Biological Approaches to Mental Representation 5595 SOME PROBLEMS FOR TELEOSEMANTICSIf the argument just offered is along the right lines, there is reason to persist in
trying to produce a good teleosemantic theory. In this last section I outline four
problems for teleosemantics, along with sketches of possible responses to them. My
treatment is sketchy and partial, of necessity. But my aim is merely to provide
readers with some idea of the challenges that teleosemantics theories face and an
idea of a few of the resources that they have available to them.^4
(1) Novel concepts. It is extremely unlikely that all of our concepts are innate.
For example, it is extremely unlikely that even the folk species concepts (CAT,
SHEEP, WOMBAT, etc.) are innate, let alone that more obviously modern con-
cepts, like ELECTRON and JET-PLANE, are.^5 What it means to say that a trait
is innate is a delicate issue, but it will suffice for present purposes to say that a
concept is innate if there was selection specifically for it, and if the selection was
ordinary natural (genic) selection. Acquired as opposed to innate concepts are a
prima facie problem for teleosemantics if it relies on a notion of function that in
turn relies on the idea that functions are what items were selected for by ordinary
natural (genic) selection.
How do proponents of teleosemantics respond to this? One type of response
is to appeal to other forms of natural (here, meaning non-intentional) selection
to underwrite the relevant notion of a proper function. Meme selection [Millikan,
1984] and conditioning [Dretske, 1986] are sometimes mentioned in this context,
and neural selection might also be a possibility. A danger with this approach is
that the notion of selection becomes so vague that it threatens to become vacuous,
but with work it might be shown that this danger can be met.
A more complex but, to my way of thinking, equally plausible idea is that
the content of acquired concepts is not derived straightforwardly from a single
selection process, but that selection is involved nonetheless. For example, it is
more plausible that a general disposition to form animal species concepts is innate
than that all of our particular animal species concepts (CAT, SHEEP, WOMBAT,
etc.) are. This disposition might take the form of a disposition to store memories
(schematic or detailed) of animals of the same species within a specialized type of
“mental folder” (a metaphor for a functional component). Along the same lines,
we might have an innate disposition to form other kinds of concepts too, such as
concepts of artifacts, natural kinds, and individuals, so that we come equipped
with a disposition to form different kinds of concepts for different kinds of kinds
(^4) I omit in what follows a resource that some proponents of teleosemantics think terribly
important: the role of the representation’s consumer. I have never been persuaded that this does
any important work for teleosemantics that cannot be done without it, but see Millikan [1989a]
and Matthen [2006] for the opposed view.
(^5) There is much debate on this, but this claim seems plausible. After all, infinitely many
different species are possible. And assuming that we could see each of them, and interact with
each of them in the usual way, there is every reason to think that we could acquire a concept of
whichever species happen to be present in our environment. Since we cannot possess an infinite
number of concepts innately, we must therefore be able to acquire particular species concepts
ontogenetically.