552 Karen Neander
who support teleosemantics are not claiming that functional norms (or “quasi-
norms” if you wish) are prescriptive; they are claiming that they are descriptive.
They are claiming that they can be analyzed in terms of what there was selection
for, not by God or by humans, but by natural processes of selection. If the norms
of content were prescriptive, and functional norms were not, then that might be
a problem for teleosemantics. Or if functional norms were prescriptive, and the
norms of content were not, then that might also be a problem. However, according
to teleosemantics, both are descriptive. On this view, neither claims about normal
functioning nor claims about correct representation necessarily imply claims about
what should be the case, either morally or pragmatically speaking. And while it
is true that, when we speak of functions, we often find ourselves speaking of what
traits are “supposed to do”, or even of what they “ought to do”, this is just sloppy
or colloquial or metaphorical talk of what they were selected for doing, on an SE
account.
We can get some sense of the role that the notion of an SE- function is intended
to play in a theory of content if we start with a crude causal theory of mental
content, and see how the problem of accounting for the possibility of misrepresen-
tation arises for it. The crude causal theory says that a representation represents
whatever causes it. It says that a token representation of typeRcorrectly repre-
sentsCs iff (if and only if)Cs and onlyCscauseRs. For example, it says that
CAT-representations represent cats iff it is the case that cats and only cats cause
CATs.
This theory has many well-known problems, one of which is the problem of
error.^2 Consider an occasion on which, when walking down the street, I see some
crumpled newspaper blown by the wind as a cat slinking. Since the crude causal
theory says that the content of a representation is whatever causes it, and the
newspaper is now among the causes of my CAT-representations, it entails that the
content of my CAT-representations includes the newspaper. This is part of what is
(after [Fodor, 1990a]) referred to as the “disjunction problem”. The crude causal
theory entails that my CAT-representations refer, not just to cats, but instead to
catsornewspapersoranything else I happen to mistake for a cat. Of course, this
is intended as a reductio of the theory since it shows that the theory makes error
impossible.
There are a number of directions one can take here. One might decide that
any attempt to develop a causal-information based psychosemantics is doomed to
fail. That would be too hasty, but this is a direction one might decide to take
anyway, once further considerations come into play. Another approach is to try
to distinguish the content-conferring causes from the non-content-conferring ones.
A teleological theory of mental content might take either tack, but I follow the
(^2) An alternative to the idea that mental representation starts with information is an isomor-
phism theory. And an alternative to the idea that information is based on causation is that a
notion of indication suffices. There are teleosemantic theories that use both of these ideas. The
problem of error could have been introduced using these ideas too, but for simplicity I here stay
with a causal-informational approach.