560 Karen Neander
of things. Genic selection might then partly determine the content of a particular
animal species concept. It might determine that the content of (e.g.) CAT is some
species or other, since that is the specialized type of “mental folder” that is used.
Then the question becomes how the content of the concept is tied to the particular
species involved.
At this point a teleological theory of content could be developed in a variety
of ways. It might appeal to an alternative form of selection, as mentioned before.
Perhaps conditioning or neural selection tunes the specialized “mental folder” to
the particular species. Or it might appeal instead to a further causal condition.
For instance, the idea might be that each new animal-species “mental folder” refers
to the species to which its triggering specimen belongs.
(2) Non-existent intentional objects. However, some concepts cannot be so
readily handled. For example, no matter what kinds of actual selection processes
we allow, our perceptual mechanisms have not been selected for producing a UNI-
CORN in response to a unicorn, and nor was any actual unicorn a triggering cause
of our UNICORN concept. Since there are no unicorns, neither we, nor our an-
cestors, could have causally interacted with them. A parallel point applies to all
concepts of things with which we have not or cannot have causally interacted at
the time of our reference to them: e.g., TOMORROW.
One response to this, insofar as it concerns non-existent intentional objects, is
to deny that concepts of them have referential content of the kind analyzed by
teleosemantics. Those who do this can still allow that UNICORN has something
like a Fregean sense — a mode of presentation, or a cognitive content. However,
this is unsatisfying in part because it doesn’t generalize (e.g., to TOMORROW).
Another option is to concede that UNICORN and TOMORRROW et al cannot
be simple. To say that a concept is not simple is not the same as saying that
it is not innate. A concept can be simple without being innate (I described two
ways in which this might happen in the preceding sub-section). If a concept is not
simple, it derives its content from the contents of other concepts, and this is the
possibility I am now pointing to. The contents of UNICORN and TOMORROW
on this view would derive from the contents of other concepts (UNICORN from
HORN, MIDDLE, FOREHEAD, etc., and TOMORROW from DAY, AFTER,
TODAY, etc.).
Examples like UNICORN and TOMORROW are not devastating for teleose-
mantics unless there is good reason for thinking that they must be simple. Argu-
ments for conceptual atomism [Fodor and Lepore, 1992] could be seen as consti-
tuting such reasons but they are not strong enough for this. This is a large claim
that cannot be adequately defended here, but bear in mind that the arguments for
conceptual atomism are negative arguments against particular theories of complex
concepts. Consider, for example, the view that complex concepts have definitions
consisting of necessary and sufficient conditions for the application of the concept.
A major problem with this view is that few of our concepts seem to have good
definitions. But the fact that few concepts seem to have good definitions is no
reason to think that none do. And TOMORROW would appear to be an excep-