It is doubtful whether an appeal to asymmetric causal dependence can
help Fodor here. For it appears that we have asymmetricdependence
between themouse-to-mouseconnection and themouse-mating-to-mouse
connection – that is, if mice had not caused me to thinkmouse, then nor
would matings of mice; but if mice-matings had not caused me to think
mouse, then nor would mice. Examples of this sort provide powerful
support for some form of teleological semantics or for some form of
functional-role semantics (see sections 3 and 4 below, respectively). Thus
one might say that what makesmousemeanmouse, is that itsfunctionis to
co-vary with mice and not mouse-matings – that is, the role ofmousein
cognition is to focus our behaviour diVerentially upon mice. Or one might
say that what makesmousemeanmouse, is the kinds ofinferencewhich the
subject is inclined to make – for example, frommousetoliving thing, and
not frommousetotemporal event.
A similar problem arises in respect of causal co-variance with events
further in. Suppose that by electrically stimulating just the right spot in my
cortex, an experimenter can make me thinkmouse. Then what shows that
mousemeansmouseand notimmediate neural cause ofmouse? But in this
case we appear to have asymmetric dependence the wrong way round – if
mice had not caused me to thinkmouse(for example, because the Men-
talese termxyzgets harnessed to do the job instead), then still the (actual)
immediate neural cause ofmousewould have caused me to thinkmouse
(wheremousewould now be harnessed to mean something else); and if the
(actual) immediate neural cause ofmousehad not caused me to think
mouse(but some other neural event had), then nor would mice have caused
me to thinkmouse(rather,mousewould have been harnessed to some
other meaning). And here, too, the solution would appear to be to move to
a form of teleological, or perhaps functional role, semantics. At any rate,
we predict that if there is any way of securing the asymmetric dependence
of theimmediate-neural-cause-of-mouse-to-mouseconnection upon the
mouse-to-mouseconnection, it will have to make at least covert appeal to
theroleofmousein normal cognition.
3 Teleo-semantics
As we have already seen, one way to motivate a form of teleological
semantics is to come to regard it as necessary to rectify the deWciencies with
informational semantics (this is Dretske’s view, 1988). But there are a
number of other ways in which one might try to motivate the attempt to
naturalise semantics into teleological, biological terms.
Teleo-semantics 167