stimuli in the occipital cortex. These stimuli are then processed by the
visual system in the normal way (Kosslyn, 1994). This means that the
visual systemcanaccess centrally stored information to enhance its pro-
cessing; and so modules need not be completely informationally encap-
sulated, after all. (Either that, or we arbitrarily conWne the visualmodule
to the processing which takes place in the occipital nerve, just as far as the
initial cortical projection-area at the back of the brain, area V1 – seeWgure
3.1 above.)
Moreover, Fodor maintains that it is only input systems which are
modular, whereas central cognitive processes are not. At this point, too, we
feel we must part company with him, and reject the division between
modular input systems and non-modular central systems. At the very least,
a modular organisation for input systems and something completely dif-
ferent for central processes would be a theoretical awkwardness; and
would also sacriWce much of the evolutionary plausibility of modular
mechanisms, unless some special origin could be postulated for the divide.
This might be something which could just about be accepted, if the
non-modularity of central cognition promised to explain how central
cognitive processes work. But it does not do so at all. On the contrary,
Fodor is deeply pessimistic about our chances of understanding central
cognition. So far as he can see, it is modular systems which we can
investigate scientiWcally, both from the experimental angle and also from
the direction of cognitive engineering (byWguring out how such systems
might compute the output they need to deliver from their proprietary
inputs). By contrast, central systems are intractable because they lack the
limiting characteristics of modules – they are not domain-speciWc and they
are not encapsulated.
It is important to note that in laying emphasis on the divide between
input systems and central systems, Fodor is not simply acknowledging the
problem of consciousness. To be sure, more of what goes on in central
cognition is available to consciousness than the inaccessible processing
within input modules. And certainly most people, on thinking about the
matter objectively, would admit that consciousness is a queer phenomenon
and a surprising one toWnd in a universe of physical causation. So: what is
consciousnessfor? and: how could it possibly be implemented? are puz-
zling questions. We will be trying to give at least the outlines of answers to
these questions in chapter 9. But the claim Fodor makes is that the sorts of
processes which go on in central systems – conscious or not – cannot be
devolved to modules. The next section considers whether the arguments
for this are convincing.
Fodorian modularity 65