5 Input systems versus central systems
Characterising their main cognitive functions in an ordinary and un-
theoretical way, central systems operate to form beliefs and decisions. And
forming both beliefs and decisions involves reasoning, whether the process
of reasoning is conscious or not. But reasoning will often require one to
bring together information from various diVerent domains. In thinking
about whether to take on a dog as a pet, for example, one will need to
weigh up such disparate things as the appeal of canine companionship, the
children’s enthusiasm, and the health beneWts of some obligatory walking
against, on the other side, the costs and responsibilities of care, the
increased dangers of infection and allergic reactions in the household, the
emotional blow of bereavement, and so on.
In outline, then, Fodor’s argument for the non-modularity of central
systems goes like this. Central systems are the area of cognition in which
we achieve integration of information from various domains. If they
integrate information across domains, then they are not domain-speciWc.
He also argues that they are ‘in important respects,unencapsulated’ (1983,
p.103). If they are not domain-speciWc and not encapsulated, then they
lack the main characteristics of modules. So the conclusion would seem to
be that central systems are not modular.
Let us take this argument a bit more slowly, starting with the claim that
central systems are not domain-speciWc. It seems obvious that information
does get integrated, in terms of belief-formation, in terms of speech produc-
tion, and in terms of the initiation of action. Surely this can only happen if
separate information-processing streams deliver their output to systems
which can somehow put the information together. So if I come to believe
that there is a pig in my garden, I might well exclaim ‘There is a pig out
there!’.Prizingthe conditionof mylawnandXower-beds,Iamgoing to take
what steps I can to render the garden pig-free as soon as possible. Granted
that I would not be in this pig-spotting state of alarm, if there was not
something looking like a pig to me – and probably also sounding and
smelling like a pig (as proof against hallucination) – but that is clearly not
enough on its own toWx a belief and raise fears of its consequences.
Considerations of this sort do establish that central systems are not
domain-speciWc in the way that input systems are. So, they are domain-
general, then? No, that does not follow. As a number of recent thinkers
have pointed out, this line of argument does not rule out the possibility of
conceptual modulesbeing deployed in thinking (Smith and Tsimpli, 1995;
Sperber, 1996). Of course, if the conceptual modules simply duplicate the
domains of the input modules, no integration could occur. But a certain
degree of integration of information can be achieved via central conceptual
66 Modularity and nativism