modules provided their domains are diVerent from the domains of input
modules, and provided such conceptual modules can take the output from
input modules as (part of)theirinput. We willWrst sketch out the case for
believing in conceptual modules, before returning to Fodor’s more for-
midable point about encapsulation.
5.1 The case for conceptual modulesAs we have already noted, evolutionary considerations militate against the
idea of an unstructured general intelligence. Rather, since evolution oper-
ates by eVecting small modiWcations to existing systems, and by adding
new systems to those which are already in place, one might expect toWnd
cognition as a whole (and not just input and output systems) to be
structured out of modular components. This is just the hypothesis which
has been taken up and developed within the relatively new movement of
evolutionary psychology(Barkowet al., 1992; Hirschfeld and Gelman,
1994; Sperberet al., 1995b). By speculating on the cognitive adaptations
which would have been advantageous to humans and proto-humans in the
environments in which they were evolving (as well as drawing on develop-
mental and cross-cultural evidence), evolutionary psychologists have pro-
posed a rich system of modular components, including systems designed
for reasoning about the mental states of oneself and others; for detecting
cheaters and social free-riders; for causal reasoning and inferences to the
best explanation; for reasoning about and classifying kinds within the
plant and animal worlds; for mate-selection; for various forms of spatial
reasoning; for beneWcence and altruism; and for the identiWcation, care of,
and attachment to oVspring. This is the ‘Swiss army knife’ model of
cognition (to be contrasted with the picture of the mind as a powerful
general-purpose computer), according to which human cognition derives
its power and adaptability from the existence of a wide range of specialist
computational systems.
The case for a ‘mind-reading’ or social cognition module will be con-
sidered in detail in chapter 4. But two points are worth noting here. The
Wrst is that the mind-reading system must plainly operate upon conceptual
inputs, rather than on low-level perceptual ones. For in general it is not
bodily movements, as such, which are the targets of folk-psychological
explanation, but ratheractions, conceptualised as directed towards speciWc
ends. Moreover, the mind-reading system can just as easily be provoked
into activity by linguistic input (which is archetypally conceptual, of
course), as when someone describes another’s actions to us, or as happens
when we read a novel and seek to understand the actions and motivations
of the characters. The second point about the mind-reading system is that
Input systems versus central systems 67