these techniques (habituation and dishabituation paradigms) deserve
recognition both for their extreme ingenuity and for the extraordinary
patience they have shown in the service of cognitive science (Spelke, 1985;
see KarmiloV-Smith, 1992, for surveys of much of the infancy data). In a
typical case, an infant is repeatedly presented with a stimulus until ‘ha-
bituation’ occurs, and the infant’s sucking rate returns to normal. Then
new stimuli can be presented, varying from the original along a variety of
dimensions, and the extent to which the infant is surprised can be meas-
ured by the change in its sucking rate.
The results of these studies show forms of awareness in infants so young
that one can hardly speak of any plausible process of learning at all. This
developmental evidence supports nativism because it greatly reinforces
considerations of ‘poverty of the stimulus’. At just a few months – or even
merely a few hours after birth – babies clearly have very limited data! But it
should also be noted that more child-adequate techniques involve inves-
tigating infants’ awareness in the sort of domain-speciWc way which is
highly suggestive of modularity. Thus, new-born babies show preferential
interest in face-like shapes (Johnson and Morton, 1991). Neonates can also
detect numerical diVerences between arrays with a small number of dots or
shapes (Gelman, 1982; Antell and Keating, 1983), with control experi-
ments indicating that it is the number of dots they are reacting to. And
whereas Piaget thought that knowledge of basic properties of physical
objects, such as their permanence, was only acquired slowly through
sensory-motor interaction and certainly not before the end of theWrst year,
habituation trials show that four-month old babies are already making
inferences about the unity of partly obscured objects, and have expec-
tations concerning the impenetrability and normal movements of objects
(Spelkeet al., 1994; Baillargeon, 1994).
3.2 Dissociation evidenceIf the developmental studies have made the idea that we are entirely reliant
on general learning mechanisms unlikely, evidence from dissociations –
both genetically caused, and due to brain damage in adults – exhibits the
modularity of the mind in a surprising but unequivocal way.
For example, compare and contrast four diVerent genetically related
conditions: speciWc language impairment, Down’s syndrome, Williams’
syndrome, and autism. (The last of these will receive extensive discussion
in chapter 4.) First, children can exhibit a whole host of impairments
speciWc to language, including comprehension deWcits and various forms
of production deWcit, while being otherwise cognitively normal (see Rapin,
1996, for a review). Second, Down’s children have general learning diYcul-
58 Modularity and nativism