whereas in learning a language one has to master complex phonetic and
syntactic systems, and the rules which assign semantic properties to senten-
ces, while also developing the physical skills necessary to articulate speech.
Yet it is reading which requires tuition and special coaching, without
which the ability will never be acquired; and many otherwise normal
children never do acquire it. However valuable as a skill, reading is a learnt
rather than a natural domain of human cognition. In contrast, capacities
for speech are developed by all children in just a few years, in the absence
of special disability.
We should stress, however, that the sort of developmental rigidity we
have in mind as supporting nativism is entirely compatible with a con-
siderable degree of developmental plasticity (contrathe crude picture of
innateness attacked by Elmanet al., 1996). This is because rigidity is a
question of thegoaltowards which the normal development of cognitive
organisation tends, whereas what can be plastic is the way such modular
organisation is implemented in the causal processes of development. Some
modules do seem to require dedicated neural structures – for example,
vision and its various sub-systems. Others, while developmentally rigid in
discharge of function, may be developmentally plastic in the way in which
they are implemented. Handedness, for example, has a marked eVect on
hemispheric specialisation. Right-handers normally have their speech cen-
tres in the left hemisphere and left-handers in the right hemisphere. But for
either handedness, if the region where speech centres would normally
develop is damaged at an early age, then the corresponding area of the
other hemisphere can be ‘co-opted’ for those modular functions. (An
obvious speculation about the developmental process is that cognitive
modules with irreplaceable and dedicated neural structures have a more
ancient evolutionary history, dating back at least six million years to the
common ancestor of ourselves and chimpanzees, in whom there was
probably a minimum of hemispheric specialisation. See Corballis, 1991.)
Whatever the pathways of development, they are compatible with our
position if in the normal case they lead to a common outcome in terms of
modular cognitive organisation, from varied input. For it is this common
outcome which will be innately pre-speciWed – at least in so far as genetic
inheritance predisposes towards the development of such a cognitive sys-
tem.
We should also stress that the sort of nativism defended here is not
seriously threatened, we believe, by the progress made by connectionist
modelling of cognitive processes. It is true that connectionism is often seen
as an anti-nativist research programme. And at least in its early days, the
hope was to discover systems which would learn to produce any output
from any input, while mimicking human performance – that is to say,
The case for nativism 55