ties,Wnding the acquisition of new skills and new information diYcult. But
they acquire language relatively normally, and the evidence is that they
have intact social cognition, or ‘mind-reading’ capacities, as well. (In fact,
Down’s children are routinely used as a control group in experiments on
mind-reading impairments in autism, with most of them passing at a rate
comparable to mental-age-matched normal children). Third, Williams
children, too, have intact – indeed, precocious – social cognition and
language, but do not suVergenerallearning diYculty. They acquire infor-
mation without diYculty, but have severely impaired spatial cognition,
and appear to have great diYculty in tasks which require theorising
(KarmiloV-Smithet al., 1995; Tager-Flusberg, 1994). Finally autistic chil-
dren can have normal language (at least in respect of syntax and the
lexicon, as opposed to pragmatics) but have poor communication skills,
and have impaired social cognition generally (Frith, 1989; Baron-Cohen,
1995). It is very hard indeed to make sense of these phenomena without
supposing that the mind is organised into a variety of pre-speciWed
modules, which can be selectively impaired.
Turning now to brain lesions in adults, these have frequently been
associated with special and unanticipated impairments, such as particular
forms of agnosia or aphasia. Prosopagnosia, an inability to recognise
faces, provides a good example (Bruce, 1988; Bruce and Humphreys,
1994). Subjects can be impaired in this ability without any corresponding
decline in their ability to recognise other objects. In this case the dis-
sociation evidence is supported by other reasons for supposing that we
have a special processing system for dealing with facial recognition. The
domain is one to which, as we have seen, infants display preferential
attention from a very early age; and the adaptive importance of interaction
with speciWc others ensures signiWcant cognitive eVects from spotting ‘The
same face again’ or judging ‘That’s a new face I haven’t seen before’.
Several other visual agnosias are well known, suggesting that visual
perception is really a hierarchy of interconnected modules. For example,
some subjects have been impaired speciWcally in their recognitional capac-
ities: they could still visually delineate objects (as evidenced by their
drawings), and they knew what a certain kind of object was (as evidenced
by their ability to supply deWnitions) – and yet puzzlingly, they still could
not recognise even the most familiar things. In other cases object-recog-
nition may be unimpaired, but the subject is no longer able to perceive
movement in the normal way. (See Sachs, 1985; Humphreys and Riddoch,
1987.)
The same general story – of various dissociations suggesting a variety of
discrete processing systems at work – can be told in relation to speech and
language-processing. Aphasia, which is an inability to produce or to
Developmental rigidity and modularity 59