The understanding of autism has come a long way since 1938 when Leo Kanner first observed that a minority of children existed in a state of "extreme autistic aloneness". Kanner accounted for this by pointing to a particular style of parenting that was cold and aloof - the "refrigerator mother". Contemporary theorists have placed the roots of autism firmly in biology. In Mindblindness, Simon Baron-Cohen elegantly relates his variant of the biological theme. His theory has had enormous impact in the field of psychology and his book is a welcome means of making this important view more widely accessible.

Baron-Cohen has set himself the ambitious task of bringing together evolutionary biology, cognitive science, developmental neuropsychology, comparative psychology, even literature. He takes us on a journey through evolutionary history, into the realm of current human mind-reading, ending up in the impoverished social world of persons with autism.

Let me begin with the evolutionary story. Baron-Cohen presents a lucid and interesting account of the genesis of mind-reading as an adaptation to the demands of living in increasingly large social groups. The larger the group, the more complex and varied are the social interactions of which to keep track. This necessitated the emergence of a more powerful tool to predict and explain behaviour: the ability to read minds - to understand the mental states of others.

While Baron-Cohen's assertions about the phylogeny of mind-reading seem credible, he glosses over the large qualitative differences that exist between non-human and human primates. Humans, for example, are language users; animals, arguably, are not, at least not in the sense of using language in the complex communicative ways that humans do. Once you can use language you have an immensely powerful tool for mind-reading. Not only can you get the thoughts in your head into the mind of another, but you can check to make sure the other has understood. Baron-Cohen clearly views language as being important in mind-reading but provides no explicit comparison of language in apes and humans. In spite of this, I think readers will find Baron-Cohen's evolutionary account stimulating, if somewhat speculative. Mind-reading does appear to have great adaptive and predictive power, giving "nature" good reason to support its development.

ÁñÁ«ÊÓƵ

Baron-Cohen goes on to explain how exactly the ability to mindread, as it has evolved, develops in humans. Building on earlier work by Alan Leslie and others, Baron-Cohen proposes that it is the maturation of specific parts or modules of the brain that enable us to understand the mental lives of others. Baron-Cohen asserts that there are at least four distinct modules that work together to allow fully fledged mind-reading. Early in infancy we start with a primitive module for detecting intention (the ID); we then acquire a module for detecting eye direction (EDD); followed by a shared attention mechanism (SAM); and, finally, all these interact to permit the coming on-line of the theory of mind module (TOMM). The reader has to work hard to keep track of this cast of characters and it might be easy for the uninitiated to get just a tad lost.

Some modules are more important than others. You can get by without EDD and still develop mind-reading, as in the case of visually impaired children. What you cannot do without is either ID or SAM. It is the absence of SAM that is said to account for the specific deficit in TOMM that Baron-Cohen maintains exists in children with autism. You cannot get to TOMM if you do not have a working SAM.

ÁñÁ«ÊÓƵ

By three to four years of age normal children, SAMs well in place, begin to understand that people have beliefs, desires, intentions and so on. In contrast, children with autism lack such mentalising ability. The big question is whether this mentalising deficit is the core deficit of autism. Autism is characterised by a number of impairments, only some of which relate to mind-reading. Children with autism also engage in repetitive behaviour, have restricted interests, and have difficulties recognising patterns. Do we need to postulate increasing numbers of faulty modules to account for this or is there some more general underlying deficit that mindblindness fails to capture?

One of the most promising aspects of Baron-Cohen's theory is the potential for early detection of autism, currently difficult to diagnose until around three years of age. With the mindblindness theory, because SAM is a necessary precursor to TOMM that emerges by about 18 months, there is new hope for early detection of autism.

Given the promissory note that such early detection holds out, I was surprised to see almost no space afforded to a discussion of intervention or treatment. Baron-Cohen gives us one poignant example of a very high-functioning woman with autism, Temple Grandin, and illustrates how she seems to have found an alternative pathway to understanding how to survive in her social world. Perhaps the next job of theorists and researchers should be to work towards a greater understanding of these different routes in order to reduce the impact of autism on children and their families. To do that we need an adequate account of the deficits faced, and Baron-Cohen's theory goes a good distance towards that goal.

Suzanne Hala is research fellow in experimental psychology, University of Cambridge.

ÁñÁ«ÊÓƵ