This may end up being another lengthy review, so here's the headline:
THIS BOOK IS INTELLECTUALLY DISHONEST
Having said that I had better justify my claim pretty quickly. The primary problem is that in presenting his thesis the author extensively cherry-picks the evidence. Supporting evidence is discussed at length. Contrary evidence is minimised in two ways. Firstly on some occasions it is simply not mentioned at all. Secondly, if mentioned it is not discussed in detail and not referenced so one cannot look it up for one's self either. Twice the author says, "There is counter evidence but let's assume I'm right anyway." Cherry picking of results in this way allows the reader no opportunity to fairly assess whether the author has a sound scientific case or not.
The author also indulges in a chapter on "evolutionary speculations." Now, whilst admitting that they are speculations a couple of times, what the author is doing here is making up plausible stories to justify his position, then using a few anthropological studies to back him up somewhat. Again, there is no presentation of any counter-arguments that might exist. The chapter appears simply in order to act as what Daniel Dennett describes as an "intuition pump." That is an attempt at rhetorical persuasion, rather than scientific discussion. The kind of approach we are talking about is rife in socio-biology - a field so derided by other scientists for its total lack of rigour that it changed its name to evolutionary psychology, in order to carry on being allowed to waste public research funding. The fundamental trouble with this type of evolutionary Just So Story is that just about any educated person with a little imagination can come up with one to cover just about any hypothesis about human psychology you care to name. The whole approach has become so laughable that there has even been a competition in a popular science magazine to invent the silliest example.
The author also uses a technique of mixing illustrative examples, case-studies, anecdotal evidence and actual scientific results in a manner that requires careful concentration in order to keep them all apart and know what's explanatory and what's evidence. (There is also an entire chapter devoted to a case study of a mathematician with Asperger's Syndrome and speculations that Paul Dirac and Albert Einstein also had AS. This appears to be just padding to make the book longer.) To give an egregious example of this sort of thing, in endeavouring to persuade the reader that the male bias in STEM jobs is due to innate sex differences, the author trots out three professors he knows who aren't biased in their selection process. This is supposed to counter all the research on the subject that says there is selection bias and that it occurs in schools and my personal anecdotal evidence of appalling sexism in academia. Another example: male suppression of emotion (particularly crying) is innate essentially because he says so, despite the abundant evidence that there is cultural variation regarding this, both now and historically.
The author has a hypothesis that autism is caused by having an "extreme male brain." His approach to this idea goes as follows: Take every observed behaviour/task where women statistically out-perform men. Define all these tasks as empathy/empathising. Take all the observed behaviour/tasks where men statistically out-perform women. Define all these tasks as "systemising" (which appears little distinguishable from STEM). Now define a person who is good at "empathising" (in quotes because it's a gigantic extension of the concept you'll find in the dictionary) as having a "female brain." Similarly define a "male brain" as being good at "systemising." Cherry-pick the evidence that these differences are biologically innate to the sexes. Next say, people with autism behave in a manner consistent with having an extreme case of the "male brain." Here comes the really horrendous and scary part of the book: The theory doesn't work. It doesn't explain all the symptoms generally associated with AS or classical autism. In order to get it to work the author says we should re-define one of the major areas of autistic behavioural difference so that the hypothesis fits and totally ignores the area of sensory differences between autistic and neurotypical people. This is a massive dis-service to people with autism. Instead of making an honest attempt to understand the condition, the author is attempting to bend the very defnition of it to his will so that it fits his hypothesis: I'm being used to further his career. This is utterly disgusting. And we're taking about the head of the Cambridge Autism Research Centre here; he must be influential.
I would like to think that the above described crimes are confined to this book. They aren't. Modern popular science publishing is rife with books that are trying to sell you the author's pet theory and are not doing so honestly. If you read this type of thing I strongly advise you to use the utmost scepticism when you are reading and to watch out for the techniques described here. It's as if the authors of such books don't believe scientific ethics (which boil down to scrupulous honesty about the strengths and weaknesses of ones work and what one did) don't apply in the popular science arena. Not only is this false, but when one is talking to potentially scientifically naive people, the onus is on the scientist to attempt to help them understand how to make their own fair assessment of the evidence.
Non-scientific criticisms of the book include: it's padded. It's patronising in tone, especially early on.
I could append here a lengthy discussion of the question of whether there are innate psychological differences between the sexes but this review is probably alarmingly long already. Suffice to say that there are some cross-cultural, repeatable statistical differences between males and females in performance on certain tests. These results, scientifically, require an explanation. An honest one, not driven by egotistical attempts to further one's reputation or by PC ideology.
Again, I could append my views on where autism research should go but I will confine myself to one point: Many people who have an autism diagnosis complain of extreme sensitivity to certain sensory stimuli; these could be specific noises, bright lights, types of touch stimulus or specific smells. These make some people's lives miserable and restricted. Nothing in the "extreme male brain" hypothesis explains why this should be the case. There are very few therapies to help with these problems and they have limited efficacy. Very little research is being done on it by anybody, but is the leading problem autistic people able to express an opinion want help with.