ego.istic.org

While I was an undergraduate, I participated in some data collection by the Department of Experimental Psychology's Autism Research Centre. They wanted to know, amongst other things, whether maths ability is genetic.

One of the papers to come out of these data is Wheelwright et al., Predicting Autism Spectrum Quotient (AQ) from the Systemizing Quotient-Revised (SQ-R) and Empathy Quotient (EQ). Baron-Cohen's larger programme is to rehabilitate the once-trendy-then-abandoned EMB theory: that systemizing is a more male than female tendency, empathizing a more female than male tendency, and that autism spectrum conditions, from what they call “classic” autism to people, well, just being a bit mathmo, is caused by having an “extreme male brain”. Asperger himself believed something similar, but without any real data on which to base a hypothesis it fell by the wayside as autism awareness rocketed. Now Baron-Cohen is the front-man for a revised, more precise hypothesis, and he and his colleagues and co-authors are trying to refine a theory using actual data.

Some aspects of this programme are lightly discussed in this Language Log article. But don't be fooled: that page also links to EQSQ and advises you to compare your score to the chart given just below, but the EQSQ scores are for the revised SQ and the chart is for the original, from an earlier paper; the results are not comparable.

The SQ and its successor the SQ-R measure “the drive to analyze, understand, predict, control and construct rule-based systems.” (I hate to mention it, but my day job atm requires me to construct a rule-based system.) The study grouped its control subjects (i.e. ones not diagnosed with an autism spectrum condition) by subject, as physical science, biological science, social science, and humanities. You may not be surprised to learn that the physical scientists scored highest on SQ-R and AQ, and lowest on EQ. The scatter plot of all participants, both controls and ASC diagnosees, is reproduced with annotations below.

Of course, the differences in mean between each subject group are small compared to the standard deviation within each group: I wouldn't like to try to predict subject by SQ-R score. And the problem with comparing control subjects with ASC diagnosees, one recognised by the authors, is that all mathmos have Asperger's anyway.