On the coprocessor model itself -- the phenomenon we're trying to explain here is people who are good at analytical thinking but bad at social interaction. I really think there's a tendency around here to automatically identify that pattern with Asperger's, or with being at some point on the "spectrum." It's way more parsimonious to think of it as a result of specialization. If you're kind of good at (or interested in) analytical things, and kind of bad at (or uninterested in) social things, you'll specialize your own brain in that direction. It may even be in your best interest to specialize to some extent, to play to your strengths.

In fact I'd hypothesize that a lot of this specialization happens in early childhood, from rather small chance events like being an early reader, or being nearsighted. It's a little easier and more pleasant to sit still and read than to go out and play, so you start specializing from the age of three or so. Or you're Robert Louis Stevenson and develop a rich inner life directly because of childhood illness.

Children naturally play to their own strengths. They wear their favorite pathways down smooth, and ignore the thornier, more unpleasant ones. If you're a "systematizing" child, you'll do more and more systematizing, and it probably won't be until adolescence or adulthood that you'll want to go back and learn the skills that always felt unnatural. It takes a certain degree of self-awareness to go against the grain of your own nature. (Optimizing globally instead of locally, I think, is the appropriate metaphor.)

The thing is, "hardware" metaphors treat neurotype as static, and I think for many people it's more likely to be dynamic, and the result of a specialization process. It goes the other way, too: people who have much stronger social than analytical skills have probably put more effort into developing social skills, and then created a positive feedback loop of playing to their strengths. Sure, there are fixed brain differences on the extremes (autism or Williams syndrome) and a specialization process may well begin with a small fixed brain difference, but I don't think the fixed stuff explains the great human variation in social vs. analytical skills.