It's important to note that it's not clear that Asperger's actually exists as a separate set of people, rather than simply as a group of people who's personality fits in some distant corner of the normal distribution.<p>I.e., I'm saying I want to see a continuous stable functional <i>with a margin</i> (1) to distinguish between asperger's and non-aspergers before I take it too seriously as a "condition". As far as I know, this has never been done.<p>(Note that I'm not disputing full fledged autism. I'm just uncertain about aspergers.)<p>(1) For those unfamiliar with stats, this is what I mean. Let x be an N dimensional vector, representing the personality/physical features of people (i.e. x[1] = propensity for antisocial behavior, x[2] = seizure frequency, x[...] = other diagnostic criteria). Then I'd like to see a continuous function f(x) such that if f(x)>1, you have aspergers, f(x)<0 you don't, and <i>the probability that 0<f(x)<1 is very small</i>.