I think one of the problems with the definition of autism is we keep expanding it. It started as "early infantile autism", and then it became "autism", and now it's "autism spectrum disorder". I'm not opposed to that from the standpoint of trying to broaden our vistas, and so forth. But from a research point of view, the term autism is lost in specificity.

I have a fairly strict definition of early infantile autism. That is not to say that people who don't meet that classic description don't have autism, but we might do well to narrow our definitions, and our samples, down to groups that are very similar, because I think you're more likely to find the cause.

With respect to why some acquired savants are musical, and some are good at the arts, and some are mathematical, why aren't they all the same; I think that has to do with this genetic endowment that all of us have.

When we come to research, if we want to find out the cause of autism, we're going to have to be much more specific, and that's why when it comes to research, I'm fairly strict with respect to criteria. When it comes to treatment, I'm much more open to not making that differentiation.

I think that if I could do any sort of research of autism that I wanted to do, at this point I would take a sample of classic, early infantile autism persons and compare them with what I call "classic late onset autism", individuals. I think we will find that the cause of those youngsters with autism who have autism from birth is probably different than those who have late onset autism.

We can't use the word normal anymore because it's sort of come to be politically incorrect, because normal implies a classification, and categorizations, and exclusions, and so forth. So neurotypical is the word that we now have to use for what I call normal behavior. Neurotypical behaviors are those kinds of behaviors within the range of usual human conduct that do not rise to the level of a disorder.

The beginning of wisdom is to call things by their right names in research.