This is the kind of topic that unfortunately often devolves into a flame war because each side simply doesn't understand the other. But there's something important that neither Silberman nor the mass media has caught onto yet: that the collection of discrete diseases referred to as "autism" has a nexus, and it's in the cerebellum.

There are a large number of recent (2013-2015 range) medical journal articles in respected publications such as Cell that confirm this. Autism is largely about the cerebellum: what happens when cerebellar cells aren't present at all (severe disability, as in Dandy-Walker Variant, aka "autism"), and what happens when they're present but not wired up properly (unusual personalities, aka the condition formerly known as "Asperger Syndrome"). It's not really that simple of course but those are both real conditions that we don't bother distinguishing right now.

This is what should be in the newspaper: that autism is really a collection of things that should be called Cerebellar Dysmorphic Disorder. Everything else is clearer when people aren't just arguing about abstract terms with no morphological basis. (Wonder why some people with autism are able to have intense debates while others can't talk at all? If you have an arbitrarily-shaped Dandy-Walker cyst that has displaced 30% your cerebellum, it's unlikely anyone else will have the exact same symptoms, but you might not be able to do a lot of things.)

The upshot is that anyone with severe deficits and an autism diagnosis should get an MRI of their brain to image the cerebellum.

This is a topic I take really seriously. E-mail me at greenspan at post.harvard.edu with any questions. I'm not a doctor but these views have been run by many MDs and MD/PhDs and I'm happy to point people to journal research if helpful.