The specific results were obtained on autistic people with normal IQs – “high functioning”, from across the Eastern US. Quoting from the press release:
Using non-invasive fMRIs, the team looked at the brains of 17 people with autism and 17 control subjects as they read and indicated their comprehension of English sentences. In both the healthy brains and in the brains with autism, language functions were carried out by a similar network of brain areas, but in the autism brains the network was less synchronized, and an integrating center in the network, Broca’s area, was much less active. However, another center, Wernicke’s area, which does the processing of individual words, was more active in the autism brains.
or as the Economist summarized:
Three differences emerged. First, Wernicke’s area, the part responsible for understanding individual words, was more active in autists than non-autists. Second, Broca’s area–where the components of language are integrated to produce meaning–was less active. Third, the activity of the two areas was less synchronised.
This led them, based on the previous research on lower “white matter” levels, to the “underconnectivity theory”:
The brain likely adapts to the diminished inter-area communication in autism by developing more independent, free-standing abilities in each brain center. That is, abnormalities in the brain’s white matter communication cables could lead to adaptations in the gray matter computing centers. This sometimes translates into enhanced free-standing abilities or superior ability in a localized skill.
[…]Autism is a system-wide brain disorder that limits the coordination and integration among brain areas. This theory helps explain a paradox of autism: Some people with autism have normal or even superior skills in some areas, while many other types of thinking are disordered.