Because of autism’s heterogeneity, there is unlikely to be a brain difference that characterizes autism as a whole. In our research, brain structure does not map onto diagnostic labels. However, new types of statistical analysis might bring to light key brain differences among autistic people and those with other neurodevelopmental conditions, which could help identify subgroups of people who share similar brain structure and function.

In a study published earlier this year, we found no significant structural differences in the brain between people with autism, those with attention deficit hyperactivity disorder (ADHD) and controls. We did find differences between individuals, but not at the group level. So we analyzed our data more agnostically using an algorithm that can cluster individuals according to the similarity of their brain features.

We found clusters, but they seem to cross diagnostic labels. So a child with ADHD may be more similar to a child with autism than she is to another child with ADHD. These findings fit with our previous work, but we still don’t understand how these clusters relate to genes and behavior. If we can find an association between these brain groups and genetics or behavior, we could use brain-structure profiles to predict biology and behavior — and to test treatments.

This approach could reveal autism subtypes as well as how developmental conditions overlap. To know whether a difference is specific to autism, it is important to compare autistic people with those who have another neurodevelopmental condition. Looking across conditions should become standard practice for these studies.