Despite almost four decades of clinical research with ever more sophisticated brain imaging techniques, no brain signature has yet been identified that is undisputedly unique to autism. Although this null result was initially attributed to technical and clinical inconsistency, it may actually reflect the heterogeneity of the spectrum.

My team has scanned the brains of more than a dozen different mouse models with mutations in genes associated with autism, using functional magnetic resonance imaging (fMRI). We hypothesized that autism features reflect an alteration in brain activity.

We found some alterations to circuits that are common to mice with different mutations. But somewhat unexpectedly, we also observed that brain activity patterns can diverge widely across autism mouse models, even among those with comparable behaviors. These results suggest that — at the macroscale — autism is unlikely to be associated with a unique set of structural or functional differences.

This doesn’t mean we should give up brain mapping in autism. Rather, our focus should shift from autism-specific brain differences to brain signatures specific to autism subtypes, defined genetically. The latter goal may be attainable and clinically useful: It could help us identify circuits or functional alterations that we can target with treatments.