At the Italian Institute of Technology's Center for Neuroscience and Cognitive Systems, researchers have cracked open one of neuroscience's most frustrating puzzles: why autism looks so different from one person to the next, and what that variation actually means in the brain.
For years, brain imaging studies of autism have produced maddeningly mixed results. Some showed reduced connectivity between brain regions; others found increased connectivity; still others uncovered complex patterns that seemed to contradict everything else. Most researchers treated this variability as noise, a sign of the reproducibility crisis plaguing modern neuroscience. But Alessandro Gozzi and his colleagues at the Italian Institute of Technology, collaborating with Adriana Di Martino and researchers at the Child Mind Institute in New York, asked a different question: what if that variability isn't noise at all, but signal?
Their answer, published in Nature Neuroscience, points to two distinct autism subtypes defined by opposite patterns of brain connectivity—one characterized by hypoconnectivity and the other by hyperconnectivity. This discovery matters because it suggests that the clinical diversity of autism, long viewed as a messy problem to be averaged away, actually reflects genuinely different underlying biological mechanisms.
To test this rigorously, the team designed a cross-species study that began with precision. They examined connectivity patterns in 20 different mouse models of autism, each genetically engineered to exhibit autistic behaviors, studied under controlled experimental conditions. This approach allowed them to link varied connectivity patterns to specific molecular and cellular processes in a way human studies alone cannot. They then checked whether the same patterns appeared in brain imaging data from autistic people themselves.
Using resting-state functional MRI—a technique that captures spontaneous brain activity in awake, unstimulated subjects—the researchers mapped connections between brain regions in both mice and humans. The consistency they found across species was striking: the same connectivity subtypes that emerged in the mouse models appeared in human patients. This wasn't coincidence; it was evidence of distinct biological pathways.
Gozzi explained the shift in thinking that made this breakthrough possible: "Rather than asking whether the autistic brain is simply more connected or less connected, we asked whether there are distinct connectivity subtypes that point to different forms of underlying biology." By reframing variability as a mechanistic clue rather than a limitation, the team transformed one of neuroimaging's persistent frustrations into a discovery.
The implications ripple outward. Autism spectrum disorder has always been understood as clinically heterogeneous—the experiences, aptitudes, and needs of autistic people vary significantly. Now there's evidence that this diversity reflects actual differences in brain organization. Understanding that autism may comprise distinct subtypes with different neurobiological foundations could eventually reshape how researchers approach treatment development, how clinicians think about intervention, and crucially, how autistic people understand their own neurology. The variability that once looked like a problem now looks like the answer.