As a neuroscientist who has spent the past decade collecting MRI scans in hundreds of sleeping toddlers and children with autism, I’m not ready to throw in the towel. I believe that there are brain differences unique to autism, but it is likely that there is more than one neural signature for autism. There are several important factors to consider.

First and foremost, autism is undeniably heterogeneous and can co-occur with many different conditions. This heterogeneity likely becomes ‘noise’ when looking at group-level differences between the brains of people with autism and those of controls or people with another condition.

Our approach is to attempt to identify more homogeneous subgroups of individuals with autism based on clinical, biological and brain characteristics. Our goal is to find clinically meaningful neural phenotypes that can aid in predicting outcomes and inform treatment.

Second, we should consider that most imaging studies focus on a narrow part of the autism population — usually older individuals without intellectual disability who can tolerate being inside a brain scanner. We haven’t fully evaluated young children at the time of diagnosis, which is more likely to elucidate brain differences related specifically to autism rather than those associated with behavioral interventions or medications. Nor have we evaluated individuals across the entire spectrum. Only a minority of MRI studies include individuals with severe traits.

And finally, we need to evaluate differences in trajectories of brain development across the lifespan. Most imaging studies focus on single snapshots of the brain. But brain development is dynamic, and longitudinal studies are critical to understanding brain differences in autism and other neurodevelopmental conditions.