Group averages can hide important differences

Brain activity in autistic adults may be far more varied from person to person than conventional group comparisons suggest, according to a study reported in JNeurosci. The work compared adults with autism spectrum disorder (ASD) with adults who do not have autism while they watched films, and it found a notable pattern: average responses across the two groups looked similar, but individual responses among autistic participants differed more widely from one another.

The finding does not identify a single brain signature for autism. Instead, it points to the limits of interpreting a heterogeneous condition chiefly through group averages. The researchers say the result supports closer attention to individual patterns and to the possibility that more meaningful subgroups may exist within the broad ASD diagnosis.

Using films to model complex experiences

The research collaboration, led by Dorit Kliemann and James Traer of the University of Iowa, analyzed imaging data previously collected at Indiana University and Caltech. Participants with and without ASD watched movies from several genres, including animation, action, romantic comedy and Hitchcockian thriller.

Films supplied a changing mix of sounds, visual information and social situations. The team used the resulting imaging data to train a computational model that searched for relationships between brain activity and features of the scenes, including auditory, visual and abstract social features. Once trained, the model accurately predicted activity across different brain systems.

That approach is useful because social behavior rarely unfolds as a single, isolated stimulus. A movie can combine dialogue, faces, motion, music, narrative context and changing social cues. Kliemann said the model made it possible to examine robust brain responses to many aspects of these complex and dynamic settings.

Similarity at the group level, diversity at the individual level

When the researchers compared group averages, the model predicted activity across different brain systems in similar ways for adults with and without ASD. On its face, that result could be read as evidence of broad similarity between the groups in how their brains responded during the viewing sessions.

But a different picture appeared when the analysis compared people within each group. Autistic individuals deviated more from their group average than non-autistic individuals did from theirs. The greater variability was seen in nearly every brain system assessed and was especially pronounced in the social brain system.

The distinction matters. A group average can provide a useful summary, but it can also smooth over substantial differences among the people whose data produce that average. In this case, similar averages did not mean that every participant responded similarly. The study suggests that variation among autistic individuals is itself a central feature that researchers need to account for.

What the result means for autism research

The study does not establish a new treatment or explain why individual autistic brain responses differ. It also does not say that any one response pattern is better or worse than another. Its principal contribution is methodological and conceptual: research designed around average differences between diagnostic groups may miss relevant individual information.

Current approaches to understanding and treating ASD often depend on such group comparisons. Traer said the similarity of the group averages was surprising and raised a different set of questions: how do individual brain responses differ, and why? Answering those questions could require study designs that preserve individual-level variation rather than treating it as noise around a typical group response.

Kliemann said identifying ASD subgroups may help create a more precise understanding of the condition and could inform efforts to develop individually tailored support. That is an ambition rather than a conclusion of the current analysis, but it follows directly from the study’s central observation that the autism group was not uniform.

A case for more precise comparisons

The work also highlights why interpretation of neuroimaging results requires care. An average finding can be statistically sound while remaining incomplete as a description of a diverse population. The new analysis shows how a model trained on rich, naturalistic experiences can expose differences that a straightforward comparison of group means does not reveal.

For autism research, the immediate implication is not to discard group-level studies. It is to pair them with tools capable of examining people as individuals. Looking more closely at which responses cluster together, which differ and how those patterns relate to lived experiences could produce a more detailed map of neurodiversity.

That shift may be especially relevant for research on social difficulties, where context is complex and people can respond to the same scene in different ways. The study’s movie-based design and computational model provide one route for investigating that complexity. Its result is a reminder that a shared diagnosis does not necessarily imply a shared neural profile.

This article is based on reporting by Medical Xpress. Read the original article.

Originally published on medicalxpress.com