When a neurotypical child looks at a face, the brain's electrical response to that face becomes more specialized and distinct as the child gets older. A Yale-led study reports that this sharpening does not appear the same way in autistic children.
The research, published in Nature Mental Health on June 29, analyzed high-density EEG recordings from 399 autistic and neurotypical children in a case-control observational study. The finding is a difference in developmental trajectory, not a defect, and the study does not explain why the difference occurs or what it predicts for any individual child.
"The neural activity that we measure at the scalp is just not as distinct in autistic children," said first author Jason Griffin, now an assistant professor of psychology at the University of Houston and formerly a postdoctoral researcher in the lab that led the work.
The Method That Made the Finding Possible
Most EEG research on face processing has focused on a small number of electrodes over specific brain regions where face-related responses appear. The best-known of those measures is the N170, a shift in brain activity roughly 170 milliseconds after seeing a face, which is delayed in autistic individuals.
James McPartland, Harris Professor in the Yale Child Study Center and director of the Center for Brain and Mind Health, offered a comparison in the university's announcement. Studying traffic through a single downtown camera tells you about one road. The new approach was closer to viewing the whole city from a satellite.
The team drew on data from the Autism Biomarkers Consortium for Clinical Trials, a multi-site research effort. Children viewed images of faces and objects while researchers recorded electrical signals from all 128 scalp electrodes. Machine learning was then used to test whether the pattern of brain activity could predict what a child was looking at.
It could, more reliably for neurotypical children than for autistic children. According to the published abstract, during specific stages of face processing, the neural representations for faces, inverted faces, and houses were significantly more distinct in neurotypical children.
The Developmental Pattern at the Center of the Result
The age-related component is what distinguishes this study from earlier work showing group differences at a single point in time.
The paper compared three developmental age bands: 6 and 7 year olds, 8 and 9 year olds, and 10 and 11 year olds. Neurotypical children showed increasing neural specialization across those bands, with face- and identity-selective representations becoming more distinct with age. That pattern was not observed in the autistic participants.
"Their face-specific processing is not following the same trajectory," Griffin said in comments carried by MedicalXpress.
In ordinary neurodevelopment, brain responses generally become more specialized and efficient with experience, a process sometimes described as neural tuning. This study reports that the tuning trajectory for face-specific signals looks different in autism.
One framing correction is worth making. Some coverage has described the autistic pattern as showing "earlier and less" refinement. The published paper describes an absence of the age-related increase seen in neurotypical children, not earlier maturation, and readers encountering the shorter phrasing should treat the published description as authoritative.
MedicalDaily Evidence Check
This is a large observational case-control study, which is a genuine strength in a field where sample sizes of 20 to 40 have been common. Nearly 400 participants and full-scalp recording give it more statistical power than most prior work on this question.
It remains an association. The study does not identify a cause of autism, explain why face-related signals develop differently, or establish that these signals are responsible for any behavioral trait. The relationship between a scalp-recorded electrical pattern and how a person actually experiences faces in daily life is not a straight line.
The design also compares children of different ages at a single point in time rather than following the same children as they grow. That kind of cross-sectional design can describe how patterns differ across age groups, but it cannot track an individual trajectory, and cohort differences can influence the picture.
Autism is highly variable, and group averages describe a group. Nothing in this research allows a conclusion about any individual autistic child, and no diagnostic or clinical test comes out of it. Autism is currently diagnosed through behavioral observation, and that has not changed.
The work was supported by the Hilibrand Foundation and by National Institute of Mental Health grants. Several authors disclosed consulting relationships with pharmaceutical and health technology companies, which is standard in this field and is documented in the paper's competing interests statement.
The Careful Case for Biomarker Research
The researchers describe the findings as a first step toward biomarkers that could eventually assist with diagnosis or with identifying candidates for supportive services. That aim deserves both context and caution.
The practical argument is about timing and access. Behavioral diagnosis often takes years, waitlists for evaluation are long in much of the country, and children in under-resourced areas are diagnosed later on average. An objective measure that could shorten that process would matter to families. Coverage of the study has noted McPartland's point that understanding developmental differences could also inform when supports are most useful.
The caution is that a biomarker measuring a group-level difference is not the same as a test that works for an individual, and the gap between the two has swallowed many promising findings in psychiatry and neurodevelopment. Any measure proposed for clinical use would need validation in independent samples, established accuracy at the individual level, and demonstration that using it improves outcomes rather than simply producing a number.
There is a framing point as well, and autistic self-advocates have raised it consistently. A difference in how a brain processes faces is a description of variation, not a deficit requiring correction. Practices such as forcing eye contact are not supported by this or any similar research, and the study offers no basis for them.
For parents, the useful takeaway is narrow. If you have concerns about your child's development, the route remains a pediatrician referral for developmental evaluation, and early access to supports a family wants is more valuable than any laboratory measure currently available. This study changes what researchers will investigate next. It does not change what happens at your next appointment.
Frequently Asked Questions
What did the study find? That brain signals associated with face processing were less distinct in autistic children, and that the age-related sharpening of those signals seen in neurotypical children was not observed in the autistic group.
How large was the study? It analyzed high-density EEG data from 399 autistic and neurotypical children, drawing on the Autism Biomarkers Consortium for Clinical Trials.
Which ages were compared? Three bands: 6 and 7 year olds, 8 and 9 year olds, and 10 and 11 year olds, compared at a single point in time rather than followed over years.
Does this explain what causes autism? No. The study describes a difference in developmental patterns. It does not identify a cause or explain why the difference occurs.
Does it mean something is wrong with autistic children's brains? The researchers describe a different developmental trajectory, not an abnormality. The findings offer no support for practices such as forcing eye contact.
Can this be used to diagnose autism? Not currently. Autism is diagnosed through behavioral observation. A group-level research finding is not the same as a validated individual diagnostic test.
What should parents do if they have concerns? Ask a pediatrician for a developmental evaluation referral. Early access to supports a family wants remains more useful than any research measure available today.
