NIMH » Attention to Geometric Images May Offer Biomarker for Some Toddlers with Autism


Research Highlight

Researchers are looking for measurable biological indicators, known as biomarkers, that could help identify children with autism spectrum disorder (ASD) within the first few years of life—before they would typically receive a diagnosis. Establishing accurate biomarkers could inform both research and clinical practice and help improve and expand early ASD screening efforts. A study supported by the National Institute of Mental Health suggests that visual attention to dynamic geometric images may offer one such biomarker for a subset of toddlers with ASD.

The study, led by Karen Pierce, Ph.D., at the University of California, San Diego, builds on previous work by Pierce’s team showing that some toddlers with ASD have a strong visual preference for geometric images. That is, when these toddlers have the option to look at different types of images, they spend a disproportionate amount of time looking at images containing geometric shapes.

Pierce and colleagues hypothesized that this geometric preference might be robust enough to serve as a biomarker for identifying some young children with ASD. They designed a study to find out whether they could replicate their previous findings in a larger, more diverse group of young children.

Notable for its large sample size, the study included 1,863 toddlers with an average age of about 2 years. The researchers recruited families to participate through community referral and population-based screening in health care settings using an approach called Get SET Early. After completing in-depth diagnostic evaluations, the research team classified the toddlers into distinct groups:

  • Children with ASD
  • Children with some features of ASD
  • Children with global developmental delay
  • Children with language delay
  • Typically developing children

The toddlers watched a 1-minute video that contained geometric images on one side of the screen and images of children doing yoga on the other side of the screen. The researchers used eye-tracking technology to follow and measure what the toddlers looked at as they watched the video—the overall proportion of time each toddler spent looking at the geometric images served as a measure of geometric preference. Details about this approach, called the GeoPref Test, and example video clips are available on the UC San Diego Autism Center of Excellence website.

The data showed that, overall, toddlers with ASD spent relatively more time looking at geometric images, and some children with ASD fixated on the geometric images more than 90% of the time. In contrast, toddlers in the other groups—children with global developmental delay, children with learning delay, typically developing children, and typically developing children who had siblings with ASD—showed a preference for the images with children.

When the researchers examined the data from toddlers with ASD, they found that their visual preferences were associated with their clinical measures. Specifically, toddlers with ASD who strongly preferred geometric images also had higher symptom scores, lower cognitive ability scores, and lower adaptive behavior scores than toddlers with ASD who strongly preferred images showing children.

The researchers found that their simple measure of geometric preference—proportion of time spent looking at geometric images—accurately distinguished toddlers with ASD from toddlers without ASD with few false positives.

Additional data suggested that geometric preference remained stable over the 12 months following the initial eye tracking session. Using modern bioinformatics techniques, the researchers confirmed the statistical robustness of their findings.

According to Pierce and colleagues, the study shows that the geometric preference measure can serve as an objective, accurate, and reliable biomarker of an ASD subtype.

The fact that the geometric preference measure was accurate in a large, diverse, community-based sample that included toddlers with non-ASD developmental delays suggests the measure may have use in real-world clinical settings and in research. If validated in further studies, this measure may offer a low-cost, scalable tool for screening and identification that could help some toddlers with ASD receive earlier diagnosis and access to services.

Additional research examining the mechanisms, including possible genetic links, that drive preferences for geometric images can help researchers better understand variation among children with ASD.


Wen, T. H., Cheng, A., Andreason, C., Zahiri, J., Xiao, Y., Xu, R., Bao, B., Barnes, C. C., Arias, S. J., & Pierce, K. (2022). Large scale validation of an early-age eye-tracking biomarker of an autism spectrum disorder subtype. Scientific Reports, 12, Article 4253.


MH118879, MH080134, MH104446, MH081755



Leave a Reply

Your email address will not be published. Required fields are marked *