Toddlers at risk for autism spectrum disorders preferred to look at geometric patterns more than social images in a 1-minute experiment, in contrast to developmentally delayed and developmentally normal toddlers, according to a report published online Sept. 6 in the Archives of General Psychiatry.

"When the percentage of time a toddler spent fixating on geometric patterns was 69% or greater, the positive predictive validity for accurately classifying that toddler as having an autism spectrum disorder was 100%," said Karen Pierce, Ph.D., department of neurosciences, Autism Center of Excellence, University of California, San Diego, and her associates.

Other investigators have used eye-tracking technology to assess differences between autistic and other infants and young children in response to pictures of faces. However, "given the active pace of brain development during the infancy period combined with high intersubject variability of eye tracking patterns to faces during this time, examining the percentage of time an infant attends to the eye region of a face may not be stable enough to make diagnostically predictive claims," Dr. Pierce and her colleagues noted.

"An alternative method to investigate early indicators of autism is to measure a very simple behavior: preference."

Developmentally normal infants and toddlers are known to show a distinct preference for faces when presented with two different images to look at. "What infants prefer to look at when given a choice between two images may turn out to be a more clearly observable indicator of risk than how they look at a single image," the researchers said.

To test this hypothesis, they used eye-tracking technology to monitor subjects’ gaze when watching split-screen moving images of children dancing or performing yoga on one side (dynamic social images) and moving geometric shapes on the other (dynamic geometric images). This movie contained 28 separate scenes, with each scene varying in duration from 2 to 4 seconds, for a total presentation time of 1 minute.

The study subjects were 110 toddlers (aged 14-42 months). A total of 37 children had autism spectrum disorders (27 with autistic disorder, 9 with pervasive developmental delay not otherwise specified, and 1 with autism spectrum features). Another 51 children, matched for age and gender, were developmentally normal. And 22 children who had developmental delay (12 with language delay and 10 with global developmental delay) were matched with the autism group based on chronological age, verbal and nonverbal developmental quotient as assessed on the Mullen Scales of Early Learning, and overall functioning.

Overall, the percentage of time that the toddlers spent viewing the geometric images was significantly different among the diagnostic groups. Toddlers with autism spectrum disorders fixated significantly longer on geometric images than did developmentally typical or developmentally delayed toddlers.

Forty percent of the autism group spent more than half their viewing time fixated on the geometric images, compared with only 2% of the developmentally typical toddlers and 9% of the developmentally delayed toddlers. Many of the children with autism spectrum disorders spent more than 70% of their viewing time watching the geometric images, and several spent more than 90% of their viewing time doing so—a percentage that was never found in either of the other groups.

When a cutoff was established at 69% of viewing time spent on geometric rather than social images, autism spectrum disorders were correctly predicted in 100% of the affected children and in none of the other groups of children.

The children with autism spectrum disorders who showed the most distinct preference for geometric images also showed a unique pattern of saccades, or abrupt eye movements, while viewing the movie. They showed significantly fewer saccades than did any of the other children when looking at their preferred geometric stimuli, and conversely they showed nearly twice as many saccades when looking at the social stimuli.

Other researchers have postulated that increased saccades while viewing faces reflects anxiety among people with autistic spectrum disorders, Dr. Pierce and her associates said (Arch. Gen. Psychiatry 2010 Sept. 6 [doi:10.1001/archgenpsychiatry2010.113]).

The findings show that a preference for watching geometric images plus aberrations in the number of saccades might indicate risk for an autism spectrum disorder in children as young as 14 months of age. In addition, "we believe that it may be easy to capture this preference using relatively inexpensive techniques in mainstream clinical settings such as a pediatrician’s office," they noted.

Infants and children found to follow these patterns of eye movement would be "excellent candidates for further developmental evaluation and possible early treatment."

However, it is important to note that 60% of the subjects with autism spectrum disorders did not exhibit these patterns, and that those who did were not necessarily the ones with the most pronounced symptoms.

Moreover, "while the discovery of a putative new early warning sign of autism is encouraging, results should be interpreted with some caution" because approximately 20% of the initial study population was excluded because of poor compliance with testing, the researchers added.

This study was funded by grants from the National Institute of Mental Health.

Eye-Tracking Suggests Autism in Toddlers