How we look at other people’s faces is strongly influenced by our genes, scientists have found in new research that may be especially important for understanding autism because it suggests that people are born with neurological differences that affect how they develop socially.

The study, published on Wednesday in the journal Nature, adds new pieces to the nature-versus-nurture puzzle, suggesting that genetics underlie how children seek out formative social experiences like making eye contact or observing facial expressions. Experts said the study may also provide a road map for scientists searching for genes linked to autism.

“These are very convincing findings, novel findings,” said Charles A. Nelson III, a professor of pediatrics and neuroscience at Harvard Medical Schooland Boston Children’s Hospital, who was not involved in the research. “They seem to suggest that there’s a genetic underpinning that leads to different patterns of brain development, that leads some kids to develop autism.”

Dr. Nelson, an expert in child development and autism who was an independent reviewer of the study for Nature, said that while autism is known to have a genetic basis, how specific genes influence autism’s development remains undetermined.

The study provides detailed data on how children look at faces, including which features they focus on and when they move their eyes from one place to another. The information, Dr. Nelson said, could help scientists “work out the circuitry that controls these eye movements, and then we ought to be able to work out which genes are being expressed in that circuit.”

Continue reading the main story

“That would be a big advance in autism,” he said.

In the study, scientists tracked the eye movements of 338 toddlers while they watched videos of motherly women as well as of children playing in a day care center. The toddlers, 18 months to 24 months old, included 250 children who were developing normally (41 pairs of identical twins, 42 pairs of nonidentical twins and 84 children unrelated to each other). There were also 88 children with autism.

Scientists study identical twins because 100 percent of their genes are the same, so if they share characteristics that are more individualized in other children, those traits are considered at least partly inherited. Nonidentical or fraternal twins share 50 percent of their DNA, so stark differences between identical and nonidentical twins suggest that those traits are strongly influenced by genes.


A study tracked the eye movements of 338 toddlers, including 42 pairs of identical twins. CreditJohn Constantino

In the study, how much one identical twin looked at the eyes of people on screen matched the other identical twin 91 percent of the time. For fraternal twins, the match dropped to 35 percent. For unrelated children, when measured as pairs of the same age and sex, the match was 16 percent. And when the unrelated children were paired at random, their time spent looking at eyes did not match at all, said Warren Jones, the study’s senior author and an assistant professor of pediatrics at Emory University School of Medicine.

How much the children looked at mouths followed a similar pattern. And although each toddler watched the videos without other children present, identical twins often moved their eyes at nearly the same moment — as close as 16.7 milliseconds apart — and in the same direction.

“It’s a really remarkable set of findings in that it really shows that genetic factors are driving differences in the way that toddlers are looking at faces,” said Brad Duchaine, a professor of psychological and brain sciences at Dartmouth, who was not involved in the study. “This suggests that genetic differences drive this important aspect of the way that we interact with others.”

Dr. Jones, whose co-authors include Dr. John Constantino, director of child and adolescent psychiatry at Washington University School of Medicine in St. Louis, and Ami Klin, director of the Marcus Autism Center at Children’s Healthcare of Atlanta, said: “When we started to get the results back, I thought that I had the wrong data because the match between identical twins was so strong. I thought I might have mistakenly matched data from the same twin.”

With the children with autism, the researchers found that, compared with typically developing toddlers, they spent significantly less time looking at faces and more time looking at objects. That difference was especially pronounced with the day care videos, scenes presenting many more things to look at than the close-up videos of women talking to the camera, Dr. Jones said. When watching the day care videos, toddlers with autism looked at faces half as often as typical children did, and at objects almost twice as much.

The difference was so consistent that researchers could identify most children with autism just by looking at the eye-tracking results, Dr. Jones said. That result reinforced previous research in which Dr. Jones, Dr. Klin and colleagues showed that babies from 2 months to 6 months old who looked less at people’s eyes in videos were more likely to be given an autisim diagnosis at age 3 and that eye-tracking could provide an early behavioral indicator of autism.

Experts said that because the study shows that a social behavior that is significantly different in children with autism is strongly influenced by genetics, it might help scientists home in on specific genes to better understand autism or at least a key autism characteristic.


The study being performed. “It’s a really remarkable set of findings in that it really shows that genetic factors are driving differences in the way that toddlers are looking at faces,” said Brad Duchaine, a professor of psychological and brain sciences at Dartmouth. CreditRobert J. Boston

Even when identical twins watched completely different videos, their results matched. “How much Twin 1 looked at the eyes in a video that Twin 2 didn’t get to see predicted how much Twin 2 would look at the eyes in a different video,” Dr. Jones said.

That suggests, he said, that the genetically driven behavior involves “seeking out” social information found in the eyes rather than “merely responding to” facial features, a finding that could help pinpoint “what is disrupted in children with autism as they develop and learn about the world.”

Dr. Nelson said one question for further research was “how specific are these phenomena to autism, or might you see them in other neurological disorders?”

Another question is “how does this actually affect longer-term development of the brain,” said Matthew Peterson, a postdoctoral researcher at the Massachusetts Institute of Technology, whose studies have found that “everybody has a preferred location or position on the face where they will always look when they identify someone” — higher or lower on the face; toward the eyes, nose or mouth.

In the new study, researchers also tested the typically developing children again at age 3 and found that identical twins still strongly matched in how much they looked at eyes and mouths. That suggests that compared with genes, “that experience they’re having in that year — that’s not having much of an influence,” said Dr. Duchaine, whose work has found genetic roots to face recognition.

Still, Dr. Nelson cautioned against overemphasizing the direct role of genes. “Twins have identical DNA, but they don’t have identical experiences and they don’t have the same brains,” he said.


Most likely, through evolution, we came to have “genes that regulate the formation of the neural circuits that underpin how we visually inspect the social world,” he said.

“That,” he said, “helps ensure that we are social beings.”