Trouble with eye contact is a hallmark of children with autism and related disorders. But according to a new study in Nature by Warren Jones and Ami Klin, atypical patterns of eye contact show up before other overt signs of trouble in kids that are later diagnosed with autism spectrum disorders. Furthermore, the study’s findings may have important implications for how we think about, study, and diagnose autism. Jones and Klin studied 110 infants; 59 were at high risk for autism spectrum disorders (since at least one of their siblings had already been diagnosed), and 51 were at low risk (with no known autistic family members). To examine and quantify the kids’ eye contact, each child was strapped into their car seat or a baby carrier and placed in front of a monitor. Here, they watched videos of a woman looking right into the camera and acting as a mom or babysitter would, talking aloud, playing patty cake, or doing some other typical activity. The researchers used an eye-tracking camera and software to monitor where on the screen the infants were looking during the videos and how long their gaze lasted.


Each child was first tested at two months of age, then tested at regular intervals until they were two years old. By the time the kids were three, 12 of the high-risk children and one from the low-risk group had been formally diagnosed with autism or a related disorder. Because of the low incidence of autism in girls (both in this cohort and in general) the researchers only included the 11 autistic and the 25 “typically developing” boys in this study. What the researchers found was far from subtle: the kids that ended up being diagnosed with autism spectrum disorders followed a very different developmental pattern in terms of eye contact than the other children did.


The children that didn’t end up being diagnosed (those that developed “typically”) spent most of their time looking at the eyes of the woman on the screen, spending much less gazing at her mouth, body, or other objects in the video. Between two and six months of age, these kids’ eye contact with the actress increased sharply. However, the kids who would later be diagnosed with autism-related disorders (let’s refer to these kids as “pre-autistic,” for ease of reading) demonstrated very different patterns of eye contact as they grew up. On average, their eye contact with the woman actually declined from the first test at two months until the very last time they were tested. By the time they were two years old, these children made about half as much eye contact as the typically developing kids did at the same age.


They were also much more fixated on objects; by the time they were two, the pre-autistic kids were spending twice as much time looking at non-human objects as the other children were. Jones and Klin found that the bulk of these differences occurred between two and six months of age. During this period, the eye contact of typically developing infants increased by 3.6 percent each month, compared to a monthly decline of 4.8 percent for the pre-autistic kids. Finally, the researchers looked at the distributions of the eye contact growth curves to determine the degree of overlap between the two groups. They found that there was far less overlap than would be expected by chance, suggesting that there are significant and major differences in the developmental paths of these children. Despite the small sample size, the results are robust; as they grew, the kids that would later be diagnosed with autism spectrum disorders spent less and less time making eye contact and became more fixated on non-human objects.


Conversely, kids developing in a more typical way were getting better and better at making eye contact during the same time period. The results illustrate what Jones and Klin call the “derailment” of various skills and behaviors that help children develop socially. Whatever is triggering the changes in eye contact (and the authors are careful not to speculate much about what these causes are) is likely changing the social experiences that these kids are having with caregivers, peers, and others. Kids that won’t—or can’t—engage in “normal” eye contact won’t have the same social experiences that other children will. And eye contact likely isn’t the only skill being “derailed,” it’s simply the only one that’s been demonstrated so far.


These derailments are important because there doesn’t appear to be one single cause of autism-related disorders. There may be as many as five hundred different genes that contribute to these disorders, so there are likely many paths to autism. The research by Jones and Klin suggests that while the initial mechanisms may be different, the ways that autistic kids diverge from a typical developmental path might be universal. Early diagnosis is one obvious—and potentially game-changing—application of this research. But there are also other ways that these findings could transform our understanding and approach to autism spectrum disorders. One particularly interesting result, which the researchers didn’t expect, is that eye contact in all the very youngest infants was very similar.

autisim baby

At two months of age, there was no difference between the eye contact of kids who would later be diagnosed with autism and that of kids who wouldn’t. Even though autism is thought to be congenital, it appears that the underpinnings of typical social development are present at birth, no matter the later diagnosis. This is an unexpected and exciting finding, since it suggests the possibility for early intervention. It’s way too early to talk in terms of cure or treatment, since we don’t know much about the mechanisms behind these disorders. But these results do suggest that there is a narrow window during which these inborn neural and behavioral foundations could potentially be preserved. ( By Kate Shaw Yoshida from )