Autism and Eye Tracking

BY BRYN FARNSWORTH
SCIENCE EDITOR
IMOTIONS

Autism was first recognized as a disorder in 1943, and is now known to affect around 1% of the population. The disorder has since been characterized as representing a spectrum in which a range of difficulties with social interaction, communication and repetitive behavior are found. Subsequent studies have shown the disorder to be neuro-developmental in origin, and diagnosis is typically made by a psychologist / psychiatrist at 4-5 years of age.

Further studies have shown that early socialization interventions can drastically improve the prognosis and development of autism, affirming the importance of identifying this disorder at an early developmental point. Of course, any early identification should be completed as reliably and accurately as possible. Recent scientific studies have shown that using eye tracking technology could be one way in which to identify consistent differences that emerge between people with autism and neuro-typical individuals.

One of the first studies completed that examined autism with eye tracking devices was done by Kevin Pephrey and others (2002) who found that people with autism were more likely to view non-feature areas of the face, than feature areas (such as the eyes, nose, mouth etc). This work was complemented by Ami Klin and colleagues (2002), who found similar results with film clips of social scenes, a setting that is more representative of real-world scenarios.

Further and more recent research by Yi and others (2013) has found that the abnormality in face-scanning can be solely attributed to examinations of the eyes, if IQ differences across participants are accounted for. Following on from these (and other) increasingly specific successes in determining differences across autistic and neuro-typical individuals, we are now at a critical juncture in which experiments have begun identifying bio-markers – reliable early indicators of this disorder that can be quantified and acted upon.

An article released just last year has shown the greatest promise of this breakthrough yet. Karen Pierce and colleagues (2016) examined 444 toddlers from various groupings (such as with autism spectrum disorder, typically developing children, developmentally delayed infants, and so on), finding that autism could be predicted among these groups with 86% accuracy. This was determined by examining the gaze responses to film clips of social scenes, and of geometric patterns. The toddlers were around 2 years of age on average, tempting the possibility that a definitive test could be administered at an even earlier age.

There are obvious advantages to reliably identifying autism at an early stage, in order to alleviate the deleterious symptoms that can later emerge. Various private sector and academic institutions are currently racing to determining the best way in which to make an early diagnosis. Other potential sources of bio-marker identification are also being explored, such as with EEG, and other brain imaging techniques, suggesting the possibility of using multiple sources for increased accuracy. However the future of autism diagnosis turns out, it is sure to use eye tracking to be faster, more accurate, and more refined than ever before – keep your eyes peeled.