The Potential of Eye-Tracking As an Outcome Measure for Autism Intervention Studies

Background: Outcome measurement is a significant issue in autism intervention research. Among other concerns, it is possible that autism intervention trials may fail to detect change because of inadequate outcome measurement – either because changes occur in domains which are not measured or because the measures used are insufficiently sensitive. Short intervention periods may induce change in domains proximal to the intervention which would take longer to become apparent at the level of global symptom change. There is a need for objective measures, that can be administered by blinded researchers, to detect such proximal intervention effects.

Eye-tracking is a method which can be applied to non-verbal populations, to chart the development of cognition in early life, and to detect differences between typical and atypical populations in childhood, adolescence and adulthood.  In both of these areas, eye-tracking is demonstrably pragmatic, objective and sensitive and can be administered blind. However eye-tracking is not normally used to evaluate response to intervention.

Objectives: To investigate whether eye-tracking can be used to detect subtle changes in eye-movements in response to repeated exposure to learning content.

Methods: Participants were 40 typically developing children, aged 20-25 months (mean = 23 months). We used a Tobii eye-tracker to assess free-viewing of: cartoon images from an iPad app called FindMe; cartoon images from other apps (close-generalisation); photographs (distant-generalisation).  All images were matched for level of complexity. Half the participants took home an iPad with the app for a period of 10 – 14 days. The FindMe app rewards participants for finding and touching the person within a scene. Then all participants returned to view the same stimuli again.

Results:   Preliminary data (n=15) indicate that repeated exposure to visual content in an iPad app does change eye-movement patterns. For children in the app-exposure group, mean looking time on the person (FindMe images) increased by 80ms on average, while the control group had a looking time change of -3ms. In the final analysis we will report data on the relationship between level of exposure (e.g. amount of game play, highest level reached) and fixation patterns, as well as evidence of effects for close- and distant-generalisation stimuli.  

Conclusions:   This proof-of-concept study indicates that repeated exposure to learning content can, even in a very short period of time, lead to changes in fundamental behaviours which are indicative of underlying cognition. We will interpret results in the light of the potential future application to meaningful outcome measure in autism intervention, especially where interventions are mediated by technology.