Assessing the Use of Blink Inhibition As a Measure of an Individual's Level of Engagement with Ongoing Content

Background: One of the guiding goals of autism research is to understand the experience of individuals with Autism Spectrum Disorder (ASD) as they navigate the social world. Our laboratory recently developed a novel method for quantifying a critical aspect of subjective experience: how engaged individuals are with what they’re viewing. This measure relies on analysis of dynamic patterns of eye-blinking. Given that blinking results in a brief loss of visual information, viewers unconsciously modulate the precise timing of their eye-blinks. Importantly, viewers are least likely to blink when looking at something they perceive to be most important or salient. While this measure provides critical insight into what is perceived as engaging by a group of viewers, quantifying what an individual perceives as engaging presents unique challenges. Individual viewers spend far more time not blinking than blinking – as a result, moments of perceived salience (indexed by statistically significant blink inhibition) are more readily quantified when examining patterns of eye-blinking in a group of viewers. Overcoming these challenges is critical, as individual metrics of perceived stimulus salience have the potential to meet several growing needs in autism research, such as parsing heterogeneity in ASD and the development of diagnostic tools and outcome measures.  

Objectives: (1) Assess the feasibility and robustness of individual patterns of eye-blinking as a measure of visual engagement; and (2) quantify deviations in visual social engagement by comparing patterns of eye-blinking in individual children with ASD to TD viewers.  

Methods: Pilot Task: Eye-tracking data were collected from 20 typical adults. Participants viewed videos that alternated between scenes of water animals and scenes of land animals. Half the participants were instructed to count the number of water animals, and the other half counted land animals. The task was designed so that the two categories of content would be differentially engaging to different groups of viewers. Natural Viewing Task: Eye-tracking data were collected from school-age TD children (n=40) and children with ASD (n=49) viewing age-appropriate scenes of social interaction.  

Results: As expected, every participant blinked less during the scenes that required counting of animals compared with scenes they viewed passively. The difference in blink rate during the task-relevant vs. task-irrelevant scenes was significant, assessed by permutation testing (p<0.001). In addition, a linear SVM classifier was trained to assign participants to one of the two experimental groups using the timing of their eye-blinks. The classifier was trained on participants’ mean blink rates during each of the water and land scenes and assigned 95% of participants to the correct group. Immediate next steps include using a similar classification approach to compare the timing of eye-blinks made by TD and ASD children during natural viewing of social scenes.  

Conclusions: The classifier results demonstrate that it is possible to classify participants by experimental group using individual patterns of eye-blinking. This is an important first step towards demonstrating the feasibility of using eye-blinking as an individual measure of visual engagement, a tool that may provide critical insight into the subjective experience of individuals with ASD.