Developing an Eye-Tracking Biomarker to Measure Social Motivation in Minimally Verbal Children with ASD

Background: Approximately 30% of persons with autism have minimal verbal ability (MVA). This subgroup of cases often experiences significantly greater morbidity and poorer outcomes relative to their ASD peers with average verbal ability (AVA). This subgroup is frequently excluded from research studies, as most ASD-related measures are not applicable to persons with cognitive / language deficits. Eye-tracking technology can be used to quantify ASD-relevant features such as social attention and motivation, can be configured to require minimal task demands of the participant. For these reasons, this method is a reasonable platform for the development of measures applicable to cases with cognitive / language deficits.

Objectives: To compare subgroups of minimal verbal ability (MVA) and average verbal ability (AVA) children with ASD on gaze measures from a passive social versus nonsocial image paired preference task.

Methods: Our preferential viewing (eye-tracking) task measures patterns of visual attention to social and nonsocial content. Social images (SOC) are paired with one of two types of objects: images of high interest to children with ASD (HAI images; e.g. trains, electronics) and images of low interest to children with ASD (LAI images; e.g. clothing, furniture). Stimuli were presented in a passive manner (no instructions) for 5 seconds each, with a variable 2-6 second interstimulus interval. Participants with ASD were classified as either MVA (Verbal IQ < 70; N = 15; mean age = 11.9 years) or AVA (Verbal IQ > 90; N = 48; mean age = 13.9 years); our sample also included typically developing peers (TYP; N = 31, mean age = 13.8 years).

Results: Task completion rate was similar across ASD groups (67% of MVA, 69% of AVA). Both MVA and AVA subgroups were slower than the TYP group to orient to faces (latency), only when paired with HAI images (F(2,71) = 6.59, p = .002) but not LAI images (F(2, 71)= .457, p = .635). Latency to face did not differ between ASD subgroups for either array type (t = -1.6, p = .22). MVA participants made more fixations to faces than both TYP and AVA participants (F(2,59) = 11.2, p < .001); however, these fixations were shorter in their average duration (F(2,59) = 17.18, p < .001).

Conclusions: Here we demonstrate that minimally verbal children with ASD can successfully complete a passive eye-tracking task at rates that are comparable to those of verbal ASD peers. Both MVA and AVA children were significantly slower to look to faces when paired with an HAI image, relative to TYP, providing evidence that social motivation deficits may manifest similarly in both ASD subgroups. Domain general attentional parameters (e.g., fixation duration) were similar in AVA and TYP but altered in MVA suggesting that MVA children may have a unique attentional impairment relative to their verbal peers with ASD.