Tailoring Eye Tracking to Specific Targets of Pivotal Response Treatment (PRT): Preliminary Results from Novel Eye-Tracking Tasks in Children with and without ASD

Background: Eye tracking has been used for almost 15 years in autism research but has only recently begun to advance as a viable method for monitoring clinical treatment(Dawson et al, 2010). In an ongoing study, we aim to track change in response to evidence-based behavioral treatment, PRT, through eye tracking. Here we report preliminary results from novel paradigms developed to measure specific PRT intervention targets, including a Dyadic Bid task(DB) examining sensitivity to overtures for engagement, a Conversation Following task(CF) examining conversational engagement, a Social Referencing task(SR) examining monitoring of nonverbal information, and a Theory of Mind task(ToM) examining perspective-taking.

Objectives: To explore baseline between-group differences in children with and without ASD in performance on four novel eye-tracking paradigms tailored specifically to assess change following PRT.

Methods: Participants included thirty-six 4 to 8 year-olds, ASD n=14 (11 males, M_DQ=88.4 sd=19.8) and non-ASD n=22(13 males; M_DQ=108.9 sd=12.4). A joint DB and CF task included clips of actors engaging in naturalistic conversation. During each clip, an actor turned to the camera and performed a bid for the participant’s attention (question or statement) then paused as though waiting for a response. DB main outcome measure was proportion of time looking at the bid actor’s face(BidActor%), and for CF, the proportion of time looking at the speaking actor(Speaker%). SR included clips of an actor performing an increasingly stressful task(e.g. inflating a balloon until it pops). SR outcome measure was proportion of time looking at the actor’s face(Face%) and activity(Activity%) during the event escalation and resolution. ToM included a hide-and-seek false belief task. ToM outcome measure was proportion of time looking in the hiding place associated with the protagonist’s false belief(FalseBelief%) and where the target object was actually hidden(TrueLocation%). Univariate ANCOVAs controlling for DAS-II GCA Standard Score were conducted to examine main effects of diagnosis(dx).

Results: DB: There was a significant main effect of dx on BidActor% during the pause following bids(F(1,36)=6.3, p<.05, η_p²=.16), and during bids that were statements(F(1,36)=4.9, p<.05, η_p²=.129), with the ASD group looking less. CF: There was a significant main effect on Speaker%(F(1,36)=4.96, p<.05, η_p²=.13) with the ASD group looking less at the speakers. SR: There was a significant main effect on Face% during the event’s resolution(F(1,36)=4.3, p<.05, η_p²=.115), and a trend toward a main effect on Activity% during the escalation(F(1,36)=3.6, p=.066, η_p²=.098) with the ASD group looking less at face and activity. TOM: There was a trend toward a main effect on TrueLocation%(F(1,36)=3.3, p=.078, η_p²=.094) with ASD looking more at the true location. There was no main effect on FalseBelief%.

Conclusions: Preliminary between-group differences provide initial validity for these novel eye-tracking paradigms. Children with ASD appear to be less sensitive to overtures for social engagement, particularly during more socially-nuanced bids. They also exhibit decreased monitoring of speakers’ faces during conversation, and less referencing of another’s face following the culmination of a stressful event. Results from ToM may reflect a deficit in perspective-taking within the ASD group. Further study is needed to assess the ability of these paradigms to measure within-subject change in response to PRT.