Examining Ecological Validity in the Use of Eye-Tracking for Toddlers with Autism Spectrum Disorders

Thursday, May 12, 2016: 11:30 AM-1:30 PM
Hall A (Baltimore Convention Center)
A. Navab1, T. Vernon1, J. Bradshaw2, E. J. Horowitz1, A. Barrett3, J. Ko3, S. Offman1, J. Gong1, A. C. Voos1, T. German1 and R. L. Koegel1, (1)University of California, Santa Barbara, Santa Barbara, CA, (2)Marcus Autism Center, Emory University, Atlanta, GA, (3)Koegel Autism Center, University of California Santa Barbara, Santa Barbara, CA

A growing body of eye-tracking literature demonstrates that when compared to TD controls, toddlers with ASD exhibit decreased attention towards socially salient stimuli and increased fixation on non-social stimuli (Pierce et al., 2015; Shic et al., 2011). The degree of deviation away from a typical gaze pattern appears to be a reliable measure of autism severity and holds promise as a potential diagnostic tool (Jones et al, 2008; Rice et al, 2012). However, there is currently little known regarding how eye-tracking task performance relates to preferential gaze during naturalistic interactions with caregivers. Such lines of inquiry may elucidate the ecological validity of eye-tracking data as a social engagement predictor in the real world—especially within the context of crucial interactions with caregivers that may serve as fundamental agents of change during early intervention efforts.


The current study presents data from an intervention study using eye-tracking as an indicator of social motivation prior to treatment onset among toddlers with ASD. Baseline levels of preferential looking were compared to TD toddler performance. Eye-tracking preferential looking was also compared to behavioral performance on measures of autism symptomology (ADOS) and gaze during parent-child interactions in a structured laboratory observation (SLO).


Participants included toddlers with ASD (18-54 months) and TD controls. Children viewed a social versus non-social preference paradigm modeled after the social preference task adapted from Pierce et al. (2011). Six 5s videos depicted clips of social interactions presented side-by-side with clips of non-social geometric patterns. Differential looking scores (DLS) for preference across videos were calculated and compared within and between groups. Analyses included associations between DLS and ADOS social affective scores and gaze toward caregiver during the SLO.


An independent samples t-test revealed that ASD toddlers preferentially gazed at non-social stimuli significantly more than TD controls (T=2.6, p<0.05). This lower mean social preference ratio found in the ASD group was moderately associated with ADOS severity of social impairment. No relationships were found between non-social preferential eye-tracking looking behavior and gaze to caregiver during the SLO.


The data suggest that compared to TD controls, toddlers with ASD displayed a preference for non-social eye-tracking stimuli that predicted severity of social affect symptoms. Findings also revealed that participants’ preference for non-social eye-tracking stimuli did not predict decreased gazing to caregivers during the SLO. Different eye-tracking paradigms may be better suited for unique functions.  The task used in this study appears to be a strong indicator of symptom severity, but alternative paradigms may be better suited as a metric of real world social performance.  In the future, naturalistic social paradigms may need to be explored as a method for predicting ecologically valid social behaviors in the real world (e.g. looking behaviors, joint attention). Social looking behaviors appear to be influenced by motivational interactions with a caregiver and may be a responsive target to early intervention efforts. Additional post-intervention data may enhance these findings and reveal how social motivation may be a prognostic indicator for treatment outcome for toddlers with ASD.