Spontaneous Goal Attribution in Children with ASD: A Comparative Eye-Tracking Study

Background:  Typically developing infants are able to spontaneously encode the actions of human agents as goal-driven (Canon & Woodward, 2011). In contrast, individuals diagnosed with Autism Spectrum Disorder (ASD) have demonstrated limited ability to interpret the actions of human agents in terms of goals and intentions (Senju et al., 2009), which may be tied to a well-documented decrease in attention to social stimuli. Early interventions seek to develop these abilities in children with ASD, but to date the majority of outcome studies have used performance-based measures to assess, leaving open the possibility of examining less directly observable mechanisms underlying the disorder.

Objectives:  The present study examines the degree to which toddlers with ASD encode the motion of a human hand as goal-directed prior to the onset of an early intervention program. Performance on an established eye-tracking paradigm is compared to that of an age-matched group of typically developing toddlers, with the aim of demonstrating a significant deficit in goal attribution.

Methods:  An eye-tracking video paradigm modeled after Woodward (1998) was used to measure goal encoding. Participants were 24-48 month-old toddlers who had previously received a diagnosis of ASD and age-matched typically developing (TD) toddlers. Each participant first watched six familiarization videos in which a hand repeatedly reached for one of two toys (a penguin and a ball). A single test trial followed in which the location of the two toys was reversed and the hand was shown resting in between as if just about to reach for one or the other. Anticipatory looks and total looking time toward each of the two toys were recorded using an SMI infrared eye tracker.

Results:  A differential looking score (DLS) was calculated to provide a comparative measure of looking times across the two toys (possible scores ranged from -1 to 1, with a score closer to zero indicating lower preference for either toy). Analysis revealed that while the typically developing group showed a strong preference for the goal-directed toy (M=0.60, SD=0.68), looking times for the children with ASD were distributed roughly equally across the two (M=-0.16, SD=0.62). This same pattern followed for anticipatory (first) looks towards either toy; on average the ASD group had fewer first fixations to the goal-directed toy (M=0.56, SD=0.53), compared to the TD group (M=1.00, SD=0.00).

Conclusions:  Taken together, these results suggest that prior to intervention children with ASD have difficulty interpreting the reaching and grasping motion of a human hand as goal-directed, compared to typically developing counterparts. Moreover, our finding that the ASD group appeared to show no preference for either toy may be driven by a more general reduction in social motivation, which is believed to be one of the core deficits of the disorder. In examining the underlying mechanisms of such social abilities, both in the present study and with the collection of post-intervention data, we ultimately hope to contribute to the existing characterization of ASD and provide direction for the further development of treatment programs.