Abnormal Joint Attention Network in Children and Adolescents with Autism: An Interactive Eye-Tracking Paradigm

Background: Social interactions form a substantial part of human daily living and develop from fundamental processes during infancy, such as joint attention (JA), to highly sophisticated processes in adolescence and adulthood. In the case of autism spectrum disorder (ASD), social interactions are typically affected with deficits in JA as one of the earliest sign to be at risk for ASD. Numerous behavioral studies have consistently demonstrated the compelling link between JA and ASD. However, to fully understand the deficits in JA and the link to ASD, it is crucial to examine the underlying neural processes during development, which are until now poorly understood. 

Objectives: We compared the neural correlates of JA and its modulation by a familiar and unfamiliar interaction partner and by self- and other-initiation in children and adolescents with and without ASD.  

Methods:   We used an interactive eye-tracking setup (Wilms et al., 2010), in which participants were looking at a face surrounded by three targets (left, right and top). In the Self-Initiated conditions, participants shifted their gaze towards a target. The interaction partner followed the participant’s gaze (JA-Self) or the interaction partner shifted her gaze downwards (Control_Self). In the JA-Other condition, participants were followed the interaction partner’s gaze. In the Control-Other condition, participants shifted the gaze towards the target that changed color while the interaction partner shifted her gaze downwards. Importantly, the gaze-contingent setup allowed for comparison of neurofunctional differences for successfully established JA irrespective of behavioral differences. We analyzed 14 participants with ASD (8-18 years) and 14 age-, gender-, and IQ-matched TD participants. Brain imaging data were analyzed with SPM8, using a flexible factorial ANOVA model (random effects analysis, threshold: p<0.05 cluster-level FWE corrected, voxel level p<0.001). 

Results: TD participants elicited greater activation in the right STS during JA than control conditions compared to participants with ASD who even showed deactivation during JA. Similarly, TD participants elicited left TPJ and left inferior parietal lobe activation specifically for JA, whereas participants with ASD elicited the left TPJ over all conditions with no differentiation between JA and control conditions. TD participants also deactivated the right temporal pole during control conditions and showed nearly no activation during JA. Participants with ASD though, deactivated the same region during JA. Furthermore, we found an overall effect of familiarity in the ASD group, i.e. increased activation when interacting with their mother in bilateral fusiform gyrus, left precuneus, left STS, ventral striatum and amygdala. 

Conclusions: The STS, TPJ and temporal pole have all been discussed as critical brain regions for social processing with documented abnormalities in ASD (Allison, Puce, & McCarthy, 2000; Olson, Plotzker, & Ezzyat, 2007; Zilbovicius et al., 2006). Specific neurofunctional differences even during successfully established JA in children and adolescents with ASD suggest persistent developmental differences in JA capabilities, contributing to social processing difficulties. Thus, for basic processes such as JA abnormalities in neural functioning appear to persist despite seemingly successful behavioral maturation. Additionally, our data supports previous studies demonstrating enhanced activation in participants with ASD when processing familiar faces.