Neural Sensitivity to Live Social Interaction Captures Developmental Variability in ASD Traits

Background: The explicit social cognitive abilities of individuals with autism spectrum disorders (ASD) are frequently greater than corresponding abilities to navigate live social interaction (Klin et al., 2003). Understanding the behavioral and brain bases of this gap is fundamental to understanding ASD. To date, however, few neuroimaging paradigms have directly compared the processing of live versus recorded social stimuli, in part due to difficulties maintaining experimental control. Novel well-controlled neuroimaging paradigms examining specialization for live interaction will elucidate both core ASD symptoms and common comorbidities that are also most acute in live contexts (e.g., social anxiety).

Objectives: This study validated a novel approach to studying the neural correlates of processing of live versus recorded speech in children and adults and examined the relation between sensitivity to live interaction and ASD traits.   

Methods: Thirty-one typical adults (13 males) and 17 typically-developing children (8 males) aged 7-13 years (mean=10.8, SD=1.6) completed a novel fMRI task in which they listened to short vignettes in two conditions: Live and Not-Live. In the Live condition, participants listened to a social partner that they believed was speaking over a live audio feed. In the Not-Live condition, participants listened to recorded speech from another speaker. Vignettes contained no references to people or social situations, and, unbeknownst to participants, Live segments were prerecorded in order to match speech characteristics and linguistic content with Not-Live speech. To match attention, participants answered a question and received feedback after each vignette. ASD traits were measured in adults with the Autism Quotient (AQ) and in children with the Social Responsiveness Scale (SRS). Children also completed the Social Anxiety Scale for Children—Revised.  

Results: Children and adults rated the Live condition as feeling more live (i.e., like the speaker was talking directly to the participant) than the Not-Live condition (p<.01). For adults, greater perceived liveness for the Live condition correlated with lower AQ scores (r=-.40, p<.05). Adult whole-brain analysis comparing Live to Not-Live audio revealed increased activity in social brain regions, with greater differential dorsal medial prefrontal cortex (dMPFC) activation associated with lower ASD traits (r=-.51, p<.01). To analyze brain-behavior relations in children, regions of interest were created from adults. For children, greater activation for Live versus Not-Live stimuli in left superior temporal sulcus (STS) and temporoparietal junction (TPJ) was associated with diminished ASD traits (ps<.05), and social anxiety was positively correlated with live sensitivity in right TPJ (r=.66, p<.01).

Conclusions: This novel neuroimaging paradigm successfully captured neural differences elicited by real-time interaction in children and adults. In adults, higher ASD traits were associated with diminished neural discrimination for live versus recorded speech and with decreased ratings of perceived liveness for live, but not recorded, speech. For children, this paradigm dissociated between ASD traits (marked by decreased live specialization) and social anxiety traits (marked by increased live specialization). These findings are consistent with the hypothesis of impaired response to live interaction in ASD, perhaps reflecting diminished resonance with social partners, and indicate that interactive paradigms are an important direction for future clinical research.