Abormal Connectivity in the Social Brain in ASD Identified Via Naturalistic Social Perception and Independent Components Analysis

Background:   There is substantial evidence that ASD is associated with functional abnormalities in the "social brain", including amygdala, medial prefrontal cortex, fusiform gyrus, and superior temporal sulcus.  Although most studies on the social brain in ASD treat each of these areas in isolation, findings of abnormal white matter development in ASD suggest that social perception deficits may have as much or more to do with diminished communication between structures as abnormalities within them.  The present fMRI study examines connectivity between nodes of the social brain via two underutilized methodological approaches.  The first involves the perception of naturalistic social videos with no specific task instructions (passive viewing), in order to maximize the possibility of observing network-wide (as opposed to task and region-specific) activation.  Participant attention was monitored via simultaneous eyetracking and an unanticipated post-scan quiz on the content of the videos.  The second was the use of spatiotemporal independent components analysis (ICA) – an approach that is sensitive to coordinated patterns of activation across networks.

Objectives:   This study tests for abnormal connectivity between nodes of the social brain in ASD during naturalistic social perception.

Methods:   ASD and typically developing control (TDC) samples were scanned for 8.2 minutes while they watched videos of two children playing.  Videos alternated between 1-minute depictions of two children playing together (joint play) or separately (parallel play).  Data collection is ongoing going, with the present sample consisting of 27 ASD and 37 TDC participants.  ASD diagnoses were established via ADOS and ADI-R.  Groups were matched on age (ASD=10.0, TDC=9.3) and cognitive ability (Differential Abilities Scale-II GCA, ASD=108, TDC=112).  Independent components were identified by combining both TDC and ASD groups, then testing for post hoc differences in component weighting as a function of group and task condition (joint versus parallel play).  ICA was carried out for two separate subsets of participants (each including TDC and ASD) who received two different counterbalanced block orders; the first order was used to identify independent components, and the second was used to replicate them.

Results:   A single spatiotemporal independent component was identified spanning amygdala, medial prefrontal cortex, fusiform gyrus, and peri-amygdala.  Weightings for this component were significantly greater for the joint than the parallel play condition (p < .001), suggesting that the network is modulated by the amount of social information in a scene.  Although the component was identified by pooling data from ASD and TDC groups, weightings for this component were significantly less in the ASD than the TDC group (p = .039), indicating abnormal network connectivity in ASD.  Analyses integrating eyetracking data with these fMRI results are ongoing.

Conclusions:   Findings from this study indicate that social perception deficits in ASD may arise from deficits in connectivity between nodes of the social brain.  These findings are particularly compelling, as they rely on a statistical procedure (ICA) that makes no a priori assumptions about functional task (joint versus parallel play) or group (ASD versus TDC), and yet can distinguish both.