A specific hypoactivation of right temporo-parietal junction/posterior superior temporal sulcus in response to socially awkward situations in autism

Abstract

People with autism spectrum disorder (ASD) often have difficulty comprehending social situations in the complex, dynamic contexts encountered in the real world. To study the social brain under conditions which approximate naturalistic situations, we measured brain activity with FUNCTIONAL MAGNETIC RESONANCE IMAGING: while participants watched a full-length episode of the sitcom The Office. Having quantified the degree of social awkwardness at each moment of the episode, as judged by an independent sample of controls, we found that both individuals with ASD and control participants showed reliable activation of several brain regions commonly associated with social perception and cognition (e.g. those comprising the 'mentalizing network') during the more awkward moments. However, individuals with ASD showed less activity than controls in a region near right temporo-parietal junction (RTPJ) extending into the posterior end of the right superior temporal sulcus (RSTS). Further analyses suggested that, despite the free-form nature of the experimental design, this group difference was specific to this RTPJ/RSTS area of the mentalizing network; other regions of interest showed similar activity across groups with respect to both location and magnitude. These findings add support to a body of evidence suggesting that RTPJ/RSTS plays a special role in social processes across modalities and may function atypically in individuals with ASD navigating the social world.

Keywords: autism; fMRI; mentalizing; right posterior superior temporal sulcus; right temporo-parietal junction; social cognition.

Fig. 1

Top: raw data for social awkwardness rating task. Each row represents an individual subject. Subjects’ ratings are represented on a gradient from white (not awkward) to black (maximally awkward). Bottom: the continuous regressor of interest for fMRI analysis: a smoothed running average of subjects’ mean awkwardness rating. The ‘halfway’ point, at which the episode was divided, is shown with a dashed line. Note: a measure of variance is intentionally absent in this bottom plot, since the mean signal reflects the actual regressor used in the fMRI analysis. A sense of variance at each point can be obtained by viewing the raster plot above.

Fig. 2

Top row: voxels positively correlated with observation of socially awkward moments (red-yellow) across all subjects (after voxel-level threshold of P < 0.01 and cluster-level threshold of P < 0.05). Outline of ‘mentalizing network’, as defined a priori by an independent study (Dufour et al., 2013), is shown in green. Bottom row: areas (red) correlated with social tasks, via the meta-analytic tool Neurosynth. The Neurosynth ‘decoder’ tool makes it possible to compare unthresholded maps of brain activation with statistical maps generated by Neurosynth across a wide variety of keywords. Of the 92 possible keyword ‘features’ made available for comparison, the ‘social’ brain map correlated best with the pattern of activation observed here (i.e. correlation with socially awkward moments). Left: lateral view (x = 52) of RTPJ and RSTS regions. Center: coronal view (y = −58), with RTPJ, PC and LTPJ visible. Right: medial view (x = 0), with PC and MPFC visible. Aside from LTPJ, there was substantial overlap between these previously defined ROIs and regions which appear to have been active with respect to our variable of interest.

Fig. 3

The cluster of voxels for which we observed a significant group difference (NT > ASD), shown on a gradient from red to orange. These sagittal cross sections go from lateral (left) to medial (right): $x=62,58,56,50$. The cluster straddles predefined RTPJ and RSTS regions; outlined here in blue and yellow, respectively.

Fig. 4

Weight of evidence (log₁₀ odds), either in favor of a difference between group means (above dotted line) or for no group difference (below dotted line). For each subject, for each ROI, a sphere of voxels was defined around the voxel most correlated with social awkwardness. The radius of this sphere could be made to vary; here, the weight of evidence is calculated under varying assumptions for the radius of this sphere, from 3 mm (smallest circles) to 9 mm (largest circles). The choice of assumptions does not seem to qualitatively affect the result of the analysis; for most regions, we find weak to moderate evidence in favor of the null hypothesis, with the exceptions of RTPJ and RSTS, for which we find weak evidence for a group difference.

Fig. 5

Clusters co-active, across all subjects, with the region of RTPJ/RSTS for which a group difference (NT > ASD) had been previously observed in response to social awkwardness. Outline of ‘mentalizing network’, as defined a priori by an independent study (Dufour et al., 2013), is shown in green. Left: lateral view (x = −52) of LTPJ and left temporal lobe. Center: coronal view (y = −56). Right: lateral view (x = 52) of RTPJ and right temporal lobe.