Specificity, reliability and sensitivity of social brain responses during spontaneous mentalizing

Abstract

The debilitating effects of social dysfunction in many psychiatric disorders prompt the need for systems-level biomarkers of social abilities that can be applied in clinical populations and longitudinal studies. A promising neuroimaging approach is the animated shapes paradigm based on so-called Frith-Happé animations (FHAs) which trigger spontaneous mentalizing with minimal cognitive demands. Here, we presented FHAs during functional magnetic resonance imaging to 46 subjects and examined the specificity and sensitivity of the elicited social brain responses. Test-retest reliability was additionally assessed in 28 subjects within a two-week interval. Specific responses to spontaneous mentalizing were observed in key areas of the social brain with high sensitivity and independently from the variant low-level kinematics of the FHAs. Mentalizing-specific responses were well replicable on the group level, suggesting good-to-excellent cross-sectional reliability [intraclass correlation coefficients (ICCs): 0.40-0.99; dice overlap at P_uncorr<0.001: 0.26-1.0]. Longitudinal reliability on the single-subject level was more heterogeneous (ICCs of 0.40-0.79; dice overlap at P_uncorr<0.001: 0.05-0.43). Posterior temporal sulcus activation was most reliable, including a robust differentiation between subjects across sessions (72% of voxels with ICC>0.40). These findings encourage the use of FHAs in neuroimaging research across developmental stages and psychiatric conditions, including the identification of biomarkers and pharmacological interventions.

Keywords: animated shapes; biomarker; fMRI; reliability; spontaneous mentalizing.

Fig. 1.

Overview over stimuli and experimental design. (A) Stimuli consisted of three types of animated video clips with increasing levels of social significance, represented by three simplified icons during the subsequent categorization (MCQ-cat: ‘Which category did the previously presented animation belong to?’). Example video clips are accessible at https://sites.google.com/site/utafrith/research. (B) ToM animations were additionally rated according to perceived emotionality (MCQ-feeling: ‘How did the little/big triangle feel at the end of the animation?’), which probed for the acquired concept of the displayed emotional states, and thus served as a rough estimation of the subject’s understanding of the cover story: ‘coaxing’ and ‘surprising’ require both triangles to be rated similarly positively, while ‘mocking’ and ‘seducing’ require the little triangle to be rated more positively than the big triangle. Emotional states were represented by schematic faces with an unhappy, neutral and happy expression, respectively. Bar graphs depict average rating scores (± SE) for each ToM animation, with ‘+1’ referring to ‘happy’ and ‘−1’ referring to ‘unhappy.’ Panel (C) depicts the temporal structure of a ToM trial. The presentation of the video clips was preceded by a jitter with variable duration (M = 995.67 ms, s.d. = 418.3). Responses were given with the right thumb, using the left, upper and right key of an MRI compatible button box (Current Designs, PA, USA). As soon as responses were given during MCQ ratings, the chosen icon was framed in red for the duration of one additional second, followed by a blank screen for the remainder of the respective MCQ phase. No feedback on response accuracy was given.

Fig. 2.

Schematic overview over the reliability analysis strategy, with A1/A2: number of supra-threshold voxels at T1/T2, A_overlap: number of voxels with supra-threshold activation at both T1 and T2, BMS: between-subjects mean square, EMS: error mean square, k: number of repeated sessions.

Fig. 3.

Functional activation (A) and reliability metrics (B–D) during spontaneous mentalizing (ToM) compared to agency perception (GD). Sections display thresholded (A) T-maps and (B) ICC(3,1)-maps (i.e. single-voxel reliability). ROIs are outlined in red. (C) Spatial reliability of group activation maps is illustrated as scatter plots of voxel-wise contrast estimates at session 1 (T1) and session 2 (T2). Voxels belonging to the respective ROI are highlighted in red. Dashed lines designate zero on each axis. (D) Sections showing the overlap (in yellow) of whole-brain networks, defined at T1 (blue) and T2 (red) at a significance threshold of P_uncorr < 0.001.

Fig. 4.

Whole-brain effects of low-level kinematic stimulus properties. (A) Sections displaying activated clusters. (B) Sections showing outlines of the pre-defined mentalizing network (red), overlaid on whole-brain effects of low-level kinematics. (C) Outlines of the whole-brain effects of low-level kinematics are projected on sections displaying whole-brain effects of mentalizing (ToM > GD). All statistical maps were thresholded at P < 0.05, uncorrected for multiple comparisons across the whole brain to minimize false negatives.