Perceiving social interactions implicitly engages typical theory of mind areas

792 

Abstract Submission 

Zizhuang Miao¹, Heejung Jung¹, Philip Kragel², Patrick Sadil³, Martin Lindquist³, Tor Wager¹

¹Dartmouth College, Hanover, NH, ²Emory University, Atlanta, GA, ³Johns Hopkins University, Baltimore, MD

Zizhuang Miao  
Dartmouth College
Hanover, NH

Heejung Jung  
Dartmouth College
Hanover, NH

Philip Kragel, PhD  
Emory University
Atlanta, GA

Patrick Sadil, PhD  
Johns Hopkins University
Baltimore, MD

Martin Lindquist, PhD  
Johns Hopkins University
Baltimore, MD

Tor Wager, PhD  
Dartmouth College
Hanover, NH

Experimental paradigms investigating the neural correlates of social interaction perception and theory of mind (ToM) have often confounded the two. For instance, animations of geometric shapes have been taken as instances of both types of process (e.g., Moessnang et al., 2016; Varrier & Finn, 2022), and the false belief stories designed to study theory of mind (Saxe & Kanwisher, 2003) include significantly more social interactions than control stories. These confounds possibly contributed to large variations across studies (e.g., Schurz et al., 2014) and obscured our understanding of the neural correlates of social interaction perception and ToM. Here, we directly compared the neural correlates of perceiving social interactions and demand on ToM during naturalistic narrative perception within a single study.

Ninety-three healthy adults listened to ("Audio") or read ("Text") narratives (each 2-4 minutes long) during fMRI scanning. Each narrative includes a mix of social interactions and nonsocial scenarios, along with a blend of characters' actions and mental activities. Social interaction scenarios explicitly involved interpersonal communication or shared activity (e.g., "he told her to take care"), whereas nonsocial scenarios involved only one person or lacked interactions (e.g., "they were friends for years"). Descriptions of characters' mental activities, marked by verbs such as "thought", "felt", and "knew", placed explicit demands on participants' ToM. We coded social interactions and explicit ToM demands on a clause-by-clause basis. The codings were relatively independent (r = -0.08 and -0.25 under the Text and Audio conditions, respectively, all variance inflation factors < 1.1), which enables modeling them in one regression model.

Across the Audio and Text conditions, group-level univariate activations of social interactions (SInt) and ToM were both significant in left temporoparietal junction (TPJ), left superior temporal sulcus (STS), bilateral precuneus (PC), and left dorsomedial prefrontal cortex (dmPFC) (false discovery rate (FDR) corrected, q < .01). Contrasting SInt and ToM indicated that SInt more strongly activated bilateral TPJ, left STS, and right dmPFC (FDR q < .01). We compared the SInt - ToM contrast maps with three a priori term maps from Neurosynth.org (Yarkoni et al., 2011), namely "tom", "mentalizing", and "social interaction", and found significant correlations with "tom" (r = .18, p < .001) and "mentalizing" maps (r = .12, p < .001) but not the "social interaction" map (r = .03, p = .26). Further, two ToM regions of interest (ROI) were defined as where z values in the "tom" and "mentalizing" map are > 3, and the average activation of SInt in both ROIs was significantly larger than ToM (t(92) = 3.49, p < .001; t(92) = 2.31, p = .023).

Consistent with previous studies, we found TPJ, STS, precuneus, and dmPFC responded to both social interaction perception and explicit demands of theory of mind. More importantly, minimal social interaction perception – listening to or reading stories in which two or more characters directly interacted – induced activity in classic ToM-related regions and more similar spatial patterns with prototypical ToM-related activation maps. Within those regions, the effects of social interaction perception were stronger than the effects of listening to or reading narratives that explicitly referred to characters' mental activities. This suggests that, even in the absence of explicit demands of ToM, participants were implicitly more inclined to engage the typical brain areas activated by previous ToM tasks when social interactions were present. Our findings underscore the necessity for improved experimental designs to better distinguish between social interaction perception and theory of mind, thereby advancing our knowledge of their neural correlates.

Social Cognition ¹

Social Interaction

Activation (eg. BOLD task-fMRI) ²

FUNCTIONAL MRI

Social Interactions

Other - Social cognition; Theory of mind; Mentalizing