Social interaction recruits mentalizing and reward systems in middle childhood

Abstract

Social cognition develops in the context of reciprocal social interaction. However, most neuroimaging studies of mentalizing have used noninteractive tasks that may fail to capture important aspects of real-world mentalizing. In adults, social-interactive context modulates activity in regions linked to social cognition and reward, but few interactive studies have been done with children. The current fMRI study examines children aged 8-12 using a novel paradigm in which children believed they were interacting online with a peer. We compared mental and non-mental state reasoning about a live partner (Peer) versus a story character (Character), testing the effects of mentalizing and social interaction in a 2 × 2 design. Mental versus Non-Mental reasoning engaged regions identified in prior mentalizing studies, including the temporoparietal junction, superior temporal sulcus, and dorsomedial prefrontal cortex. Moreover, peer interaction, even in conditions without explicit mentalizing demands, activated many of the same mentalizing regions. Peer interaction also activated areas outside the traditional mentalizing network, including the reward system. Our results demonstrate that social interaction engages multiple neural systems during middle childhood and contribute further evidence that social-interactive paradigms are needed to fully capture how the brain supports social processing in the real world.

Keywords: fMRI; mentalizing; middle childhood; social interaction; social reward; theory of mind.

Figure 1

The interactive mentalizing task. Children completed 24 trials of each condition (Peer Mental, Character Mental, Peer Non‐Mental, Character Non‐Mental) in an event‐related design. Mental trials required reasoning about mental states, while Non‐Mental trials did not. In the Peer trials, children believed they were interacting with a child being scanned in another laboratory, whereas in Character trials, they believed they were answering questions about a fictional character provided by a computer. All trials had predetermined peer or computer responses. A smiley face (Peer) or check mark (Character) in the Feedback period indicated a match between the child's response and the peer or computer response [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 2

Behavioral results. (a). In‐scanner performance by condition. Mean values are plotted for RT (seconds) and accuracy (% correct) for each of the four conditions. Repeated measures ANOVA indicated a significant main effect of social interaction on RT such that children responded more quickly on Peer than Character trials. Error bars represent 95% CIs. #p < .1; **p < .005. (b) Posttest questionnaire. For Peer and Character conditions separately, children rated on a Likert‐type scale of 1–5 how much they enjoyed interacting with their partners (Peer) and answering questions from the computer (Character), how much they liked guessing what their partners would pick (Peer) and what came next in the story (Character), and how much they paid attention when interacting with their partners (Peer) and when answering questions from the computer (Character). Wilcoxon signed rank tests were used to compare ratings between Peer and Character conditions. #p < .1; **p < .005 [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 3

Whole‐brain analysis of the main effects of mentalizing and social interaction (cluster corrected p < .05). Mentalizing (Mental vs. Non‐Mental) activated regions previously identified in the mentalizing literature (dMPFC, TPJ, STS, and ATL). Social interaction (Peer vs. Character) activated similar regions, as well as additional cortical midline regions and subcortical structures associated with reward (e.g., amygdala, striatum) [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 4

Whole‐brain pairwise comparisons between the four conditions (cluster corrected p < .05). Offline mentalizing (Character Mental vs. Character Non‐Mental) elicited a pattern of activation similar to the main effect of mentalizing (Figure 3). In contrast, no regions were significantly more active for Peer Mental than Peer Non‐Mental. Social interaction without explicit mentalizing demands (Peer Non‐Mental vs. Character Non‐Mental) recruited similar regions as in the main effect of social interaction (Figure 3), whereas a smaller subset of these regions was more active for mentalizing within social interaction than offline mentalizing (Peer Mental vs. Character Mental). CM, Character Mental; CNM, Character Non‐Mental; PM, Peer Mental; PNM, Peer Non‐Mental [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 5

Shared regions for mentalizing and social interaction. Binarized, cluster‐corrected maps for offline mentalizing (Character Mental > Character Non‐Mental; green) and social interaction without mentalizing demands (Peer Non‐Mental > Character Non‐Mental; blue) are shown along with their conjunction (red), which reveals both overlapping and distinct regions of activation [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 6

ROI analysis of mentalizing regions during social interaction. ROIs were defined by the Character Mental > Character Non‐Mental contrast. Individual beta values for each condition within each ROI were extracted; average values are plotted with error bars representing 95% CIs. Paired t tests indicated nonsignificant differences between Peer Mental and Peer Non‐Mental in all ROIs except right ATL, as well as nonsignificant differences between Character Mental and Peer Non‐Mental in all ROIs. ATL, anterior temporal lobe; IFG, inferior frontal gyrus; LOFC, lateral orbitofrontal cortex; STS, superior temporal sulcus. * p < .05 corrected; #p < .1 corrected [Color figure can be viewed at http://wileyonlinelibrary.com]

Figure 7

Effects of age on neural response to social interaction. (a) Whole‐brain effects of age on social interaction (Peer vs. Character; cluster corrected p < .05). Differences in activation to Peer versus Character conditions decreased with age in several frontal, temporal, insular, and subcortical areas. (b) Effect of age on mentalizing ROIs. ROIs were defined by the main effect of mentalizing (Mental > Non‐Mental). All regions showed a significant negative correlation between age and difference in activation to Peer versus Character conditions. ATL, anterior temporal lobe; dMPFC, dorsomedial prefrontal cortex; LOFC, lateral orbitofrontal cortex; STS, superior temporal sulcus; TPJ, temporoparietal junction. * p < .05 corrected [Color figure can be viewed at http://wileyonlinelibrary.com]