Network structure underlying resolution of conflicting non-verbal and verbal social information

Abstract

Social judgments often require resolution of incongruity in communication contents. Although previous studies revealed that such conflict resolution recruits brain regions including the medial prefrontal cortex (mPFC) and posterior inferior frontal gyrus (pIFG), functional relationships and networks among these regions remain unclear. In this functional magnetic resonance imaging study, we investigated the functional dissociation and networks by measuring human brain activity during resolving incongruity between verbal and non-verbal emotional contents. First, we found that the conflict resolutions biased by the non-verbal contents activated the posterior dorsal mPFC (post-dmPFC), bilateral anterior insula (AI) and right dorsal pIFG, whereas the resolutions biased by the verbal contents activated the bilateral ventral pIFG. In contrast, the anterior dmPFC (ant-dmPFC), bilateral superior temporal sulcus and fusiform gyrus were commonly involved in both of the resolutions. Second, we found that the post-dmPFC and right ventral pIFG were hub regions in networks underlying the non-verbal- and verbal-content-biased resolutions, respectively. Finally, we revealed that these resolution-type-specific networks were bridged by the ant-dmPFC, which was recruited for the conflict resolutions earlier than the two hub regions. These findings suggest that, in social conflict resolutions, the ant-dmPFC selectively recruits one of the resolution-type-specific networks through its interaction with resolution-type-specific hub regions.

Keywords: PPI; brain; conflict monitoring; empathy; human; theory of mind.

Fig. 1

Task design and behavioral results. (A) Task design. In each trial during fMRI scanning, a 1.5 s video stimulus was presented with audio. The subjects were instructed to freely judge whether a professional actor who appeared in the video was a friend or foe by pressing a different button in the following 2 s response period. The average duration of a single trial was 7 s. (B) Types of video stimuli. In each video stimulus, 1 of 20 professional actors spoke an emotional word (positive or negative verbal information) with an emotional facial expression and voice prosody (positive or negative non-verbal information). (C) Differences in response time between congruent and incongruent stimuli. Responses to incongruent stimuli were significantly longer than responses to congruent stimuli (**P < 0.01, paired t-test). (D) Difference in response time between non-verbal–cue-biased and verbal–cue-biased resolutions. Friend/foe judgments of incongruent stimuli were classified into non-verbal–cue-biased or verbal–cue-biased resolutions of incongruence (see Introduction). There was no significant difference in response time between non-verbal–cue-biased and verbal–cue-biased resolutions. (E). Difference in the number of incongruent resolutions. Participants made significantly more non-verbal–cue-biased resolutions than verbal–cue-biased resolutions (***P < 0.001, paired t-test).

Fig. 2

Brain regions related to V and NV resolutions. (A) Brain regions specific to V resolutions. The bilateral ventral posterior inferior frontal gyrus (pIFG) showed significantly larger activity during V resolutions than during NV resolutions (P < 0.05, FDR-corrected). (B) Brain regions specific to NV resolutions. The posterior dorsal medial prefrontal cortex (post-dmPFC), bilateral anterior insula (AI) and dorsal pIFG showed significantly larger activity during NV resolutions than during V resolutions (P < 0.05, FDR-corrected). (C) Brain regions common to V and NV resolutions. Conjunction analysis of V resolutions (P < 0.001, uncorrected) and NV resolutions (P < 0.001, uncorrected) revealed brain regions common to V and NV resolutions (P < 10⁻⁶, uncorrected). The regions include the anterior dmPFC, bilateral fusiform gyrus and bilateral superior temporal sulcus (STS).

Fig. 3

PPI-based networks and comparison with behavioral pattern. (A) PPI-based network topology. The solid arrows show PPIs enhanced during the NV resolutions compared with during V resolutions, whereas the dashed arrows indicate PPIs enhanced during the V resolutions (P < 0.05, FDR-corrected, Table 2). The boxes show the V regions; the circles, the NV regions; the gray boxes, the V&NV regions. (B) Degree distribution. The bars show the number of the significantly enhanced PPIs for each of the regions. The right ventral pIFG had the largest degree (i.e. the largest number of significant PPIs). The post-dmPFC had the largest degree among NV regions, and the ant-dmPFC had the largest degree among V&NV regions. (C) Comparison between the brain activity and the tendency of conflict resolutions. Left panel: Among the V regions, only the right ventral pIFG showed the significant positive correlation between its brain activity and the number of V resolutions (P < 0.05, Bonferroni-corrected). Right panel: Among the NV regions, only the post-dmPFC showed the significant positive correlation between its brain activity and the number of NV resolutions (P < 0.05, Bonferroni-corrected). (D) Comparison between the brain activity and the response time. Among the V&NV regions, only the ant-dmPFC showed significant positive correlations between its brain activity and the response time spent for conflict resolutions (P < 0.05, Bonferroni-corrected).

Fig. 4

Characterization of the central regions based on MVPA. (A) Representative result of the MVPA using the brain activity in the resolution-type–specific hub regions. In the case of one of the subjects in the fMRI experiment II (S1), the MVPA using brain activity of resolution-type–specific hub regions accurately predicted whether the subjects chose a V or NV resolution in response to the incongruent stimulus (90%, P < 0.001 in a permutation test). (B) MVPA results in all the seven subjects. The MVPA using the brain activity in the resolution-type–specific hub regions predicted types of conflict resolutions (V or NV resolutions) in all the seven subjects in the fMRI experiment II (*P < 0.05; **P < 0.01; ***P < 0.001). (C) Comparison of time courses of classification accuracy among the three central regions. Another MVPA estimated which period had the largest information about the difference between the congruent and incongruent stimuli trials. For the ant-dmPFC, the classification accuracy peaked at 2 s after the stimulus onset, whereas, for the post-dmPFC or right ventral pIFG peaked at 4 s after the stimulus onset (P < 1 × 10⁻⁴ in a permutation test). * shows the period during which there was significant difference in the classification accuracy between the ant-dmPFC and post-dmPFC/right ventral pIFG (P < 0.05, Bonferroni-corrected, in a paired t-test). Error bars: s.d.