The mentalizing network orchestrates the impact of parochial altruism on social norm enforcement

Abstract

Parochial altruism--a preference for altruistic behavior towards ingroup members and mistrust or hostility towards outgroup members--is a pervasive feature in human society and strongly shapes the enforcement of social norms. Since the uniqueness of human society critically depends on the enforcement of norms, the understanding of the neural circuitry of the impact of parochial altruism on social norm enforcement is key, but unexplored. To fill this gap, we measured brain activity with functional magnetic resonance imaging (fMRI) while subjects had the opportunity to punish ingroup members and outgroup members for violating social norms. Findings revealed that subjects' strong punishment of defecting outgroup members is associated with increased activity in a functionally connected network involved in sanction-related decisions (right orbitofrontal gyrus, right lateral prefrontal cortex, right dorsal caudatus). Moreover, the stronger the connectivity in this network, the more outgroup members are punished. In contrast, the much weaker punishment of ingroup members who committed the very same norm violation is associated with increased activity and connectivity in the mentalizing-network (dorsomedial prefrontal cortex, bilateral temporo-parietal junction), as if subjects tried to understand or justify ingroup members' behavior. Finally, connectivity analyses between the two networks suggest that the mentalizing-network modulates punishment by affecting the activity in the right orbitofrontal gyrus and right lateral prefrontal cortex, notably in the same areas showing enhanced activity and connectivity whenever third-parties strongly punished defecting outgroup members.

Figure 1

Design and decision screen. A: Depicted is the third‐party punishment paradigm with group manipulation (ingroup/outgroup). During Stage 1, which took place in the training course of the Swiss army, Player A and Player B (officer candidates) played a simultaneous Prisoners' Dilemma Game (PD), in which they were free to decide whether to cooperate (transfer the points) or to defect (keep the points). During Stage 2, which took place in the fMRI‐scanner, some of the officer candidates in the role of a third‐party (Player C) were confronted with the decisions of Player A and B and had the opportunity to assign (costly) punishment points to one of the players. For that purpose, Player C was endowed with 10 points for each judgment trial. One point assigned for punishment reduced the income of the punished player by three points. Note that we recoded all of Player C's decisions in such a way that A always refers to the player that C can punish, whereas B always refers to the player that C cannot punish. Crucially, Player A (whom C could punish) was either from the same group/platoon as Player C, as in the group constellations ABC (in green color) and AC (in blue color), or was from a different group/platoon as in the group constellation BC (in red color). Thus, ABC and AC are ingroup constellations, whereas BC is an outgroup constellation. B: Depicted is an example for a decision screen third‐parties saw during the scanning session. In this particular case, third‐parties were confronted with the outgroup constellation BC and an outgroup member who defected against a cooperating ingroup member. The group affiliation of Player A and B was indicated both verbally (other group/your group) and schematically (in black or gray color). Please note that we reversed the color for the schematic depiction of ingroup and outgroup members and the punishment scale (9 6 3 0 instead of 0 3 6 9) for half of the subjects.

Figure 2

Punishment behavior. Analyses of third‐parties' punishment behavior of Player A revealed, as expected, a strong impact of group membership on punishment behavior when Player A defected and Player B cooperated (DC), a weak effect when both players defected (DD), and no significant impact of group membership when Player A cooperated (CC, CD). Thus, third‐parties' punishment behavior revealed the expected parochial pattern, qualified by increased punishment of outgroup members and reduced punishment of ingroup members for the same defective behavior.

Figure 3

Outgroup effects: Punishment network. Depicted is the increased activation in the brain (at P < 0.005, voxel extent threshold: 10 voxels, activity in all regions survives small volume family‐wise‐error (FWE) corrections at P < 0.05, except the dorsal caudatus which just falls short of the threshold with P = 0.057, see Methods section for details) contrasting the outgroup (BC) minus the ingroup (AC+ABC) constellations, when Player A defected and Player B cooperated (behavioral pattern DC). Consistent with the increased punishment pattern in the outgroup condition (BC), increased activity was mainly found in brain areas involved in punishment‐related decision processes, including (A) right orbitofrontal gyrus (BA 11/47, x = 33, y = 39, z = −9), (B) right lateral prefrontal cortex (BA 44/45, x = 57, y = 12, z = 15), and (C) right dorsal caudatus (x = 15, y = 24, z = 9). Bar plots representing contrast estimates (in/outgroup vs. baseline) of functional ROIs (see Method section for details) revealed in accordance with the similar punishment pattern that the two ingroup constellations (AC, ABC) show a highly similar activity pattern (P > 0.25 for all paired t‐tests between the two ingroup constellations). Asterisks denote increased activity compared with baseline at P < 0.05 (*), P < 0.01 (**), P < 0.005 (***), or P < 0.001 (****). Please see Supporting Information Figure S1 for event‐related BOLD time courses of the depicted brain regions.

Figure 4

Connectivity analyses within the punishment network. A: Physio‐Physiological Interaction (PPI) analyses using the rOFG and right caudatus as seed regions revealed evidence (at P < 0.005, voxel extent threshold: 10 voxels) that the rOFG positively modulates the functional connectivity between right caudatus and rLPFC—notably in the same area of the rLPFC showing increased activity when third‐parties strongly punish outgroup members who defect against cooperating ingroup members (the same activation as illustrated in Fig. 3B). These highly specific connectivity and activity patterns provide evidence for a functionally connected neural network orchestrating punishment behavior. Color coding: connectivity effect depicted in red, activation level effect depicted in yellow, overlap depicted in orange. B: Connectivity analyses using the rOFG as seed region revealed (at P < 0.005, voxel extent threshold: 10 voxels, in violet color) that the functional connectivity between the rOFG and rLPFC depends on third‐parties' punishment level, that is the higher third‐parties punish defecting outgroup members in the DC condition, the stronger is the functional connectivity between these two regions. The scatter plot depicts this effect using a functional ROI of the rLPFC (BA 45/46, x = 48, y = 21, z = 21). In order to visualize the spatial proximity of all activation and connectivity effects in the rLPFC, the same activity and connectivity patterns described in (A) are also depicted here in (B) in the same colors. For display purposes, all activation and connectivity patterns in (A) and (B) are depicted at P < 0.01.

Figure 5

Ingroup effects: Mentalizing network. Depicted is the increased activation in the brain (at P < 0.005, voxel extent threshold: 10 voxels; activity in all regions survives small volume family‐wise‐error (FWE) corrections at P < 0.05, see Methods section for details) contrasting the ingroup (AC+ABC) minus the outgroup (BC) constellations, when Player A defected and Player B cooperated (behavioral pattern DC). Increased activity was mainly found in brain areas involved in mentalizing processes, including (A) dorsomedial prefrontal cortex (DMPFC; BA 9, x = 6, y = 54, z = 30), (B) left temporo‐parietal junction (lTPJ, BA 39/40/22, x = −45, y = −60, z = 21) and (C) right temporo‐parietal junction (rTPJ, BA 39/40, x = 57, y = −60, z = 30). Bar plots (color coding as in Fig. 3, red = outgroup BC, blue = ingroup AC, green = ingroup ABC) representing contrast estimates (in/outgroup vs. baseline) of functional or spherical ROIs (see Method section for details) revealed in accordance with the similar punishment pattern that the two ingroup constellations (AC, ABC) show a highly similar activity pattern (P > 0.25 for all paired t‐tests between the two ingroup constellations). Asterisks denote increased activity compared with baseline at P < 0.05 (*), P < 0.01 (**), P < 0.005 (***), or P < 0.001 (****). Please see Supporting Information Figure S2 for event‐related BOLD time courses of the depicted brain regions.

Figure 6

Connectivity analyses within the mentalizing network. A: Connectivity analyses using the DMPFC as seed region revealed (at P < 0.005, voxel extent threshold = 10 voxels) that the functional connectivity between the DMPFC and lTPJ depends on third‐parties' punishment level, that is the less third‐parties punish defecting ingroup members in the DC condition, the stronger is the functional connectivity between these two regions. This finding provides additional evidence that the mentalizing network is recruited in order to reduce the punishment of defecting ingroup members. Color coding: connectivity effect depicted in red, activation level effect depicted in yellow (the same activation as depicted in Fig. 5B), overlap depicted in orange. B: The scatter plot visualizes the effect explained in (A) using a functional ROI of the lTPJ (x = −57, y = −54, z = 24).

Figure 7

Connectivity analyses between the mentalizing‐network and punishment‐network. We applied Physio‐Physiological Interaction (PPI) analyses using the DMPFC and lTPJ as seed regions in order to reveal the effective connectivity between the two networks shown to orchestrate the parochial nature of altruistic norm enforcement. Findings revealed evidence (at P < 0.005, voxel extent threshold: 10 voxels) that the lTPJ modulates the effective connectivity between the DMPFC and (A) the evaluation system in the lateral (BA 10/11, left: x = −42, y = 54, z = −9; right: x = 24, y = 60, z = −6, depicted in blue) and medial OFG (BA 10/11, x = −15, y = 45, z = −9, depicted in red) as well as (B) the cognitive control system in the rLPFC (BA 45/46, x = 54, y = 36, z = 15, depicted in blue). In detail, medial areas of the OFG (depicted in red) show an enhanced positive connectivity with the DMPFC whenever the lTPJ is strongly activated. In sharp contrast, lateral areas of the OFG and the rLPFC (depicted in blue) show an enhanced negative connectivity with the DMPFC whenever the lTPJ is strongly activated. These highly distinctive connectivity effects in medial and lateral areas of the OFG support our hypothesis that a justification process in the mentalizing network might change the evaluation of ingroup members' defective behavior (making it less negative and/or more positive). Notably, the negative connectivity effects are localized in neighboring and overlapping areas of the punishment‐network depicted in Fig. 3 and 4. In order to visualize this spatial proximity, the same activity (depicted in yellow) and positive connectivity patterns (depicted in red and violet in the zoom view of Fig. 7B) are shown here. We in particularly want to point out (see zoom view in Fig. 7B) that the negative connectivity effect in the rLPFC (in blue) is localized in the same area showing an enhanced positive connectivity with the rOFG, whenever third‐parties strongly punish defecting outgroup members (depicted in violet, overlap depicted in dark violet). For display purposes all activation and connectivity patterns in (A) and (B) are depicted at P < 0.01, except for the zoom view in (A) on which the patterns are depicted at P < 0.05.

Figure 8

Summary: The analysis of the neural underpinnings of the parochial nature of altruistic norm enforcement revealed the following activity and connectivity pattern. First, the increased punishment of defecting outgroup members is associated with increased activity in a functionally connected network of brain areas involved in punishment‐related decision processes (red circles and red lines with arrows). Second, the stronger the connectivity within areas of this punishment network, the stronger defecting outgroup members are punished (violet lines with arrows). Third, the reduced punishment of ingroup members' defective behavior is associated with increased activity in the mentalizing network of the brain, suggesting that third‐parties try to understand and justify ingroup members' defective behavior (blue circles). Fourth, the stronger the connectivity within areas of this mentalizing network, the less third‐parties punish defecting ingroup members (orange lines with arrows). Fifth, the analysis of connectivity between the punishment and mentalizing/justification network suggests that the mentalizing/justification process reduces the punishment behavior by modulating the activity in areas of the punishment network associated with negative evaluation processes (rOFC) and the assignment of an appropriate punishment level via the weighting of economic‐self‐interests (rLPFC, blue lines with arrows).