Neurocomputational Mechanisms Linking Future Interaction Prospects to Reactive and Proactive Costly Punishment

Costly punishment in social bonds includes reactive penalties for transgressions and proactive sanctions against fair behaviors; however, a unified neurocomputational framework is lacking. Using neuroimaging and computational modeling, we tested how future interaction prospects modulate both forms of punishment. Participants, acting as second- or third-party punishers, responded to offers from partners with or without future interactions, imposing harsher punishment on both fair and unfair offers from non-future-interacting individuals, consistent with lower self-reported closeness. Heightened aversion to disadvantageous inequity, indexed by dorsal anterior cingulate activity, mediated increased reactive punishment toward non-future-interacting wrongdoers, whereas punishment of future-interacting transgressors engaged the dorsolateral prefrontal cortex (dlPFC). A stronger preference for relative advantage and reduced harm sensitivity, linked to the dlPFC and temporoparietal junction, mediated greater proactive punishment toward non-future-interacting individuals. Proactive punishment also exhibits collective retaliation against others' unfairness. These findings establish a unified neurocomputational framework that integrates the mechanisms by which future interaction prospects regulate costly punishment within social bonds.