AI study: Personality traits shape cooperation in dynamic networks

A new study published in Scientific Reports (arXiv:2607.23168) models how stable personality differences shape cooperation when social ties can form and dissolve. Researchers David Abián, Jorge Bernad, Sergio Ilarri, and Raquel Trillo-Lado built a repeated Prisoner's Dilemma on endogenous networks where agents' Big Five traits—extraversion (target number of partners), openness (search breadth beyond friends-of-friends), and agreeableness (baseline cooperation willingness)—drive local decisions. At each encounter, agents combine their baseline with a partner's directly observed history, without trait labels, gossip, or global reputations. Network ties form when under-connected and cut when over-connected, prioritizing defectors for removal. The study varied network size (N=30–200), population composition, and the balance between trait-driven and history-driven behavior.

Three robust patterns emerged. First, 'cooperate first, then reciprocate'—high initial willingness to cooperate combined with history-sensitive response—produces systems that are simultaneously more prosperous, fairer, and safer. Second, personality has predictable conditional effects: agreeableness helps when history matters but hurts when behavior is mostly trait-driven; extraversion amplifies the environment; openness has little net payoff effect. Third, the network reorganizes accordingly: degree assortativity stays near zero, but agreeable agents increasingly connect to one another when cooperation takes hold. These findings have implications for designing AI agents that cooperate in dynamic social networks.