Cooperation Regulation in Coupled Populations via Dynamic Aspiration and Justice Strategy

Le résumé fourni par la source

Promoting prosocial behavior across interconnected groups is a key problem in networked multi-agent systems. We study this problem using a two-population evolutionary-game framework that decouples payoff interaction from strategy learning and integrates dynamic aspiration as a tunable regulator. Each agent chooses among cooperation (unconditional prosocial behavior), defection (exploitation), and justice (conditional cooperation with costly punishment); total payoff combines intra-population and weighted inter-population interactions through a coupling parameter ω ; aspiration adapts to recent payoff experience; and dissatisfied agents update by within-population Fermi imitation. Strategy decomposition analysis reveals that under baseline aspiration (A = 1.6), prosocial outcomes are maximized at low-to-moderate coupling (ω* = 0.2 on lattice networks), driven primarily by justice-mediated defector suppression rather than pure cooperator proliferation. At this optimum, justice agents dominate (fJ = 0.63), defectors are minimally present (fD = 0.15), and prosocial fraction reaches fC + fJ = 0.85. Crucially, initial aspiration level systematically modulates optimal coupling: across A ∈ [0, 4], the preferred ω* shifts from 0.1 to 0.8, validating dynamic aspiration as a system-level control mechanism. Topology strongly modulates regulation: lattice yields highest prosocial outcomes; small-world requires higher coupling to compensate for long-range defector spread. Sensitivity analysis shows robustness to secondary parameters (a, K, l) while confirming the central role of aspiration-driven modulation.

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.