New 'Willed Agent' model boosts cooperation by ignoring cost-benefit fluctuations

Standard rational actor models often fail to explain cooperation in social dilemmas because continuous utility maximization destabilizes outcomes. In a new paper accepted at CogSci 2026, researchers Yizhe Huang, Bin Ling, Song-Chun Zhu, and Xue Feng introduce a framework of 'will'—a mechanism that persistently pursues goals while ignoring local cost-benefit fluctuations. They formalize Willed Agents as potential minimizers, distinguishing them from cumulative utility maximizers. Dynamical analysis of infinite populations shows willed agents shrink the feasible state space, acting as boundary constraints that accelerate convergence in canonical social dilemmas.

Through multi-agent simulations in a spatiotemporal Stag Hunt Game, the team demonstrates that willed agents function as 'cooperation catalysts,' enabling groups to surmount high-risk thresholds where purely utility-maximizing agents fail. Notably, heterogeneous will strength promotes cooperation, and agents that autonomously suspend rational re-evaluation significantly outperform continuous optimizers. These findings suggest that successful cooperation relies on the cognitive capacity to strategically constrain calculation, offering new insights for multi-agent systems, AI alignment, and game theory.