LIDS Seminar Series: Eric Mazumdar

“Behavioral Economics as a Foundation for Principled Multi-Agent Reinforcement Learning: Tractability, Robustness, and Other Free Lunches”

Emerging applications in AI are fundamentally multi-agent, yet little guidance exists for designing agents for strategic settings. Indeed, strategic interactions can break conventional intuition in ML and AI. Furthermore, classical game-theoretic concepts like Nash equilibria are well known to be intractable to compute and have failed to yield the algorithmic foundations needed for scalable multi-agent learning.

In this talk, I will show, using both theory and experiments, that behavioral economics provides principled foundations for multi-agent learning. In particular, imbuing agents with features of human decision-making — namely strategic risk-aversion and bounded rationality — gives rise to new game-theoretic equilibria that are provably robust and computationally tractable across all games. This allows us to develop principled and scalable multi-agent learning algorithms. Beyond the computational benefits, these approaches yield surprising free lunches: in cooperative settings, our agents can achieve outcomes strictly better than Nash, exhibit less free-riding, and collaborate more consistently with new partners — with empirical gains extending even to preliminary experiments on collaborative tasks between LLMs. These results suggest that moving beyond classical game-theoretic concepts can give us new foundations for principled and scalable strategic decision-making.