Abstract

This paper presents a large-scale simulation study evaluating the relative success of competing moral theories in strategic interactions. Rather than relying on abstract philosophical argumentation, we adopt a game-theoretic framework in which agents governed by distinct moral principles interact across the entire range of symmetric 2-player, 2-strategy, normal form games, including the Prisoner’s Dilemma, Stag Hunt, and Hawk-Dove. We analyze 32 moral theories, including five foundational frameworks—average consequence utilitarianism, Kantian deontology, selfishness, maxi-min, and empathy-based morality—as well as 26 synthesized moralities formed by combining these base theories. Through Monte Carlo simulations, we determine which moral rules yield the highest expected material payoffs, both in homogeneous interactions (where all agents follow the same rule) and in mixed populations that include selfish agents. Our findings reveal that synthesized moralities systematically outperform their foundational counterparts, highlighting the practical advantages of moral autonomy. In particular, the synthesis of utilitarianism and deontology (UD) emerges as the 'optimific champion', achieving the highest payoffs under the assumption of universal adherence. However, in mixed societies where agents must coexist with selfish individuals, a different moral rule, what we call the 'realist’s champion', prevails: the synthesis of deontology and selfishness (SD). These results have significant implications for moral philosophy, challenging the traditional focus on isolated moral theories and suggesting that moral rules are best understood in relation to their interactive success. The study also underscores the essential role of deontology in viable moral syntheses, raising fundamental questions about the evolutionary robustness and practical viability of moral systems.