The End of Utility: a Coherence-Centric Game Theory for Intelligent Systems

Classical and evolutionary game theory, rooted in utility maximization and static equilibrium logic, collapse under the demands of recursive intelligence, multi-agent feedback, and dynamic coherence. This paper introduces Recursive Coherence Game Theory (RCGT)—a first-principles reformation of strategic logic where the core invariant is not payoff, but coherence: the persistent self-organization of agents across time, scale, and semantic layer. Grounded in Recursive Choice Theory, Layered Intelligence, and Coherence Ethics, RCGT models strategy as the modulation of coherence gradients rather than the pursuit of discrete rewards. It introduces Adaptive Coherence Equilibria—dynamic attractor states where agents align recursive stability rather than static dominance—and frames multi-agent interaction as the co-evolution of coherence fields. This paradigm resolves foundational paradoxes in classical game theory, including the Frame Problem, bounded rationality, and the Symbol Grounding Problem, while enabling ethically grounded and empirically testable architectures for intelligent systems. Simulations demonstrate that coherence-driven agents outperform utility-maximizing counterparts in resilience, adaptability, and emergent alignment. The implications extend across AI, governance, and civilization design, proposing a post-utility strategic logic for systems that must persist through entropy, uncertainty, and recursive evolution. RCGT offers not just a new kind of game theory—it offers a map of reality’s own intelligence: recursive, adaptive, and coherence-first.