An Axiomatic and Decision-Theoretic Framework for Rational Dating Under Uncertainty

We model romantic dating as a sequential decision-making problem under uncertainty, subject to informational constraints, bounded rationality, and adversarial behavioral patterns. An axiomatic system is introduced to formalize fundamental principles governing stable interpersonal interaction, including mutual enthusiasm, character dominance, behavioral supremacy, and resource preservation. On this basis, we construct a dynamic state-transition model and a Bayesian belief-updating framework for candidate evaluation. An expected utility formulation is derived, incorporating investment costs and stability-adjusted rewards. We further analyze the robustness of the proposed system against three pathological latent types---avoidant agents, intermittent reinforcement agents, and high-chemistry/low-character agents---and show that naive policies fail under these regimes. By introducing variance penalties, repair constraints, and chemistry exclusion rules, we establish sufficient conditions for early termination and optimal escalation. The resulting framework yields provably conservative, dignity-preserving policies that minimize long-term regret and susceptibility to behavioral manipulation, while maintaining tractability for real-world application.