Existing academic research on lying predominantly focuses on neural mechanisms, behavioral characteristics, game-theoretic benefits, and moral psychology, while failing to explore the fundamental axiomatic logic that underpins the emergence of human lying behavior. Based on the universal axiom of the binary excluded middle law, the dual theoretical architecture of ontology entropy and cognitive entropy, and the cognitive entropy threshold convergence theory, this study reconstructs the essential definition of lying at the fundamental logical level. This paper defines lying as a deliberate logical decision-making behavior in which the cognitive subject completes deterministic binary truth-value judgment of objective facts and actively selects the opposite truth-value for external expression. This study systematically distinguishes the essential logical boundaries among truthful expression, unintentional error, cognitive ambiguity, and deliberate lying. It rigorously demonstrates that proficient liars possess stronger capabilities in logical decomposition, parallel reasoning, and contradiction verification, which constitute the core endogenous advantages of high-skill lying behavior. Furthermore, this paper reveals the essential defects of current probability-fitting artificial intelligence models in simulating human deceptive logic and explains why mainstream large models cannot generate real human-like lying behavior. On this basis, a novel identification framework for AI false behavior is constructed, providing feasible industrial implementation paths for AI trusted security detection. Breaking the traditional subjective behavioral research paradigm, this study incorporates human deceptive behavior into a unified objective logical analysis system, filling long-standing interdisciplinary research gaps and delivering dual theoretical innovation and engineering application value.
Xiangsheng Yu (Sat,) studied this question.