This paper develops a formal model of moral balancing and applies it to deceptive communication in a sender–receiver environment. Agents are characterized by a stable moral identity (MI) and a fluctuating moral self-image (MSI). The discrepancy between the two defines a moral-balance state that shifts the private cost of lying and evolves with past behavior. In a one-shot interaction, this mechanism yields a threshold structure in which low-MI types lie whenever deception is materially attractive, whereas high-MI types lie only after accumulating sufficient moral surplus. The model thereby provides a unified formal account of truth-telling bias, moral licensing, and moral cleansing reconciling them as different regimes of the same self-regulation process, determined by the agent’s position relative to their identity benchmark. In repeated interaction, asymmetric moral updating implies a critical lie-frequency threshold above which deceptive paths become psychologically unsustainable for high-MI types. Repeated deception is thus bounded not only by reputational considerations but also by an internal self-regulation constraint. The framework yields testable predictions about heterogeneity in honesty, the role of initial moral balance, the temporal clustering of deception, and the conditions under which identical moral manipulations produce opposite behavioral responses. • Moral balancing formalized as state-dependent lying costs in signaling games. • High-MI types lie only after accumulating sufficient moral surplus. • Asymmetric updating bounds long-run deception via morality creep. • Critical lie-frequency threshold separates sustainable from unsustainable paths. • Model explains mixed replication results in moral licensing literature.
Konrad Kober (Fri,) studied this question.