Why are lies small?

Deception is common, yet lies are typically small and not easily spotted. The dominant explanations for why this is the case are psychological in nature, such that small lies are thought to allow individuals to maintain their self-concept or that individuals have a general preference for truth-telling. In this paper, we suggest that lies are typically small because small lies allow individuals to retrieve benefits associated with lying while preventing punishment from interaction partners. We present two game-theoretic models in which agents are incentivized to misreport private information. In a two-player game, we show that the equilibrium size of the lie is predicted to be smaller when the reputational costs of lying are larger, and smaller when the state of the world is more observable. In a three-player game, we show that lies can remain stable in group settings provided that the equilibrium size of the lie is small enough, so that the sender retains plausible deniability about lying. A unique and central prediction of our models is that biased self-presentation will be modulated by the degree of observability of the trait. We review the empirical literature on positive illusions, and argue that the findings support this prediction. We propose that the aversion to lying observed in laboratory experiments reflects the workings of a psychology well-adapted to the incentive structure of natural environments and suggest that models of human social behavior should focus on capturing relevant parameters from such environments.