We develop a framework in which individuals' preferences co-evolve with their abilities to deceive others regarding their preferences and intentions. We show that a pure outcome is stable, essentially if and only if it is an e¢cient Nash equilibrium. All individuals have the same deception ability in such a stable state. In contrast, there are non-pure outcomes in which non-Nash outcomes are played, and di¤erent deception abilities co-exist. We extend our model to study preferences that depend also on the opponent's type.