whether or not to continue interacting with player 1. We find con- ditions in which there is an equilibrium where player 2 chooses to interact with player 1 only if player 1 cooperates without looking (CWOL) and player 1 chooses to CWOL. We show that this equi- librium is robust to both high degrees of rationality, as modeled by subgame perfection and learning or evolutionary dynamics, as mod- eled by the replicator dynamic. Using computer simulations, we also show that it emerges with high frequency, even in the presence of other equilibria. Additionally, we show that the ability for player 1 to avoid looking, and the ability for player 2 to detect looking increases cooperation. We propose this model as a possible expla- nation for a number of interesting phenomena, and thereby derive novel predictions about these phenomena, including why we dislike “flip-flopping” politicians and respect principled people more gen- erally, why people cooperate intuitively, and why people feel disgust when considering taboo trade-offs, and why people fall in love. Cooperation occurs when we take on costs to help others. A key mechanism by which cooperation is sustained is reciprocity: individuals condition their behavior on others’ past actions [1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14]. In reality, we not only condition on others’ past actions, but also on the decision-making process that lead to cooperation: we place more trust in cooperators who do not take time to carefully weigh the costs of cooperation and who do not try to collect data on such costs before deciding whether to cooperate. For example, we are impressed by colleagues who agree to proofread a paper without thinking twice, and view with suspicion those who ask, “how long will it take?” before agreeing to attend a practice talk. Such considerations are left out of standard 2 HOUSE_OVERSIGHT_026522