This article studies defense strategies for protecting objects of interest when Attackers exhibit behavioral biases commonly documented in behavioral economics. To this end, I conduct a lab experiment in which participants play the Patrolling Game to capture the main trade-offs in the strategic interaction between Patroller and Attacker. The data, together with structural analysis, allow me to determine the extent to which Attackers exhibit behavioral biases and to test the effectiveness of differing defense strategies. Three key findings emerge: first, defense strategies exploiting the estimated irrationality of the Attackers improve patrolling effectiveness by up to 10%; second, such strategies remain optimal even when the Attacker is perfectly rational; and third, participants tasked with designing defense strategies fail to adopt such strategies even after ample experience.

This paper studies coordination when people do not share common meanings. We conduct a laboratory experiment that varies the extent of shared meaning and the information participants have about their partners' actions. Three main findings emerge. First, without shared meanings or observable actions, coordination is no better than under random choice, whereas either shared meanings or observable actions substantially improve coordination. Second, poor coordination is associated with participants' tendency to pursue role-dependent equilibria, which require asymmetric roles to be assigned without prior agreement. Third, information acquired through a shared history facilitates substantially better coordination than the same information provided without such a history. These findings illustrate both why coordination can fail in the absence of shared meanings and how shared experience can help overcome this problem.

Studies on commitment products report high failure rates. These rates are concerning as failures leave individuals worse off than if they had not taken the contract. We explore whether some failures are planned, with individuals expecting future uncertainty to make it sometimes optimal to fail. We term this planning for the possibility of failure, which differs from failing to plan (mistakes made when taking contracts). We run a new controlled laboratory experiment to identify planning patterns. According to our estimates, more than a third of total commitment take-up reflects them. This result suggests that planning for the possibility of failure is widespread, so high failure rates do not necessarily stem from mistakes, and they alone cannot undermine the use of commitment contracts.