“Once? No. Twenty times? Sure!” Uncertainty and precommitment in Social Dilemmas

In many real-world social dilemmas, interdependent actors must decide whether to invest in preventive measures against uncertain losses. For example, when consumers decide whether to get the flu shot each year, the likelihood of getting sick depends not only on the actions of the individual consumer, but also on the actions of those around her. If everyone invests in protective measures, there is herd immunity (and some people can free-ride without getting the flu shot). Uncertainty about losses lowers cooperation between individuals quite substantially. For example, while cooperation rates in a loss-framed, repeated prisoner’s dilemma (PD) may hover around 60-75%, uncertainty lowers the cooperation rate to around 25-40% (Gong, Baron, & Kunreuther, 2009; Kunreuther, Silvasi, Bradlow, & Small, 2009). Thus, cooperation rates under uncertainty are up to 3 times lower. Because individuals are less likely to cooperate under uncertainty, and because so many real-world situations of interest resemble interdependent-security (IDS) situations (with uncertain, negative prospects), behavioral research should explore ways to increase cooperation under uncertainty. One possibility comes from Slovic, Fischhoff, and Lichtenstein (1978) work on single vs aggregate probabilities: consumers were not concerned with the likelihood of being killed in a single ride when presented with the probability of around 0.00000025 of being killed and about 0.00001 of becoming injured, whereas they were more likely to use their seatbelt when the aggregated probabilities for a lifetime were presented (about 0.01 and 0.33 for death and injury, respectively).