Cultural evolutionary processes are often hypothesized to reduce behavioral variation within groups and stabilize variation between groups. This idea has been at the cen- ter of a longstanding debate about how culture affects the evolution of human social behavior. More recently, however, the idea has also figured repeatedly in development programs intended to promote improvements in human health and well-being. Female genital cutting is the most prominent example. Cutting affects millions of girls and women. It can lead to serious lifelong health problems, and governments and development agencies spend considerable resources promoting the abandonment of cutting. Programs promoting abandonment often assume that cutting persists because cultural evolutionary processes have trapped cutting societies in a self-reinforcing cutting equilibrium. If so, individual decision makers cannot afford to unilaterally deviate from the local cutting norm. An outside agency, however, with the right intervention, could po- tentially recruit the cultural evolutionary forces that currently favor cutting so that they favor abandonment. Although this view of female genital cutting has been extremely influential, it has not been rigorously tested. In a first study, we do so with data we collected using novel methods in 45 communities in Sudan. In contrast to the prevailing view, we find tremendous variation in attitudes and behavior at very local scales. In effect, cutting and non-cutting families live door-to-door, which is inconsistent with the notion that cutting persists because of local norms. In a second study, we developed interventions that exploit extreme local heterogeneity in attitudes about cutting. Specifically, we produced various telenovela-style movies that depict the members of an extended family as they confront each other with divergent views about whether the family should continue cutting. The movies dramatize discord within a family, and as a result they focus on heterogeneity that is as local as possible. The movies serve as treatments in two separate randomized controlled experiments in 127 communities in Sudan, and in our talk we will present results from these two experiments.