Homophily, the tendency to interact with others of similar type, is widely observed in nature. Sex- and age-related homophily, for example, shapes the formation of clusters of preferred companionships in zebras1, dolphins2, and predicts both the quantity and quality of many primate interactions3, 4. Meerkats tend to assortatively associate with other group members of similar attributes in dominance and foraging networks5. And across many dimensions of phenotypes, humans exhibit high levels of homophily in social tie formation6, 7, 8. In fact, recent evidence suggests that humans may even exhibit genotypic homophily, meaning that individuals with a certain genotype are more likely to be friends with others of the same genotype9. Heterophily, the tendency to interact with others of different type, also exists in nature at both the cellular10, 11, 12 and organismic levels. For example, research on collaboration networks suggests that people are likely to form heterophilic task-related ties with those who are complementary to their own skill sets8. Analogously, hunter-gatherer life is characterised by long-term imbalances in productivity and consumption, and by the division of labour13; hence, one might possibly expect that social interactions would, at least in part, be heterophilic, offering complementary advantages to interacting parties; but they are not7.