Friends bolster health and happiness, with friend preferences directing us toward friends who can facilitate this. Intuition and research alike suggest people prefer friends to be kind and trustworthy and disfavor viciousness and indifference (and befriend the similar, familiar, nearby). Taking a target-specific approach, we predict and find people possess preferences not only for how friends should behave toward us, but—because our friends interact with others, and these interactions can affect us—people also possess nuanced, distinct preferences for how friends should behave toward others: (a) When targets of friends’ behavior are unspecified (reflecting previous work), preferences track how people want friends to behave toward us. In line with that work, (b) people want friends to be kinder and more trustworthy than not. But (c) people also want friends to be more prosocial toward us than toward others and (d) sometimes want friends to be more vicious than prosocial—toward our rivals.

Research on close relationships has tended to focus on the dyad (e.g., friends, romantic partners, rivals). Less attention has been paid to the myriad third parties who impact our social lives through their own relationships with our dyadic partners. What drives our feelings toward such third parties? A classic formalist theory, Balance Theory, suggests we like third parties who share our feelings toward our existing partners (and dislike those who do not) because of affective balance. Here, we propose a new embedded dyad framework which foregrounds the substantive indirect effects that third parties can have on our outcomes through their relationships with our partners. Consistent with the embedded dyad framework, we find that people like third parties who share our hatred for our rivals and our love for our friends (as predicted by both views); but we dislike those who share our love for our spouses (countering Balance Theory). Further supporting predictions uniquely derived from an embedded dyad framework, (a) greater perceived exclusivity in positive dyadic relationships (e.g., friendships) drives dislike toward third parties who share our love for our positive partners; (b) greater perceived welfare suppression by our negative partners (e.g., rivals) drives liking toward third parties who share our hatred of our negative partners. This framework thus critically extends cognitive consistency views by emphasizing the real costs and benefits of navigating dyadic relationships within larger social networks.