Not just social networks: How people infer relations from mutual connections.

Abstract

People can infer relationships from incomplete information about social networks. We examined whether these inferences depend on domain-specific knowledge about social relationships or instead depend on domain-general statistical reasoning. In five preregistered experiments, participants (total N = 1,424) saw two target entities and their connections to others in social, semisocial, and nonsocial networks. In Experiments 1 and 2, participants made similar judgments across social and nonsocial networks: with greater proportion of mutual connections and number of connections, the two entities were judged as more likely to be connected to each other. These findings support the domain-general account. The next experiments provided further support for this account, while also investigating the question of whether people use mutual connections to infer the broader structure of networks. In Experiments 3 and 4, participants were asked whether entities connected to both targets were connected to each other, and judgments were hardly affected by network information. In Experiment 5, participants judged connections were more likely when entities were connected to both targets rather than when they were connected to only one. Overall, the findings support the domain-general account of network inferences and further suggest that participants' inferences primarily concerned target entities and not the broader structure of the network.