Human behavior is embedded in social networks. Certain characteristics of the positions that people occupy within these networks appear to be stable within individuals. Such traits likely stem in part from individual differences in how people tend to think and behave, which may be driven by individual differences in the neuroanatomy supporting socio-affective processing. To investigate this possibility, we reconstructed the full social networks of three graduate student cohorts (N = 275; N = 279; N = 285), a subset of whom (N = 112) underwent diffusion magnetic resonance imaging. Although no single tract in isolation appears to be necessary or sufficient to predict social network characteristics, distributed patterns of white matter microstructural integrity in brain networks supporting social and affective processing predict eigenvector centrality (how well-connected someone is to well-connected others) and brokerage (how much one connects otherwise unconnected others). Thus, where individuals sit in their real-world social networks is reflected in their structural brain networks. More broadly, these results suggest that the application of data-driven methods to neuroimaging data can be a promising approach to investigate how brains shape and are shaped by individuals’ positions in their real-world social networks.

Trustworthiness is a fundamental dimension underlying trait impressions of individual faces, and these impressions predict real-world social consequences. Building on ensemble coding research from the vision sciences, we explored to what extent statistical information about trustworthiness is gleaned from rapid exposure to crowds of faces. We showed that with half-second exposures to sets of eight faces, perceivers are sensitive to the set’s average level of trustworthiness (Study 1). Moreover, this group-level sensitivity biases individual group member evaluations (Study 2), as well as downstream social behavior related to those evaluations (Study 3), toward the mean of the group. Together, the findings add to a growing body of “people perception” research and show that even high-level social characteristics such as personality traits may be spontaneously gleaned from rapid exposure to crowds of faces.

Knowledge of someone’s friendships can powerfully impact how one interacts with them. Past research suggests that information about others' real-world social network positions–e.g., how well-connected they are (centrality), 'degrees of separation’ (relative social distance)–is spontaneously encoded when encountering familiar individuals. However, many types of information covary with where someone sits in a social network. For instance, strangers’ face-based trait impressions are associated with their social network centrality, and social distance and centrality are inherently intertwined with familiarity, interpersonal similarity, and memories. To disentangle the encoding of social network position from other social information, participants learned a novel social network in which social network position was decoupled from other factors, then saw each person’s image during functional magnetic resonance imaging scanning. Using representational similarity analysis, we found that social network centrality was robustly encoded in regions associated with visual attention and mentalizing. Thus, even when considering a social network in which one is not included and where centrality was unlinked from perceptual and experience-based features to which it is inextricably tied in naturalistic contexts, the brain encodes information about others’ importance in that network, likely shaping future perceptions of and interactions with those individuals.