Social avoidance can be quantified as navigation in abstract social space

We navigate social relationships daily, making decisions that can change our affiliation and power relations with others. People high in social avoidance report perceiving little affiliation and power in their social lives. Do they also make low affiliation and low power interaction choices in actual social interactions? We hypothesized that social avoidance can be quantified as navigation in an abstract social space framed by power and affiliation. To test this, we recruited two large samples of online participants (Initial sample n = 579, Validation sample n = 255) to complete a naturalistic social interaction game where they form relationships with fictional characters, and a battery of questionnaires. Factor analysis of the questionnaires revealed a social avoidance factor that related to a low affiliation and low power interaction style, which was reflected in large social distance between the participants and characters. This distance, in turn, was related to smaller and less complex real-world social networks-suggesting that this abstract behavioral geometry reflects real-life behavioral tendencies. Language analysis of post-task character descriptions found semantic representations that mirrored the relationships formed in the task, with social avoidance relating to more negative impressions. This approach suggests that social avoidance can be thought of as an abstract, two-dimensional navigational strategy, potentially leading to effective strategies for social skills training and therapy.