Convergence of dyadic similarity ratings predicts similarity in neural representations of others within social networks

Presenter(s): Youri Benadjaoud—Human Physiology/Psychology

Faculty Mentor(s): Taylor Guthrie, Rob Chavez

Session: Prerecorded Poster Presentation

A history of classic research in social psychology has demonstrated that human social groups are highly homophilous- people tend to associate with others similar to themselves . More recently, researchers showed that the brain shows similar effects of homophily, with close individuals showing greater neural response similarity to naturalistic stimuli than unfamiliar individuals (Parkinson et al ., 2018) . It is an open question, however, whether a similar degree of neural homophily exists when close individuals think of other specific members of their social group . The current study investigated this question by recruiting multiple social network groups that consisted of several close-knit individuals . Using a round robin fMRI design, individuals completed a standard self/other trait judgement task in which each participant was both the perceiver and a target . Similarity among dyadic pairs were calculated within multivoxel response patterns to each other member of their group . Using the correlation distance between multivoxel pattern response vectors combined with euclidean distance calculations between perceiver ratings of similarity with the target, we fit a multilevel linear effects model that predicted neural similarity from the convergence of dyadic similarity ratings. Our results indicate that the degree of similarity between multivoxel response patterns while individuals were rating the same target were significantly predicted by the the degree to which the perceivers agreed on how similar they were with the target . These findings suggest that people who agree on how similar a person is to themselves tend to have greater similarity in neural representations of that particular other .

Leave a Reply

Your email address will not be published. Required fields are marked *