People tend to see themselves as similar to their ingroup, and people often accomplish similarity with others by projecting their self-beliefs onto their perceptions of others. However, existing research on self-anchoring has not considered the within-person cognitive mechanisms facilitating this process. The current study aims to establish the similarity-based (i.e., if I am outgoing, my group ought to be characteristic of semantically similar traits such as sociable and fun) and contrastive (i.e., what is characteristic of my ingroup in contrast to a given outgroup) mechanisms by which people's self-evaluations on traits generalize to ingroup evaluations. Across three studies using minimal groups (N = 61), university groups (N = 283), and racial groups (N = 265), we find that people use semantic similarity among traits to infer the extent to which traits ought to be characteristic of their group if related traits are characteristic of themselves. We further find that this tendency is primarily driven by a motivation to achieve similarity with the ingroup rather than dissimilarity from the outgroup. However, in the racial context, racial minority participants contrasting against the racial majority were driven moreso to achieve dissimilarity from the majority outgroup. We fit a computational model measuring the extent to which people convert self-beliefs into ingroup-beliefs prior to generalization, and find that this tendency was weaker when people contrasted their ingroup against an outgroup that they felt more positively about (i.e., the higher status university in Study 2 and the fellow minority racial group is Study 3), reflecting that self-anchoring may be more pronounced when contrasting against majority or more disliked outgroups. In fact, this projection rate was correlated with self-reported intergroup bias in studies 2 and 3 and social identification in all three studies, reflecting that the extent to which individuals generalize about their groups based on themselves may depend on how biased and affectively attached they are to their social groups. Findings reflect that how people generalize from the self to the group may enact similarity-based classification processes that are amplified under particular intergroup contexts.

A core principle of deliberative democracy posits that information exchange enhances the accuracy of group beliefs. However, in the face of unprecedented access to both first-hand empirical information and second-hand estimates from social networks, partisans often disagree on fundamental facts supported by data. In this dissertation, I integrate research on sampling models and motivated reasoning to examine the mechanisms driving partisan disagreements concerning gun control policies and their impact on the wisdom of partisan crowds. Across two studies, partisans learned about the impact of a policy increasing access to guns on subsequent crime rates, with manipulated access to first-hand empirical information, second-hand social estimates, or both. Our findings show that collective error reduces when individuals sample empirical data (Studies 1-2) and further decreases when they consider the average estimates of others (Study 1). The wisdom of crowds is also enhanced when partisans learn from viewing the estimates of fellow partisans, even when they have the choice to decide from which communities to sample (i.e., from Democrats or Republicans). However, collective accuracy was attenuated when partisans sampled social information due to biases in their sampling behavior (e.g., prioritizing in-group members) and the systematic bias of estimates along party lines (Study 2). These findings emphasize the importance of encouraging individuals to diversify their information sources and expand their social networks to include a wider range of perspectives. Furthermore, they reveal the boundary conditions of partisan social influence on the wisdom of crowds, indicating that while social influence can increase collective accuracy to a certain degree, it may also introduce systematic partisan biases that amplify divides, especially when social estimates propagate across networks and become increasingly detached from empirical evidence. In summary, this dissertation highlights the potential benefits and limitations of social influence in shaping collective judgments and offers valuable insights into how individuals gather information and update their beliefs. Ultimately, we hope this research can inform interventions aimed at fostering informed decision-making, bridging ideological divides, and paving the way for collaborative problem-solving in the face of persistent partisan motivations.

Self-disclosure is a powerful way for people to connect, yet few studies have examined how disclosers can build a sense of shared meaning with listeners. We investigated possible neural, linguistic, and vocal mechanisms through which listeners may come to understand disclosers more intimately. We found that across numerous brain regions, greater neural asynchrony between speakers and listeners was associated with greater shared meaning-making. We also found that people who felt more connected at baseline had greater neural asynchrony, and that asynchrony partially explained the positive relationship between baseline connection and shared meaning-making. In addition, we found that neural asynchrony during key moments of speech may facilitate shared meaning-making. In particular, greater neural asynchrony when speakers (1) talked more slowly and (2) paused more was associated with greater shared meaning. These findings shed light on key processes that enable successful communication, and contribute to a more nuanced theoretical framework for how people build shared understanding through self-disclosure.