Belief revision in a micro-social network: Modeling sensitivity to statistical dependencies in social learning

AbstractIn both professional domains and everyday life, people must integrate their own experience with reports from social network peers to form and update their beliefs. It is therefore important to understand to what extent people accommodate the statistical dependencies that give rise to correlated belief reports in social networks. We investigate adults’ ability to integrate social evidence appropriately in a political scenario, varying the dependence between the sources of network peers’ beliefs. Using a novel interface that allows participants to express their probabilistic beliefs visually, we compare participants against a normative Bayesian standard. We find that they distinguish the value of evidence from dependent versus independent sources, but that they also treated social sources as substantially weaker evidence than direct experience. The value of our elicitation methodology and the implications of our results for modeling human-like belief revision in social networks are discussed.

Return to previous page