Motivated Model Persuasion

Abstract

We theoretically study model persuasion in settings where motivated beliefs influence model selection and belief updating. Building on Schwartzstein and Sunderam (2021), we consider a sender proposing a model to interpret signals and a receiver choosing between a default model and the sender’s model. The receiver differs from a rational Bayesian receiver in the framework of model persuasion. We propose two different mechanisms through which a receiver holding motivated beliefs differs from a rational Bayesian receiver, yielding biased posteriors which are consistent with the motivated belief.