Biases in Belief Updating Within and Across Domains

Why do people sometimes overreact to new information and other times underreact? We develop a model in which the strength of a signal—how much one should update their beliefs with new information—depends on multiple features of the information environment. Limited attention to these features leads to misperceptions of signal strength: people approach a problem with an experience-based prior, which they adjust only partially based on how much attention they pay to different features. This mechanism explains a wide range of belief-updating patterns. Insensitivity to a single feature generates underreaction to strong and overreaction to weak signals, and more neglected features amplify this. Insensitivity to multiple features can instead break that pattern: insensitivity to one feature can generate excess sensitivity with respect to another, leading to overreaction to both weak and strong signals. A series of experiments provides support for the model and its underlying mechanism.