A Memory Model of Belief Formation

Co-authors: Maxim Bakhtin and Muriel Niederle

Abstract
The agent in our model retrieves memories and combines them with the prior to form a belief. The agent is fully Bayesian and rational but faces a constraint on memory retrieval—she can only sample observations one at a time instead of retrieving all of them simultaneously. Retrieval is primarily random, but the agent can partially target retrieval using an index. The index splits the database of memories into two (or more) groups based on the values of one (or more) attribute. The agent chooses which indexed group to sample in each period to ensure that her beliefs are as accurate as possible. We show that the agent will generically oversample one group and characterize three forces that determine which group the agent samples more intensely. We then show that oversampling translates directly into belief distortion. We use this insight to explain well-known biases in beliefs across individuals, such as the “depression babies” effect, rational stereotypes, and the dependence of beliefs on the history of previously encountered problems.