Belief patterns with information processing

TL;DR

This paper introduces a model of decision-making that explains belief polarization and biases as rational responses to costs in information processing, reconciling observed biases with Bayesian inference.

Contribution

It presents a novel framework where belief biases emerge naturally from rational, cost-aware information acquisition strategies.

Findings

Belief polarization can occur even with shared evidence.

Biases like disconfirmation and confirmation are compatible with Bayesian reasoning.

Costly information processing explains belief updating phenomena.

Abstract

This paper presents a model of costly information acquisition where decision-makers can choose whether to elaborate information superficially or precisely. The former action is costless, while the latter entails a processing cost. Within this framework, decision-makers' beliefs may polarize even after they have access to the same evidence. From the perspective of a Bayesian observer who neglects information processing constraints, the decision-makers' optimal behavior and belief updating may appear consistent with biases such as disconfirmation, underreaction to information, and confirmation bias. However, these phenomena emerge naturally within the model and are fully compatible with standard Bayesian inference and rational decision-making when accounting for the costs of information acquisition.