🤖 AI Summary
This paper investigates how rational inattention endogenously generates echo chambers in environments characterized by information overload and algorithmic personalization, where limited attention forces agents to strategically select information sources, thereby inducing selective exposure and opinion homogenization.
Method: We develop a unified game-theoretic and Bayesian learning model that jointly captures strategic source selection and belief updating under attention constraints.
Contribution/Results: We formally prove—without assuming preexisting biases or malicious filtering—that echo chambers emerge spontaneously solely from attention scarcity and dynamic source choice. We derive a sufficient condition for echo chamber formation and identify a critical threshold relationship between attention cost and platform personalization intensity. Our results provide a quantifiable theoretical benchmark for algorithmic governance and pinpoint precise intervention levers for mitigating polarization.
📝 Abstract
We propose a theory of echo chamber based on rational inattention, i.e., the ability to allocate limited attention capacities across information sources in a rational, flexible, manner. Such a premise has become increasingly relevant in today's digital age, as people are inundated with information on the one hand, but can selectively choose which information sources to visit using personalization technologies on the other hand. Since [Sunstein, 2007] and [Pariser, 2011], it has been long suspected that rational inattention could engender a selective exposure to content and a formation of homogeneous opinion clusters. This paper develops a novel model to formalize this idea.
