Information Costs and Sequential Information Sampling

Abstract

We propose a new approach to modeling the cost of information structures in rational inattention problems, the cost functions. These cost functions have two properties that we view as desirable: they summarize the results of a sequential evidence accumulation problem, and they capture notions of distance. The first of these properties is connected to an extensive literature in psychology and neuroscience, and the second ensures that neighborhood-based cost functions, unlike mutual information, make accurate predictions about behavior in perceptual experiments. We compare the implications of our neighborhood-based cost functions with those of a mutual-information cost function in a series of applications: security design, global games, modeling perceptual judgments, and a linear-quadratic-Gaussian tracking problem.