Attention and Evidence

Abstract

In traditional models, agents have ample reason to gather cost‐free evidence. Rational agents expect evidence gathering to weakly improve the quality of their decisions and the accuracy of their credences. These results are linked to more general principles of reflection and the value of knowledge taken to characterize genuine learning experiences. For attentionally limited agents, I show that matters are more complex. Rational attentionally limited agents still cannot expect to make worse decisions or hold less accurate credences as a result of gathering some rather than no evidence. But they can expect to make worse decisions and hold less accurate credences as a result of gathering more rather than less evidence about the world. This result highlights a neglected distinction between two versions of the requirement to gather cost‐free evidence and a gap between evidence gathering and learning experiences.