Cost of Reasoning and Strategic Sophistication

Abstract

I designed an experiment to study the persistence of the prevailing levels of reasoning across games. Instead of directly comparing the k-level(s) of reasoning for each game, I used cognitive load to manipulate the strategic environment by imposing variations on the subject’s cost of reasoning and their first- and second-order beliefs. Subjects have systematic changes in k-level(s) of reasoning across games. That finding suggests that subjects are responsive to changes in the strategic environment. Changes in k-level(s) of reasoning are mostly consistent with the endogenous depth of reasoning model when subjects are more cognitively capable or facing less cognitively capable opponents. Subjects have cognitive bounds, but often choose a lower-type action due to their beliefs about their opponents. Finally, cognitive ability plays a significant role in subjects making strategic adjustments when facing different strategic environments.