Causality and self-signaling in economic games

Abstract

Our ability to cooperate is one of the cornerstones of our success as a species, and the story of how humans have been able to put aside immediate personal gain in favor of a longer view is widely studied. We add to this literature by exploring certain seemingly irrational behaviors observed in economic games. This is a set of behaviors that look like misunderstandings of causality, and which are irrational or must be the result of a misunderstanding on a classical economic reading. However, modes of cognition such as those reflected in self-signaling theory may serve to explain how the seemingly irrational might sometimes be quite sensible. We elicit these behaviors using real-time multiplayer economic games and suggest mechanisms whereby players may incorporate the value of receiving certain signals themselves into their utility calculations, thus making for rational behavior--and rational inference--in cases where it is not obviously so. In particular, we demonstrate order effects in economic games with no information flow and we show systematic biases in estimates of what one's partners have done versus the population. Both of these phenomena are consistent with a combination of self-signaling and a limit on the direction of inference in time.