BOUNDRAT

“Bounded rationality” refers to the idea that people typically do not perfectly optimize, but choose sensible heuristics that are easy to implement and tend to perform well enough. A growing body of literature in economics and psychology documents that different situations evoke different heuristics, and so we propose a series of experiments to study the forms of bounded rationality that are relevant in diverse real-world situations. Our experiments are guided by theory, but inspired by everyday events. Our first experiment aims to provide a unifying explanation for contradictory findings concerning biases in choice over multi-attribute goods (e.g. goods defined by price and quality). Our second experiment seeks to determine if, and why, people tend to update beliefs differently following rare (as opposed to common) events, e.g. natural disasters. Our third experiment investigates sender-receiver games in which the sender has an incentive to lie in order to appear to be the "good" type whereas the receiver simply wishes to make an accurate judgement. For such games, we also propose a model of behavior based on limited strategic sophistication. Our fourth experiment leverages a recently introduced theoretical framework to empirically document the relationship between a player's role in a game (their payoffs and action-space) and how "noisy" their behavior is. The results will inform models relating complexity and noise in games.