Moving Past Rational Behavior: A Game Theory Experiment

Wednesdays@NICO | 12:00-1:00 PM, March 2, 2016 | Chambers Hall, Lower Level

Julia Poncela-Casasnovas, Northwestern Institute on Complex Systems

Abstract

Both in real life situations and behavioral experiments alike, people often behave against game-theoretical predictions. However, little is known about whether or not individuals display consistency when facing various games, or, if they do, what would be a possible classification of these different-but-consistent behaviors. I present the results of a lab-in-the-field experiment, with individuals playing multiple games under different incentives, with different, anonymous partners. Then, by running a robust clustering algorithm on the data, I am able to classify the subjects in distinct, well-defined groups (–phenotypes–) according to their actions in the different scenarios presented to them. I identify these phenotypes as Trustful, Pessimist, Envious, Optimist, and Clueless. None of these phenotypes correspond to the expected behavior from the Nash equilibria, however, I am able to clearly infer their behavioral rules, and also to connect them with a measure of risk-aversion. My classification is applicable in contexts beyond social dilemmas, and opens the door for more realistic modeling of human behavior, along with practical applications: from business management and human resources, to global policy making.

Bio

I work with Professor Brian Uzzi at NICO, studying the emergence of scientific fields, collaboration dynamics at institutions, and how errors propagate in networks. I received my BS, MS, and PhD in Physics from the University of Zaragoza, where I studied how complex network topologies affect game theoretic outcomes. Since then I have expanded my research interests into the social sciences, studying how medical innovations are adopted and how an individual's network of contacts could help achieve weight-loss goals.

Related Info