A Neural Model of Stochastic Choice in a Mixed Strategy Game

31 Pages Posted: 26 Oct 2011 Last revised: 10 Apr 2013

See all articles by Ryan Webb

Ryan Webb

University of Toronto

Michael Dorris

Queen's University

Date Written: April 9, 2013


In strategic games with a unique mixed strategy equilibrium, players face both an incentive to best-respond to valuations and to act unpredictably. We developed a model of how neural circuitry represents a balance between these two incentives in the course of a decision. Choice is modelled as the result of the interaction between action value input from upstream brain areas and the noise inherent in neuronal networks: large differences in action valuations between options lead to reliable best-response choices whereas small differences result in a choice selection process dominated by noise. Action value input was measured in superior colliculus activity while monkeys played a saccade version of matching pennies. We found that model simulations based on these measures exhibit similar choice biases as found in behavioural data. Deviations from the mixed equilibrium strategy were predicted by the action value measurements within the model. This yields a neural choice mechanism that is capable of implementing both critical aspects of equilibrium formation in strategic games: best-response and stochastic behaviour.

Keywords: Neuroeconomics, Random-Utility, Quantal Response Equilibria, Game Theory, Strategic Decision Making

Suggested Citation

Webb, Ryan and Dorris, Michael, A Neural Model of Stochastic Choice in a Mixed Strategy Game (April 9, 2013). Available at SSRN: https://ssrn.com/abstract=1949732 or http://dx.doi.org/10.2139/ssrn.1949732

Ryan Webb (Contact Author)

University of Toronto ( email )

105 St. George Street
Toronto, Ontario M5S 3E6 M5S1S4

Michael Dorris

Queen's University ( email )

Kingston, Ontario K7L 3N6

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Abstract Views
PlumX Metrics