Use this URL to cite or link to this record in EThOS: http://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.658518
Title: Distributed dynamics and learning in games
Author: Pradelski, Bary S. R.
ISNI:       0000 0004 5354 3535
Awarding Body: University of Oxford
Current Institution: University of Oxford
Date of Award: 2015
Availability of Full Text:
Access through EThOS:
Full text unavailable from EThOS. Restricted access.
Access through Institution:
Abstract:
In this thesis we study decentralized dynamics for non-cooperative and cooperative games. The dynamics are behaviorally motivated and assume that very little information is available about other players' preferences, actions, or payoffs. For example, this is the case in markets where exchanges are frequent and the sheer size of the market hinders participants from learning about others' preferences. We consider learning dynamics that are based on trial-and-error and aspiration-based heuristics. Players occasionally try to increase their performance given their current payoffs. If successful they stick to the new action, otherwise they revert to their old action. We also study a dynamic model of social influence based on findings in sociology and psychology that people have a propensity to conform to others' behavior irrespective of the payoff consequences. We analyze the dynamics with a particular focus on two questions: How long does it take to reach equilibrium and what are the stability and welfare properties of the equilibria that the process selects? These questions are at the core of understanding which equilibrium concepts are robust in environments where players have little information about the game and the high rationality assumptions of standard game theory are not very realistic. Methodologically, this thesis builds on game theoretic techniques and prominent solution concepts such as the Nash equilibrium for non-cooperative games and the core for cooperative games, as well as refinement concepts like stochastic stability. The proofs rely on mathematical techniques from random walk theory and integer programming.
Supervisor: Young, H. Peyton; Tarres, Pierre Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.658518  DOI: Not available
Keywords: Mathematics ; Game theory,economics,social and behavioral sciences (mathematics) ; Operations research,mathematical programming ; Social influence ; Economics ; Microeconomics ; game theory ; bounded rationality ; decentralized markets
Share: