Learning to play Bayesian games

By Eddie Dekel, Drew Fudenberg, David K. Levine. 2004

246 citations recorded by the source corpus.

This paper discusses the implications of learning theory for the analysis of games with a move by Nature. One goal is to illuminate the issues that arise when modeling situations where players are learning about the distribution of Nature's move as well as learning about the opponents' strategies. A second goal is to argue that quite restrictive assumptions are necessary to justify the concept of Nash equilibrium without a common prior as a steady state of a learning process.

Subfields: Economics and Econometrics, Management Science and Operations Research, Safety Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Economics and Econometrics

In Management Science and Operations Research

In Safety Research

Citations in the corpus

Discovering Diverse Multi-Agent Strategic Behavior via Reward Randomization (2021, background): “Besides, RPG helps train adaptive policies against a set of opponents, which is related to Bayesian games (Dekel et al., 2004; Hartline et al., 2015).”
Opinions as Incentives (2009, background): “The posteriormi of player i is that , where2 2 2 2˜qFs ∼ N(rs (1 r)m , j ) r :p j /(j j )i 0 0 1 6 We acknowledge that learning justifications for equilibrium are more difficult than usual when players have heterogeneous priors (Dekel, Fudenberg, and Levine 2004).”
Autonomous agents modelling other agents: A comprehensive survey and open problems (2017, background): “) Subsequent works studied the impact of prior beliefs on equilibrium convergence and showed that if players have different prior beliefs, their play may converge to a subjective equilibrium which is not a Nash equilibrium [90,210].”
Learning and Equilibrium (2009, background): “In the setting of a static Bayesian game, Dekel et al. (2004) extended the definition of SCE to allow for these sorts of coarser maps from outcomes of the game to observations.”
Advancing Beyond Advances in Behavioral Economics (2006, background): “14 Eddie Dekel et al. (2004) provide a learning-theoretic analysis of how the outcomes of Bayesian games depend on what agents observe when the game is played.”
The Identification Power of Equilibrium in Simple Games (2008, background): “and Morris (2007) and Dekel, Fudenberg, and Levine (2004) .”
Rationalizability in infinite, dynamic games with incomplete information (2003, background): “11 Dekel et al. (2001) shows that the Bayes–Nash equilibrium concept is very hard to justify in terms of learning even for games with asymmetric information where the ex ante stage is real, but players have subjective heterogeneous priors on the state of Nature, which is drawn at random in each…”
A game-theoretic model and best-response learning method for ad hoc coordination in multiagent systems (2013, background): “Since then, there have been several works on learning in Bayesian games, e.g. [Jordan, 1991,Kalai and Lehrer, 1993,Dekel et al., 2004].”
Recency, records and recaps: learning and non-equilibrium behavior in a simple decision problem (2014, unknown): “…and best respond to this information, while allowing players to have incorrect beliefs provided those beliefs are consistent with their observations, so that players can only have wrong beliefs “off the equilibrium path” [Battigalli & Guatoli 1997, Dekel et al. 1999, 2004, Fudenberg & Levine 1993].”
Aggregation of Information and Beliefs: Asset Pricing Lessons from Prediction Markets (2009, unknown): “See, for instance, Forsythe et al. (1992) and Berg et al.”
Prediction, learning, and games (2006, background): “See also Jordan [169, 170], Dekel, Fudenberg, and Levine [87], Fudenberg and Levine [117, 119], and Nachbar [221].”
A Dual-Self Model of Impulse Control. (2006, unknown): “In Benabou and Tirole [2004], for example, the agent knows the distribution of costs but does not know his hyperbolic parameter � . Both Bodner and Prelec [2003] and Benabou and Tirole [2004] analyze Bayesian equilibria of their models, which raises the question of whether a plausible non-equilibriu”