1. Reinforcement learning with foregone payoff information in normal form games.
- Author
-
Funai, Naoki
- Subjects
- *
REINFORCEMENT learning , *NASH equilibrium , *GAMIFICATION , *GAMES - Abstract
• We investigate the reinforcement learning model of Erev and Roth with additional payoff information on unchosen actions. • We provide conditions under which the learning process almost surely converges to a unique regular quantal response equilibrium in normal form games. • Given some payoff normalisation, the convergence is obtained in any two-player game. • The same convergence results are obtained in an adaptive learning model which nests several other learning models under the linear choice rule of the reinforcement learning model. • Under the logit choice rule, the reinforcement learning model almost surely converges to a strict Nash equilibrium in some games. This paper studies the reinforcement learning of Erev and Roth with foregone payoff information in normal form games: players observe not only the realised payoffs but also foregone payoffs, the ones which they could have obtained if they had chosen the other actions. We provide conditions under which the reinforcement learning process almost surely converges to a regular quantal response equilibrium (Goeree et al. 2005). We also show that the reinforcement learning model and an adaptive learning model which nests experience-weighted attraction learning, payoff assessment learning and stochastic fictitious play learning models share the same asymptotic behaviour under the linear choice rule of the reinforcement learning model. In addition, we provide conditions under which the reinforcement learning process under the logit choice rule almost surely converges to a Nash equilibrium. [ABSTRACT FROM AUTHOR]
- Published
- 2022
- Full Text
- View/download PDF