Expected Motion in 2x2 Symmetric Games Played by Reinforcement Learners
Reinforcement learners tend to repeat actions that led to satisfactory outcomes in the past, and avoid choices that resulted in unsatisfactory experiences. This behavior is one of the most widespread adaptation mechanisms in nature. This Demonstration shows the expected motion of a system where two players using the Bush–Mosteller reinforcement learning algorithm play a symmetric game. Mathematical analyses conducted by the contributors of this Demonstration show that the expected motion displayed in the figure is especially relevant to characterize the transient dynamics of the system, particularly with small learning rates, but, on the other hand, this expected motion can be misleading when studying the asymptotic behavior of the model. Further information at http://luis.izqui.org and http://segis.izqui.org.