Limiting Dynamics for Q-Learning with Memory One in Symmetric Two-Player, Two-Action Games

Figure 3

All possible edges in the BRN of the prisoner’s dilemma. See Table 2 for conditions for the existence of edges.