We study the evolution of behavior under reinforcement learning in a Prisoner's Dilemma
where agents interact in a regular network and can learn about whether they play one-shot …