This paper focuses on emulating human behavioral patterns found in the play of the Stochastic Prisoner’s Dilemma, using a randomized, teacherless training framework applied to a Q-Learning algorithm. The stochastic setup, borrowed from literature, is simulated in the Axelrod Library and under the right incentives and parameters, this paper shows that the QLearner is capable of reproducing behaviors akin to pessimism/optimism usually found in human play. Under different parameters, the QLearner is shown to mimic how human agents update their beliefs about the future once the final form of the game gets realized.