In this paper, we introduce two new learning models: impulse-matching learning and action-sampling learning. These two models together with the models of self-tuning EWA and reinforcement learning are applied to 12 different 2 x 2 games and their results are compared with the results from...