The investigation of response time and behavior has a long tradition in cognitive psychology, particularly for non-strategic decision making. Recently, experimental economists have also studied response time in strategic interactions, but mostly for one-shot games. This is the first study to investigate the determinants of response time in a repeated game with fixed partner matching, specifically a constant-sum game with a unique mixed strategy Nash equilibrium. I specify and test hypotheses of the relationship between response time and the history of play, such as patterns in past joint actions and payoffs, the degree of randomization and the difference in expected payoffs of the available actions. The main finding is that actions that are consistent with the win-stay/lose-shift heuristic are markedly faster than inconsistent actions. This was found in virtually the whole population supporting previous findings based on action data only that this heuristic is widespread as an intuitive strategy in both humans and animals. Furthermore, I relate the magnitude of response time differences to the degree of regret and disappointment experienced (or avoided) in the prior round, calculated relative to the appropriate counterfactuals. The higher the experienced regret and disappointment from a loss in the prior round, the faster the subsequent response time when subjects shifted away from the previous action. Furthermore, disappointment had a stronger effect on response times than regret. These results highlight the important role of emotions arising from the history of game play, and how they may drive the implementation of strategic behavior.