We studied collaborative skill acquisition in a dynamic setting with the game Co-op Space Fortress. While gaining expertise, the majority of subjects became increasingly consistent in the role they adopted without being able to communicate. Moreover, they acted in anticipation of the future task state. We constructed a collaborative skill acquisition model in the cognitive architecture ACT-R that reproduced subject skill acquisition trajectory. It modeled role adoption through reinforcement learning and predictive processes through motion extrapolation and learned relevant control parameters using both a reinforcement learning procedure and a new to ACT-R supervised learning procedure. This is the first integrated cognitive model of collaborative skill acquisition and, as such, gives us valuable insights into the multiple cognitive processes that are involved in learning to collaborate.

Game playing is an excellent domain for researching interactive behaviors because any time the outcomes of the interactions between people are associated with payoffs the situation can be cast as a game. Because it is usually possible to use game theory (von Neumann & Morgenstern, 1944) to calculate the optimal strategy, game theory has often been used as a framework for understanding game-playing behavior in terms of optimal and sub-optimal playing. That is, players who do not play according to the optimal game theory strategy are understood in terms of how they deviate from it. In this chapter we explore whether or not this is the right approach for understanding human game-playing behavior, and present a different perspective, based on cognitive modeling.