Learning when to switch tasks in a dynamic multitasking environment

When performing concurrent tasks of non-trivial durations, people balance task processing by interleaving segments of one task with another. In this paper we describe a cognitive model of multitasking and interleaving for such extended concurrent tasks, specifically focusing on learning when to switch from one task to another. To explore this issue, we performed an empirical study of “discrete driving” in which participants used a keyboard to steer a vehicle while entering navigation information as a secondary task. We then compare and contrast two models of this task that learn taskswitching behavior based on the (changing) characteristics of the discrete driving task.