Humans routinely perform multiple tasks simultaneously. Understanding the capabilities and limits of human multitasking not only helps design efficient devices to enhance human performance, but also helps uncover the nature of human cognition. Research on multitasking is difficult because multitasking performance is inevitably influenced by many factors such as a personâs perceptual, cognitive and motor capabilities, a as well as the strategies adopted for managing the conflicts among tasks. Decades of experimental-psychology research has identified many of the invariable factors that influence multitasking, but it is cognitive modeling that shows the potential in integrating the factors and making quantitative predictions about the performance.

This paper reviews the results of past research on multitasking performance, and stresses the increasingly important role that cognitive modeling has played in this research endeavor. The paper discusses multiple resource theory in detail, which is the current predominant theoretical framework in psychology that incorporates various sensorimotor and cognitive factors. Cognitive architectures (primarily EPIC and ACT-R) are shown here to appropriately implement the majority of such factors in the form of computational simulation. The benefits of this integrated, computational approach is revealed in modeling studies of PRP (psychological refractory period) tasks.

The paper also discusses the influence of task strategies on multitasking performance. This type of exploration is uniquely enabled by cognitive modeling, because the production system adopted by most cognitive architectures provides a means to formally express task strategies. The research on task strategies has shed light on many cognitive functions such as executive processing and human adaptation to a task environment.

Studies on driving are discussed because driving is a major application domain of multitasking. Several driving models are reviewed and their advantages and drawbacks are analyzed. The paper concludes by summarizing the advantages of cognitive modeling over traditional experimental approach and suggesting future research directions for cognitive modeling of multitask performance.