Salvucci, Taatgen, and Borst describe an ongoing effort to create a computational model that accurately emulates how people handle multitasking and task switching.
The adaptive control of thought--rational (ACT-R) model is based on a theory about cognitive resource sharing between tasks. The primary resources are declarative memory and procedural problem rehearsal. The model is tested against human behavior in the areas of concurrent multitasking--performing several things at once with attention given to each task in short bursts--and sequential multitasking--performing longer tasks with less frequent task switching. The primary measure of “fit”--between the model and observed human behavior--concerns the delays encountered during task switching. The paper mentions that another plausible measure might focus on error rates, although it does not explore the suggestion in depth.
The value of this work is twofold: first, it confirms the importance of cognitive resource competition as an important limiting factor in human performance during multitasking; second, the model has some predictive power about when, how, and if task interruptions should be allowed to occur. ACT-R is not the final word on multitasking, but it is a useful contribution to the growing body of work on task interruption and resumption. This paper will be of interest to designers and researchers in the field of human factors.