Toward an ACT-R General Executive for Human Multitasking

While cognitive models of complex tasks have begun to incorporate increasingly sophisticated models of human multitasking, most models have utilized customized executives (Kieras et al., 2000) that fine-tune specific multitasking mechanisms for particular applications. This paper proposes a general executive for multitasking that facilitates the integration of separate task models and subsequent prediction of the effects of multitasking and task interaction. Developed in the ACT-R cognitive architecture, the general executive is specified as a new goal module that orders goals by urgency and their own requested running times. The paper demonstrates the predictive power of the general executive in the driving domain by integrating separate models of control and monitoring and predicting drivers’ gaze distributions across various regions of the driving environment.