Queueing Network-ACTR Modeling of Concurrent Tasks Involving Multiple Controlled Processes

Abstract

An important research question of human performance modeling is how to model concurrent task performance and mental workload. In this paper, we introduce a new method to model concurrent tasks involving multiple controlled processes in need of dedicated cognitive resources, which previous methods have difficulties to model. The key concept is a filtering discipline implemented in Queueing Network-ACTR cognitive architecture that allows cognitive resources to be exclusively occupied by one of the concurrent tasks when necessary, instead of switching between the tasks as frequently as possible. In the simulations of dual-tasks involving diagnostic decision making and patient status tracking, we found that the new discipline is necessary to model human performance and mental workload. Implications and practical applications of Queueing Network-ACTR in system evaluation and design are discussed.