Synopsis: This page describes theories and models of human (individual and team) multitasking behavior.

Keywords: multitasking, timesharing, attention, multiple task, strategic workload management, cockpit task management, agenda management

Last update: 23 Jun 97

A task is a goal-directed activity performed by a human. Operators of complex systems, including aircraft, must generally pursue more than one goal at a time. This is called multitasking. Psychologists and human factors scientists have recognized multitask performance as key to the performance of complex systems and to help come to an understanding of the phenomenon, they have developed a number of theories and models of multitasking behavior.

There is a large literature on multiple-task performance (e.g., Damos, 1991). These studies often involve experiments in which a subject attempts to perform two abstract tasks simultaneously. For example, the subject may perform a pursuit tracking task (in which he/she attempts to track a moving object on a computer screen with a joystick-controlled cursor) while at the same time add sets of numbers presented to him/her aurally. Task difficulty and other factors are altered to determine the effects on tracking accuracy, arithmetic response time and accuracy, and other performance measures.

A major finding resulting from such studies is called multiple resource theory (MRT, Wickens, 1992). According to MRT, human mental "capacity" can be viewed as a collection of limited, differentiated resources that must be allocated among competing tasks. If a task receives a full allocation of required resources, performance (that is, speed and accuracy) is good. As needed resources are withdrawn from a task, performance deteriorates.

The concept of dual-task behavior has been extended to multitask behavior. Many multitask theories of human behavior are based on a multitasking computer operating system metaphor (e.g., Johannsen & Rouse, 1979), where mental resources are likened to computer memory and processor time, tasks are likened to processes, and some sort of executive 'routine' allocates resources to tasks. Theories and models based on engineering methods have been developed to describe these behaviors (Pattipati & Kleinman, 1991), even connectionist models have been proposed (Detweiler & Schneider, 1991).

The concept of mental workload has emerged from this line of research. Workload is loosely defined as the resource requirements imposed by a set of concurrent tasks. There is little agreement about the exact definition of workload or about how to measure it, but experiments typically show that there is an optimal level of workload, below which and above which individual and composite task performance degrades (Wickens, 1992). Methods to measure workload have been introduced (e.g., summarized in Gawron, 1991) to ascertain the loading imposed by a set of tasks and to infer performance based on that loading.

Using workload as a working assumption, several groups of researchers have posited theories of strategic workload management as bases for explaining multitask behavior (e.g., Hart, 1985; Raby & Wickens, 1994). According to these theories, humans attend to tasks in such a way as to balance workload and therefore achieve acceptable levels of performance. In some cases, normative scheduling models from the industrial engineering and operations research literature have been applied (Moray et al, 1991). The conclusion is that humans do not schedule tasks optimally with respect to most measures.

Another line of research has focussed on task management, the management of tasks rather than the management of workload (Adams, Tenney, & Pew, 1991; Rogers, 1996; Schutte and Trujillo, 1996). According to these theories, human operators controlling complex systems (typically aircraft) engage in a process called task management (TM). TM is a high-level mental process which continuously prioritizes concurrent tasks and allocates resources to them. The research described here is along these lines.

 References (Also, see Jennifer Wilson's Agenda Management Bibliography.)