The recent growth of interest in the application of mental workload techniques has coincided broadly with the increasing use of semi-automated human-machine systems in the workplace. These changes in work design have attracted concern over their potential impact on both the operator's role in system reliability and safety and personal health and well-being (Hockey, Briner, Tattersall, and Wiethoff, 1989; Sheridan, 1981; Wall and Jackson, 1995). Assessment of the mental demands of complex systems is a primary requirement for isolating design problems and for reducing the threat to the plant and to the human operator.
Typically, complex systems require the operator to cooperate with automatic computer systems in monitoring a large number of sources of information so as to detect critical system states and diagnose and rectify faults. In addition to responding routinely to events, operators may need to prepare plans for future action, predict impending system states, and perform other tasks using information acquired through monitoring and search. The work is characterized by considerable moment-to-moment variation in the level of mental activity required of the operator over a continuous period and by frequent shifts in the level of control and effort required to carry out task-relevant actions. Such variations are difficult to capture within the framework provided by traditional workload techniques, though portable psychophysiological recording methods now offer the opportunity to monitor changes nonintrusively during normal work activities (e.g., Kramer, 1991; Wierwille and Eggemeier, 1993).
In the present paper workload is operationalized, in the work of flight engineers, in terms of the level of control associated with different kinds of flight systems maintenance actions. The analysis is based on Rasmussen's (1981, 1983, 1986) hierarchy of cognitive control activity in fault diagnosis and management: skill-based, rule-based, and knowledge-based. We report a study of systematic changes in heart rate variability (HRV) as a function of frequent and unpredictable variations in control level across continuous 3-h periods of work. The study is presented, on one level, as a further validation of the use of frequency-domain HRV methods for the assessment of mental effort (Mulder, 1980), particularly in naturalistic field studies that permit no behavioral intervention. In addition, it provides convergent validity for the observational method adopted here for classifying and analyzing continuous process activity in field situations. In combination, the two techniques provide a model for the detailed, continuous analysis of complex work, in terms of the time course of individually referenced operational demands and the psychophysiological response to demands.
Workload and Control Level in Operational Contexts
Standard workload techniques are limited in their applicability to operational contexts. Moment-to-moment variations in demand may be expected to attract corresponding changes in effort on the part of the operator, at least in situations in which errors cannot be tolerated by the system. This requires a technique capable of tracking changes in both objective demand and the operator's response to these changes over the course of the task.
Recent developments in the use of subjective workload scales (e.g., SWAT and NASA-TLX) have proved sensitive to a broad range of between-task manipulations of demand, particularly in aviation (see Wierwille and Eggemeier, 1993); however, they are designed primarily for global posttask assessments. Subjective ratings and analog control responses have sometimes been used successfully to provide short-term assessments of load (e.g., Wierwille, 1988). Such techniques, though limited in diagnostic value, may be useful in detecting peak workload levels in work contexts in which frequent verbal or motor responses may be obtained from the subject. …