טכניון מכון טכנולוגי לישראל
הטכניון מכון טכנולוגי לישראל - בית הספר ללימודי מוסמכים  
Ph.D Thesis
Ph.D StudentYakobi Ofir
SubjectThe Determinants of Association and Dissociation
Between Measures of Workload
DepartmentDepartment of Industrial Engineering and Management
Supervisor Professor Emeritus Daniel Gopher
Full Thesis textFull thesis text - English Version


Abstract

Researchers in cognitive psychology and Human Factors are frequently interested in the capacity and limitations of the human mind. Several theoretical frameworks have been suggested over the years to account for the empirical findings related to human performance and workload, leading to the development of various tools for measuring mental workload. These include direct performance measures, subjective reports, neurophysiological indices, and secondary-task procedures. Empirical data show that these measures, although generally in agreement, demonstrate dissociation in some instances. Surprisingly, the factors contributing to the association and dissociation of measures, as well as their theoretical foundations, are almost not discussed in the literature.

In this work, I present a systematic investigation of the determinants of association and dissociation between subjective and objective estimates of workload. Specifically, I review theories of resources, metacognition and dual-process models, and link them to the operationalization of workload. In four experiments, an oddball task was performed individually or concurrently with a tracking task while the basic task-characteristics and goal driven efforts were manipulated. Subjective workload estimates, the P3 event-related potential and performance in the two tasks were collected. Overall, workload measures responded differently to each manipulation in a distinct matter. When tasks were altered such that effortful, controlled processing demands were increased, subjective estimates, P3 amplitude and performance tended to associate. When performance was data-limited (in contrast to resource limited), workload measures showed an inconsistent pattern of associations. These findings suggest that subjective workload measures are sensitive to the general workload experienced by the performer, a product of both performance monitoring, and control of efforts. In contrast, performance measures are task- and context-specific. They respond to task-characteristics, which in turn predict when competition for scarce, shared resources is expected as suggested by Multiple Resources theory, and to the distinction between processing qualities as postulated by dual-process theories.

When these efforts stem from instructions (i.e., instructions to emphasize a task in dual-task settings), they will be reflected in performance and in P3 amplitude, but not in subjective workload. This is because the subjective measure, a conscious metacognitive measure, is sensitive to both monitoring and control of performance and efforts and by that reflects a general measure of workload. However, when these efforts resulted in observable performance changes, subjective workload, P3 amplitude and performance associated.

Workload measures are often employed interchangeably in the literature and by practitioners, regardless to the context and the tasks that are at the heart of the investigation. The present work emphasizes the importance of theory driven considerations when choosing workload measures, and demonstrates how and when different workload measures are expected to yield different results.