Ben B. Morgan

Effects of Temporal Stressors on Vigilance and Information Processing

The data of two control and ten experimental studies of the effects of continuous work and sleep loss on sustained multiple-task performance, representing more than 9 man-years of synthetic work by 89 different subjects, were reanalyzed to permit comparisons of the performances of three watchkeeping tasks and two active tasks that are time-shared in a multiple-task performance battery (MTPB). Specific comparisons were made of the two control groups, of the effects of 48 hours of continuous work and sleep loss with and without the employment of pulse rate and EEG-theta bio feedback and autoregulation, of the effects of the duration of continuous work and sleep loss (36, 44, or 48 hours), of the effects of the duration of rest and recovery following 36 hours of continuous work (12, 6, 4, 3, and 2 hours), and of circadian rhythm and 36-hour continuous work interactions. Eight specific conclusions are reached, the most general of which is that the relevance of typical laboratory research with single-task watchkeeping tests, including the capability of its findings being generalized to, and implemented in, practical situations involving monitoring performances within operational manmachine systems is seriously questioned, if not compromised, by the findings.

The Range and Consistency of Individual Differences in Continuous Work

Two five-man teams were trained to asymptotic levels of synthetic-work performance on a multiple-task performance battery (MTPB) which required subjects to time share five individual-performance tasks and one group-performance task. Each team was then exposed four times to continuous-work and sleep-loss stresses in a repeated-measures design. Performance levels were measured in terms of 13 individual-task measures and a general mean-percentage-of-baseline measure. Individual performance decrements were found to range from 0 to 40% below prestress levels. The findings also indicated that subjects responded consistently across the four exposures to continuous work and sleep loss. Furthermore, the subjects experienced a general performance decrement across all tasks rather than a task-specific one. These findings are discussed in terms of their important implications for the design of man-machine systems and the selection of operators who are most resistant to sleep-loss stress.

Methodological issues in the assessment of sustained performance

In both military and industrial work settings, it has become desirable, and in many instances necessary, to require sustained performance beyond the apparently optimal 8-h workday. Research in this area may be viewed historically as having been focused in three major areas: (1) determining appropriate methodologies for measuring sustained performance, (2) investigating factors that influence sustained performance, and (3) searching for ways to enhance performance during sustained operations. On the basis of discussions held in Toronto in August 1984, observations are drawn and recommendations are made concerning the directions and content of future research.

Individual and Group Performances as Functions of the Team-Training Load

The effects of team training load on both individual-and group-skill performances and training were measured in the laboratory with 10 five-member teams. Each team consisted of a specific five-member combination drawn from a total of 20 undergraduate volunteer male subjects. During the first 48 h of work, subjects were trained to perform the five individual- and one group-performance tasks presented with the Multiple-Task Performance Battery employed in the synthetic-work approach to performance assessment (see Alluisi, 1964; Chiles, Alluisi, and Adams, 1968; Morgan and Alluisi, 1972). Subsequently, trained subjects were combined in different numbers with untrained subjects to provide teamtraining loads (percentages of untrained team members) ranging from 0 to 100% in 20% steps. The five subjects in each of these teams worked together for 48 h over six consecutive days. Both individual and team performance effectiveness was reduced in direct proportion to the percentage of untrained personnel substituted into a trained team. However, neither the individual performances of the trained personnel nor the skill acquisition rates of the untrained personnel were adversely affected by increasing the team-training load.

Instructional Processes in Team Training

The present research effort involved the development of a descriptive instructional process model of an operational Navy team training system. The model focuses on the team oriented instructional assessments, decisions, strategies, approaches, feedback processes, etc. used by instructors during team training. It served to clarify and enhance the understanding of team training instructional processes, and permitted evaluation of the strengths and weaknesses of the instructional system. Based on these, several potential interventions are discussed.

Influence of Task and Situational Variables on the Applicability of Vigilance Data to Physical Security

Vigilance is one of the most thoroughly researched areas of human performance. Volumes have been written concerning vigilance performance in both laboratory and real-world settings, and there is a clear trend in the literature toward an increasing emphasis on the study of operational task behavior under environmental conditions that are common to real world jobs. Although a great deal of this research has been designed to test various aspects of the many theories of vigilance, there is a general belief that vigilance research is relevant and applicable to the performances required in real-world monitoring and inspection tasks. Indeed, many of the reported studies are justified on the basis of their apparent relevance to vigilance requirements in modern man-machine systems, industrial inspection tasks, and military jobs. There is a growing body of literature, however, which suggests that many vigilance studies are of limited applicability to operational task performance. For example, Kibler (1965) has argued that technological changes have altered job performance requirements to the extent that laboratory vigilance studies are no longer applicable to real-world jobs. Many others have simply been unable to reproduce the typical “vigilance decrement” in field situations. This has led Teichner (1974) to conclude that “the decremental function itself is more presumed than established.”

INFORMATION PROCESSING IN A BINARY CLASSIFICATION TASK

Three choice-reaction time studies were conducted to investigate whether information processing is exhaustive or self-terminating, serial or parallel, and N-dependent or N-independent. A total of 54 subjects were requited to make key-pressing responses to one, two, or three digits presented in a circular display; one key was pressed if the display contained one or more target digits and another key was pressed if the display contained only nontarget digits. The first two studies utilized within-subjects designs in which the displays were consttucted from only one target and one nontarget item (Study 1) or from three target and seven nontarget items (Study 2). The third study used a between-subjects design in which different groups of subjects responded to one-, two-, or three-element displays. In general, the results indicate that CRTs increased as the total number of display elements increased and decreased as the number of target elements (or the ratio of target to nontarget items) increased for a given display size. When only target elements were presented, CRT was independent of the number of elements displayed, and when only one target was presented, CRT increased as total number of elements increased. These combined results are interpreted as support for the inference that information processing in visual search tasks tends to be self-terminating, serial, and N-dependent (of limited capacity).

Dynamic Measures of Potential Training Success

The Armed Forces and other large organizations often use scores from standardized aptitude batteries as indicators of cognitive aptitude. However, aptitude may also be demonstrated by the learning that occurs during training and measured by parameters such as initial ability levels and time needed to acquire information or skills. By using computer administered Complex Experimental Learning Tasks (CELTs), learning rate parameters recently have proved to be pragmatic as well as theoretical indicators of final performance. Potential advantages of this approach include ease and economy of computer administration, testee acceptance of job relevant tests, and potential benefits of shortened training schedules. The current research compared rate measures derived from learning on four CELTs with a paper-and-pencil battery designed to include static aptitude measures of the same domains. Performance measures were computed from stimulus display times, subject response time, and item accuracy. Overall final performance was computed using the average of the last five minutes. Correlational analyses and regression indicate that, with some qualifications, learning rate measures are predictors of individual and overall levels of performance on each CELT. Implications of these findings are that the current practice of using static aptitude tests for selection to training programs may not provide the most accurate picture of an individuals potential success or failure in that program, and that, given the trend towrds new computer-assisted training technologies, individuals may be selected on the basis of their potential for rapid learning, thus making use of the least expensive and most efficient training methods possible.

Performance Assessment, Cognitive Abilities, and Technology: An Application

The increasing technological sophistication of organizations and the concomitant requirement for individuals to process large quantities of information has raised questions concerning the efficacy of traditional approaches to performance assessment within the personnel subsystem. Traditional approaches to aptitude testing have focused on relating psychometric profiles to subsequent job performance. These approaches are subject to criticism in that they use static measures (test scores) to predict the dynamic processes of learning and are heavily influenced by prior experience without reflecting the ability to acquire new information. An alternative to traditional psychometric approaches to personnel selection is the use of rate parameters (reflecting change in performance over time) derived from relatively complex cognitive tasks as predictors of training success. The use of microprocessor-administered tasks to assess cognitive skills and abilities is an integral part of this alternative approach. The use of computer-administered Complex Experimental Learning Tasks (CELTS) for the assessment of learning abilities illustrates the computerized, rate-based approach to performance assessment. The purpose of this paper is to describe an initial study to determine the predictability of learning rate measures for performance assessment and to suggest needed future developments in training as well as job performance research.

Individual Differences and Consistency of Performance during the Acquisition of a Complex Skill

Acquisition of complex multidimensional skills as presented by the MTPB was examined in this investigation. Results indicate individual differences in the acquisition of tasks as measured within this multiple-task battery. In addition, subjects had different acquisition rates across tasks. An additional finding was that there was a fair degree of consistency with respect to the subjects’ relative level of performance on the different tasks at any given point in training.