The effect of work load history on operational errors in air traffic control simulation: The hysteresis effect -- expectancy perseverance or short-term memory overload?
Smolensky, Mark Wayne
MetadataShow full item record
Most laboratory investigations of simple vigilance tasks and complex monitoring tasks have assessed performance with the underlying premise of an invariable work load. That is, signal probabilities within sessions have traditionally been held constant. This, however, limits the generalizability of the results since real-life vigilance and complex monitoring tasks require operators to work under widely varying levels of work load. Air Traffic Control Specialists (ATCSs), for example, must monitor their radarscopes during varying levels of air traffic activity and complexity within a work shift and consequently their work load varies accordingly. Among the handful of laboratory investigations that have assessed monitoring performance within the context of intrasession work load variability, the most dramatic behavioral patterns noted have been either a performance decrement in signal detection efficiency under moderate or low work load immediately subsequent to a period of high work load (Colquhoun and Baddeley, 1964, 1967; Krulewitz, Warm and Wohl, 1975) or a disproportionate recovery of performance to reduction in work load (Gumming and Croft, 1973; Goldberg and Stewart, 1980). These studies emphasized the importance of work load history in assessing monitoring performance for a given time period. Evidence for these performance decrements in operational settings is best represented by its documented occurrence in air traffic control (ATC) where operational errors (OEs) are reported to occur most often under low to moderate levels of work load (Aviation Safety Institute, 1974; Allnutt, 1976) and particularly when immediately preceded by a period of high work load (Biggs, 1979). Two predominant explanations have been offered for the occurrence of these performance decrements in the laboratory. One explanation is that the operator establishes certain expectancies concerning signal:event ratios during a monitoring session that persevere even after signal probabilities shift downward (Gumming and Croft, 1973). An alternate explanation is that there exists a temporal disparity between short-term memory (STM) information processing and transmission that causes the STM buffer to become overloaded (Goldberg and Stewart, 1980). The STM buffer continues to be overloaded for a time even after signal probabilities shift downward. The purpose of this dissertation is to investigate, through the process of controlled simulation, the role of expectancy perseverance and STM overload in the performance decrements that are collectively referred to as the hysteresis effect. Forty-five psychology students served as ATCSs for two onehour sessions. Students were selected on the basis of their scores on a former Federal Aviation Administration (FAA) ATCS entrance exam. Those students passing the exam were instructed in basic ATC techniques and in the use of the PC-based ATC simulation package, TRACON. This software served as the radarscope in a full-scale, low-to-moderate fidelity mockup of a Terminal Radar and Approach Control (TRACON) workstation. An ATC session began with low task demand, rose to either a high or moderate task demand approximately 20 minutes in the session, remained at that level for 20 minutes and then decreased sharply to low task demand and remained at the low task demand for the remainder of the session. To test the explanation that expectancy perseverance is the cause of the hysteresis effect, each participant experienced a cue treatment indicating either a forthcoming downward shift in task demand from the current demands of the task or a continuance of the current task demand, and a no cue treatment. For the cue treatment, both a visual cue and a verbal cue were given simultaneously immediately prior to the downward shift in task demand. To test the explanation that STM overload is the cause of the hysteresis effect, participants experienced either a high to low, moderate to low, or high to high (no shift) task demand session. Sessions were recorded on videotape. OEs were recorded, transcribed and coded from the videotapes. OEs were categorized using a taxonomy of Handoff, Keying, Navigational, Pilot and Memory errors. It was reasoned that if the hysteresis effect is due to the perseverance of expectancies, then performance should be effected by cueing since the establishment and maintenance of expectancies is primarily a matter of perception based on information regarding the recent history and present circumstances of the situation and any aids that provide information regarding forthcoming parameters of the situation should serve to assist performance. For the data to support a perseverance of expectancies theory, a performance decrement would have to occur under both high to low task demand and moderate to low task demand and the decrement would be ameliorated by cueing in both cases. It was reasoned that if the hysteresis effect is due to the cognitive hardware limitations of the human information processing system, namely, the STM store, then performance should not be dramatically effected by perceptual aids such as cueing. For the data to support a STM overload theory, a performance decrement would have to occur under high to low task demands and not under moderate to low task demands (since the STM memory buffer would never be overloaded in the latter case) and would not be ameliorated by cueing under the high to low task demand shift. LThe pattern of results do not clearly indicate support for either theory. Rather, the results for Handoff, Keying, Navigational, and Pilot OEs provide support for the perseverance of expectancies hypothesis and the results for Memory OEs provide support for the STM overload hypothesis. Furthermore, results indicate presence of the two patterns of the hysteresis effect. Non-Memory OEs display a tendency for the OE rate to lag behind a reduction in task demand while Memory OEs display a tendency to rise immediately after a sudden downward shift in task demands. Although information concerning OEs in ATC has been gathered since 1964, there have been few studies designed to analyze the existing information. Therefore, relatively little is known of the types of errors that ATCSs make and why they make them. At least in the case of ATC, it can be concluded that there is support for the both explanations for the hysteresis effect that is dependent upon the category of OEs being considered. This information could be invaluable to the enhancement of future ATCS performance. Successful ATCS performance depends, to a great degree, on the reliable recall of relevant information in STM (e.g., aircraft, destinations, altitudes, etc), as well as prior knowledge of forthcoming changes in work load. A clear understanding of the role of memory and establishment of expectancies in the ATCS's strategic model could lead to the development of effective cueing and memory aids that can help ensure the availability of accurate essential data/information when it is needed. A memory lapse, or even a delay, during a critical point such as a downward work load shift can lead to serious consequences such as conflicting tracks and inadequate separation between aircraft, or flight into terrain obstacles. Similarly, sudden, unexpected changes in work load could disrupt the flow, or, mental picture that the ATCS has established to perform the job efficiently. Thus, memory and cueing aids which would efficiently facilitate recall of critical information and inform of forthcoming changes in work load would be of great value.