Gerald A. Hadley Lawrence J. Prinzel III1 Frederick G. Freeman Peter J. Mikulka Old Dominion University
The presence of automation has pervaded almost every aspect of modern life. The use of automation has made work easier, more efficient, and safer, as well as giving us more leisure time. However, research has also shown that the increase in automation has not been without costs. These costs have resulted in increased interest in advanced automation concepts. One of these concepts is automation that is adaptive or flexible in nature and can respond to changing situational demands ( Hancock & Chignell, 1987; Scerbo, 1996).
One challenge facing those seeking to implement adaptive automation technology concerns how changes among modes of operation will be accomplished ( Scerbo, 1996). This question requires research into the various issues surrounding the use of adaptive automation. These issues include the frequency that task allocations should be made, when automation should be invoked, and how adaptive automation changes the nature of the human-automation interaction.
Although many proposals have been made concerning adaptive automation, few studies have been directed at determining how adaptive automation will impact human performance. One such study is by Scallen, Hancock, and Duley ( 1995) who reasoned that situations could exist with adaptive automation wherein a certain threshold for triggering an automation change is exceeded. This would result in an allocation in automation mode. However, the automation change prompted by the system may result in a further increase in workload leading again to another automation change. Such a situation could then lead to a number of automation changes within a short period of time. Therefore, these researchers examined how various short-cycle schedules of automation affects correlates of mental workload. They reported that performance was significantly better but workload was higher under a 15-sec schedule of automation than 60-sec. The results were interpreted in terms of a micro-tradeoff; that is, participants performed better under the 15-sec schedule at the expense of working harder.
The present study was an attempt to further the findings reported by Scallen, Hancock, and Duley, ( 1995). Vidulich and Wickens ( 1986) noted that greater effort can result in higher subjective ratings, but greater effort can also improve performance. This is a workload dissociation because increased subjective ratings and degraded performance together is what is usually indicative of increased workload. Such a dissociation may prove troublesome for system designers seeking to develop adaptive automation systems. Therefore, we sought to examine how other measures of workload are impacted by schedules of short-cycle automation. Specifically, we gathered EEG and ERP measures under the 15, 30 and 60-sec schedules of automation investigated by Scallen, Hancock, and Duley. In addition, because a concern voiced by many has been directed at how quickly and effectively operators can transition from automation to manual control when needed to do so, another goal of the study was to examine the impact of manual reversions on these correlates of workload.
Nine graduate students served as participants for this experiment. The ages of the participants ranged from 18-35. All participants had normal or corrected-to-normal vision.