Components of Vigilance Tasks

Performance efficiency in tasks requiring vigilance is clearly a product of several factors and a significant challenge in developing a better understanding of the construct requires the elaboration of the factors influencing efficient vigilant performance. In an attempt to provide an answer to this question, Jerison (1959) proposed a framework suggesting that performance on watchkeeping tasks was a function of factors including:

• The sensory modality of the target signal (auditory or visual) • The salience or detectability of those signals

• Uncertainty in position, time or nature of the signal to be detected • Characteristics of the non-signal background events

• Complexity of the task

4.2.1 Signal Modality

Vigilance experiments have been conducted in which the signal to be detected involved presentation of auditory, visual, or cutaneous stimulation. However, almost without exception it has been found that tasks involving auditory modality stimulation (e.g., audio alarms) are performed with greater efficiency and are more stable over time, than tasks involving purely visual or

better coupling between the observer and display to be monitored. For example, the provision of helmet-mounted displays can be used to elevate performance levels for tasks involving visual signals to levels obtained for tasks where detection of audio signals is required (Galinsky et al, 1990). In addition, task performance in systems that employ dual mode alerting methods (e.g., visual signal accompanied by audio alarm) is superior to single mode performance, irrespective of the primary modality method (Doll and Hanna, 1989).

4.2.2 Signal Salience

A common finding in the plethora of experiments results reporting sustained attention task performance effects is the idea that target detection is considerably enhanced by increasing the amplitude and duration of the sought for signal. Interestingly, Corcoran and his colleagues found that the prototypical performance decrement curve can be reversed when the amplitude of the input signal is abruptly increased midway through the vigil (Corcoran et al, 1977). This performance improvement tends to remain active for the remainder of the session. Increases to the duration of the signal can also improve vigilance. Radar signals of brief duration, for example, are more likely to be missed than those that remain visible for some time. Significant reductions in signal detection failure can be obtained by increasing the duration of target signals up to a limit of around four seconds. Few, if any performance gains are likely to result from increases to the duration of the target signal beyond this range (e.g., Warm et al, 1970).

4.2.3 Stimulus Uncertainty

Subjects in vigilance tasks are likely to be exposed to varying degrees of uncertainty with regard to the onset of the target event. Not surprisingly, research has shown performance to vary significantly as a function of the predictability of target signal. Signal uncertainty can be manipulated in an experimental situation in two ways. Temporal uncertainty arises when the probable timing of event signals are unknown. Spatial uncertainty occurs when the precise location of the event in the visual field is unknown. Performance decrement in tasks involving both temporal and spatial uncertainty is especially prone to detection failures. Experiments have been performed in which the density of target signals was manipulated. When critical signals are presented more frequently – in effect lowering the temporal uncertainty associated with signal rates – the greater the performance efficiency in the signal detection task (Warm and Jerison, 1984). Furthermore, response times to detection increase as a linear function of uncertainty.

Temporal uncertainty has also been examined in experiments that manipulate the time duration between presentations of target signals. When performance is compared for three conditions in which intervening times are described as predictable, irregular or unpredictable, it was found that both the speed and

accuracy of detection was far greater in the context of regular as opposed to irregular signal conditions (Warm, 1984).

The effect of spatial uncertainty on vigilance has been tested using tasks in which signals can appear on any one of several display terminals. Under such conditions, performance efficiency is lowered and subjects tend to bias their attention towards locations in which the likelihood of signal appearance is perceived to be the greatest (Joshi et al, 1985).

With regard to the issue of signal uncertainty, several studies have revealed an important effect. Subjects trained initially under conditions of high signal probability tend to be more resistant in subsequent testing to performance decrement effects. This occurs irrespective of actual signal probability in the work domain (Griffin et al, 1986). This finding is strongly suggestive of the potential of training to help minimise the influence of signal uncertainty in practical contexts.

4.2.4 Background Context

Signal detection experiments tend to use stimulus materials consisting of relatively few target signals embedded within the context of a wide range of non-target background events. For example, the subject may have to react to bright flashes of light that appear within an array of dimmer lights or respond to audio events of a particular tone presented in a sound recording of many different tones. In one obvious sense, background events are neutral with regard to performance of the vigilance task. They do not require an overt reaction. At another level, however, they are extremely influential in terms of their effect on a person’s ability to maintain focus of attention and a number of studies have revealed that performance decrement is more pronounced in situations where the occurrence of background events is high frequency (Lanzetta et al, 1987). Interaction effects between the frequency of background events and the temporal regularity of the target signal have also been observed. Synchronous relationships – those in which a regular target appears within stable background of non-target events – tend to provide the most beneficial arrangement for vigilant performance. Asynchronous relationships, on the other hand (those in which an irregular target appears within a background of varying regularity non-target events), provides the most challenging conditions for effective target detection and tends to yield the greatest number of detection failures. The effects of both event rate and event asynchrony are not trivial. They reveal that the efficiency of signal detection in a vigilance task is determined as much by what transpires in the background to the target event as it is by the critical event itself.

operating conditions can of course be much more complex than this and might be expected to require operators to cope with multiple signal sources and types. At an intuitive level it would be anticipated that increases to the complexity of the stimulus configuration would lead to loss of performance efficiency and this is exactly what early studies found. In a well-cited study reported by Jerison (1963), for example, subjects monitored three displays simultaneously with the result that the decrement function was greatly enhanced. Indeed, significant detection failures were obtained within the first 2-3 minutes of commencement of the vigil – a period in which performance efficacy is usually quite good. In contrast to this, contemporary studies have revealed a more complex picture. Several studies have shown that performance decrement effects may be reduced by increasing task complexity if the overall workload (the total number and rate of target signals and background events) remains quite low. In these studies, observers were asked to monitor between 6-36 screens over a 6 hour period. The apparent improvement in task performance, however, is short lived and increases to the overall task complexity restore the decrement function (Loeb et al, 1987). In an interesting variation on this basic theme, Fisk and Schneider (1981) approached the issue of performance decrements as a function of task complexity from the human information processing perspective of controlled and automatic processing. According to this view, automatic processes represent fast, effortless, skilled-based behaviours, whilst controlled processes are relatively slow, effortful and capacity-limited. Using a methodology in which vigilance skills were acquired by operators over several hundred trials, Fisk and Schneider were able to demonstrate that changes to the vigilance decrement function as a result of changes to task complexity, are largely restricted to people reliant upon controlled, capacity-limited processing, that is, the unskilled. Consequently, some authors conclude that the responsiveness of vigilance to task complexity might be overcome via the provision of training and practice.

4.3 Summary

The study of vigilance reveals that there are significant performance effects when people are asked to remain alert to target signals over prolonged periods. Specifically, experiments reveal that up to 50% of performance efficiency can be lost in the first 60 minutes following commencement of the vigil when operating conditions or the design of the task are unfavourable. Several factors can influence a person’s ability to react to task demands. The salience of the target object, the levels of uncertainty associated with the onset of the target event, the character and makeup of background materials, objects and events, and the complexity of the task are all factors likely to influence the quality of performance. Changes to any of these parameters can affect performance efficiency for the better or for the worse.