the accuracy, if not the speed, of conscious processing, such as logical
reasoning and decision making, does not necessarily follow this pattern.
Monk et al. (1983) found that the verbal reasoning performance of a temporally
isolated subject was not directly associated with the circadian
rhythm observed in body temperature. In a larger study, also involving
constant conditions, Monk et al. (1997) demonstrated that reasoning
accuracy did not conform to the typical circadian rhythm shared by other
performance measures. It is possible that speed/accuracy trade-offs may
account for some of these results; however, Williamson et al. (2001)
found in a laboratory study that both the speed and accuracy of logical
reasoning and other complex tasks were scarcely impacted by time-of-day
(or sleep deprivation up to 28 h). They proposed that, in comparison to
simple monotonous tasks, complex or difficult tasks are more interesting
and arousing, and the increased engagement may serve to mask or even
overcome the effects of fatigue. These findings are in line with recent laboratory
evidence suggesting that executive functions are more resistant to
impairment due to sleep deprivation than vigilance and other components
of cognition (Tucker et al., 2010). Outside the laboratory, Monk
and Folkard (1983) noted that in some industrial and medical settings,
performance on simple processing and monitoring tasks tends to degrade
on the night shift, whereas performance on more complex tasks does not
always do so. Anecdotal reports suggest that medical personnel on night
shift are most likely to experience difficulty with routine tasks, or tasks
requiring sustained attention, than tasks involving judgment (Olson &
Ambrogetti, 1998).
If in industrial settings, different errors exhibit different circadian patterns,
awareness of these patterns will have practical implications for the
design of interventions to prevent or capture errors. In the current study,
error reports from airline maintenance technicians were classified using a
taxonomy based on the SRK model, and the daily patterns of each type
of error were examined.