Shiftworkers are often required to sleep at inappropriate phases of their circadian timekeeping system, with implications for the dynamics of ultradian sleep stages. The independent effects of these changes on cognitive throughput performance are not well understood. This is because the effects of sleep on performance are usually confounded with circadian factors that cannot be controlled under normal day/night conditions. The aim of this study was to assess the contribution of prior wake, core body temperature, and sleep stages to cognitive throughput performance under conditions of forced desynchrony (FD). A total of 11 healthy young adult males resided in a sleep laboratory in which day/night zeitgebers were eliminated and ambient room temperature, lighting levels, and behavior were controlled. The protocol included 2 training days, a baseline day, and 7 x 28-h FD periods. Each FD period consisted of an 18.7-h wake period followed by a 9.3-h rest period. Sleep was assessed using standard polysomnography. Core body temperature and physical activity were assessed continuously in 1-min epochs. Cognitive throughput was measured by a 5-min serial addition and subtraction (SAS) task and a 90-s digit symbol substitution (DSS) task. These were administered in test sessions scheduled every 2.5 h across the wake periods of each FD period. On average, sleep periods had a mean (+/- standard deviation) duration of 8.5 (+/-1.2) h in which participants obtained 7.6 (+/-1.4) h of total sleep time. This included 4.2 (+/-1.2) h of stage 1 and stage 2 sleep (S1-S2 sleep), 1.6 (+/-0.6) h of slow-wave sleep (SWS), and 1.8 (+/-0.6) h of rapid eye movement (REM) sleep. A mixed-model analysis with five covariates indicated significant fixed effects on cognitive throughput for circadian phase, prior wake time, and amount of REM sleep. Significant effects for S1-S2 sleep and SWS were not found. The results demonstrate that variations in core body temperature, time awake, and amount of REM sleep are associated with changes in cognitive throughput performance. The absence of significant effect for SWS may be attributable to the truncated range of sleep period durations sampled in this study. However, because the mean and variance for SWS were similar to REM sleep, these results suggest that cognitive throughput may be more sensitive to variations in REM sleep than SWS.
