Impact of Sleep Fragmentation on Cognition and Fatigue

Abstract

Sleep continuity and efficacy are essential for optimal cognitive functions. How sleep fragmentation (SF) impairs cognitive functioning, and especially cognitive fatigue (CF), remains elusive. We investigated the impact of induced SF on CF through the TloadDback task, measuring interindividual variability in working memory capacity. Sixteen participants underwent an adaptation polysomnography night and three consecutive nights, once in a SF condition induced by non-awakening auditory stimulations, once under restorative sleep (RS) condition, counterbalanced within-subject. In both conditions, participants were administered memory, vigilance, inhibition and verbal fluency testing, and for CF the TloadDback, as well as sleep questionnaires and fatigue and sleepiness visual analog scales were administered. Subjective fatigue increased and sleep architecture was altered after SF (reduced sleep efficiency, percentage of N3 and REM, number of NREM and REM phases) despite similar total sleep time. At the behavioral level, only inhibition deteriorated after SF, and CF similarly evolved in RS and SF conditions. In line with prior research, we show that SF disrupts sleep architecture and exerts a deleterious impact on subjective fatigue and inhibition. However, young healthy participants appear able to compensate for CF induced by three consecutive SF nights. Further studies should investigate SF effects in extended and/or pathological disruption settings.

Keywords: cognitive fatigue; cognitive functions; sleep fragmentation.

Figure 1

Experimental protocol. Note: PSG = Polysomnography. VAS = Visual Analogue Scales (fatigue, sleepiness, stress, and motivation). LCL = Low Cognitive Load Condition. HCL = High Cognitive Load Condition.

Figure 2

Course of the TloadDback task in both cognitive load conditions. ISI (interstimulus interval) in the LCL condition was determined as 1/3 longer than in the HCL condition (i.e., ISI (LCL) = ISI (HCL) + 1/2 ISI (HCL)).

Figure 3

Subjective sleep variables in RS and SF conditions. (A–C) Sleep parameters for each of the three nights in the RS (blue) and SF (red) conditions. (D) Sleep satisfaction across nights and sleep conditions. N1 = Night 1, N2 = Night 2, N3 = Night 3. ** p < 0.01, *** p < 0.001.

Figure 4

Sleep architecture and sleep efficiency across the three nights in each sleep condition. RS = regular sleep; SF = sleep fragmentation; SE = Sleep Efficiency. SOL = Sleep Onset Latency; WASO = Wake After Sleep Onset; REM = time spent in REM stage; N3 = time spent in N3 stage; N2 = time spent in N2 stage; N1 = time spent in N1 stage.

Figure 5

Polysomnographic parameters in control (RS) and fragmented (SF) sleep during the whole night. SPT = Sleep Period Time; SL = Sleep Latency; SE = Sleep Efficiency. WASO = Wake After SleepOnset; Wake (%) = percentage of wake; N1 (%) = percentage of N1; N2 (%) = percentage of N2; N3 (%) = percentage of N3; REM (%) = percentage of REM; REM Phases = number of REM phases; NREM Phases = number of NREM phases; stage transitions = number of stage transitions; Intra-Sleep Awakenings > 2 min = number of wake periods with a minimal duration of 2 min; * p < 0.05, ** p < 0.01, *** p < 0.001. Error bars represent standard errors.

Figure 6

Polysomnographic parameters in control (RS) and fragmented (SF) sleep during the first sleep cycle. Wake in Cycle 1 (%) = percentage of Wake; N1 in Cycle 1 (%) = percentage of N1; N2 in Cycle 1 (%) = percentage of N2; N3 in Cycle 1 (%) = percentage of N3; REM in Cycle 1 (%) = percentage of REM. Error bars represent standard errors. * p < 0.05.

Figure 7

SF-related interference effects (Stroop task). ** p < 0.01. Error bars represent standard errors.

Figure 8

Visual analog scale for fatigue (VASf) prior to the TloadDback task. Note: ** p < 0.01. Error bars represent standard errors.

Figure 9

Performance (accuracy percentage) across the four 4 min segments in the 16 min duration TloadDback task. Note: LCL = low cognitive load; HCL = high cognitive load; * p < 0.05, ** p < 0.01, *** p < 0.001. Error bars represent standard errors.