No persisting effect of partial sleep curtailment on cognitive performance and declarative memory recall in adolescents


Ulrich Voderholzer, Department of Psychiatry and Psychotherapy, University Medical Center Freiburg, Hauptstrasse 5, 79104 Freiburg, Germany. Tel.: 49-761-270-6603; fax: 49-761-270-6667; e-mail:


Growing evidence indicates that sleep facilitates learning and memory processing. Sleep curtailment is increasingly common in adolescents. The aim of this study was to examine the effects of short-term sleep curtailment on declarative memory consolidation in adolescents. A randomized, cross-over study design was chosen. Twenty-two healthy subjects, aged 14–16 years, spent three consecutive nights in the sleep laboratory with a bedtime of 9 h during the first night (adaptation), 4 h during the second (partial sleep curtailment) and 9 h during the third night (recovery). The control condition consisted of three consecutive nights with bedtimes of 9 h. Both experimental conditions were separated by at least 3 weeks. The acquisition phase for the declarative tests was between 16:00 and 18:00 hours before the second night. Memory performance was examined in the morning after the recovery night. Executive function, attention and concentration were also assessed to control for any possible effects of tiredness. During the 4-h night, we observed a curtailment of 50% of non-rapid eye movement (non-REM), 5% of slow wave sleep (SWS) and 70% of REM sleep compared with the control night. Partial sleep curtailment of one night did not influence declarative memory retrieval significantly. Recall in the paired-associate word list task was correlated positively with percentage of non-REM sleep in the recovery night. Declarative memory consolidation does not appear to be influenced by short-term sleep curtailment in adolescents. This may be explained by the high ability of adolescents to compensate for acute sleep loss. The correlation between non-REM sleep and declarative memory performance supports earlier findings.


There is growing evidence that sleep promotes learning and memory consolidation in humans (for reviews, see Born et al., 2006; Marshall and Born, 2007; Stickgold and Walker, 2005). Moreover, there is also evidence that different components of memory processing may be related to different sleep stages. A recent study reported that not only nocturnal sleep but also short daytime naps have a beneficial effect on declarative memory (Lahl et al., 2008).

Rapid eye movement (REM) sleep seems to promote procedural memories for skills (for reviews see Maquet, 2001; Smith, 2001), whereas hippocampus-dependent declarative memory apparently benefits more from slow wave sleep (SWS) (Smith, 2001). Some studies have shown that early night sleep, rich in SWS, may play a role in the consolidation of declarative memory (Gais and Born, 2004a, 2004b;Plihal and Born, 1997). Further work suggests that non-REM sleep state characteristics may be correlated with the acquisition of declarative material (Gais et al., 2002; Molle et al., 2004). These and other studies advocate and support the ‘dual process’ hypothesis, while others propose a ‘two-step’ hypothesis requiring the participation of both SWS and REM (Ficca and Salzarulo, 2004; Stickgold et al., 2000a). While it has been shown that stabilization of some forms of procedural motor memory already begins during waking hours, the situation remains less clear for declarative memory (Stickgold and Walker, 2005). Although there is some evidence proposing that Stages 3 and 4 or non-REM sleep in general are necessary for declarative memory consolidation, non-REM sleep may in fact simply be occurring at the same time as other physiological factors involved in memory processing. Preliminary results also suggest that the length of the non-REM–REM sleep cycle may be important for declarative memory formation (Mazzoni et al., 1999; Smith, 2001).

What does this mean in the context of increasing sleep deprivation, which has been described for all age groups in our society over the past decades? In adults, negative effects of chronic partial sleep loss on health and neurocognitive performance have been demonstrated in several studies (for review see Durmer and Dinges, 2005). Sleep deprivation studies revealed an impairment of memory and numerous neuropsychological functions (Haavisto et al., 2006; Van Dongen et al., 2003). In a meta-analysis, Pilcher and Huffcutt (1996) concluded that long-term and partial sleep deprivation in adults resulted in a poorer performance in simple rather than more complex memory tasks. However, the effects of sleep deprivation on vigilance and mood seem to be more pronounced than on cognition. Remarkably, cognition and mood were affected more by partial sleep deprivation than by total sleep deprivation. Jones and Harrison (2001), on the other hand, suggested in their review that sleep reduction impaired performance of complex executive function tasks; for example, creativity and planning skills.

Comparatively few studies have been conducted so far in adolescents. Circadian rhythms and sleep patterns in adolescents have been shown to be different from those in preadolescents or adults (Carskadon, 1990). Sleep continuity and architecture undergo characteristic changes during puberty because of both psychosocial and biological factors (Carskadon, 1990; Crowley et al., 2007; Dahl and Lewin, 2002; Meijer et al., 2000; Ohayon et al., 2004; Saarenpaa-Heikkila et al., 2000; Spruyt et al., 2004; Wolfson and Carskadon, 1998).

Various surveys have concluded that sleep curtailment is common in adolescence, impacting negatively on health, wellbeing and daytime functioning (Carskadon et al., 1998). In turn, adequate sleep was associated positively with favourable health status and health-related behaviour (Carskadon et al., 1998;Chen et al., 2006; Wolfson and Carskadon, 2003), as well as with learning capacity and academic performance (Curcio et al., 2006).

The findings on the relationships between sleep deprivation and cognitive functions obtained from the limited number of studies are inconsistent. In trials with total sleep deprivation significant effects were observed in memory performance and computational speed, whereas attention and sustained motor activity were unimpaired (Carskadon et al., 1981b). College students performed significantly worse on cognitive tasks compared to non-deprived subjects (Pilcher and Walters, 1997).

Studies with partial sleep curtailment found impairments of higher cognitive functions, such as verbal creativity and abstract thinking, whereas less complex functions were hardly affected. After three nights of mild sleep restriction to 7 h per night in adolescents aged between 10 and 14 years no impairments in working memory, computational accuracy and planning ability were detected (Randazzo et al., 1998). Sadeh et al. (2003) noticed poorer performance in memory, attention and vigilance tasks after sleep restriction compared with sleep extension in children aged from 9 to 12 years. Carskadon et al. (1981a) found no significant effects after a single restricted night (4 h) on the results of a word memory test, addition test or a listening attention task (11–13 years). Recently, Gais et al. (2006) showed that sleep following learning has a beneficial effect on declarative memory, i.e. vocabulary lists (high school students, 17.4 ± 0.2 years). Backhaus et al. (2008) found that sleep facilitated declarative memory consolidation in children (9–12 years), whereas a wake period of the same duration did not. In a study by Sadeh et al. (2002), examining associations between sleep and neurobehavioural functioning in children (7.2–12.7 years), relationships between fragmented sleep and performance in more complex neurobehavioural tasks were observed, particularly in the younger age group. Simple memory tasks, motor speed and reaction time were not correlated with any of the sleep measures. Steenari et al. (2003) reported that sleep quality and quantity were associated with results of working memory tasks in school-age children (6–13 years). Some studies included subjects with different pubertal stages; however, in the majority of publications the maturational stage of participants was not described.

Considering the inconsistency and paucity of studies in adolescents, this study aimed to investigate the effects of partial sleep deprivation on declarative memory performance and cognitive functioning in adolescents. It should be pointed out that in most of the above-cited studies the study design did not allow for differentiation between the impact of increased tiredness/sleepiness and the influence of sleep deprivation per se on the dependent variables. This important methodological consideration was addressed in our study. The main research question, therefore, was whether a single night of partial sleep deprivation has a negative impact on adolescents’ declarative memory performance, assessed after a recovery night. Secondly, relationships between declarative memory consolidation and objective sleep parameters were investigated. We hypothesized that (i) sleep curtailment after a learning phase would result in significantly poorer memory performance but no differences in attention and executive tasks and (ii) there would be a relationship between declarative memory performance and objective sleep parameters during both the restricted and the control nights.



Twenty-two healthy subjects between the ages of 14 and 16 years (12 females, 10 males, mean age 15.5 ± 0.74, mean body mass index 21.7 ± 3.1 kg m2) were recruited via advertising with posters displayed at schools in Freiburg. Exclusion criteria were history of a psychiatric, neurological or other relevant medical disorder, drug abuse, sleep or memory disturbances, intake of any medication and smoking. All participants had a regular sleep–wake cycle with an average sleep time of about 8–9 h per night. This was determined by medical history, physical examination, including routine blood and urine investigations, psychological questionnaires, sleep diaries and actigraphy during the preceding week. All subjects and their parents provided written informed consent. Subjects were paid for participation. The study was approved by the Ethical Committee of the University of Freiburg and carried out according to the Helsinki Declaration.


This was a randomized, cross-over study with two blocks of three consecutive nights in the sleep laboratory. The two blocks were separated by a 3-week interval. The blocks consisted of one adaptation night (bedtime 20:00–07:00 hours), a partial sleep deprivation night (bedtime from 03:00 to 07:00 hours) or a control night (bedtime from 20:00 to 07:00 hours), and a third night (bedtime 20:00–07:00 hours) (Fig. 1). A bedtime of 9 h was chosen because studies of sleep habits in adolescents concluded that this is the physiologically optimal sleep duration in this age group. We also wanted to avoid a mild sleep curtailment in the control group. For the sleep restriction condition, a late bedtime from 03:00 to 07:00 hours was chosen based on the adolescents’ tendency for sleep-onset delay (Carskadon, 1990; Carskadon et al., 1998). The order of the conditions was randomized across subjects.

Figure 1.

 Study design; 9 h bedtime; 4 h bedtime; all nights registered polysomnographically. Cross-over design: After a 3-week interval subjects returned and participated in the opposite condition. VVM: Visual and Verbal Memory test; VLMT: auditory Verbal Learning and Memory Test; ToL: Tower of London; TMT: trail making test; TAP: Test for Attentional Performance.

Throughout the study period, all participants remained in the sleep laboratory, supervised continuously by research assistants. In addition, their sleep–wake rhythms were monitored by actigraphs. Subjects were not allowed to take any naps, to exhaust themselves physically or consume any caffeine. On the day before the experimental or control night between 16:00 and 10:00 hours subjects attended the learning session for all declarative memory tests (refer to description of tests below). Retesting was scheduled on the day after the recovery night between 08:00 and 10:00 hours and comprised the recall for the memory tests and the application of executive and attention/concentration tests (see Fig. 1).

Neuropsychological test battery

The neuropsychological assessment included different declarative memory tasks to assess the relationship between memory consolidation and sleep. Furthermore, tests for attention, concentration and executive functioning were conducted to control for any effects of tiredness, which might have persisted until the morning after the recovery night and acted as confounding variables. Test administration followed standardized procedures and was conducted by trained staff. Recall was conducted in the morning to reduce the effects of fatigue or meals on performance. A 2-min rest period was given after each test. To control for sequence effects, we used parallel versions of the tests.

Declarative memory tasks

The visual and verbal memory test (VVM) by (Schellig and Schaechtele, 2001) was applied to test short- and long-term memory of visuospatial and verbal material. By comparing short- and long-term memory, the decrease of memory performance and the rate of forgetting can be calculated. The VVM contains two subtests; in the subtest ‘City map’ subjects had to memorize the course of a route and then mark it on the same map during recall. In the subtest ‘Construction’, a description of a building was presented to participants, worded in syntactically simple sentences. Participants had to learn names, numbers and propositional contents. Recall was tested by written answers to specific questions. The time for both acquisition and reproduction was limited.

The auditory verbal learning and memory test (VLMT, an adapted German version of the Rey auditory verbal learning test; Helmstaedter et al., 2004) was used to assess secondary verbal learning and memory. Fifteen concrete items were presented repeatedly to the subject, and after each presentation as many items as possible had to be recalled. After the fifth recall, an intrusion list was presented that consisted of another 15 concrete items. This intrusion list had to be recalled as well; afterwards recall of the original word list was tested a sixth time without additional presentation. After the recovery night the word list had to be recalled a seventh time. Additionally, a recognition test was presented, which consisted of the 15 original items intermixed with 15 items of an intrusion list and 15 further concrete items not presented previously. Participants had to identify the 15 original words. The number of correctly recognized words was evaluated.

The paired-associate word list task by Plihal and Born (1997) was used in an adapted version consisting of 32 related word pairs, each presented randomly in black on a white background on a 15′′ computer screen for 1500 ms, followed by a 5000-ms period. The word list was presented repeatedly until the subject remembered at least 60% of word pairs in a cued-recall test, i.e. stating the matching word to the first words of the previously learned pairs. The recall score consisted of the number of identically correct responses. Score values were converted into percentages (% = response score/32 × 100) and are referred to subsequently as ‘memory performance’.

Executive functioning

In the computerized version of the Tower of London (ToL) problem-solving trials (Kuelz et al., 2006; Shallice, 1982) subjects had to rearrange a set of balls presented at the bottom half of the screen to reach a goal state as shown on the top half. The number of moves needed to complete the trial varied between two and six moves. Participants were told to repeat the moves while following three rules: (i) move only one ball at a time, (ii) do not move a ball in the lower row when another ball is on top and (iii) three balls can be placed on the tallest peg, two balls on the middle and one ball on the shortest. The outcome criterion was the number of correctly conducted trials (Kuelz et al., 2006).

The Trail-Making Test, Part B (TMT) (Reitan, 1956) is a widely used test of psychomotor speed, attention, sequencing, mental flexibility, visual scanning and attention-shifting executive control. Subjects had to connect a series of consecutively numbered and lettered circles distributed in a visuospatial array on a sheet of paper by alternating between the numbers and letters (i.e. 1-A-2-B, etc.).

Attention and concentration

In the Test for Attentional Performance (TAP), subjects underwent a computerized standardized test battery to assess alertness and vigilance (Fimm and Zimmermann, 1993). The TAP includes the following subtests: (i) ‘divided attention’, a dual task with visual and acoustic stimuli and (ii) ‘flexibility’, a task requiring shifting attention.

The d2 test has been developed and described in detail by Brickenkamp and Zillmer (1962), and is considered to be a measure of attentional processes. Briefly, the test consists of 14 lines of 47 randomly mixed letters (either d or p). Subjects were instructed to mark, within 20 s for each line, only the letter d with two additional strokes (d’’) and to ignore all other letters. A value for sustained attention and concentration was calculated, and the difference between concentration performance in the evening and in the morning was used for analysis.

During the experimental period, participants were asked to estimate their subjective tiredness in the mornings using a six-point scale (1 = not at all tired, 6 = very tired).


Polysomnography was recorded during the three nights for each experimental block using a standardized procedure. Details of polysomnography in our laboratory have been described previously (Voderholzer et al., 2003). Sleep recordings were scored visually by experienced raters using standard criteria (Rechtschaffen and Kales, 1968).

Data analysis

After completion of data collection and quality checks, data were analysed using the Statistical Package for Social Sciences (SPSS) version 14 (SPSS Inc., Chicago, IL, USA). Descriptive values are given as mean and standard deviation. To compare both conditions (regular sleep duration and sleep deprivation nights), the following variables were analysed statistically; for memory consolidation: ‘rate of forgetting of VVM verbal and visual tasks’, ‘delayed recall of VLMT’ and ‘memory performance in percentage of the paired-associate word list task’; for executive function and attention: ‘correctly conducted trials in the ToL’, ‘D2 attention performance’, ‘TMT drawing time’, ‘TAP flexibility, ability to shift attention’ and ‘TAP divided attention’. Due to non-normally distributed data and no differences between both experimental blocks at baseline, non-parametric (distribution-free) Wilcoxon signed rank tests were used.

The following sleep parameters: sleep period time (SPT), total sleep time, time in bed, sleep efficiency, sleep latency, sleep latency Stages 3 and 4, REM latency, awakening time, Stage 1, Stage 2, Stage 3, Stage 4, SWS, REM sleep, non-REM sleep (calculated as sum of sleep Stages 1, 2, 3 and 4; in minutes and as percentage), number of wake periods and subjective estimates of tiredness were also analysed using Wilcoxon tests. For both the restricted and the control night, correlations between objective sleep variables and neuropsychological parameters were calculated using Spearman’s correlations. The significance level was set at < 0.004 with Bonferroni-adjustment for multiple comparisons.

For all neuropsychological and polysomnographic parameters, the magnitude of the effect size (d) was calculated to detect smaller effects which might have been missed by group comparisons. Effect size interpretations were based on Cohen’s descriptive guidelines (Cohen, 1988).


Memory performance

No significant group differences were found between both experimental blocks at baseline. When comparing test results after the recovery night in the sleep deprivation to the control condition no significant differences were found in any of the tests (see Table 1). We found no difference in the overnight changes between both experimental conditions. Small effect sizes were found for the neuropsychological outcome parameters.

Table 1.   Mean neuropsychological results after control condition versus partial sleep deprivation (PSD) after night 3 (recovery night in the experimental group or second night of regular sleep duration in the control condition)
 Recall session on second day after normal sleep, mean ± SDRecall session on second day after PSD, mean ± SDWilcoxon’s paired ranked testEffect size
ZPCohen’s d
  1. Bonferroni-adjusted significance level P = 0.004.

  2. VVM, visual and verbal memory test; VLMT, auditory verbal learning and memory test; TMT, trail making test; ToL, Tower of London; SD, standard deviation; PSD, partial sleep deprivation.

Declarative memory
 VVM verbal, rate of forgetting0.3 ± 0.10.2 ± 0.2−0.4380.6610.339
 VVM visuospatial, rate of forgetting0.2 ± 0.20.1 ± 0.2−0.7130.4760.367
 VLMT: delayed recall14.6 ± 0.914.5 ± 0.9−0.8570.3910.207
The paired-associate word list task, memory performance in percentage73.4 ± 13.269.0 ± 9.8−1.0290.3040.379
Executive functions
 D2, attention performance238.6 ± 34.8224.7 ± 40.5−1.4940.1350.369
 TMT, time36.7 ± 8.035.7 ± 8.6−0.3730.7090.130
 ToL, number of correctly conducted trials15.4 ± 1.115.2 ± 1.6−0.4650.6420.095
 TAP, flexibility, ability to shift attention614.8 ± 62.5613.0 ± 70.4−0.1460.8840.430
 TAP, divided attention514.8 ± 111.3517.4 ± 155.4−0.9900.3220.019

Executive functioning

After the recovery night, no significant differences between the experimental and the control condition were found for the d2, TMT, ToL and TAP parameters.

Subjective tiredness

Subjective tiredness was significantly higher on the day after partial sleep curtailment (4.5 ± 1.2) compared to the control condition (2.6 ± 1.0) (Z = −3.693, P = 0.001). After the recovery night, subjectively estimated tiredness in the morning was still higher in the partial sleep curtailment condition (3.8 ± 1.2) compared with the control condition (2.7 ± 1.0) (Z = −3.129, P = 0.002).


At baseline, polysomnographic parameters were not statistically different in both experimental blocks. In the sleep deprivation night, we found a reduced percentage of REM sleep referred to SPT, and an increased percentage of non-REM and SWS compared to the control night. When comparing the absolute time spent in sleep stages there was an approximately 50% reduction of non-REM and an approximately 70% reduction of REM-sleep during the short night compared with the control night (Fig. 2, Table 2). There was no significant diminution of the absolute time spent in SWS during the short night compared with the control night (Table 2, Fig. 2). In the recovery night after sleep curtailment a slightly, but significantly, increased total sleep time and a decreased sleep latency could be observed (Table 2).

Figure 2.

 Sleep curtailment and absolute time spent in sleep stages: 4-h versus 9-h night in adolescents (n = 22). REM: rapid eye movement.

Table 2.   Polysomnographic data of experimental and recovery nights, under experimental (partial sleep deprivation) and control condition (regular sleep duration)
 Experimental condition (partial sleep deprivation)Control condition (regular sleep duration)Wilcoxon’s paired ranked testEffect size
Experimental night (night 2) (03:00–19:00 hours) mean ± SDRecovery night (night 3) (10:00–19:00 hours) mean ± SDNight 2 (10:00–19:00 hours) mean ± SDNight 3 (10:00–19:00 hours) mean ± SDZP Cohen’s d
  1. Night 1 was the adaptation night; SD: standard deviation; all parameters calculated on sleep period time (SPT); comparisons between the recovery night after partial sleep deprivation and the second night (night 3) of regular sleep duration; significance level P = 0.004, Bonferroni-adjusted for multiple comparisons. Significant effects are given in bold.

  2. REM, rapid eye movement; SWS, slow wave sleep; NREM, non-REM.

Sleep period time (min)235.3 ± 4.1532.2 ± 6.6518.6 ± 17.6525.1 ± 9.1−2.7440.0060.893
Total sleep time (min)232.4 ± 5.2523.7 ± 7.4503.5 ± 22.5511.5 ± 10.3−3.5070.0001.360
Time in bed (min)240.3 ± 1.4541.2 ± 1.6540.1 ± 1.0541.6 ± 1.9−0.7690.4420.228
Sleep latency (min)4.6 ± 3.37.8 ± 6.513.2 ± 5.513.9 ± 7.3−3.3290.0010.883
REM latency (min)75.5 ± 29.3103.4 ± 47.792.8 ± 36.098.4 ± 39.0−0.5840.5590.115
Awakening time (min)2.9 ± 2.48.5 ± 5.015.1 ± 8.613.6 ± 6.1   
Awakening time (%)1.2 ± 1.01.6 ± 0.92.9 ± 1.72.6 ± 1.2−2.8080.0050.943
Wake periods (number)4.5 ± 3.511.9 ± 6.818.8 ± 10.017.5 ± 7.6−2.4050.0160.777
Stage 1 (min)6.4 ± 4.819.0 ± 10.529.6 ± 14.528.1 ± 14.8   
Stage 1 (%)2.8 ± 2.13.6 ± 2.05.7 ± 2.95.3 ± 2.8−2.7430.0060.699
Stage 2 (min)85.2 ± 18.8261.2 ± 31.8251.3 ± 23.4256.0 ± 27.2   
Stage 2 (%)36.2 ± 8.249.1 ± 6.048.5 ± 4.848.7 ± 5.0−0.3730.7090.072
Stage 3 (min)40.2 ± 14.062.0 ± 17.655.1 ± 19.159.0 ± 23.8   
Stage 3 (%)17.1 ± 6.011.6 ± 3.310.6 ± 3.511.2 ± 4.5−1.1200.2630.101
Stage 4 (min)61.5 ± 30.960.1 ± 27.451.9 ± 25.752.8 ± 27.2   
Stage 4 (%)26.1 ± 13.011.3 ± 5.19.9 ± 4.910.1 ± 5.2−1.5100.1310.233
SWS (min)101.8 ± 23.5122.0 ± 32.8107.0 ± 33.5111.8 ± 39.8   
SWS (%)43.2 ± 9.822.9 ± 6.120.5 ± 6.121.3 ± 7.5−2.3380.0190.234
REM sleep (min)33.6 ± 15.5115.9 ± 20.4112.3 ± 18.7111.4 ± 16.4   
REM sleep (%)14.3 ± 6.721.8 ± 3.921.6 ± 3.521.2 ± 3.3−0.4060.6850.166
NREM sleep (min)193.3 ± 3.5402.3 ± 20.9387.8 ± 26.8395.9 ± 21.5   
NREM sleep (%)82.2 ± 7.575.6 ± 3.774.8 ± 4.375.4 ± 3.4−0.4380.6610.056


Memory performance in percentages in the paired associate word list task correlated significantly positively with non-REM sleep (r = 0.521, P = 0.003) after sleep curtailment (Fig. 3, Table 3). REM sleep was correlated negatively with memory performance in percentage in the paired associate word list task, but not significantly (r = −0.498, P = 0.054). Memory performance after the restricted night was not correlated significantly with SWS (r = 0.349, P = 0.336).

Figure 3.

 Bonferroni-adjusted correlation between the paired-associate word list task (memory performance in percentage) and non-rapid eye movement (REM) sleep referred to sleep period time (SPT), recovery night (night 3) in the sleep deprivation condition (n = 22).

Table 3.   Correlations of memory performance (paired-associate word list task) and sleep parameters of the night 2 (experimental or control condition with regular sleep duration) and night 3 (recovery night or control condition with regular sleep duration)
 Control condition (regular sleep duration)Experimental condition (partial sleep deprivation)
Memory performance in %Memory performance (%)
  1. Night 1 was the adaptation night; Spearman’s correlations, significance level P = 0.004, Bonferroni-adjusted for multiple comparisons.

  2. REM, rapid eye movement; SW, slow wave sleep; NREM, non-REM.

NREM sleep %, night 2−0.2050.0900.4190.156
NREM sleep %, night 3−0.1640.1170.5210.036
REM sleep %, night 20.1190.149−0.4300.138
REM sleep %, night 30.1290.142−0.4980.054
SWS %, night 2−0.1140.1530.2180.990
SWS %, night 3−0.1160.1520.3490.336

No significant correlations were found between memory performance and any polysomnographic variables in the control condition (see Table 3). All correlations were Bonferroni corrected.


Our study in healthy adolescents showed no effects of one night of partial sleep deprivation on the consolidation of declarative memory. Similar to studies with adults (Stickgold et al., 2000b), recall was tested after a recovery night to reduce the impact of possible confounders such as sleepiness after sleep deprivation. Our results do not necessarily contradict previous studies – most of them in adult subjects and only a few in children – which demonstrated that sleep facilitates memory consolidation (Backhaus et al., 2008; Plihal and Born, 1997). In these studies, memory consolidation was improved when subjects slept shortly after a learning phase compared to a condition without sleep (Gais et al., 2006). However, the design of our study attempted to mimic a real-life situation, with learning throughout the day and sleep curtailment during the subsequent night (Carskadon et al., 1981a; Fallone et al., 2001) with an interval of at least 5–7 h between the acquisition of new material and sleep. In our study, the restriction of nocturnal sleep to 4 h did not diminish the recall after the recovery night compared with the control condition.

There are two possible explanations for our results: first, one could assume that the consolidating effect of sleep happens within a sensitive period of a few hours after a learning phase. Gais et al. (2006) demonstrated that memory consolidation was enhanced after a 3-h compared to a 15-h time interval between acquisition and sleep. Furthermore, consolidation might be influenced by the activities during that time interval. In the majority of studies, subjects learned immediately prior to sleep and were tested shortly after the experimental night, resulting in a higher level of interference, whereas in our study there were at least 5 h between acquisition and sleep onset. A long retention interval as in our study, i.e. recall after the recovery night, apparently has no impact on memory consolidation, because other studies show even longer retention intervals (Gais et al., 2007). Similarly, there is obviously no circadian influence on already consolidated memories (Backhaus et al., 2008; Ellenbogen et al., 2006).

Secondly, adolescents might be different from adults when coping with sleep restriction. Consistent with the two-process model of sleep regulation, partial sleep curtailment to 4 h in one night was not associated with a significant reduction of the absolute time spent in SWS compared with a 9-h control night. In the sleep deprivation condition, the adolescents stayed awake for 20 h compared to 15 h in the control group, resulting in an increased homeostatic SWS pressure. This might explain why a bedtime restriction to 4 h and placing the sleep period in the second half of the night did not diminish the absolute time spent in SWS (Horne, 1988). Because declarative memory consolidation has been associated with SWS (Born and Fehm, 1998; Born et al., 2006), this ability for immediate compensation could be one reason for our negative findings. Another reason might be that perhaps our study protocol was not completely suitable to demonstrate SWS-dependent memory consolidation. We did not expect the adolescents’ pronounced ability to compensate for sleep deprivation immediately, i.e. already during the first night. In addition, we wanted to investigate explicitly sleep deprivation consistent with real-life conditions and actual sleep curtailment in adolescents. The study design was chosen to reproduce this scenario.

Plihal and Born (1997) reported a better consolidation of the paired-associate words after sleep compared to wakefulness in healthy adult men. The authors described that word recall was improved compared to the final recall during the learning phase, whereas our subjects showed poorer performance in both conditions. In a study by Gais et al. (2007), healthy adults also exhibited worse results in a word pair memory task between pre- (immediate after learning) and post- (after two nights of experimental condition) recall session. They forgot more words after sleep deprivation than after normal sleep; however, memory decay happened in both conditions.

As to be expected, subjective sleepiness ratings on the day after sleep restriction showed an acute response, which is supported by previous studies (Fallone et al., 2001; Hood and Bruck, 2002). A slightly increased sleepiness persisted until the morning after the recovery night after sleep restriction, which is in line with earlier findings in children (Carskadon et al., 1981a). Despite the increased self-reported tiredness, concentration and attention test results were not impaired following sleep curtailment. The study by Hood and Bruck (2002) identified significant differences in the impact of sleepiness on performance between non-pathological sleep-deprived subjects and narcolepsy patients. In the healthy subjects, there was no significant decrease in complex performance, which indicates that there are quantitative and qualitative differences in the nature of sleepiness between pathological and non-pathological subjects (Hood and Bruck, 2002).

Another interesting result was the significant positive correlation between non-REM sleep in the recovery night after sleep deprivation with the performance in the word pair task. The correlation was found only in the restricted night, suggesting that subjects who respond to sleep restriction with a relative preservation of non-REM are the ones with a better memory consolidation than the ones who respond to sleep restriction with a loss of non-REM. REM sleep was related negatively to memory performance in the paired-associate word list task; however, this was not statistically significant and most probably secondary to the positive correlation with non-REM sleep in percentage. This is in line with results from a recent study by Backhaus et al. (2008), who also reported a positive correlation between the retention of word pairs and the proportion of non-REM sleep, and negative correlations with the percentage of REM sleep.

Sleep during school age and adolescence shows a characteristic pattern of change, with reduced sleep periods on school nights and extended sleep periods of sleep during the weekend and holidays (Fredriksen et al., 2004; Giannotti et al., 2002; Gibson et al., 2006; Loessl et al., 2008). Furthermore, there is considerable individual variation in sleep requirements. Obviously, balancing their ‘sleep-account’ is routine for adolescents, and they have many problem-solving skills to compensate for sleep deficits. Nevertheless, the negative finding of one night of sleep curtailment in our study may not be predictive of the more extended effects of chronic sleep loss on memory performance in adolescents.

For the classroom setting, we would assume that a state of emotional arousal or increased effort due to increased motivation can overcome sleepiness and fatigue for a short period of time without observed changes in performance. Nevertheless, complex tasks that require simultaneous abstract thinking, creativity, integration and planning might be impervious to compensatory motivational strategies (Dahl, 1996).


To understand more clearly the developmental aspects of the relationships between learning, memory and sleep, future research needs to translate current approaches to the study of children and adolescents. Memory consolidation in adolescents, especially after sleep deprivation during their daily routine, has not yet been investigated systematically.


This research was supported by the German Research Foundation (‘Deutsche Forschungsgemeinschaft’; Grant VO 542/7-1) and the Freiburg University Scientific Society (‘Wissenschaftliche Gesellschaft Freiburg i. Br.’). The authors wish to thank the technical staff of the Sleep Laboratory at the University Hospital Freiburg.