Edited by: Fannie Onen, Hôpital Bichat-Claude-Bernard, France
Reviewed by: Jiu Chen, Nanjing Medical University, China; Michael Craig, Heriot-Watt University, United Kingdom
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
An extensive psychological literature shows that sleep actively promotes human episodic memory (EM) consolidation in younger adults. However, evidence for the benefit of sleep for EM consolidation in aging is still elusive. In addition, most of the previous studies used EM assessments that are very different from everyday life conditions and are far from considering all the hallmarks of this memory system. In this study, the effect of an extended period of sleep was compared to the effect of an extended period of active wakefulness on the EM consolidation of naturalistic events, using a novel (What-Where-When) EM task, rich in perceptual details and spatio-temporal context, presented in a virtual environment. We investigated the long-term What-Where-When and Details binding performances of young and elderly people before and after an interval of sleep or active wakefulness. Although we found a noticeable age-related decline in EM, both age groups benefited from sleep, but not from active wakefulness. In younger adults, only the period of sleep significantly enhanced the capacity to associate different components of EM (binding performance) and more specifically the free recall of what-when information. Interestingly, in the elderly, sleep significantly enhanced not only the recall of factual elements but also associated details and contextual information as well as the amount of high feature binding (i.e., What-Where-When and Details). Thus, this study evidences the benefit of sleep, and the detrimental effect of active wakefulness, on long-term feature binding, which is one of the core characteristics of EM, and its effectiveness in normal aging. However, further research should investigate whether this benefit is specific to sleep or more generally results from the effect of a post-learning period of reduced interference, which could also concern quiet wakefulness.
Episodic memory (EM) refers to personally experienced events, located in time and space, that are unique and whose retrieval depends on mentally traveling back in time to re-experience the previous encoding context (
According to the
Although there is ample evidence to support the specific implication of sleep stages, particularly SWS, in memory consolidation, other findings (
Both theories support the involvement of sleep in memory consolidation, either through stabilization (opportunistic hypothesis) or enhancement (the active hypothesis). Moreover, both consider active wake as conducive to new encoding and unfavorable for consolidation.
It should be noted that, with increasing age, there are substantial changes in the quantity and quality of sleep, such as reduced sleep duration, reduced SWS and REM sleep, changes in sleep timing and spindle density but also changes in sleep continuity/fragmentation (
Intriguingly, studies that examine the effect of aging on declarative memory consolidation have provided conflicting evidence. Some of them have reported that the benefits of sleep for EM consolidation were reduced in middle-aged adults (
Thus, the main aim of the present study was to evaluate the effect of daily life conditions on EM consolidation, through an ecological approach focused on the incidental encoding of everyday life-like events and the retrieval of associative aspects of EM performances in younger and older adults after a period favoring the consolidation process (sleep) vs. a period that does not favor it (active wake). We investigated whether sleep benefits memory in a What-Where-When task in older adults compared to younger adults.
To this end, we used Virtual Reality (VR) technology, which creates immersion in situations close to daily life with experimental control of What-Where-When information and perceptual details (
First, whatever the interval type (Sleep or Active Wake), we predicted that younger adults would outperform older adults in binding performance (
A total of 40 younger adults (22 ± 3 years) and 40 older adults (69 ± 5 years) took part in this study. Four younger adults were excluded because they were naturally short sleepers (<6 h), leaving a final group of 36 younger participants. Younger adults were recruited from Paris Descartes University and through flyers placed around the university. Older adults were recruited from the University of the Third Age at Paris Descartes University. They were paid at a rate of 10€/h or with course credit. This study was carried out in accordance with the ethics recommendations of Paris Descartes University and was approved by the local ethics committee of the Institute of Psychology at Paris Descartes University. All participants were informed of the academic nature of the study and gave their written informed consent for their participation in the study in accordance with the Helsinki Declaration.
We ensured that all participants had unimpaired or corrected-to-normal vision. None of them had any prior history of drug or alcohol abuse or neurologic, psychiatric, or sleep disorders. Participants were instructed to be drug, alcohol, and caffeine free for 24 h prior to and during the experiment. Participants were fluent in French, and were matched on their verbal IQ as assessed by the Mill Hill test (≥percentile 50, French translation;
Participant characteristics: shown here are the means of demographic, inclusion, and neuropsychological measures across experimental groups.
Active wake group | Sleep group | Age effects | Interval type effects | Interaction | |||
---|---|---|---|---|---|---|---|
Younger adults | Older adults | Younger adults | Older adults | ||||
Participants (M/F) | 18 (5/13) | 20 (10/10) | 18 (10/8) | 20 (7/13) | |||
Age |
23 (4.85) | 70.30 (7.89) | 22.05 (3.03) | 69.20 (5.45) | 1319.02 | 0.62 | 0.004 |
η2 = 0.95 | η2 = 0.00 | η2 = 0.00 | |||||
Education∗ | 6.78 (0.55) | 6.20 (1.10) | 6.78 (0.73) | 5.30 (1.66) | 16.08 | 3.08 | 3.08 |
η2 = 0.18 | η2 = 0.04 | η2 = 0.04 | |||||
Mill Hill | 31.94 (4.77) | 38.05 (4.20) | 32.11 (5.93) | 35.85 (6.15) | 16.62 | 0.7 | 0.937 |
η2 = 0.18 | η2 = 0.00 | η2 = 0.02 | |||||
BDI | 3.33 (3.5) | 2.10 (2.45) | 4.2 (2.07) | 3.70 (2.68) | 1.9 | 3.75 | 0.25 |
η2 = 0.02 | η2 = 0.05 | η2 = 0.00 | |||||
MMSE | – | 29.10 (1.02) | – | 29.85 (0.98) | – | 0.65 | – |
TMT |
33.44 (18.08) | 62.15 (28.68) | 31.17 (16.78) | 68.05 (56.64) | 16.88 | 0.05 | 0.26 |
η2 = 0.19 | η2 = 0.00 | η2 = 0.00 | |||||
Digit span | 16.61 (4.03) | 13.90 (3.16) | 15.83 (2.57) | 13.30 (3.48) | 11.6 | 0.80 | 0.01 |
η2 = 0.14 | η2 = 0.01 | η2 = 0.00 | |||||
Family picture test (standard EM test) | – | 33.25 (7.93) | – | 32.25 (7.00) | – | 0.17 | – |
η2 = 0.00 | |||||||
What-Where-When span | 9.33 (2.20) | 5.75 (1.55) | 9.50 (2.43) | 5.95 (2.78) | 46.25 | 0.12 | 0.00 |
η2 = 0.39 | η2 = 0.00 | η2 = 0.00 | |||||
FAB | – | 17.20 (0.70) | – | 16.70 (1.26) | – | 2.41 | – |
Within each age group (young and older adults), participants were randomly assigned to either an Active Wake or a Sleep interval group and were individually tested (
Experimental design.
The sex ratio was equivalent across the four groups (χ2 = 3.7772,
Sleep characteristics were assessed in the first session during recruitment via the Pittsburgh Sleep Quality Index (PSQI;
Results of sleep measures: shown here are the mean questionnaire scores across experimental groups.
Active wake group | Sleep group | Age effects | Interval type effects | Interaction | |||
---|---|---|---|---|---|---|---|
Sleep measures | Younger adults | Older adults | Younger adults | Older adults | |||
PSQI∗ | 5.39 (3.55) | 5.75 (3.93) | 6.67 (3.53) | 6.80 (4.65) | 0.08 | 1.75 | 0.02 |
η2 = 0.00 | η2 = 0.02 | η2 = 0.00 | |||||
St. Mary’s Hospital (sleep duration hrs) |
7 (1) | 7 (1.6) | 6.3 (1.2) | 7.15 (1) | 2 | 0.82 | 2.37 |
η2 = 0.02 | η2 = 0.01 | η2 = 0.03 | |||||
SSS1∗∗∗ | 2.41 (0.9) | 1.85 (1.04) | 2.67 (0.84) | 1.74 (0.87) | 11.9 | 0.11 | 0.72 |
η2 = 0.15 | η2 = 0.00 | η2 = 0.01 | |||||
SSS2 | 2.25 (0.77) | 1.85 (0.93) | 2.66 (.78) | 1.74 (0.80) | 10.25 | 0.4 | 1.42 |
η2 = 0.13 | η2 = 0.00 | η2 = 0.02 | |||||
The virtual environment was created with Virtools Dev 3.0 software
The virtual environment is a multimodal urban environment created from photos of Paris based on previously validated virtual reality cities used in aging studies (
The virtual urban environment.
Using this virtual urban environment, we developed a What-Where-When VR task based on previously validated (VREM) tasks in normal aging (
Subjects underwent a training session in an environment devoid of relevant events and containing only general elements (e.g., building, trees, etc.) (
Training environment (familiarization phase).
Subjects were immersed in the VR environment, the light in the room was switched off in order to increase the immersion and sense of presence but also to ensure that all participants experienced the same room-condition. They were asked to visit the city and to pay attention to all the details in order to tell us afterward if they would recommend living in this city to a friend. They were also told that they would be asked to give an assessment of the virtual environment. The task involved incidental encoding as the participants were unaware that their memory would be tested afterward. The navigation lasted on average 10 min.
Five minutes after the encoding phase, subjects were asked to verbally report the maximum of events encountered during their navigation and the associated elements:
The accuracy of the recall of the what, where, when and perceptual details assigned to each of the 20 scenes was computed. We calculated a
To take one of the above-mentioned examples, if the participant correctly reported having seen a car accident, one point was given for factual information (car accident, What score) and for each correct piece of associated information: spatial location (in front, What-Where score), temporal situation (halfway through the navigation, What-When score), perceptual details (blue and gold cars, fumes, etc., What-Details), for binding score (What-Where-When) and for high binding score (What-Where-When and Details). If the perceptual details were incorrectly recalled, but factual, spatial, and temporal contents were correct, each recall was scored 1, except for high binding which was scored 0 (for detailed scoring, see
During the second session (after 12 h), the participant began by watching a film “The mysteries of the cosmos: the sun king” during 9 min 50 s in order to ensure that delayed free recall was performed in the same context for each participant. The delayed free recall was carried out and scored in a similar manner to the first immediate free recall.
During the second session (after 12 h), and a few minutes after the delayed free recall test, each participant underwent a visual recognition test: a series of 35 stimuli with 20 old stimuli (snapshots from the virtual environment, stimuli that participants had already seen) and 15 new stimuli (snapshots from another virtual environment, 8 of which were semantically related to the environment already seen and 7 not related) was presented to the participants on the laptop and they had to decide which items they had seen during immersion in the virtual environment. Then, for each item recognized, they were requested to say whether they could mentally relive the spatio-temporal encoding context of the event or whether they just knew it (i.e., a remember versus a know judgment,
We calculated the percentage of
At the end of the experiment, subjects completed a self-administered questionnaire to evaluate their immersion, sense of presence in the virtual environment, navigation difficulties, and assessment of the environment (
Evaluation of the virtual environment: shown here are the means of Virtual Reality navigation duration and debriefing scores across experimental groups.
Active wake group | Sleep group | Age effects | Interval type effects | Interaction | |||
---|---|---|---|---|---|---|---|
Younger adults | Older adults | Younger adults | Older adults | ||||
Navigation duration (s) | 596.05 (110.17) | 728.3 (187.76) | 609.56 (132.56) | 595.50 (107.10) | 3.42 | 3.48 | 5.24 |
η2 = 0.05 | η2 = 0.05 | η2 = 0.07 | |||||
Use of laptop∗ | 6.89 (0.32) | 6.22 (1.93) | 6.67 (0.84) | 5.50 (2.42) | 5.78 | 1.53 | 0.43 |
η2 = 0.08 | η2 = 1.53 | η2 = 0.00 | |||||
Navigation appreciation |
7.67 (1.78) | 7.20 (2.35) | 7.05 (2.01) | 7.80 (2.04) | 0.04 | 0.00 | 1.83 |
η2 = 0.00 | η2 = 0.00 | η2 = 0.03 | |||||
Presence |
18.28 (6.6) | 17.44 (8) | 13.28 (6.26) | 18.77 (8.29) | 1.86 | 1.15 | 3.43 |
η2 = 0.03 | η2 = 0.02 | η2 = 0.05 | |||||
Task difficulty |
2.33 (1.19) | 4.33 (3) | 3.28 (2.35) | 4.30 (2.8) | 7 | 0.34 | 0.73 |
η2 = 0.09 | η2 = 0.00 | η2 = 0.01 | |||||
All the analyses were performed using
We first assessed the baseline difference between
Only a predictable age difference was revealed on each test, but no effect of Interval type (Wake vs. Sleep) or Age group × Interval type interaction was found. For each age group, the cognitive performances on tests assessing executive function, flexibility and working memory performances did not differ according to the Interval type. Older adults did not differ on standard tests assessing EM and frontal functions according to the Interval type (see
The main effect of Age on the amount of sleep overnight using the St. Mary’s Hospital scale was not significant. Similarly the effect of interval type and the Age group × Interval type interaction were not significant. Subjective sleep quality assessed by the PSQI did not differ across Age groups and interval type and no Age group × Interval type interaction was found. Subjective measures of alertness and sleepiness assessed during the two sessions via SSS1 and SSS2 scales revealed a main effect of Age. Younger participants reported being less alert than older adults for both Interval types (Active Wake vs. Sleep). However, no effect of Interval type or Age group × Interval type interaction was significant for either session. Thus the two age groups were well-matched for sleep measures across interval type (Active Wake vs. Sleep) (see
Concerning navigation duration, no effect of Age and Interval type was found. However, we observed an Age group × Interval type interaction. As the
A preliminary check of initial performance at encoding (session 1) across interval type (Active Wake vs. Sleep) and Age group (Older adults vs. Younger adults) indicated an expected Age effect on
An expected Age effect was revealed for all the
In sum, younger adults performed better than older adults at the encoding session whatever the interval type, the participants for each age group were well-matched across interval type and no difference in performances occurred depending on when the encoding was performed (in the morning or in the evening).
For both age groups, EM recalls were diminished in the delayed recall relative to the immediate recall in the Active Wake interval and interestingly, they were enhanced following a sleep interval.
First, for both
Binding performance (number of What-Where-When and What-Where-When-Details associations) through Interval type (Active Wake vs. Sleep) across Age group (Younger vs. Older). Error bars represent standard errors of the mean. NB: for reasons of readability, the effect of age is not reported here.
Second, concerning the
EM subscores (What, What-Where, What-When, and What-Details) through Interval type (Active Wake vs. Sleep) across Age group (Younger vs. Older). Error bars represent standard errors of the mean. NB: for reasons of readability, the effect of age is not reported here.
Concerning
Concerning the
In sum, on the one hand, for older adults, all types of EM aspects as well as both binding performances were significantly poorer after the active wake interval and significantly enhanced after a sleep interval. For younger adults, binding performances and more especially factual-temporal associations were significantly poorer after an active wake interval and significantly enhanced after sleep.
Results and analyses are presented on
ANCOVA results for recognition performances: shown here are the percentages of correct recognition of factual information (What), contextual information (What-Where, What-When) and Remember judgments (R) correctly associated to factual recognition and the percentage of correct rejections of neutral and semantically related distractors.
Active wake group | Sleep group | Age effects | Interval type effects | Interaction | |||
---|---|---|---|---|---|---|---|
Younger adults | Older adults | Younger adults | Older adults | ||||
What recognition % | 73 (11.78) | 51 (15) | 68.6 (18.7) | 48.25 (21) | 0.43 | 0.78 | |
η2 = 0.32 | η2 = 0.00 | η2 = 0.01 | |||||
What-Where recognition % | 63 (12) | 43.75 (14.5) | 60.9 (19) | 41.25 (21) | 28 | 0.11 | 0.39 |
η2 = 0.28 | η2 = 0.00 | η2 = 0.00 | |||||
What-When recognition % | 45 (12.7) | 26.5 (12) | 42.5 (16) | 29 (15) | 24.6 | 0.00 | 0.73 |
η2 = 0.25 | η2 = 0.00 | η2 = 0.01 | |||||
53 (15) | 45 (15) | 54 (20) | 52 (17) | 1.05 | 0.42 | 0.0 | |
η2 = 0.01 | η2 = 0.00 | η2 = 0.00 | |||||
Neutral % | 90 (11) | 92 (17) | 97.7 (7) | 88.6 (16) | 1.35 | 0.32 | 2.76 |
η2 = 0.02 | η2 = 0.00 | η2 = 0.04 | |||||
Semantically associated % | 82.7 (17) | 93.8 (8.6) | 95 (7.5) | 89.4 (14) | 0.94 | 1.8 | 8.6 |
η2 = 0.01 | η2 = 0.02 | η2 = 0.11 | |||||
When computing the
Concerning the
In the present study, a naturalistic (What-Where-When) EM task, rich in details and spatio-temporal context, was implemented in a virtual environment and was used to evaluate the effect of extended overnight sleep vs. extended active wakefulness on the consolidation of personally experienced events close to daily situations, in younger and older adults. As expected, the findings showed an age-related decline in EM performance for older adults compared to their younger counterparts, but most importantly, they revealed for both age groups a decline in memory performances following a period of active wakefulness and enhancement following sleep. We will briefly discuss the age-related effects on EM functioning and consider forgetting performances following the wakefulness condition, then we will focus on the effect of sleep on EM consolidation.
Basically, it is assumed that age-related differences in contextual memory are greater than those in memory for content (
It should be mentioned that during the debriefing, older adults more frequently reported that the navigation was difficult compared to their younger counterparts; nevertheless, both age groups manifested an equivalent sense of presence in the virtual environment and their appreciation of the navigation was similar. The results of binding performance are in agreement with previous findings, suggesting an impaired binding in aging (
Thus, our results confirm a noticeable decline of EM in the elderly reported in several studies (
For both age groups, EM free recall was diminished in the delayed recall relative to the immediate recall in the Active Wake interval while it was strengthened following a sleep interval. This pattern can not be attributed to some confounding effects, as participants were well-matched according to the type of interval on basic neuropsychological performance and sleep measure. In addition, we checked for the elderly that there was no difference concerning standard EM assessment, depending on when the test was done (in the morning or in the evening).
When evaluating the effect of active wake on EM retention, our data from delayed free recall highlight a significant forgetting following the Active Wake interval for both age groups. All types of information (i.e., factual, contextual, and details associations) and their related binding were affected by forgetting in older adults, but it only concerned the association of factual and temporal information in younger adults. These findings support the idea of Active Wake as an unfavorable period for consolidation (
Retrospective interference is an explanation for forgetting in long term memory while memory consolidation is understood as a process increasing resistance to interference rather than permitting performance enhancement (
However, this deleterious effect of Active Wake on free recall disappeared on recognition and remember judgments for both age groups. This finding may suggest that the memory trace was still present after a period of active wakefulness (about 12 h), but less spontaneously accessible in free recall, maybe because of a reduction in executive functions at the end of the day. Alternatively, it may indicate that active wake did not protect against forgetting, but rather that our recognition task was less sensitive to detect deficits than delayed free recall.
When investigating the effect of sleep on EM retention, our results from free recall showed that for both age groups, sleep compared to active wakefulness enhanced feature binding which is one of the main characteristics of EM.
The consolidation of temporal information appears therefore crucial in long-lasting EM. This is in keeping with the role of sleep that has been found to favor the replay of new temporal information in a forward direction, which strengthens its integration to the EM trace in young adults (
In addition, the enhancement of binding in younger adults is in keeping with Rauchs’ study (
We may also suggest that studies using somewhat simple EM tasks (e.g., only factual information) or too remote from the interests of the elderly fail to demonstrate the effect of sleep while other studies using more complex EM tasks (e.g., factual and context) or a more naturalistic approach are able to reveal the benefit of sleep in aging (e.g.,
Indeed, the hippocampus plays a crucial role in binding together different components of the new memory representation, giving rise to a multimodal representation which includes the conceptual, perceptual, and emotional components of the experience (
Most noteworthy for our purpose, no effect of sleep was found in correct factual recognition and associated spatio-temporal context and remember judgments whatever the age group. However, we noted a beneficial effect of sleep in younger adults for the rejection of semantically related distractors, which is consistent with the fact that following sleep, enhancement of details and spatio-temporal recalls may lead to improved discrimination between the studied items and hence improved source memory (
Overall, the present study has the merit of directly comparing (i) the effect of an extended period of overnight sleep and active wakefulness on memory for naturalistic experience, and (ii) the performance of young and older adults on a What-Where-When task. Moreover, the present findings are based on a virtual reality version of a What-Where-When VR task rich in perceptual details which allowed us to measure in a more ecologically valid way the hallmarks of EM (incidental encoding of events, the long-term associative memory of factual with perceptual details and spatiotemporal context, free recall, and sense of remembering). This advantageously gives a controlled measure of EM, and is enjoyable for young and older people. This is important since previous studies showed that performance in this type of virtual What-Where-When task correlates with the memory complaints in everyday life of older people, while a standard EM test does not (
Nevertheless, the study has some limitations. First, we did not use a fully immersive and interactive device, thus our results could be different with other material such as a VR head-mounted-display (HMD) or treadmill, especially for the active wakefulness condition (
Regarding the specific role of sleep, the present study cannot determine whether the observed EM benefits using a naturalistic What-Where-When task are linked to a specific stage of sleep (
The present study suggested that, compared to a post-learning period of daytime active wakefulness, a post-learning period of overnight sleep not only protects against forgetting but can induce actual improvements in memory for a naturalistic experience. Most interestingly, the study showed that despite EM decline in elderly compared to younger adults, older adults still benefit from night sleep in supporting the consolidation of EM what-where-when and perceptual details components and their binding. Thus, the most important message is that sleep may still sustain efficient EM consolidation in healthy elderly people, at least concerning spontaneous accessibility to content memory and its context after a delay of 12 h and when the participant does not report sleep problems or disorders. Further studies should investigate the effect of sleep on long-lasting EM consolidation and the neural underpinning of this benefit of sleep in the elderly. This may enable future research to establish whether this profit on memory is only attributable to the special status of sleep or more generally to the presence of an extended post-learning period of reduced interference that may also concern quiet wakefulness.
KA and PP contributed to the conception and design of the study. EO and AG-B created the virtual environment. NH and KA contributed to the acquisition of the data. KA and PP organized the database and performed the statistical analysis. SN and VLC provided suggestions in the different drafts of the present manuscript. All the authors have read and approved the submitted version.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank all the participants for their time and E. Rowley-Jolivet for the grammar and style corrections of the manuscript.
The Supplementary Material for this article can be found online at: