The Interactive Role of Sleep and Circadian Rhythms in Episodic Memory in Older Adults

Abstract

Adequate sleep is essential for healthy physical, emotional, and cognitive functioning, including memory. However, sleep ability worsens with increasing age. Older adults on average have shorter sleep durations and more disrupted sleep compared with younger adults. Age-related sleep changes are thought to contribute to age-related deficits in episodic memory. Nonetheless, the nature of the relationship between sleep and episodic memory deficits in older adults is still unclear. Further complicating this relationship are age-related changes in circadian rhythms such as the shift in chronotype toward morningness and decreased circadian stability, which may influence memory abilities as well. Most sleep and cognitive aging studies do not account for circadian factors, making it unclear whether age-related and sleep-related episodic memory deficits are partly driven by interactions with circadian rhythms. This review will focus on age-related changes in sleep and circadian rhythms and evidence that these factors interact to affect episodic memory, specifically encoding and retrieval. Open questions, methodological considerations, and clinical implications for diagnosis and monitoring of age-related memory impairments are discussed.

Sleep is vital to our physical (1,2), emotional (3), and cognitive health (4,5) and as such, being deprived of sleep has far-reaching consequences. One such consequence is impaired memory processing. These impairments are not limited to memory consolidation, which is commonly studied in the context of sleep (6). Sleep difficulties can also lead to impaired memory encoding and retrieval. Encoding and retrieval processes often require attentional control (7) that may be influenced not only by the amount of sleep or quality of sleep but also by circadian rhythms. Circadian rhythms are 24-hour cycles in physiological arousal, hormone fluctuations, and core body temperature, which differ widely between individuals and heavily influence when a person feels most cognitively alert (8,9). These endogenous rhythms are modulated by a cluster of neurons called the suprachiasmatic nucleus, located in the anterior hypothalamus (10). This internal clock influences physiology throughout the body (11). In rodents, lesioning of the suprachiasmatic nucleus results in disruption of these predictable 24-hour cycles as well as impaired hippocampal-dependent memory (10). In humans, disruption of normal circadian rhythms including operating outside of one’s optimal hours of alertness (such as in the case of jet lag), can affect cognitive functions including episodic memory performance (12), particularly with increasing age (13). Moreover, because sleep is deepest and most continuous at night, extreme or irregular circadian rhythms (eg, shift work) may exacerbate age-related reductions in sleep abilities (14). Because sleep and circadian rhythms are modifiable, understanding how they interact to influence memory is essential to investigations of episodic memory in older adults.

Although the standard sleep recommendation for adults is 7 or more hours of sleep per night (15), sleep demands can vary markedly between individuals and change throughout the lifespan. On average, sleep difficulties become more pronounced with age. Beginning in middle age (generally defined as between the ages of 40 and 60) (16), people spend more time in lighter stages of sleep (17), less time in deeper stages of sleep such as slow-wave sleep (18,19), and awaken more frequently throughout the night (20). As such, many studies have investigated whether age-related alterations in habitual sleep contribute to memory and other cognitive deficits (5,21–23). Findings from a recent meta-analysis suggest that less continuous sleep (measured with wake after sleep onset (WASO)) in older adults (those above the age of 60) (16,24,25) is positively correlated with poorer episodic memory performance (26). Specifically, WASO was more strongly correlated with older adults’ memory performance than other sleep parameters such as total sleep time, time spent in slow-wave sleep, and time spent in REM sleep (26). Thus, based on the research to date, more continuous sleep is most consistently associated with better memory performance in older adults (5,23,26–29).

Nonetheless, many studies of sleep and memory have not accounted for potential circadian factors which may heavily influence both sleep and cognitive processes critical for memory encoding and retrieval. The interaction between circadian rhythms and the drive for sleep, or sleep pressure, is illustrated by the two-process model of sleep regulation (30). In this model, drive for sleep is at its nadir in the early morning hours and increases throughout the day as sleep pressure accumulates. Sleep pressure reaches its peak at the onset of sleep and gradually dissipates throughout the night as sleep progresses. Overlapping with the drive for sleep are the fluctuations of the endogenous circadian clock, which is sleep-independent and instead ebbs and flows in accordance with the 24-hour day. Thus, the degree of sleepiness felt at any given point is due to the interaction between the circadian clock and drive for sleep, with sleep pressure being strongest when there is a greater distance between the temporal waves of these two processes (30).

Critically, circadian rhythms typically shift as we age, with older adults exhibiting a stronger preference toward morningness (13). One practical impact of this shift is that research studies taking place in the afternoon occur outside the average cognitive peak of healthy older adults (31). Thus, studies comparing performance between younger and older adults without regard to time of day may be artificially skewed toward poorer performance in older adults (31). For example, Maylor and Badham found that performance on an associative memory test decreased for both younger adults and older adults when tested at their non-optimal times of day (32). However, this effect was more pronounced in the older adults such that their performance was a near chance (32). Surprisingly, most research involving sleep and memory performance does not account for these time-of-day effects, or participants’ individual tendencies towards being “morning” or “evening” types, otherwise known as circadian typology or chronotype.

Much of the research examining chronotype and cognition also does not consider differences in sleep quality and quantity. For example, in a foundational study on chronotype and memory (31), both chronotype and accuracy of episodic memory retrieval were recorded, but subjects were not probed on their normal sleep patterns or how well they had slept prior to testing. This early experiment established the importance of both chronotype and time of day effects on episodic memory. However, now that consistent associations concerning the effects of sleep (26,33) and circadian rhythm factors (10) on memory processing have been established, it is imperative to understand how these factors interact with each other and age to modulate age-related deficits in episodic memory.

This review will examine the intersection between habitual sleep and circadian rhythms on episodic memory performance in older adults. Specifically, the first sections will discuss how differences in habitual sleep and circadian factors each modulate episodic memory functioning in terms of encoding and retrieval. We will review the small number of studies that have concurrently examined sleep and circadian rhythms in relation to episodic memory performance in older adults. The following section will discuss the interactive effects of sleep and circadian rhythms and how memory abilities for older adults in the current literature may be underestimated due to a lack of attention to this interaction. We discuss how assessments of sleep and circadian rhythms in research and clinical practice can provide a more accurate estimation of memory abilities and cognitive status in older adults. Finally, we raise methodological issues and outstanding research questions regarding the interaction of sleep and circadian rhythms to inform our understanding of age-related memory decline.

Episodic Memory Declines in Older Adulthood

Episodic memory is a form of conscious memory for events that are grounded in a specific time and place (34). In essence, episodic memories are the memories related to subjective human experience, providing individuals with a sense of their past, and allowing them to adapt their learning and behavior for the future (34). Episodic memory involves three main states: encoding, consolidation, and retrieval. Encoding is the process through which new information is initially learned (34,35). Consolidation involves the transfer of information from a labile, hippocampal-dependent state during encoding, to a stable long-term memory trace stored in the cortex (36). Retrieval occurs when disparate encoded information is integrated into a reconstructed mental representation, or “episode,” along with an awareness that the episode occurred in the past (34).

Distinguishable forms of memory content are also influenced by aging. There is ample behavioral and neurological evidence that episodic memories can be either based on familiarity or based upon recollection (37). Familiarity-based memory is often described as a feeling that something has been experienced before, but without memory of the specific details or context of the episode. In contrast, recollection is the retrieval of detailed memories. Throughout the course of healthy aging, recollection memory performance decreases, but recognition-based memory performance generally remains intact (37,38), providing evidence for a dissociation between these two forms of memory.

Episodic memory abilities can decline with normal aging. Encoding difficulties may be the result of degradation of sensory systems (39), impairments in selective attention (40), ineffective strategy use (41,42), or an inability to effectively bind target items together with related information, an effect described by the associative deficit hypothesis (43). According to the associative deficit hypothesis, item memory (as measured using a standard old/new recognition memory test) is comparatively preserved with aging, but associative memory is impaired due to the reduced ability to bind related information together (38,43). Evidence for the associative deficit hypothesis is robust and has been supported across a variety of associate types (eg, words, pictures, context, source) (38). The associative deficit may in part be due to differences in grey matter volume in the prefrontal cortex. In one study, older adults with the greatest associative memory had greater grey matter volume in both the dorsal lateral prefrontal cortex and the ventrolateral prefrontal cortex when compared with older adults with lower associative memory accuracy (44).

Memory consolidation is facilitated by sleep (6), and this relationship is moderated by age (45). A meta-analysis conducted by Gui and others found that sleep facilitated episodic memory consolidation, as measured by retrieval success, to a significantly greater degree in younger adults than older adults (45). Although sleep is critical for memory consolidation, older adults also showed a memory consolidation deficit when compared with younger adults during the same period of wakefulness (45), suggesting that consolidation during both sleep and wakefulness changes with age. It is worth noting that in this research, memory consolidation is inferred by comparing measures of retrieval during equivalent periods of sleep and wakefulness, which may or may not adequately inform consolidation success. Due to this, and the extensive body of literature devoted to the effects of sleep on memory consolidation (see review by Muehlroth et al. (46), for further discussion), it is not considered further in this review.

Retrieval difficulties may result from decreased executive control (47,48), implicit learning interfering with explicit retrieval (49,50), ineffective memory search strategies (51,52), or impaired retrieval monitoring (53). As an example of implicit interference, when told to ignore distractor words in favor of learning target pictures, older adults tended to encode both the distractors and the target items, an effect known as hyper-binding (49). Further, they were unaware that they had implicitly linked the items together when they were tested on these associations later (49), which suggests that older adults may encode irrelevant distractors that may impair the ability to retrieve relevant information in the future. More recently, Greene and Naveh-Benjamin found that immediate retrieval of picture pairs was not impaired in older adults, but only when the pairs were intact at test (52). After a delay, older adults had difficulty with a recognition memory test, but recollection was comparable to younger adults (52), indicating that specific, previously encoded details were no longer retrievable in this older adult sample. This is an instance where ineffective memory search strategies impair retrieval. Together, these, and other experiments suggest that changes at various stages of memory processing can result in age-related episodic memory performance deficits. Further, these changes have been shown to be influenced by sleep (45,54).

Sleep and Episodic Memory Abilities in Older Adulthood

Decreased sleep quality in old age has been associated with a subsequent decline in health status (22), and increased likelihood of cognitive decline and dementia (21). Two recent meta-analyses analyzed the effects of habitual sleep on cognitive performance in healthy aging. Lo and others found that cognitive performance as measured by verbal memory, working memory, and executive functioning, was impaired in both long sleepers (ranging from 8.5 to 11 hours of sleep per night) and short sleepers (less than 6 hours of sleep per night) (55). However, this meta-analysis examined only self-reported total sleep time, which limits the conclusions which can be drawn, as other parameters such as sleep continuity have been shown to have stronger positive correlations with episodic memory in older adults than total sleep time (26). Nonetheless, this finding intriguingly demonstrates that more sleep time is not in itself better for cognition. For example, someone with longer self-reported sleep duration may be spending a greater proportion of time in lighter stages of sleep, and as such, their sleep may be less restorative than someone who has a shorter sleep duration but spends a greater proportion of time in restorative slow-wave sleep.

A variety of studies have shown associations between slow-wave sleep, spindle activity, WASO, sleep efficiency, and total sleep time with memory performance in older adults. A meta-analysis by Hokett et al., demonstrated significant relationships between sleep architecture and memory performance when collapsed across various sleep and task measures among both older and younger adults (26). This included measures of slow-wave activity or delta power (<4 Hz EEG activity) (56), as well as sleep spindle activity which is highest during stage N2 sleep (57). However, analyses examining the moderating effects of age on sleep-memory associations found stronger relationships between sleep architecture and memory in younger adults, particularly for slow-wave sleep (26). In contrast, greater WASO was more strongly associated with poorer memory in older adults (26). Nonetheless, both decreased delta power (29) and reduction in fast spindle density (58) have been shown in individual studies to be associated with memory ability in older adults.

These findings support the notion that habitual sleep affects episodic memory processing in older adults independently of memory-task domains, but the importance of sleep architecture may shift with age. Hokett et al. speculated that slow-wave sleep may be less beneficial for memory in older adults, or a functional reorganization of sleep stages could explain these age differences (26). Alternatively, such a shift may in part be due to decreased generation of slow-waves in older adults compared with younger adults (59), which makes it more difficult to detect significant relationships with slow-wave sleep in older adults. Additionally, the association between WASO and memory may be stronger in older adults because healthy young adults tend to have little variation in WASO (60,61). Regardless, it is clear that age-related changes in both global sleep parameters such as WASO and slow-wave sleep, as well as quantitative EEG sleep measures such as delta power and spindle activity appear to be important for memory processing in older adults. Critically, all of these sleep measures discussed can be negatively impacted by disrupted circadian processes (62–64). Thus, a closer examination of how circadian rhythm changes interact with sleep to affect memory in older adulthood is warranted.

Circadian Stability, Cognition, and Memory in Older Adulthood

Circadian rhythms are daily fluctuations of the body’s regulation of internal processes such as arousal, body temperature, and hormone cycling (8,9). The circadian system also has a considerable influence on memory encoding and retrieval ­processes. For instance, rodent models demonstrate that the hippocampus receives direct and indirect feedback from the suprachiasmatic nucleus through changes in neurotransmitter (eg, GABA) and hormone (eg, glucocorticoids, melatonin) release as well as clock gene rhythms (10). Alterations in clock gene function result in decreased hippocampal-dependent memory performance and reduced neurogenesis in the hippocampus in rodents (10,65). This relationship suggests that hippocampal-dependent memory processes can be adversely affected by a disruption in circadian processes independently of sleep.

Manipulations of circadian rhythms also have a negative impact on cognition in humans. Wright et al., found that a manipulation of a 24.6-hour day instead of a 24-hour day desynchronized participants’ circadian rhythms with their sleep schedules over a month, leading to a significant impairment in learning (66). These detrimental effects of circadian misalignment on cognitive performance in humans are well illustrated by research on shift workers, who routinely have a ­mismatch between their circadian rhythms and sleep-wake times (12). Chellappa and others manipulated circadian alignment by having participants complete cognitive tasks during both a simulated “day shift” and “night shift (12).” Participants showed decreased performance on measures of attention, information processing, and motor abilities (12), suggesting that some forms of cognition may be more dependent on the 24-hour clock. Although this experiment did not find an effect on memory performance between those assigned to “day shifts” and “night shifts,” is important because it is one of the few to simulate the effects of circadian misalignment on cognition.

Correlational studies have similarly found cognitive impairments associated with shift work. Firefighters without sleep disorders had poorer verbal, visual, and composite memory scores when tested after an overnight shift compared with when tested before beginning a day shift (67). Similar memory deficits have been found cross-sectionally, with night shift workers performing worse than day shift workers on verbal memory (68), immediate recall (68,69), and total number of items remembered (68,69). However, research examining length of shift (eg, 12, 18, or 24 hours) rather than type of shift (eg, day shift or night shift), generally fails to find significant memory differences (70–72). Although fatigue is a serious concern for those working long shifts, circadian misalignment may have a more profound impact on memory performance. Critically, the chronic effects of circadian misalignment may have long-term consequences for cognitive health into older adulthood and negatively affect an individual’s cognitive trajectory.

Although circadian rhythms tend to be stable within a single individual, there is variability in the rhythm timing across individuals (9). Thus, the time of day of peak physical and cognitive performance varies across people. Circadian activity rhythms are a way to measure the stability of one’s daily circadian functioning and include measures such as interdaily regularity (stability of rest-activity from day-to-day), intradaily fragmentation (irregular rest-activity patterns over the course of a day), and amplitude (cosine function comparing the most/least active periods during a 24-hour day) (73).

Poor circadian activity rhythm stability (74–76) and amplitude (77–79) are commonly related to cognitive impairment. One study using structural neuroimaging found a strong positive relationship between intradaily circadian activity rhythm fragmentation and medial temporal atrophy (80). Sherman et al., scanned participants using functional MRI while they took a paired associates test and found that more stable circadian rhythms were correlated with better memory performance and that this relationship was mediated by fMRI ­activity in the hippocampus (81). Thus, older adults with more stable daily sleep/activity rhythms also had greater hippocampal activation when engaged in a memory task. This greater hippocampal activation in turn led to significantly better memory performance, providing further evidence for the link between circadian rhythm stability and episodic memory in older adults.

Effects of Chronotype on Cognition and Memory in Older Adulthood

Chronotype, or the propensity toward morningness or eveningness, is another circadian process that changes as a ­function of age. Starting between the ages of 30 and 35, chronotype trends increasingly towards morningness with each subsequent decade (82,83). Between the ages of 60 and 75, 73% of older adults are classified as morning types, 25% as intermediate types, and only 2% are evening types (13). These findings are robust and have been demonstrated both cross-sectionally (82–85) and longitudinally (86). The consistency of these findings suggests that this shift towards ­morningness throughout adulthood is not due to social or environmental influences. Rather, it is the result of biological changes which are linked to the aging process (87,88).

Coinciding with this age-related shift towards morningness is an increase in what is known as the synchrony effect, whereby participants tend to perform better during their optimal time of day in terms of alertness (31). According to the synchrony effect, people who feel more cognitively alert in the morning will perform optimally during morning hours, but people who feel more alert in the evening will perform best during evening hours. There is evidence for synchrony effects on some aspects of memory, in particular for associative memory (31,32). However, synchrony effects have not been consistently found for item memory (32,89,90). Whereas associative memory tasks are more dependent on effortful strategic memory abilities, item memory tasks can also be accomplished accurately using automatic processes that may be less affected by lifestyle factors (43). The task dependency of the synchrony effect may in part be due to shifts throughout the day in cognitive load, or the ability to remember and manipulate information over a short period of time. As such, it would be expected that the synchrony effect would be stronger for tasks that require more cognitive resources. These synchrony effects parallel the robust evidence that associative memory specifically declines with healthy aging (38). Thus, it is critical to understand the role of synchrony effects in aging on episodic memory performance.

Could Synchrony Effects Account for Age-Related Memory Deficits?

Among the few studies comparing synchrony effects of episodic memory in older adults (31,89,91), researchers have found that older morning-type and intermediate-type adults have comparable memory performance to younger evening-type adults in the morning hours, but older adults’ performance deteriorates throughout the course of the day (31,89,91). In one study, intermediate-type older adults’ delayed recall was significantly impaired in the evening hours when compared with memory recall performed in the morning and mid-day (90). Thus, the synchrony effect is not limited to those with strong morning tendencies. Because laboratory experiments can take place in the afternoon and early evening hours, cognitive differences measured between young adults and healthy older adults may be inflated (31). Much of the current memory research does not record circadian typology or time of testing, calling into question whether the memory deficits which older adults typically show are truly the result of declining memory, or rather, if they are the result of being tested at a non-optimal time of day.

From a practical standpoint, individual differences in chronotype also have implications for both diagnosis and monitoring of memory impairments in clinical settings. Walters and Lesk (92) compared the performance of older adults on a battery of neuropsychological tests used in the screening of dementia and found that measures of processing speed peaked at 0800 and sharply declined as the day progressed (92). This relationship was further moderated by age; the older half of participants exhibited worse performance in the afternoon compared with the young-old (92). Although the Walters and Lesk study (92) did not find time-of-day effects for measures of episodic memory, it highlights the need for researchers and clinicians to be cautious of potential time-of-day effects when drawing conclusions about cognitive performance in older adult populations.

The Interactive Effects of Habitual Sleep and Circadian Factors

Although sleep homeostasis and circadian rhythms are independent processes, disruption in one can substantially impact the other. For instance, sleep stages can be reduced or altered if sleep occurs during the day (66). Further, poor sleep at night can lead to napping and reduced activity during the day, which can blunt circadian amplitude (93). In a study with drosophila, Le Glou et al. found that circadian factors influenced the effects of sleep deprivation (94). Specifically, memories that were weakened by sleep deprivation could be recovered if tested at the peak time for wakefulness (94).

Using a multidimensional assessment of sleep, as with the concept of sleep health (95,96), can help identify the specific and interacting sleep and circadian processes that influence the various components of memory. The National Sleep Foundation created an initiative to develop such an assessment of sleep health, resulting in the Sleep Health Index (95). Factor analysis has since established that the Sleep Health Index is comprised of three factors- sleep quality, sleep duration, and disordered sleep (95), and is negatively correlated with assessments of disordered sleep such as the Insomnia Severity Index and the Pittsburgh Sleep Quality Index (97). The RU SATED model is another multidimensional measure of sleep health that encompasses dimensions such as regularity, satisfaction, alertness, timing, efficiency, and duration (96). A recent principal components analysis of the RU SATED model of sleep health found that all 6 dimensions have varying degrees of correlation, but also load on three distinct components—global sleep health, sleep timing and regularity, and sleep quality (98). The second component of sleep timing and regularity encompasses differences in circadian activity rhythms (98). Thus, examining multidimensional sleep health may be one approach to exploring the ­individual, interactive, or additive effects of the drive for sleep and circadian components of sleep described previously. Multidimensional sleep health may also elucidate the combined contributions of sleep pressure and alterations to the circadian pacemaker to various health conditions. Lee and Lawson found that a composite measure of sleep health explained more variance in both stress and chronic health conditions in middle- and older-aged adults than examining each dimension independently (99). Similar findings have been shown for depression (100), and risk of heart disease (101). However, there are no studies to our knowledge examining sleep health in relation to cognition or memory abilities, providing a promising and currently unexplored opportunity to investigate the interaction of habitual sleep and circadian rhythms in episodic memory.

Current Methodological Issues

There are potential methodological issues when examining the independent or interactive effects of sleep and circadian rhythms on episodic memory in older adults. For instance, sleep manipulations can influence circadian rhythms, and circadian manipulations such as light and sleep timing can negatively influence sleep (62,64). Ensuring that these factors are measured and kept constant within experiments is critical for understanding their independent influences. For instance, the effects of circadian rhythm disruption are commonly found with memory access broadly (102,103), whereas total sleep deprivation has been shown to impair next-day memory retrieval and the binding of relational memories (104,105). However, these distinctions may be partly due to differences in measurement between circadian versus sleep-focused studies. Nonetheless, as some studies show that pathological memory changes beyond that of healthy aging include impaired item and familiarity-based memory processing (106), an important avenue for further investigation is whether there are distinctions in how poor sleep versus circadian disruptions such as routinely having shifted, fragmented, or variable 24-hour schedules, lead to these pathological memory deficits.

Further, within the chronotype literature, there is a lack of consensus as to when people should be tested to best capture synchrony effects. Although many studies conduct morning testing between 0730 and 0900, some subjects are tested as late as 1030; similarly, evening testing of participants has been reported as early as 1700 and as late as 2100 (31,32,90,107). Greater precision of experiment timing, particularly during the evening hours is warranted because there are considerable changes in both body temperature and activity levels even within chronotypes throughout the course of the evening (108). Based on differences in body temperature shifts, recent research suggests that the most non-optimal time of testing for morning people is between 2130 and 2300, whereas the most non-optimal time of testing for evening people is between the hours of 0700 and 0900 (108). The difference between the time of day in which evening research is conducted compared to when cognitive performance is at its lowest may underestimate the role of the synchrony effect on memory performance, particularly for younger morning types.

Although manipulating the time of day in an experimental setting is both time and resource intensive, it is important for researchers who work with aging populations to be aware of and consider these time-of-day effects. If researchers want to equate older and younger adult performance, both populations should be tested early in the morning. If researchers instead are interested in testing optimum performance for older and younger adults, older adults should be tested early in the morning, whereas younger adults should be tested in the afternoon. Regardless, at a minimum, researchers should record the time of testing to use as a covariate.

Another external environmental factor known to influence circadian rhythms is ambient light. Research has found that there are age (109) and chronotype (110) differences in the amount of ambient light exposure throughout the course of the day. Healthy older adults who are exposed to more bright light throughout the day have greater sleep efficiency and nap less than those exposed to less ambient light (111). The length of light waves, time of light exposure, and length of bright light exposure may also affect next-day cognitive performance (112). Healthy older adults who were exposed to blue light for several hours in the evening performed better on a working memory task upon awakening the next morning compared with when they were exposed to white light the night before testing (112). Thus, like time-of-day, even if ambient light is not a variable of interest, it is recommended that this data be collected and considered as a model covariate.

To better understand the role of circadian typology and time-of-day effects on memory, more careful representation of chronotypes should be considered. For example, extreme evening and extreme morning types should be examined separately from those who are moderate evening or moderate morning people. It is currently unclear whether moderate evening and moderate morning people are more similar to intermediate types, or to their extreme chronotype counterparts. Treating circadian typology as a continuous variable could provide clarity to this question by allowing for a more nuanced examination of how the degree of morningness/eveningness is affected by time of testing within chronotype.

Methodological issues may also stem from the variety of memory tasks used to assess episodic memory. How episodic memories are evaluated varies greatly between studies, and thus poses challenges with generalizing the degree and types of memories which are most sensitive to disrupted sleep. For instance, standardized global measures of cognition such as the Mini Mental State Examination (113) or the Montreal Cognitive Assessment (114) are typically administered by a neuropsychologist or other practicing clinician with the intent of identifying global cognitive impairments and inform diagnoses of mild cognitive impairment and dementia (113,114). Although these measures may be useful in some research settings, such questionnaires provide a limited assessment within each cognitive domain (eg, episodic memory, processing speed). Restricting the number of questions specific to each cognitive domain has the benefit of being fast and easy to administer (113,114), but does not provide researchers with a more nuanced assessment of memory functioning, such as being able to evaluate recognition vs. recall, or encoding vs. retrieval success. It is recommended that researchers carefully select memory tasks based on the memory processes of interest and the conclusions one wishes to draw. For example, a graded old/new recognition memory task with options to respond based on a subjective judgment of whether retrieval involves recollection (R) or various degrees of familiarity strength (eg, F4, F3, F2, F1) can provide powerful measures of recognition performance, recollection and familiarity, and memory strength. A simpler and less time consuming Remember/Know paradigm provides the first 2 of these measures but lacks a memory strength measure. Thus, depending on the hypotheses being tested, researchers may need to design their own experimental paradigms; in others, the use of existing standardized measures of episodic memory may suffice.

Optimally, objective measures of habitual sleep and circadian activity rhythm patterns would be included in longitudinal population studies, as current research often consists of small, cross-sectional samples. However, these population studies are very time and resource intensive. An alternative source of population-level data could be obtained from consumer sleep and activity trackers. Current consumer sleep tracking models (eg, Fitbit) are comparable with research grade accelerometers (115,116), but overestimate sleep efficiency and underestimate sleep latency (ie, the time it takes one to fall asleep) and transitions between sleep stages when compared with traditional polysomnography (115–117).

Additional methodological considerations particularly for meta-analyses, include more standard definitions for what constitutes “older adults.” For example, of the studies cited in this review, most have older adult populations between the ages of 60 and 80. Such a wide age range allows for considerable variability in habitual sleep patterns, as both sleep quality and sleep efficiency begin to decline in a person’s 50s and continue throughout the rest of older adulthood (82,118). Similarly, there are inconsistencies in the literature as to the definitions of “short” and “long” sleep. For instance, studies included in the Lo et al., meta-analysis defined “long sleep” as being at least 8.5 hours, to as much as 11 hours of sleep, and “short sleep” as being less than 6.5 hours of sleep, to as little as less than 4 hours of sleep per night (33). A lack of consistency in operational definitions of short-sleepers and long-sleepers makes it challenging to discern what range of total sleep times is problematic for older adults.

Conclusions, Future Directions, and Outstanding Questions

Episodic memory impairments are a normal part of aging. However, the factors that allow us to maintain memory abilities into older adulthood as well as the factors that lead to pathological memory deficits remain poorly understood. The potential interaction between sleep homeostasis and circadian factors on memory performance in older adults emphasizes the need to merge the study of sleep and circadian rhythms. Since both sleep pressure and circadian rhythms influence sleep regulation (30), future research should aim to account for both dimensions of sleep and circadian rhythms, particularly when studying the effects of age on cognition.

The development of wearable technologies such as activity trackers and in-home EEG headbands offers the potential for early identification and treatment of sleep-related disorders in large, diverse samples of human participants. Activity trackers allow the wearer to monitor sleep patterns and bring sleep disturbances to the attention of health care providers. The use of in-home EEG headbands is a promising recent development that would expand accessibility for examining sleep architecture in larger-scale studies.

Further, preliminary research has found utility in the use of the more inclusive survey measure of sleep health for subjective measures of sleep related to both the drive for sleep (eg, sleep efficiency) and circadian processes (eg, regularity and timing). Researchers studying this intersection of sleep, circadian rhythms, and episodic memory should include rigorous methods for assessing and manipulating sleep and circadian measures, and include more detailed memory assessments that span encoding and retrieval, as well as item and associative memory processes.

Moving forward, it will be imperative to understand both modifiable lifestyle factors and the underlying neurobiology which distinguish healthy agers from those who go on to develop neurodegenerative disorders such as mild cognitive impairment and Alzheimer’s disease. Because sleep is known to be important for learning and is impaired in both healthy older adults, and to a larger degree, those suffering from dementia, this is a promising area of future aging and memory research.

Funding

This work was supported by the National Science Foundation (grant 1460682 to M.E.W.).

Conflict of Interest

None declared.