Introduction

We remember life’s important moments especially well; whether it be a happy childhood memory, or a fear-inducing experience, these life events seem to be retained and recalled. These emotional memories are the core of our personal history. Emotional memory refers to the encoding and storage of stimuli and events that elicit an emotional response. This aspect of memory can be measured by arousal (level of excitement) and valence (negative or positive)[1]. The mechanisms through which sleep impacts the various components of emotional memory are evasive. Existing studies point to a range sleep of characteristics that may be contributing to the brain’s ability to encode and consolidate emotional content, however, there have been few replicable studies [2,3,4]. In addition to sleep characteristics, the timing of the sleep opportunity itself, relative to the learning process has not been consistent across studies [5]. Methodological variances across existing studies make it difficult to determine whether sleep preferentially enhances emotional memory consolidation above and beyond the consolidation of neutral memories [5, 6]. Further, the specific impacts of a common sleep disorder, obstructive sleep apnoea (OSA), on emotional memory are poorly understood [7]. To this end, this review will consolidate existing evidence pertaining to the models and mechanisms through which sleep is theorised to impact emotional memory and apply it to a specific sleep disordered population.

Sleep and Encoding

There is strong evidence that sleep prior to learning neutral information is a critical process for memory consolidation [8,9,10]. Whether sleep prior to learning has a greater effect on emotional memory consolidation has not been elucidated. Van der Helm, Walker [11] discuss the impact of prior sleep on the brain’s ability to encode information, particularly emotionally charged information. When tested on positive, negative, and neutral material after 36-h of sleep deprivation, the sleep deprived group showed a 40% memory deficit across all valences compared to controls [8]. Further, while neutral and positive content showed a 59% retention deficit, negative memories remained resistant to deterioration [8, 11]. This effect is supported by Tempesta et al. [12], who found that one night of sleep deprivation before learning was associated with a deterioration in recall ability when compared to participants who slept during the previous night. When comparing the effects of sleep loss on neutral, negative, or positive material, negative stimuli were also the most resistant to deterioration [12]. Evidence from neuroimaging has shed light on the potential mechanisms of this negative memory bias. A functional MRI study by Yoo et al. [13] found that amygdala activation in response to negative stimuli was increased by approximately 60% in sleep deprived participants, compared to those allowed to sleep prior to scanning, suggesting that this heightened neural activation may enhance encoding.

Kaida and colleagues [9] examined the specific contribution of REM sleep on emotional memory encoding [9]. They assessed memory recall to emotional images in 16 healthy males under two experimental conditions: one after total sleep deprivation and the second after REM deprivation (induced by playing tones contingently upon initial signs of REM sleep). Total sleep deprivation (TSD) resulted in deterioration of emotional memory encoding independent of valence. Contrary to expectations, REM deprivation did not selectively impair encoding of emotional memories, suggesting that encoding ability, independent of the valance of the stimuli, depends instead on prior NREM sleep. Van Der Werf et al. [14] suggest that slow wave sleep (SWS) spindles are a vital precursor to encoding, and that deprivation of this activity leads to reduced hippocampal activation during encoding and subsequent learning impairment [15].

It has been suggested that encoding deficits may be due to attentional impairments associated with sleep loss. Kaida et al. [9] found similar deficits on the psychomotor vigilance task in both the TSD and REM deprivation conditions, suggesting that attentional impairments are not the only explanation for sleep deprivation-related encoding impairments. Interestingly, other studies have also shown that despite participants reporting less subjective sleepiness when caffeine is administered during periods of sleep loss, the negative effect on memory is not overcome [11]. This suggests that a specific characteristic of sleep is contributing to this effect above and beyond an attention deficit that may be incurred by sleep deprivation [14].

These studies highlight the importance of sleep prior to learning to prepare the brain for retention of emotional information. More critically, a lack of sleep prior to encoding may lead to a negative memory bias, however more evidence is needed to confirm this conclusion.

Sleep and Consolidation

Emotional Versus Neutral Content

In the past decade, the role of sleep in consolidating emotional memories has received increased attention. Emotionally arousing content is generally better consolidated than neutral content [16] but whether sleep further enhances this effect remains to be confirmed. Some studies suggest that a sleep opportunity post-learning has an enhancing effect on consolidation compared to wakefulness, with a preferential effect on emotional over neutral content [3, 17]. Cunningham et al. [18] found that when subjects have a sleep opportunity following the encoding of negative scenes, recognition of objects in the scenes was significantly better compared to a daytime wakefulness group. A comparison to neutral content was not included in this study.

A meta-analysis conducted by Lipinska et al. [19] suggested that a preferential effect for the consolidation of emotional material over neutral only occurred under certain methodological conditions, with potential moderating factors being the type of stimulus and the measurements used to assess memory, such as recall or recognition tasks [19]. Similarly, Davidson et al. [6] discuss how sleep preferentially enhancing the consolidation of emotional memories over neutral memories has not been consistently replicated. The authors also point to methodological discrepancies across studies as a likely contributing factor, as well as sample size and statistical analyses [6].

Existing studies have explored the effect of sleep on emotional memory consolidation by comparing sleep and daytime wakefulness, and sleep and nighttime wakefulness (i.e., sleep deprivation) [12]. While it seems logical that if sleep enhances the consolidation of emotional memory, then sleep deprivation would impair it, however this is would be an oversimplification. The consolidation of emotional memory can be assessed by performance on recognition and recall tasks, and by assessment of affective tone [20]. The preservation of affective tone and specific details of emotional content do not appear to be affected by sleep in the same way [11, 20, 21]. Van der Helm, Walker [11] proposed a “sleep to remember, sleep to forget” model where they hypothesise that sleep after encoding enhances the retention of detail but reduces the intensity of affective tone. This finding was not replicated Tempesta et al., [20], who found that ratings of valence on positive, neutral, and negative content did not change before and after a sleep opportunity. Tempesta et al. [20] compared the sleep opportunity group to a sleep deprivation group and found that the sleep deprived group rated positive, neutral, and negative content as more negative following sleep deprivation, suggesting that sleep deprivation universally depresses affective tone and good sleep preserves it. The effect of sleep deprivation on emotional memory consolidation may have considerable clinical implications, which are discussed further in the chapter in this series by Ogden & Drummond. Baran et al. [17] also found that a sleep opportunity preserved affective tone of emotional memories, however, the comparison group in this study experienced daytime wakefulness, as opposed to sleep deprivation. It is important to consider the timing of wake and sleep relative to circadian rhythms and the learning process in the study of emotional memory, as it is not known whether this influences the processing of detail and affective tone. Cunningham et al. [5], in their review, discuss how the temporal location of sleep across the memory consolidation process may be a key factor in explaining the elusive effect of sleep on emotional memory. A further extension on this may be to investigate the temporal location of sleep relative to both the learning process and circadian rhythms. Studies suggest that the ability to consolidate information fluctuates with circadian rhythms, with some times of day being more optimal than others [22]. Future studies might consider a methodology that builds on the participants individual body clock.

Positive Emotional Content

The broad hypothesis that a post-learning sleep opportunity enhances memory for emotional content implies that all emotional content will be enhanced or impaired in the same way, but there is evidence to suggest that this is not the case. Tempesta et al. [12] identified a gap in the literature wherein most studies investigating the effect of sleep on the consolidation of emotional content used negative stimuli, leaving positive stimuli relatively unexplored. Tempesta et al. [12] incorporated and compared content with a negative and positive valence in their study investigating emotional content consolidation after nocturnal sleep or wakefulness. Across both groups, emotional material was better recognised than neutral material, however, there was no preferential effect for positive or negative content in particular [12].

Lipinska et al. [19] found that when emotional stimuli consisting of both negative and positively charged content is compared with neutral stimuli, there is no benefit of sleep on the consolidation of emotional content over neutral content. These findings contrast with a study conducted by Chambers, Payne [23], who found that a 12-h period of sleep following learning, compared to the same period of wakefulness, preferentially enhanced memory for positive over neutral information. This study used humorous and non-humorous content to represent positive and neutral valence, respectively. Other studies have utilised facial expressions [24], clips from romantic historical movies [12], or even erotic images [25]. The range of ways “positive” valence has been operationalised may be contributing to the inconsistency in findings and should be addressed in future studies.

Ultimately, the existing literature exploring the role of sleep in the consolidation of emotional memory is inconsistent. While an enhancing effect of sleep may exist, further investigation into the specific variables that may be contributing to the effect is required. Lipinska et al. [19] note that the method of recall varies across studies and the most pronounced preferential effect for emotional content over neutral content was reported when participants completed a free recall task, as opposed to a forced recognition task. Other inconsistencies across the literature include whether participants in sleep groups had a full night of nocturnal sleep, or a daytime nap, and whether participants in the wake group experienced daytime wakefulness or sleep deprivation overnight [4,5,6, 19].

Effect of Age

Episodic memory is known to decline as we age, and it appears that the retention of emotional information undergoes changes throughout the lifespan. For example, when emotional and neutral stimuli are presented at a learning opportunity, the effect of post-learning sleep on consolidation is not consistent across age groups [26]. Jones et al. [27] found that a post-learning sleep opportunity enhanced recognition for negative and neutral content in young adults (aged 18–30), but not for middle aged adults (aged 35–50), suggesting a negative memory bias in younger adults. The authors posit that the enhancing effect of sleep on this negative memory bias, therefore, declines with age [27]. Huan et al. [26] built upon this finding by incorporating emotional content of both a negative and positive valence in a sample of 60 young adults (aged 19–23) and 60 older adults (aged 61–71). Older adults displayed an enhancement to positive, but not negative, emotional content when sleep followed learning; thus concluding that the effect of sleep on emotional memory does not decline with age, only the negative memory bias seen in younger adults [26]. This finding was consistent with an earlier study conducted by Gui et al. [28] who also reported a trend of older adults experiencing enhancement to positive emotional content when sleep follows learning.

The variation of the influence of sleep on emotional memory across the lifespan suggests that the mechanism contributing to the effect likely involves a cognitive component. Huan et al. [26] note the socioemotional selectivity theory, which posits that older adults will prioritise emotionally nourishing experiences more so than younger adults, who will prioritise education and the pursuit of long-term goals, as they become aware of a shrinking time horizon [29]. Increased positive mood, emotional regulation, and emotional resilience in older age [30, 31] may contribute to this shift in positive memory consolidation, however, further studies are needed to fill this knowledge gap.

Mechanisms

Below we discuss three mechanisms through which sleep may influence the encoding and consolidation of emotional material: first, the influence of physiological arousal at encoding and the existing evidence for its capacity to predict recognition accuracy; second, how specific neural circuits and neurotransmitters, whose activity is modulated by sleep, may be related to emotional memory; and finally, the role of sleep macro-architecture in the encoding, consolidation, and recognition of emotional content.

Emotional Valence and Physiological Arousal

Cunningham et al. [18] theorised that physiological reactions to emotional content were a crucial first step in the process of encoding emotional memory. In their 2014 study, the authors measured skin conductance response (SCR) and heart rate deceleration at encoding and recognition for two groups who were exposed to neutral and negative emotional stimuli [18]. Both groups had an encoding and recognition session, which were separated by 12 h of either daytime wakefulness or a nocturnal sleep opportunity [18]. The magnitude of the physiological reaction measured at encoding was related to better recognition at a subsequent memory test, but only when the delay between the sessions included a sleep opportunity [18]. Cunningham et al. [18] posit that the physiological response “tags” the experience for later sleep-based consolidation processes. This finding supports earlier work by Buchanan et al. [32], and Abercrombie et al. [33] who note that increased physiological reaction at encoding is associated with enhanced memory, while also extending the paradigm to incorporate sleep as a potential mechanism. Abercrombie et al. [33] reported that stimulus words that elicited a greater physiological response were associated with better recall performance two days later. Memory consolidation processes that occur during sleep may also contribute to this learning enhancement Cunningham et al. [18].

Other studies extended the work of Cunningham et al. [18] and added that the stress hormone cortisol may be responsible for the enhancement observed when physically arousing stimuli are better retrieved after a period of nocturnal sleep compared to daytime wake Bennion et al. [34] Payne, Kensinger [35]. While Bennion et al. [34] note that cortisol has been linked to memory in previous literature, the authors add a novel perspective by investigating cortisol at encoding, as opposed to during sleep as other studies have done. This study found that increased cortisol at encoding predicted memory performance after a 12-h delay, however, this effect was only true when a period of nocturnal sleep occurred. The authors note these findings as further evidence of the “emotional tagging” theory, wherein emotional memories are prioritised for consolidation during sleep opportunities Bennion et al. [34]. Payne, Kensinger [35] introduce a model describing how sleep and stress interact to enhance the consolidation of emotional memories. The intricate mechanisms behind the “tagging” phenomenon require further investigation but begin to depart from the concept of physiological arousal, so are further discussed below.

The relationship between emotional memory enhancement and physiological arousal is unlikely to be linear [36]. According to the well-known Yerkes-Dodson law, there may be a point at which stimuli of a high emotional valence are associated with decreased memory, however, this has not been tested on humans. Investigating a ceiling to the relationship between memory enhancement and physiological arousal would be a worthwhile future direction to elucidate the mechanisms behind this effect. Bennion et al. [36] also posit that the attention-narrowing effect might also be contributing to the heightened physiological responses, which could potentially be further explored with distractor tasks and eye tracking technology.

Neural Circuits and Neuromodulators

Two neuromodulators that have consistently appeared as relevant to the encoding and consolidation of emotional memory are cortisol and norepinephrine [34]. As discussed above, Bennion et al. [34] found that cortisol at encoding predicted recall after a 12-h delay only when a period of nocturnal sleep occurred. Neuroimaging data from the same study showed different patterns of neural activation during recall between the sleep and wake group [34]. Notably, the brain regions active in the sleep group appeared to be more ordered and refined than those active in the wake group [34]. From this, the authors suggest that the increased cortisol at baseline, combined with the nocturnal sleep opportunity, facilitated the refinement of neural circuits utilised for recall, but stress that this process only occurs during sleep. This echoes the study conducted by Payne, Kensinger [37] who discuss how different neural circuits are engaged for recall after sleep and wake. It has been observed that after sleep, the connection between regions associated with emotional memory—the amygdala—have stronger connections to the hippocampus and ventromedial prefrontal cortex [35, 37]. Studies investigating exactly how this strengthening occurs are yet to be conducted [34].

Norepinephrine (NE), commonly known as the stress hormone, has been extensively shown to be involved in the process of consolidating emotional memory [38]. A well replicated finding is the administration of propranolol (an NE inhibiting medication) prior to the encoding of emotional material, and a subsequent decrease in memory performance compared to a placebo [38]. This effect of NE on encoding emotional material is well established, however the way NE interacts with sleep to consolidate emotional memory is a relatively new direction for the field [39]. NE production and circulation decreases during most stages of sleep, with the exception being SWS [39]. In SWS, regions of the brain associated with the production of cortical NE have been observed to be active and projecting to neural structures associated with emotional memory such as the amygdala and hippocampus [39]. Based on this pattern, Groch et al. [39] investigated the effect on emotional memory when NE is inhibited by intravenously administering clonidine. The disparity between correct recall of neutral and emotional content was diminished, suggesting that the preferential consolidation of emotional material was impaired by the inhibition of NE during sleep Groch et al. [39]. Further studies investigating the role NE of on emotional memory consolidation during sleep may shed further light on how neuromodulators are involved in nocturnal neural processes.

Sleep Architecture

The role of the unique sleep stages in the consolidation of emotional memories has gained attention in recent years [4]. In particular, the importance of REM sleep and SWS are hypothesised to have unique contributions to the preferential effect for consolidating emotional memories over neutral ones [4]. Morgenthaler et al. [3] investigated the effect of REM-specific sleep deprivation on emotional memory consolidation and found no significant effect, suggesting that SWS is more involved in emotional memory consolidation. However, Wiesner et al. [40] reported that REM specific sleep deprivation was associated with decreased retention of negative stimuli encoded during a prior learning session. Cairney et al. [41] reported that SWS and sleep spindles seem to be associated with faster response time on an emotional memory test following sleep, but not accuracy. Evidently, results across the literature are mixed.

SWS and REM sleep may facilitate distinct but complementary steps in the consolidation of emotional memory, moving away from the idea that one of these sleep stages is inherently more important than the other [41]. Cairney et al. [41] reported that SWS appears to be involved in a change of retrieval location for emotional memories, with a preferential effect for aversive memories over positive ones. Following this retrieval relocation, activity in REM sleep strengthens the neuronal connectivity in the hippocampus, theorised to be facilitated through cholinergic activity and theta oscillations [34, 41]. This framework is in part supported by the meta-analysis conducted by Schafer et al. [4], who found evidence supporting previous findings about the role of theta oscillations in neural structures concerned with emotional memory. The role of SWS, however, continues to be inconsistent across the literature, with some studies suggesting it is more involved with consolidating neutral memories [4, 34], and others finding a preferential effect for negative memories [41].

Implications for Clinical Populations

The significance of sleep on the consolidation of emotional memory is clearly supported by the literature, despite exact mechanisms remaining elusive. A natural line of questioning that then follows is the impact of long-term sleep disruption on emotional memory in the context of sleep disorders.

Obstructive sleep apnoea (OSA) is a common sleep disorder characterised by disruptions to breathing while asleep by way of partial or total airway collapse resulting in decreased oxygen saturation, and frequent arousals from sleep [42,43,44]. These frequent awakenings often prevent the individual from achieving adequate time in deep stages of sleep, leading to an array of daytime consequences, including memory impairment [45]. Deficits in multiple domains of memory, such as autobiographical, procedural, and episodic, have been reported in the literature [46,47,48], but research on emotional memory in OSA is sparse [49].

Cunningham et al. [49] recently reported that individuals with OSA experienced a recognition deficit for content encoded the previous night compared to healthy controls. Recognition for negative content was particularly impaired in the OSA group, and that across groups, recognition accuracy was positively correlated with REM sleep [49, 50]. Given the known reduced REM sleep in OSA [42], and the emerging evidence supporting REM sleep as a crucial step in the consolidation of emotional memory, it is logical that a deficit in this type of memory has been observed. This finding is likely to be clinically significant, as there are well-supported negative consequences of impaired emotional memory on health and wellbeing [42]. Further, as discussed earlier, sleep deprivation prior to encoding can lead to a negative emotional memory bias [11, 12]. Consistent REM sleep deprivation is also associated with decreased emotional regulation [51]. Experiencing impairment to emotional memory, and in particular, a negative memory bias, may increase the risk of mood disorders in OSA patients [52]. These consequences are all in addition to other medical and physical consequences that OSA can incur, such as daytime sleepiness, hypertension, and cognitive decline [42, 53, 54], which can interact to increase the risk of psychopathology. OSA is just one clinical population in which the impacts of sleep on emotional memory processes are important to investigate; there are other clinical conditions in which individuals experience difficulties with sleep and changes in emotional processing, including insomnia, depression and PTSD that are under investigation.

Future Directions

The effect of sleep on the consolidation of emotional memories remains an elusive concept, with an extensive range of potential influencing factors. In order to make sense of the inconsistent results across studies there are few future directions we believe are worth considering. Firstly, and as suggested by previous authors Cunningham et al. [5], Davidson et al. [6], consistent methodology across studies will be imperative to solving the replicability crisis this field is currently experiencing. Factors such as sample size, the type of stimulus, time of day of testing, the measures selected to assess memory, and statistical analysis, are important considerations, but there are more conceptual considerations that may also benefit future studies. Cunningham et al. [5] raise a strong point regarding the consideration of when the sleep or sleep deprivation window occurs relative to the learning process. Existing studies appear to have predominantly had a “top-down” approach, in that different methodologies have been tested with the aim to infer mechanistic possibilities based on results. We propose that a “bottom-up” approach should also be considered. Sleep and learning are broad concepts, each with their own intricacies, which may be better studied by breaking down their respective stages and building up. Existing studies investigating the effect of different sleep stages [2, 41], brain activation [13, 34], and hormone fluctuations [34, 39] on emotional memory provide a strong foundation for further exploration.

Finally, understanding the role of sleep in the consolidation of emotional memories in the context of different clinical conditions has important implications. The relationship between sleep and memory in the context of sleep disorders such as insomnia and OSA, as well as emotional processing disorders, such as depression and PTSD, need further investigation, the results of which, may contribute to patient care, prevention, and management.

Conclusions

There is a growing pocket of literature investigating the role of sleep in the encoding and consolidation of emotional memory, however, results are mixed [4,5,6, 19, 23, 34]. Broadly speaking, sleep deprivation has been associated with an impairment to encoding emotional content, with negative stimuli being more resistant to deterioration [11, 12]. The influence of sleep after encoding has been more variable [4], with some studies finding a preferential effect for the consolidation of emotional content over neutral when a sleep opportunity is present [11, 37, 55] and others reporting no such effect [4, 6]. Evidence for the preferential effect for negatively toned content over positive content is also mixed [12, 19, 23]. Moderating factors of this effect may include age [26, 27], recall method [19], and timing [5]. When investigating the mechanisms that may be contributing to this effect, the “tagging” hypothesis consistently appears in the literature [18, 36]. As discussed above, this hypothesis posits that physiological arousal and elevated cortisol at encoding “tags” information for preferential processing during sleep [18, 36]. The exact processes that occur during sleep are widely debated [4]. Examining sleep architecture and its role in emotional memory consolidation mostly points to SWS being involved in the relocation of information, followed by the strengthening of connectivity with emotional centres of the brain in REM sleep [18, 34]. The importance of sleep in the consolidation of emotional memories prompts the investigation of clinical populations who experience chronic sleep disturbance or deprivation, such as the OSA population [42]. Emerging evidence suggests that individuals with OSA may experience impairments in emotional memory encoding and consolidation, which may place them at increased risk for negative wellbeing outcomes [47, 49, 50]. Remaining knowledge gaps include the nocturnal neural processes involved in the consolidation of emotional material, the role of physiological arousal at encoding, and the long term effect of sleep disturbance on emotional memory processing and mental health.