Feeling memories: What is the role of sleep in the processing of emotional memories?
Literature Thesis Mona Zimmermann
11099119 Supervisor: Elsa Juan Second Assessor: Umberto Olcese
Abstract
Sleeping is an integral part of being alive. However, up to this day, all the functions of this state of lowered consciousness are yet to be uncovered. Sleep disturbances are implicated in a myriad of psychopathologies, indicating sleep’s functions as important avenue for research. Evidence has pointed to a role of sleep in memory consolidation and, subsequently, sleep’s involvement in emotional memory processing was suggested (e.g. by the ‘Sleep to forget, Sleep to remember’ model). Is sleep – and particularly REM-sleep - involved in the attenuation of emotionality that we attach to our memories? This review set out to answer this question by discussing studies investigating the neural, physiological and subjective levels of emotionality. Overall, while, neuroimaging studies do point to an attenuation of emotionality via sleep in general, physiological and subjective findings are more mixed. The nature of sleep’s involvement in comparison to wakefulness remains unclear. Looking at specific sleep stages, SWS instead of REM-sleep seems to be a plausible alternative sleep stage, potentially involved in attenuating processes, while experiencing more REM-sleep might protect the emotionality of memories. Some studies suggest that the attenuating role of sleep might need more time to unfold. Future research should address the limitations of current studies, and focus on direct replications to achieve a clearer picture of sleep’s role in emotional memory processing. Keywords: Sleep | REM-sleep | Emotional Memories | SFSR model | Emotional tone
Table of Contents
INTRODUCTION ...4
1. EMOTIONAL MEMORY ...6
1.1THE MEMORY BENEFIT FOR EMOTIONAL EVENTS ...6
1.2THE DEVELOPMENT OF EMOTIONALITY OF EMOTIONAL MEMORIES ...8
2. SLEEP AND EMOTIONAL MEMORY ...11
2.1THE NEUROPHYSIOLOGY OF SLEEP ...11
2.2SLEEP AND EMOTIONAL MEMORIES:THE ‘SLEEP TO FORGET, SLEEP TO REMEMBER’ HYPOTHESIS ...13
BOX 1–REM VS.SWS AS OPTIMAL STAGES FOR EMOTIONAL MEMORY PROCESSING? ...14
BOX 2–TYPICAL STUDY DESIGN:EMOTIONAL MEMORY CONSOLIDATION DURING SLEEP ...15
2.2.1 Sleep and the processing of emotional memories: Neuroimaging studies ...17
2.2.2 Sleep and the processing of emotional memories: Physiological measures ...18
2.2.3 Sleep and the processing of emotional memories: Subjective Measures ...19
2.2.4 Sleep and the processing of emotional memories: Other study designs ...21
BOX 3–DREAMING ...22
DISCUSSION ...24
POSTULATE (1):NEURAL MARKERS OF EMOTIONALITY SHOW ATTENUATED REACTIVITY TO EMOTIONAL MEMORIES AFTER A PERIOD OF SLEEP...24
POSTULATE (2):PHYSIOLOGICAL MEASURES OF EMOTIONALITY DO NOT SHOW CLEAR ATTENUATED REACTIVITY TO EMOTIONAL MEMORIES AFTER A PERIOD OF SLEEP ...24
POSTULATE (3):SUBJECTIVE MEASURES OF EMOTIONALITY SHOW DO NOT SHOW CLEAR ATTENUATED REACTIVITY TO EMOTIONAL MEMORIES AFTER A PERIOD OF SLEEP ...25
LIMITATIONS ...26
DIRECTIONS FOR FUTURE RESEARCH...27
CONCLUSION ...28
Introduction
Sleeping is an integral part of being alive. Not surprisingly then, people have been fascinated by this mysterious state of lowered consciousness over many, many decades. Upon awakening from a period of sleep, we sometimes seem to remember, for example, study material better than before going to bed, might experience changes in our mood when we wake up from a bad dream or might feel well rested and more relaxed than we felt when going to bed. Up to this day, researchers are working on uncovering the functions of sleep - with heated debates governing the scientific discourse.
What is known, is that sleeping is characterised by an elaborate change in the neural signature of the brain in comparison to wake (for review see: Carskadon & Dement, 2005). Throughout the night, this signature goes through cycles of alternating ‘sleep stages’,which are characterised by complex oscillatory activity and neurohormonal changes. These patterns of changes are hypothesised to play an important role in a myriad of processes that support our functioning selves, both on a biological and behavioural level (e.g. Stickgold, 2005). For example, parts of the homeostatic housekeeping of the brain takes place during sleep (Wang, Grone, Colas, Appelbaum & Mourrain, 2011).
On a behavioural level, researchers have argued that the state of the sleeping brain provides an excellent environment for the consolidation and processing of declarative and nondeclarative memory (for reviews see Klinzing, Niethard & Born, 2019; Stickgold, 2005, Walker & van den Helm, 2009). Neuroscientific and behavioural evidence in rats and humans have provided evidence for such memory related processing. Furthermore, sleeping seems to play a role in emotion regulation processes.
Due to sleep’s role in memory processing and emotion-related processes, some researchers have hypothesised that sleep might be involved in the consolidation and
processing of memories that are associated with an emotional tone (i.e. emotional memories) (e.g. Walker & van der Helm, 2009). Generally, emotional memories are often remembered better than ‘neutral’ memories (for review see: LaBar & Cabeza, 2006). Whether sleep plays an active role in such preferential processing of emotional memories has led to an ongoing debate in the research community. Some postulate that sleep is not only actively involved in the preferential consolidation of emotional memories, but at the same time helps to attenuate the emotional tone associated to such memories in an adaptive manner (‘Sleep to forget, sleep to remember model’ by Walker & Van der Helm, 2009).
Understanding the role of sleep in such processes could be very relevant for the treatment of several psychopathologies such as anxiety disorders (e.g. PTSD) and depression. Interestingly, these psychopathologies oftentimes include sleep related symptoms (e.g. Germain, 2013; Tsuno, Besset & Ritchie, 2005). While, the question of directionality of this symptom-psychopathology relationship remains, it opens the intriguing possibility that sleep is an integral part of processes involved in emotionality of memory and as such implicated in the development of these pathologies.
This review set out to shed light on the question whether and to what extent sleep is involved in emotional memory processing by providing an integrative overview and
discussion of the current literature. Specifically, the aim was to investigate whether sleep is involved in the processing of the emotional tone associated with a memory.
Main
1. Emotional Memory
Events in life that are emotional often signal important learning opportunities, turning points for our personal development and impact the way in which we experience our daily life (Dolan, 2002). For example, experiencing a burglary at home will trigger a significant emotional response. We might experience a lot of fear and react accordingly by screaming. However, the experience teaches us how to behave in the future and might impact how we feel in certain situations. We might remember to lock the door more thoroughly at night and are more anxious when we are alone and think about the event.
Memories of such emotional events are termed emotional memories and are the focus of this review.
1.1 The Memory Benefit for Emotional Events
Anecdotal evidence indicates that emotional events are remembered better and more vividly than neutral events. Research investigating this in the lab has found evidence to support this claim (e.g. Dolcos, LaBar & Cabeza, 2005; for reviews see: Dolan, 2002; LaBar & Cabeza, 2006; Yonelinas & Ritchey, 2015).
Especially in the long term, emotional stimuli seem to be remembered better than neutral ones (Kleinsmith & Kaplan, 1963; Dolcos et al., 2005; for review see: Yonelinas & Ritchey, 2015). While there is no memory enhancing effect of emotion at immediate recall, this effect emerges over the course of hours to weeks. It seems that emotionality has a specific benefit for the recollection of memories, while not benefitting the feeling of familiarity with stimuli (e.g. Ochsner, 2000; for review see Yonelinas & Ritchey, 2015).
The mechanisms underlying enhanced memory for emotional events are closely tied to the structure of emotionality. The emotionality of an event or stimulus can be evaluated on two dimensions: valence and arousal (LaBar & Cabeza, 2006). Valence is the tone of the experienced emotion and is usually a spectrum from positive (pleasant) to negative (unpleasant), with neutral being the middle judgement. Arousal is the judgement of how
exciting (or intense) a stimulus is. This dimension has calm and excitement as the two opposing extremes. Arousal can alternatively be measured through psychophysiological measures, such as heart rate deceleration and skin conductance. It is related to autonomic nervous system activity.
Especially the dimension of arousal has been investigated as a key player for the memory enhancement of emotional events. Arousal is supposed to play a role for the
encoding of new information and has been shown to additionally have subsequent effects on the consolidation and retrieval of memories (Dolcos et al. 2005; Dolan, Lane, Chua & Fletcher, 2000, for review see: LaBar & Cabeza, 2006). The amygdala, as modulated by arousal, is strongly implicated in this effect. According to the memory-modulation
hypothesis, arousal through amygdala activity has substantial influences on memory systems in the medial temporal lobe, that are responsible for the encoding, consolidation and retrieval of all kinds of declarative and non-declarative memories (Dolcos et al., 2005; McGaugh, 2004; for review see: LaBar & Cabeza, 2006). For example, long-term memory for emotional stimuli is predicted by amygdala activity at encoding (Dolcos, LaBar & Cabeza, 2004).
The amygdala is said to directly and indirectly project to many of those systems and triggers neurohormonal changes (e.g. the release of the stress hormone cortisol). In turn, these changes are implicated in the long-term systems consolidation of such memories (McGaugh, 2004; Ritchey, Dolcos & Cabeza, 2008, for review see: LaBar & Cabeza, 2006). Researchers argue that this amygdala effect is in line with the delayed memory benefit for emotional memories, as such physiological changes need time to unfold (for review see: LaBar & Cabeza, 2006). Amygdala activity at retrieval further predicts successful recollection of emotional stimuli, indicating that the amygdala is not only involved in the encoding and consolidation, but also remembering of emotionally arousing stimuli (Dolcos et al., 2005, Dolan et al., 2000). Specifically, the active recollection (in comparison to familiarity) of emotional stimuli is associated with activity in the amygdala (and hippocampus) (Dolcos et al., 2005).
But what role – if any - does the valence of an emotional experience play in the emotional memory effect and is there an interaction with arousal? Studies have shown that there is no difference between positively or negatively valenced stimuli for the memory benefit and that this benefit is present regardless of arousal (i.e. valence is enough to benefit memory in the absence of arousal; Adelman & Estes, 2013). However, negative memories and high arousing memories are often remembered more vividly, albeit not necessarily more
accurately, than positive or non-arousing emotional memories. The latter have been shown to be processed on more a schematic level (Ochsner, 2000; Bowen, Kark & Kensinger, 2017; Gomes, Bainerd & Stein, 2013).
When looking at the neural correlates, an interaction effect between arousal and valence has been found. Generally, differential brain areas are recruited at encoding depending on the level of arousal and the valence of a stimulus (Mickley Steinmetz & Kensinger, 2009). The amygdala is implicated in the processing of both, negative and positive stimuli (Bowen et al., 2017; Kensinger & Schacter, 2006). However, while memory encoding for negative, arousing stimuli is related to visual (occipital) and temporal areas, enhanced memory for positive and non-arousing stimuli is related to more frontal areas (Kensinger & Corkin, 2004; Mickley Steinmetz & Kensiger, 2009; Mickley & Kensinger, 2008). Furthermore, depending on the level of arousal and the valence of the stimulus, the amygdala shows differential connectivity patterns to other brain areas at encoding, that is predictive of later memory performance (Mickley Steinmetz, Addis & Kensinger, 2010). Mickley et al. (2010) showed that for negative items, arousal leads to enhanced amygdala connectivity to the inferior frontal gyrus and occipital areas, potentially in line with such memories being remembered in a detailed and vivid fashion. For positive items on the other hand, these connections are weakened for arousing stimuli and strengthened for non-arousing items. While the latter finding was somewhat surprising to the authors, they overall proposed that the found connectivity might be in line with the nature of the later remembered memories (i.e. while negative memories are remembered vividly, positive memories are more schematic and conceptual).
1.2 The development of emotionality of emotional memories
As discussed above the two dimensions of emotionality – arousal and valence – are at the core of the memory benefit for emotional memories.
When considering the above example of a burglary, it would be adaptive if
remembering the event led to an attenuated – but not eradicated – emotional response after some time has passed. The arousal experienced, while going through the stressful experience, and the negative valence associated with it should be down-tuned to ensure normal
functioning in life. It would be very straining to re-experience the event over and over again, through our emotional reaction when remembering the event – this would be considered a
pathological state and is a hallmark of for example PTSD and anxiety. However, the tone should not vanish all together, as it reminds us of dangerous situations from which we can learn. On the other hand, when considering positively valenced events, such memories might be dear to the heart and it might be adaptive to keep the emotionality at a similar level as at encoding. For example, this might help to encourage certain positive behaviors (Walker & Skowronski, 2009).
Research in the lab has indeed found that the emotional intensity of memories at retrieval changes over time in a similar manner as just described. The intensity of the emotional tone of autobiographical memories (i.e. memories of events and semantic knowledge regarding one’s own life) fades with time.
Interestingly, there is a difference in such fading for positively valenced and
negatively valenced memories. While there are still methodological difficulties that need to be addressed in this line of research, an abundance of results has shown that the emotional affect of positively valenced memories tends to fade more slowly, than that of negatively valenced memories (e.g. Ritchie, Skowronski, Harnett, Wells & Walker, 2009; Walker, Skowronski, Gibbons, Vogl & Thompson, 2003; for reviews see: Walker & Skowronski, 2009; Skowronski, Walker, Henderson & Bond, 2014). This effect is the ‘fading affect bias’ (FAB) (Walker & Skowronski, 2009).
Several characteristics of this effect have been identified (for review see: Walker & Skowronski, 2009). Firstly, the difference in positively valenced and negatively valenced emotional memories is independent of the initial arousal at encoding (Ritchie et al., 2009). However, the general rate of fading of emotional intensity depends on the arousal
experienced at encoding (e.g. Ritchie, Skorowski, Hartnett, Wells & Walker, 2009; Walker, Skowronski, Gibbons, Vogl & Thompson, 2003). Secondly, the magnitude of the FAB depends on inter-individual differences and emotion regulation capacities (Muir, Madill & Brown, 2017; Walker et al., 2003). For example, people who are mildly depressed, show a smaller FAB than healthy controls i.e. the negative emotionality of memories fades less strongly than the negative emotionality of healthy controls in comparison to positive emotionality (Walker et al., 2003). Thirdly, different types of social and private memory rehearsal have differential effects on the FAB (Ritchie et al., 2006). More social and private rehearsal to uncover the meaning of an event, is associated with a stronger fading of negative events, while the affect of positive memories is protected. Cued, involuntary private rehearsal
and rehearsal that is done to keep a memory alive or to influence one’s mood, leads to negative affect fading less, the more the subjects engages in such rehearsal.
2. Sleep and emotional memory
2.1 The Neurophysiology of Sleep
Generally speaking, sleep is a state of lowered consciousness in which our awareness for the surrounding environment is down-tuned, albeit not entirely dissolved (Dang-Vu et al., 2010). When we are asleep, our brain goes through cyclic, mostly global, neurobiological changes that have been generally described as sleep stages. Each stage is characterised by specific features such as muscle tone, eye-movement, heart rate, oscillatory behaviour and neurotransmitter (NT) composition (Carskadon & Dement, 2005; Dang-Vu et al., 2010). These stages go hand in hand with the depth of sleep that is experienced. Overall, the stages fall into two categories: rapid eye movement (REM) sleep and non-REM sleep, which alternate in cycles of approximately 90 minutes (Carskadon & Dement, 2005).
REM sleep has its name based on the pattern of eye movement during that stage. During REM-sleep, periods of rapid eye movements are alternated by periods of no
movements. It is one of the lighter sleep stages that is especially prevalent towards the end of the night (Carskadon & Dement, 2005). During REM sleep, the neurotransmitters (NTs) found in higher concentrations are acetylcholine and cortisol (McGinty & Szymusiak, 2010). Arousal related NTs such as dopamine, histamine, orexin, noradrenaline, and serotonin are at their lowest during REM sleep (in comparison to wake and non-REM sleep). When looking at recordings of the sleep-EEG, REM sleep is overall more similar to wake states than non-REM sleep. It is governed by frontal and hippocampal theta activity (4 -7Hz) and ponto-geniculate-occipital waves.
The global level of neural activity is more reminiscent of the wake state than other sleep stages in non-REM (for review see Dang-Vu et al., 2010). For some regions, brain activity is higher during REM sleep than during wake, while other regions show decreased activity in comparison to wake. Areas showing increased activity include, for example, the thalamus, hippocampus, amygdala, anterior cingulate cortex, dorsomedial prefrontal cortex and temporal-occipital areas (Corsi-Cabrera et al., 2016; Dang-Vu et al., 2010, Ioannides, Kostopoulos, Liu & Fenwick, 2009). Thus, regions that are part of the emotional processing circuit, show enhanced activation during REM sleep. In contrast, areas such as the
dorsolateral prefrontal cortex, precuneus, posterior cingulate cortex and inferior parietal cortex show decreased activity.
Non-REM sleep is divided into three stages. The first two stages describe light sleep and the last stage is deep sleep (Carskadon & Dement, 2011). The oscillatory activity seen during non-REM sleep is governed by interactions between the thalamus and the cortex. In contrast to REM-sleep, the oscillatory patterns are rhythmic and synchronized over the cortex. The most prevalent patterns seen characteristic of the different non-REM stages are sleep spindles (all non-REM stages), K-complexes (stage 1 and stage 2), sharp-wave ripples (stage 3 and stage 4), delta waves (stage 3 and stage 4) and slow waves (stage 2, stage 3 (20-50% of activity) and stage 4 (>(20-50% of EEG activity)) (Carskadon & Dement, 2005;
Diekelmann & Born, 2010; McGinty & Szymusiak, 2010). Slow wave sleep (stage 3) which is characterized by slow-wave and delta wave activity, is the deepest form of sleep and can predominantly be measured in the first third of the night (Carskadon & Dement, 2005).
Generally, non-REM sleep is characterized by a lower level of neural activity in different areas of the brain in comparison to activity observed during wake. For example, during SWS, brain activity is thought to decrease by 40% in comparison to wake (Dang-Vu et al., 2010). Such reduction of activity occurs in both cortical (e.g. dorsomedial prefrontal cortex, anterior cingulate cortex, precuneus) and subcortical (e.g. thalamus, basal ganglia, brainstem and basal forebrain) brain regions (for review see Dang-Vu et al., 2010; Ioannides et al., 2009). Furthermore, during non-REM sleep the NT concentration of acetylcholine, cortisol, serotonin, histamine, noradrenaline, orexin and dopamine are lower than during wake but somewhat higher than during REM (Diekelmann & Born, 2010; McGinty & Szymusiak, 2010). Contrary to REM sleep, the NTs acetylcholine and cortisol are lower during non-REM than during wake (McGinty & Szymusiak, 2010).
Interestingly, sleep and the circadian rhythm – which is the rhythm generated in the suprachiasmic nucleus and underlies wake-sleep cycles throughout the body - have been shown to introduce changes on the synapse level (for review see Wang et al., 2011). For example, gene expression in the cerebral cortex is altered during sleep – independent of when sleep occurs during the day (Cirelli, Guitierrez & Tononi, 2004; for review see Sun, Zhou, Cichon & Yang, 2020). This altered gene expression is said to particularly enhance neural plasticity in the brain. The circadian rhythm in general has been shown to have important influences on synaptic plasticity. For example, protein synthesis and breakdown of proteins important for such plastic changes have been found to be differentially affected according to the circadian rhythm in combination with sleep (for reviews on synaptic plasticity during sleep see Wang et al., 2011 and Stickgold, 2005)
2.2 Sleep and emotional memories: The ‘Sleep to forget, sleep to remember’ hypothesis
Generally, sleep is an important player in the consolidation of memories (for review see Diekelmann & Born, 2010; Stickgold, 2005). Studies in rodents and humans have shown that the sleeping brain exhibits a replay of activity that is reminiscent of activity seen during the encoding of experiences. According to the active system consolidation hypothesis, such processes are particularly dependent on SWS and promote the consolidation of such
experiences (for review see Sun et al., 2020; Diekelmann & Born, 2010). This replay activity, and sleep in general, induce plastic changes on the synapse level (Diekelmann and Born, 2010). Genes important for the consolidation of memories are upregulated – depending on experiences and behaviors during wakefulness (Stickgold, 2005). Additionally, the synaptic homeostasis hypothesis postulates that slow oscillations are involved in downscaling of the strength of synapses, that were strengthened through new experiences during wakefulness. Such downscaling eliminates very weak connections, while relatively strengthening more important connections, thereby highlighting important memories.
Based on this role of sleep in memory consolidation in general, it was proposed that sleep might play a mediating role for the benefit of emotionally tagged memories (see Box 2 for typical study set up). Indeed, several studies have found that sleep, and specifically REM-sleep, supports the consolidation of emotional memories and emotional features of stimuli (e.g. Hu et al., 2006; Nishida, Pearsall, Buckner & Walker, 2009; Payne, Stickgold, Swanberg & Kensinger, 2008, for review see: Cunningham & Payne, 2017; Goldstein & Walker, 2014; Stickgold & Walker, 2013). Another question emerged following these findings. Might sleep additionally be involved in the housekeeping of emotionality of such memories? Does sleep influence what we feel when remembering?
The ‘Sleep to remember, sleep to forget’ (SRSF) hypothesis by Walker and van der Helm (2009), postulates that sleep is not only involved in strengthening the memory for emotional events, but at the same time helps in attenuating the ‘emotional tone’ of such memories. ‘Emotional tone’ here can be understood as arousal, operationalized as amygdala reactivity and connected to autonomic nervous system activity. According to Walker and van der Helm (2009), the network dynamics between the amygdala, hippocampus and cortex at
encoding are replayed during REM sleep, allowing the reprocessing of such experiences during REM sleep. These reactivations are potentially coordinated by theta oscillations between the involved brain regions. Importantly, during the REM sleep stage, aminergic neurotransmitter (such as noradrenaline, which is involved in high stress and anxiety states) concentrations are low and cholinergic concentrations are high, which is the opposite of what the concentrations were during encoding. Cholinergic NTs are involved in the consolidation of amygdala-dependent emotional memories via their influence on the amygdala (for review see: Walker & van der Helm, 2009). The ‘new’ neurochemical milieu in which the encoded memories are reprocessed - according to Walker and van der Helm (2009) - allows the emotional tone of the memory to be attenuated, while still allowing the process of systems consolidation to take place. These processes might take several nights of sleep to unfold. According to Walker and van der Helm (2009) they might be mirrored in sleep mentation, and dreaming might in turn be directly connected to emotional processing (see Box 3).
Several postulates can be derived from the SFSR hypothesis, spanning behavioral, physiological and neural levels. Firstly (1), amygdala and more generally limbic reactivity at retrieval of emotional memories should be lower after a period of sleep than after a period of wakefulness. Secondly (2), physiological arousal as measured with for example, skin
conductance response (SCR), heart rate deceleration (HRD) and electromyography (EMG) measures should show less pronounced autonomic nervous system activity after a period of sleep, in comparison to wakefulness. Thirdly (3), this attenuated autonomic reactivity might be accompanied by an attenuation of the subjective evaluation of valence and arousal (i.e. more cognitively controlled processes). All of these effects should especially be pronounced when looking at REM sleep specifically.
The following sections discuss research investigating each of these postulates. The first section includes findings from neuroimaging research. Following, studies looking at physiological measures and studies focusing on subjective measures are discussed.
Box 1 – REM vs. SWS as optimal stages for emotional memory processing?
REM-sleep has emerged as the potential sleep stage for emotional processing. It was suggested that REM-sleep is involved in the ‘emotional housekeeping’ of the brain, including preparing individuals for future emotional behavior, being actively involved in the consolidation of emotional memories and potentially helping in the attenuation of the emotional tone of experiences (Goldstein & Walker, 2014, Walker & van der Helm, 2009). Walker and colleagues (2009a, 2014) pointed out several characteristics of REM sleep that have afforded such proposals: (1) Some limbic and related areas show increased activity during REM-sleep in comparison to wakefulness, (2) Theta activity that coordinates activity between subcortical and
cortical areas involved in emotional processes are prevailing during REM-sleep and (3) the neurochemical milieu shows an absence of arousal related, aminergic NTs, contrasting the milieu during which emotionality is usually experienced in wakefulness, while cholinergic NTs important for emotional learning are at their highest.
On the other hand, SWS has been proposed as an alternative sleep stage for emotional processing. SWS is involved in memory reactivation and consolidation during sleep and has been hypothesized to lead to a downscaling of synaptic connections during sleep, efficiently organizing the consolidation of only the most important memories, thereby regulating their memory strength (Sun et al., 2020). Potentially, such downscaling might apply to emotional processing as well.
Furthermore, SWS is generally seen as a state involved in sleep homeostasis, regulating the demand for sleep, following life’s challenges (Achermann & Borbély, 1999; Pace-Schott et al., 2011). Some researchers argued that a potential ‘emotional homeostat[tic]’ role of this sleep stage might be analogous to such other homeostatic roles (p. 33, Pace-Schott et al., 2011; Talamini et al., 2013).
Another characteristic of SWS might additionally point towards an emotional role of SWS. Noradrenaline, a NT involved in emotional memory consolidation, such as fear conditioning, is highest during SWS in comparison to other sleep stages. It is released in transient bursts from a region (locus coeruleus) highly connected to limbic structures (Groch et al., 2011).
Box 2 – Typical Study Design: Emotional memory consolidation during sleep
In a typical study investigating sleep’s role in emotional memory consolidation and affective tone processing, participants view a set of emotional photos or videos, which are high in arousal. In most studies, the stimuli are negatively valenced but some have included positive stimuli. Participants are later tested on their recognition or recall memory for these stimuli. When the affective tone processing is evaluated, participants are asked to rate the seen stimuli on measures such as valence or arousal, at encoding and retest or only at retest. An often-used scale is the Self-assessment Manikin (SAM) scale (Bradley & Lang, 1994, see Figure 1). The SAM is a pictorial scale consisting of three subscales, two of which – valence (pleasurable to unpleasable) and arousal (arousing to un-arousing) – are frequently used in studies evaluating emotionality of emotional memories. It should be noted that the subjective evaluation of stimuli is based both on indirect automatic processes but also on top-down control (for discussion see Bolinger et al., 2019).
Some studies use physiological and neuroimaging measures in addition to or instead of subjective measures. Skin conductance response (SCR), heart rate deceleration (HRD) and sometimes facial electromyography (EMG) are physiological measures to operationalize a more automatic emotional response (for discussion see Bolinger et al., 2019). To investigate the neural mechanisms, Electroencephalography (EEG) and particularly functional MRI (fMRI) are frequently used.
Generally, studies can be categorized into several types of designs. Here, these will be referred to as ‘classical’, ‘sleep-stage enhancement’ and ‘sleep deprivation’ designs. Some studies investigate the effect of sleep in general, while others focus specifically on REM sleep in comparison to other sleep stages. Furthermore, in some study designs, the immediate effects of
Figure 1 The valence (upper pane) and arousal (lower pane)
a nap or night of sleep after encoding are investigated, while in other designs the immediate and delayed (delay of one week for example) effects are investigated.
In most ‘classical’ sleep study designs (see Figure 2), it is investigated whether a period of sleep between encoding and recognition of emotional stimuli has any effect on recognition or recall accuracy and emotionality associated with the stimuli. Also, it is investigated whether there is a difference in these measures between emotional (arousing and valenced stimuli) and neutral stimuli. The sleep period can be short as in a nap and is compared to wakefulness.
In ‘sleep stage enhancement’ designs (see Figure 3), researchers try to manipulate the amount of REM and Non-REM sleep that participants experience. One method is to set the time at which the participant goes to sleep. Sleeping at a later time is associated with more REM sleep in comparison to earlier sleep in which SWS predominates (e.g. Sopp, Michael, Weeß & Mecklinger, 2017).
In ‘sleep deprivation’ studies (see Figure 4) on the other hand, the effect of sleep deprivation on emotional memory processing in comparison to normal sleep is investigated. Depending on the research question (e.g. concerning sleeps role in general or the role of a specific sleep stage), participants either get sleep deprived for a whole night or are not allowed to enter certain sleep stages, as tracked by sleep-polysomnography. Sleep deprivation studies do have their disadvantages. It has been argued that the deprivation introduces an additional stressor into the experiment, that may have an effect on other processes and therefore may confound the effects (e.g. Lipinska et al., 2019). Also, it is difficult to really achieve a period of sleep that entirely lacks the targeted sleep stage.
Figure 2 Schematic visualization of a ‘classical’ study design.
Figure 3 Schematic visualization of a ‘sleep stage enhancement’
Regardless of which paradigm is used, in a good study design, researchers control for possible confounding factors, that are due to the circadian rhythm. For example, physiological measures show variations throughout the day. Arousal systems show an increase in activity over the day that is associated with the experience of emotionality (Hot, Leconte & Sequeira; 2005). Such daily fluctuations can therefore majorly confound findings in emotion research. To control for such effects, the study should be conducted at different times.
2.2.1 Sleep and the processing of emotional memories: Neuroimaging studies
As mentioned above, the SFSR hypothesis postulates that the activity of areas
associated with emotional processing and connected to autonomic networks should be down-tuned after a night of sleep for emotional memories. Only a few studies have investigated these potential neural markers of attenuated emotionality (e.g. van der Helm et al., 2011; Sterpenich et al., 2007; Wassing et al., 2019; Bolinger et al., 2019).
Using a ‘classical sleep study design’ (see Box 2) van der Helm et al. (2011) found direct evidence for their SFSR model as a night of sleep indeed led to a decreased amygdala reactivity to emotional stimuli at retest. Also, the functional connectivity between the amygdala and the ventromedial prefrontal cortex – a region implied in emotional processing such as fear extinction - increased after sleep (Phelps, Delgado, Nearing & LeDoux, 2004).
Other studies indirectly testing the SFSR hypothesis, by investigating sleep deprivation and subjects with insomnia disorder, found somewhat corroborating results. Sleep deprivation – in comparison to sleep - has been associated with increased amygdala reactivity to negatively valenced (but not positively valenced) remembered stimuli
(Sterpenich et al., 2007). In a similar vein, insomnia disorder has been linked to activity in the ACC when remembering personally relevant memories from years ago, while such activity is not detectable in normal sleepers. The ACC is an area involved in both emotion regulation and expression and cognitive appraisal of such emotion. Thus, disturbed sleep as seen in insomnia disorder might be detrimental for the processing or attenuation of
emotionality associated with specific memories, while ‘normal’ sleep might be important for such processing (Wassing et al., 2019).
The SFSR model further hypothesizes that particularly REM-sleep may play a role in the emotional tone processing of memories. Some evidence for this comes from the finding that REM-sleep deprivation (REM-D) is associated to preserved activity in emotional processing related areas in comparison to undisrupted sleep. Areas such as the prefrontal cortex (important for e.g. emotion regulation), occipital and temporal areas show preserved or
increased activity after REM-D, while normal sleepers show a decrease in emotional reactivity (Rosales-Lagarde et al., 2012). These findings indicate that REM sleep may be involved in the attenuation of emotionality.
2.2.2 Sleep and the processing of emotional memories: Physiological measures
Following potential changes in neural activity in areas related to emotional processing and physiological responses, physiological measures, reflecting autonomic reactivity, should show an attenuation to previously encoded emotional memories after sleep, especially after a period rich in REM-sleep - if the SFSR hypothesis holds true.
Some studies have indeed found evidence for such an attenuation after sleep in comparison to wakefulness for a variety of physiological measures (e.g. Cunningham et al., 2014; Lipinska & Thomas, 2019, Pace-Schott et al., 2011). For example, Cunningham et al. (2014) found that HRD and SCR responses are attenuated when encoding is followed by a night of sleep but the measures are preserved by a day of wakefulness. Similar results were obtained by Lipinska and Thomas (2019), albeit for slightly different measures, and SCR was not affected. Interestingly, already a short nap can be sufficient to achieve such an attenuation (Pace-Schott et al., 2011).
While the above studies found evidence for an attenuation of emotionality of emotional memories as operationalized by physiological measures, others have found the opposite effect. Some studies show that sleep actually preserves the reactivity of HRD to emotional memories in comparison to wakefulness (Ashton, Harrington, Guttesen, Smith and Cairney, 2019; Bolinger et al., 2019). Interestingly, potential changes seen in physiological measures might take time to emerge, potentially underlying the negative findings. Bolinger et al. (2019) found no evidence for an attenuation of physiological arousal in the short term, however, having slept immediately after encoding was associated with attenuated emotional responses in the longer term. This finding is reminiscent of the assumption of Walker and van der Helm (2009), that the effects might unfold over time.
When looking at specific sleep stages and their influence on the emotionality as operationalized by physiological measures, the prediction of the SFSR model – that REM sleep is the important sleep stage - does not seem to be supported. Generally, several studies have found that more time spent in REM-sleep is associated with less habituation of
physiological measures i.e. REM-sleep might actually preserve the emotionality attached to a memory, when looking at physiological measures (Gilson et al., 2016; Werner et al., 205;
Pace-Schott et al., 2011). Furthermore, experiencing SWS might be associated with more habituation than experiencing REM sleep during a period of sleep. In a study by Pace-Schott et al. (2011) SWS was correlated with more habituation of the EMG, while REM sleep was associated with less habituation of the SCR between sessions. Contrary to the hypothesis of Walker and van der Helm (2009), these physiological findings might indicate that REM sleep is either not the most important player in emotional memory processing (but SWS might be), is actually counterproductive for such processing and adds to a preservation of emotionality, or that an attenuation takes more time to be detected on the physiological level.
2.2.3 Sleep and the processing of emotional memories: Subjective Measures
A third postulate that could be derived from the SFSR hypothesis is that cognitively controlled subjective perceptions of emotionality (e.g. valance and arousal) might also reflect an attenuation, if there are changes on the autonomic level (i.e. neural and physiological level). Both arousal and valance might show such an attenuation, if the SFSR hypothesis holds true.
Very few studies have indeed found that subjective ratings are attenuated after sleep (e.g. van der Helm et al., 2011; Gujar et al., 2011). For example, concomitantly to finding reduced amygdala reactivity after sleep (see section 2.3.1), van der Helm et al. (2011) found that subjective emotionality ratings went down after sleep. Others, have found a similar effect for valence ratings only – albeit in children (for who emotional processing might be different than for adults; Bolinger et al., 2018).
However, a wealth of research has not found such attenuation of subjective ratings. There is evidence that rather than showing an attenuation, subjective ratings display an enhancement or preservation of emotionality of memories after sleep or are not influenced differently by sleep than by wake (Ashton et al., 2019; Baran, Pace-Schott, Ericson and Spencer, 2012; Bolinger et al., 2019; Pace-Schott et al., 2011; Tempesta, De Gennaro, Natale and Ferrera, 2015). Baran et al. (2012) even proposed a new hypothesis that is in stark
contrast to the SFSR model – the emotional salience account. In this model sleep is proposed to protect the salience of emotional memories, similar to how it protects the memory itself.
Interestingly, the influence sleep may have on the evaluation of emotionality, might be associated with age (Jones, Schultz, Adams, Baran & Spencer, 2016). In older people the positive (but not negative) valence of memories is preserved over sleep, while in younger
people there is a bias towards a preservation of negative affect. This suggests that the
mechanisms underlying sleep’s role in emotional memory processing might be dynamic and might additionally involve different mechanisms for differently valenced memories
The lack of findings in support of the SFSR, when looking at subjective measures, has alternatively been hypothesized to be due to sleep primarily affecting NMDA-dependent mechanisms that are involved in autonomic learning instead of more controlled processes dependent on cortical processes (i.e. subjective evaluations). Sleep may primarily influence autonomic arousal and its underlying mechanisms, but this type of habituation might not be detectable in subjective measures.
Others have suggested that subjective ratings are first not influenced by a period of sleep, but that the evaluation of emotionality attenuates over time and needs several nights of sleep to evolve (Bolinger et al., 2019). This is in line with findings on the physiological level discussed above (Bolinger et al., 2019).
Mixed results similarly emerge for the specific role of REM sleep, for studies
operationalizing emotionality through subjective measures. Few studies find that REM sleep is associated with attenuated evaluations of emotionality (e.g. Gujar et al., 2011; Rosales-Lagarde et al., 2012), while others find indications that more REM-sleep preserves, if not enhances the subjective evaluation of emotionality of memories (Baran et al., 2012; Gilson et al., 2016; Jones et al., 2016; Lara-Carassco et al., 2011; Werner, Schabus, Blechert &
Wilhelm, 2020).
Werner et al. (2020) argued that the effects of REM sleep might take more time to unfold, explaining the mixed findings. In their study, longer time spent in REM sleep was associated with higher aversiveness ratings of negative pictures at retrieval in the short-term. However, when looking at aversiveness ratings, duration and number for intrusive memories of these pictures after two days, participants that got REM sleep after encoding showed a better adaptation towards such memories than participants that did not experience REM sleep directly after encoding. This finding is reminiscent of the findings by Bolinger et al. (2019) for physiological measures and subjective measures, when looking at sleep in general.
Alternatively, as discussed for physiological measures, some studies using subjective measures to operationalize emotional processing, have found an association of SWS instead of REM sleep in the potential processing of emotionality during sleep (Groch et al., 2011; Jones et al., 2016; Talamini, Bringmann, de Boer and Hofman, 2013). For example, in a study by Talamini et al. (2013), the percentage of SWS was heightened following an
emotionally negative movie and the percentage of REM sleep decreased towards the end of the night. Importantly, a higher percentage of SWS was associated with a heightened attenuation of emotionality ratings, suggesting that increased SWS after a negative experience might act as an adaptive mechanism.
More indirect evidence for a role of SWS comes from a study by Groch et al. (2011), who injected clonidine during SWS – an agonist that blocks the release of Norepinephrine (NE) from a region strongly connected to the amygdala and hippocampus. NE is implicated in emotional memory consolidation and its levels are decreased during SWS but not as low as during REM. During SWS, NE is released from the locus coeruleus in transient bursts. While the placebo group showed a reduction of arousal and negative valence ratings at recall of emotional stimuli, in comparison to new pictures, this effect was not present in the group that was injected with the agonist. The authors interpreted the finding as an indication for a role of SWS in emotional tone processing.
2.2.4 Sleep and the processing of emotional memories: Other study designs
The above discussed studies used (1) neuroimaging measures, (2) physiological measures and (3) subjective measures to operationalize emotional processing during sleep. However, another approach by Deliens and collegues to studying this processing yielded interesting results to evaluate the SFSR hypothesis as well.
Deliens et al. (2013, 2014) used a mood-induction procedure to influence the context in which subjects learned word pairs. If sleep is successful at attenuating the emotional tone associated with a memory, the mood context at recall should not interfere with such memory if it is incongruent with the mood context at encoding. In a first study, Deliens, Gilson, Schmitz and Peigneux (2013) investigated this using a sleep deprivation design. Three full nights of sleep led to lower interference of emotional (positive and sad) context at recall, than a night of sleep deprivation, followed by two full nights of sleep. This indicated that sleep deprivation had a negative effect on the inhibition of encountered emotional context at encoding. In a follow up study, Deliens, Gilson and Peigneux (2014) then investigated whether one night of sleep is already sufficient to achieve such emotional unbinding. Instead of using a sleep deprivation design, sleep was compared to a day of wakefulness. Results showed that in both conditions, there was an interference effect at recall, indicating that one night of sleep preserved the emotional tone in which the stimuli were initially encoded.
Neu and Peigneux (2013) showed that emotional context still interferes at recollection after both a period of sleep rich in REM or non-REM sleep, indicating that REM-sleep specifically protects the emotional tone associated with a memory.
Therefore, these three studies provide evidence that sleep may in fact be involved in the emotional tone attenuation of emotional memories as hypothesized by the SFSR model. Again, these studies show that such attenuation might need more than one night of sleep to unfold – as has been suggested by the SFSR model.
Box 3 – Dreaming
A still very mysterious sleep experience, that has drawn in many generations of researchers and philosophers is dreaming. Dreaming – often interchangeably referred to as sleep mentation - reintroduces a form of consciousness which allows the dreamer to immerse in experiences and thoughts while sleeping. There is no definite definition of dreaming, but sleep mentation in general has been described as “the occurrence of any subjectively experienced cognitive event during sleep” (Nielsen, 2010; p. 576). Most likely, every human (and non-human animal) dreams.
Sleep-awakening studies have shown that dreaming can occur both in Non-REM and REM sleep, with dream recall being higher for REM sleep than non-REM sleep (Nielsen, 2010). Furthermore, the form of dreams experienced seem to be different for the different sleep stages, with REM dreams being reported to be more vivid, emotional and bizarre, which to some extent is in line with brain areas being active during this stage of sleeping (for reviews see, Nielsen, 2010 and Nir & Tononi, 2010). Furthermore, non-REM dreams have been reported to incorporate more episodic memories than REM dreams do, in which memory sources are more related to semantic memory for example (for reviews see Wamsley & Stickgold, 2011). However, non-REM dreams can become more similar to REM dreams throughout the night, calling the dichotomy between REM and non-REM dream features somewhat into question (for review on dream content see Zadra & Domhoff, 2010).
Years of dream content research have established that sleep mentation can take on many forms of complexity, with experiences spanning all senses and levels of emotionality (Smith et al., 2004). Also, episodic memory aspects from wake are frequently incorporated into dreams (Nielsen & Stenstrom, 2005; for review see Wamsley & Stickgold, 2011). The events that are incorporated seem to be personally important to the subject, while everyday activities are rarely incorporated (Eichenlaub et al., 2019; van Rijn et al., 2015).
Dreaming is often overlooked when it comes to the functions of sleep in general. The effects of sleep might not simply be detachable from the effects that dreaming - as a part of sleep - has during this period of lowered consciousness. As dreams often incorporate episodic and personally relevant information (Eichenlaub et al., 2019; van Rijn et al., 2015; Nielsen & Stenstrom, 2005; for review see Wamsley & Stickgold, 2011), some have argued that dreaming might reflect memory and emotional processing taking place, while others postulate that it might play an active role in the emotional processing of personally relevant information and emotional coping (Cartwright et al., 1991; Cartwright et al., 2005; for review see Cartwright, 2005; Wamsley & Stickgold, 2011). Cartwright et al. (2006) for example found that dreaming about a negative event, helps to cope with such an event in the future. Similarly, experiencing fear during dreams seems to be associated with attenuated subjective and neural reactivity (amygdala, insula, midcingulate cortex) to negative emotional stimuli (Sterpenich, Perogamvros, Tononi & Schwartz, 2020).
A lot of different models on the underlying mechanisms for the emotional processing during dreaming have been proposed (for integrative overview see Malinowski & Horton, 2015). These include dreams serving as a simulation for possible future
scenarios and thereby helping to prepare an appropriate emotional reaction; or dreaming might help to compare new emotional experiences with old memories to understand ones’ own emotions better (see overview Malinowski & Horton, 2015). Walker and van der Helm (2009) mention that these potential emotional processing roles of dreaming fit very well with their SFSR model, which provides potential mechanisms for such attenuation to take place. However, as mentioned above, they restrict the potential role of dreaming to REM-sleep, while dreaming occurs during other sleep stages as well.
Additionally, dreaming might act as a cue to rehearse emotional memories. Waking from an emotional dream that reiterates relevant personal information might serve as a starting point to reflect more on such experiences during wakefulness. Such rehearsing of emotional memories has been shown to influence the FAB and thus emotional tone attenuation (see section 1.2). Whether dreaming really does influence emotional memory processing and what the exact underlying neural correlates are, remains to be elucidated.
Discussion
The encoding of emotional memories is dependent on the arousal and valance experienced during an event (e.g. Adelman & Estes, 2013; Dolcos et al., 2005). The subsequent benefit for consolidation of such memories is aided by sleep, as the current literature suggests (e.g. Hu et al., 2006). But is sleep involved in how we feel when remembering certain life events from our past? Does it influence the development of
emotions attached to such memories? Generally, the emotionality of memories does seem to change over time, potentially differently for positive in comparison to negative memories (Walker & Skowronski, 2009). To understand the role of sleep in such processes, in this review, studies investigating the SFSR model by Walker and van der Helm (2009) were discussed in light of three postulates concerning neural (1), physiological (2) and subjective levels (3). The conclusions are summarized below:
Postulate (1): Neural markers of emotionality show attenuated reactivity to emotional memories after a period of sleep
The few studies that have investigated the underlying neural markers of possibly attenuated emotional reactivity have found somewhat converging evidence for the SFSR hypothesis. The here discussed studies have found indications that sleep benefits the
adaptation of reactivity to emotional memories. Direct and indirect evidence from disrupted sleep somewhat corroborate that sleep is beneficial for the processing of emotional memories and that REM-sleep specifically might be the main player in such processes.
Postulate (2): Physiological measures of emotionality do not show clear attenuated reactivity to emotional memories after a period of sleep
Studies investigating physiological measures as autonomic markers of emotionality of emotional memories have yielded mixed results. Some show an attenuation of emotionality over a period of sleep as short as a nap, while others find that sleep preserves the
emotionality – at least in the short term. Potentially the dynamics take time to unfold to attenuate the emotionality (Bolinger et al., 2019).
Making matters more difficult, some of the studies did not control for circadian influences, which might have confounded the physiological measures (Cunningham et al., 2014; Lipinska & Thomas, 2019; Hot et al., 2005). Also, the physiological measures that were used and for which an effect was found wildly differed between studies, making it harder to really compare findings. Therefore, the postulate derived from the SFSR
hypothesis, that sleep attenuates emotionality as captured in physiological reactivity can thus not be supported or rejected based on the current research.
However, the picture seems clearer when looking at specific sleep stages. Studies have not found evidence for a crucial role of REM-sleep. Instead, studies seem to converge on the finding that REM-sleep is associated with more preservation of emotionality, rather than attenuation. Instead, SWS might be a more important player for the attenuation of emotionality as measured through physiological measures.
Postulate (3): Subjective measures of emotionality show do not show clear attenuated reactivity to emotional memories after a period of sleep
Based on subjective measures of emotionality, the SFSR hypothesis can not confidently be supported. As with physiological measures, the results are mixed, and lean towards no involvement or a protective function of sleep in the subjective evaluation of emotionality. It remains unclear whether changes can just not be detected by subjective measures as they concern mostly autonomic processes, or whether more sleep or time is needed to detect such attenuation on a subjective level – as has similarly been suggested by studies looking at the physiological level (e.g. Bolinger et al., 2019). As with studies using physiological measures (see section 2.3.2), matters are made more complicated through problems in study designs and the use of and findings for different subjective measures.
Looking at the potential role of specific sleep stages on subjective measures, results are mixed as well. However, SWS instead of REM sleep as an important player in emotional processing is suggested – similar to what studies using physiological measures to
Limitations
Generally, the very mixed results might be due to limitations of the discussed studies such as the widely differing study designs used. While some studies seem to follow a
classical design construct as described in Box 2, others employ very unusual and new designs to investigate these questions. Even though the latter is important to potentially triangulate findings from other studies, it makes comparisons between findings very difficult. Especially, as the studies using a similar paradigm often differ in smaller things such as the measures they take, which stimuli they pick and what control conditions they include. The fact that direct replications of studies is rare or even nonexistent, makes it very difficult to reach a somewhat definite conclusion of whether and how sleep is involved in the emotional processing of memories. It seems that researchers aimed for conceptual replication1 rather than experimental replication. Some researchers such as Chambers et al. (2019) however argue that the latter from of replication is equally important and necessary to understand whether an effect is indeed present.
Apart from these differences in study designs, there are a few major limitations that seemed to plague several of the discussed studies.
First of all, some studies did not account for circadian effects (e.g. Cunningham et al., 2019; Jones et al., 2016; Lipinska & Thomas, 2019). Controlling for the effects of time of day is however crucial when investigating sleep related cognitive processes and when measuring physiological reactivity. These measures can be sensitive to the time of day and can thus, majorly influence the observed effects (Hot et al., 2005; see Box 2).
Secondly, some studies - specifically studies investigating the effect of specific sleep-stages (e.g. Werner et al., 2015) – did not include a wake control group. This might make it difficult to really draw conclusions on whether sleep is a substantial player in the processing of emotionality in comparison to wakefulness (Walker & van der Helm, 2009), or whether both wakefulness and sleep play a complementary or overlapping role in such processing.
Thirdly, it is questionable whether the common use of pictures or videos as stimuli is really ecologically valid. Can the processes regarding personally meaningful and emotional memories really be compare to the exposure and re-exposure to (standardized) stimuli? It should further be considered that the direct re-exposure to such stimuli might differ from the
1 Conceptual replication is the replication of a finding or principle in many different contexts using different techniques.
way in which we are cued and remember emotional memories. This discussion might be rather philosophical but should be considered nevertheless to conduct meaningful research.
Finally, most of the studies use aversive (i.e. negatively valenced) stimuli, while positively valenced stimuli are often ignored. It is unclear whether the findings for negatively valenced stimuli apply to positively valenced stimuli as well, or whether the effects might be very dissimilar. As discussed in the sections above, the arousal at encoding, but also the valence of the encountered event trigger specific neural interactions that allow for the memory benefit and potentially the encoding of the affective tone of such memories. Importantly, the neural signature at encoding differs between positively valenced and negatively valenced stimuli (e.g. Mickley Steinmetz et al., 2010). Also, as suggested by the FAB effect – the emotional tone of positive memories seems to fade more slowly than that of negative events (Walker & Skowronski, 2009). These two findings imply that the processing of the emotional tone is dependent on the valence of a memory, which might trigger
differing, complex mechanisms. This difference between positive and more negative memories is independent on the initial arousal experienced at encoding. Some differential effects for positively versus negatively valenced memories were found in studies including both types of stimuli indicating that it might indeed be relevant in sleep-related processes (e.g. Gujar et al., 2011; Jones et al., 2016). Others did not treat the valence as different factors, and as such a potential differing effect was hidden (e.g. Deliens et al. , 2013; Deliens et al., 2014). Arguably, such findings regarding the differential processing of positively versus negatively valenced memories should be taken into consideration when trying to understand how sleep is involved in the processing of emotionality.
Directions for future research
Future research should focus on addressing the above shortcomings of the current research. It is important that researchers focus on directly replicating specific studies instead of relying on conceptual replication. Without robust, repeated observable effects, it will remain difficult in the future to draw clear conclusions on whether sleep is substantially involved in the attenuation of the emotional tone of emotional memories. Connected to that, more attention should be payed to the details of the study designs, such that the mixed effects that are simply due to differences in study design can be resolved. This does not mean that
improved or novel study designs should not be considered in the future, but more replication of specific designs might help to observe more robust effects.
Apart from these study-related future directions, there are several topics that will be interesting to address in future research. Firstly, investigating the temporal dynamics of sleep’s involvement in emotional tone attenuation further, might be interesting to understand whether such processing indeed takes several nights of sleep (as suggested by Bolinger et al., 2019; Deliens et al., 2013, 2014; Werner et al., 2020). Furthermore, investigating the neural correlates of emotional tone processing in general and with the involvement of sleep will be important to understand whether and how the underlying neural signature changes. Few studies have investigated this for now (e.g. van der Helm et al., 2011).
With regards to the sleep stages involved in the processes, maybe it would be
interesting to investigate whether and how the different stages cooperate to process emotional memories instead of perceiving each stage as a separate entity. For example, maybe one could investigate whether a specific balance of REM and SWS is optimal to attenuate the emotionality of memories.
All in all, there is an abundance of possible future directions for the study of sleep and emotional memory processing. These will require an interdisciplinary approach, combining research from purely psychological, cognitive neuroscientific and potentially clinical
psychology/neuroscience. Findings from these different disciplines will allow a triangulation of results and will ultimately help to get a well-rounded understanding of the processes involved in the development of emotional memories.
Conclusion
This review set out to investigate whether sleep is involved in the processing of emotional memories – specifically in the attenuation of the emotional tone of such memories.
Overall, it seems that sleep is indeed involved – however the nature of such involvement is unclear. Drawing definite conclusions is hindered by problems in study
designs and a reliance on conceptual replication. The few neuroimaging findings and findings from studies using context interference designs support the SFSR hypothesis directly.
However, physiological and subjective findings do not clearly show whether such involvement of sleep takes the form of attenuation, protection or enhancement of emotionality. The time dynamics of such mechanisms seem to be unclear and potential
attenuation might take more time to unfold – as suggested by Walker and van der Helm (2009).
The mixed results of sleep in general might be due to the possibility that the emotional processing present during sleep is less strong than processes occurring during wakefulness. This might have hidden subtle processing occurring during specific sleep stages which might nevertheless be necessary for the adaptive processing of emotionality. Contrary to the hypothesis of Walker and Van der Helm (2009), most of the studies investigating specific sleep stages, on a neurophysiological and subjective level, suggest that REM-sleep does not seem to be an important player to attenuate the emotionality of memories. Rather, SWS has been suggested as a more relevant sleep stage to consider for potential emotional attenuation to take place during sleep (see Box 1 for consideration REM vs. SWS).
References
Achermann, P. & Borbérly (1999). Sleep homeostasis and models of sleep regulation. Journal of biological rhythms, 14(6), 559-570. doi: 10.1177/074873099129000894.
Adelman, J. S., & Estes, Z. (2013). Emotion and memory: A recognition advantage for positive and negative words independent of arousal. Cognition, 129(3), 530-535. doi: 10.1016/j.cognition.2013.08.014
Ashton, J. E., Harrington, M. O., Smith, A. K., & Cairney, S. A. (2019). Sleep preserves physiological arousal in emotional memory. Scientific reports, 9(1), 1-10. doi: 10.1038/s41598-019-42478-2.
Baran, B., Pace-Schott, E. F., Ericson, C., & Spencer, R. M. (2012). Processing of emotional reactivity and emotional memory over sleep. Journal of Neuroscience, 32(3), 1035-1042. doi: 10.1523/JNEUROSCI.2532-11.2012
Bolinger, E., Born, J., & Zinke, K. (2018). Sleep divergently affects cognitive and automatic emotional response in children. Neuropsychologia, 117, 84-91. doi:
10.1016/j.neuropsychologia.2018.05.015.
Bolinger, E., Cunningham, T. J., Payne, J. D., Bowman, M. A., Bulca, E., Born, J., & Zinke, K. (2019). Sleep's benefits to emotional processing emerge in the long
term. Cortex, 120, 457-470. doi: 10.1016/j.cortex.2019.07.008.
Bowen, H. J., Kark, S. M., & Kensinger, E. A. (2018). NEVER forget: negative emotional valence enhances recapitulation. Psychonomic Bulletin & Review, 25(3), 870-891. doi: 10.3758/s13423-017-1313-9.
Bradley, M. M., & Lang, P. J. (1994). Measuring emotion: the self-assessment manikin and the semantic differential. Journal of behavior therapy and experimental
psychiatry, 25(1), 49-59.doi: 10.1016/0005-7916(94)90063-9.
Carskadon, M. A., & Dement, W. C. (2005). Normal human sleep: an overview. Principles
and practice of sleep medicine, 4, 13-23.
Cartwright, R. D. (1991). Dreams that work: The relation of dream incorporation to adaptation to stressful events. Dreaming, 1(1), 3.
Cartwright, R. (2005). Dreaming as a mood regulation system. In Principles and practice of
sleep medicine (pp. 565-572). WB Saunders.
Cartwright, R., Agargun, M. Y., Kirkby, J., & Friedman, J. K. (2006). Relation of dreams to waking concerns. Psychiatry research, 141(3), 261-270. doi:
10.1016/j.psychres.2005.05.013.
Chambers, C. (2019). The seven deadly sins of psychology: A manifesto for reforming the
culture of scientific practice. Princeton University Press.
Cirelli, C., Gutierrez, C. M., & Tononi, G. (2004). Extensive and divergent effects of sleep and wakefulness on brain gene expression. Neuron, 41(1), 35-43. doi: 10.1016/s0896-6273(03)00814-6.
Corsi‐Cabrera, M., Velasco, F., del Río‐Portilla, Y., Armony, J. L., Trejo‐Martínez, D., Guevara, M. A., & Velasco, A. L. (2016). Human amygdala activation during rapid eye movements of rapid eye movement sleep: an intracranial study. Journal of sleep
research, 25(5), 576-582. doi: 10.1111/jsr.12415
Cunningham, T. J., Crowell, C. R., Alger, S. E., Kensinger, E. A., Villano, M. A., Mattingly, S. M., & Payne, J. D. (2014). Psychophysiological arousal at encoding leads to reduced reactivity but enhanced emotional memory following sleep. Neurobiology of
Cunningham, T. J., & Payne, J. D. (2017). Emotional memory consolidation during sleep. In Cognitive Neuroscience of Memory Consolidation (pp. 133-159). Springer, Cham.
Dang-Vu, T. T., Schabus, M., Desseilles, M., Sterpenich, V., Bonjean, M., & Maquet, P. (2010). Functional neuroimaging insights into the physiology of human
sleep. Sleep, 33(12), 1589-1603. doi: 10.1093/sleep/33.12.1589.
Diekelmann, S., & Born, J. (2010). The memory function of sleep. Nature Reviews
Neuroscience, 11(2), 114-126.
Dolan, R. J., Lane, R., Chua, P., & Fletcher, P. (2000). Dissociable temporal lobe activations during emotional episodic memory retrieval. Neuroimage, 11(3), 203-209. doi: 10.1006/nimg.2000.0538.
Dolan, R. J. (2002). Emotion, cognition, and behavior. science, 298(5596), 1191-1194. doi: 10.1126/science.1076358
Dolcos, F., LaBar, K. S., & Cabeza, R. (2004). Interaction between the amygdala and the medial temporal lobe memory system predicts better memory for emotional events. Neuron, 42(5), 855-863. doi: 10.1016/s0896-6273(04)00289-2.
Dolcos, F., LaBar, K. S., & Cabeza, R. (2005). Remembering one year later: role of the amygdala and the medial temporal lobe memory system in retrieving emotional memories. Proceedings of the National Academy of Sciences, 102(7), 2626-2631. doi: 10.1073/pnas.0409848102
Deliens, G., Gilson, M., Schmitz, R., & Peigneux, P. (2013). Sleep unbinds memories from their emotional context. cortex, 49(8), 2221-2228. doi: 10.1016/j.cortex.2012.11.014
Deliens, G., Gilson, M., & Peigneux, P. (2014). Sleep and the processing of
emotions. Experimental brain research, 232(5), 1403-1414. doi: 10.1007/s00221-014-3832-1.
Deliens, G., Neu, D., & Peigneux, P. (2013). Rapid eye movement sleep does not seem to unbind memories from their emotional context. Journal of sleep research, 22(6), 656-662. doi: 10.1111/jsr.12065
Eichenlaub, J. B., van Rijn, E., Phelan, M., Ryder, L., Gaskell, M. G., Lewis, P. A., ... & Blagrove, M. (2019). The nature of delayed dream incorporation (‘dream‐lag effect’): Personally significant events persist, but not major daily activities or
concerns. Journal of sleep research, 28(1), e12697. doi: 10.1111/jsr.12697.
Germain, A. (2013). Sleep disturbances as the hallmark of PTSD: where are we now?. American Journal of Psychiatry, 170(4), 372-382. doi:
10.1176/appi.ajp.2012.12040432
Goldstein, A. N., & Walker, M. P. (2014). The role of sleep in emotional brain
function. Annual review of clinical psychology, 10, 679-708. doi: 10.1146/annurev-clinpsy-032813-153716. doi: 10.1146/annurev-clinpsy-032813-153716
Gomes, C. F., Brainerd, C. J., & Stein, L. M. (2013). Effects of emotional valence and arousal on recollective and nonrecollective recall. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 39(3), 663. doi: 10.1037/a0028578.
Gilson, M., Deliens, G., Leproult, R., Bodart, A., Nonclercq, A., Ercek, R., & Peigneux, P. (2016). REM-enriched naps are associated with memory consolidation for sad stories and enhance mood-related reactivity. Brain sciences, 6(1), 1. doi:
10.3390/brainsci6010001.
Groch, S., Wilhelm, I., Diekelmann, S., Sayk, F., Gais, S., & Born, J. (2011). Contribution of norepinephrine to emotional memory consolidation during
sleep. Psychoneuroendocrinology, 36(9), 1342-1350. doi: 10.1016/j.psyneuen.2011.03.006
