Steriade et al. (1993b) have reported that SOs persisted in the cortex even after the connected thalamus was destroyed, indicating that they originated from recurrent cortical connectivity (Timofeev and Steriade, 1996). SOs represent alternations in the neuronal network between synchronized membrane depolarization, known as “up-states,” and prolonged hyperpolarization, known as “down-states.” They exhibit non-stationary bistability even during the physical disconnection of the cerebral cortex.

Where in the cortex and how are slow waves generated? Although the cellular mechanisms underlying the generation of slow waves are still not fully understood, many important findings have been reported to date (Sanchez-Vives, 2020). In vivo and in vitro studies have consistently demonstrated that deep or infragranular layers, particularly the internal pyramidal cell layer of the cortex (layer V), play a crucial role in initiating up-states (Chauvette et al., 2010; Fiáth et al., 2016). During up-states, layer V neurons have more intense and longer discharge than layer II/III neurons (Capone et al., 2017). In addition, several studies have indicated that activation of layer V neurons by optogenetic stimulation can initiate up-states and induce SOs (Beltramo et al., 2013; Stroh et al., 2013).

Up-states may be triggered by the summation of miniature excitatory postsynaptic potentials (EPSPs) formed by the action-potential-independent release of transmitter vesicles (Timofeev et al., 2000). These EPSPs may originate from a residual synaptic activity resulting from stimulus processing during prior wakefulness and become sufficiently large to generate repetitive up-states, especially after intense learning. These intrinsic excitability in layer V neurons leads them to begin firing during down-states (Sanchez-Vives and McCormick, 2000; Compte et al., 2003; Sakata and Harris, 2009) and may enable them to function as a specific “pacemaker” (Le Bon-Jego and Yuste, 2007). After initiation, the up-state may be amplified by intrinsic currents such as persistent Na⁺ and high-threshold Ca²⁺ currents. Synchronization of activity during depolarizing states may be partially maintained via corticocortical glutamatergic synaptic connections. Since the early characterizations of SOs both in vivo and in vitro, a contribution of N-methyl-d-aspartate (NMDA) and α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid (AMPA) receptors to establishing long-range synchrony has been reported (Steriade et al., 1993b; Sanchez-Vives and McCormick, 2000). The transition from the up- to down-state ends the synaptic reverberation that generates the up-state. The induction of this process has been mainly linked to a combination of outward K⁺ currents and disfacilitation resulting from the depression of excitatory synapses (Galarreta and Hestrin, 1998; Bazhenov et al., 2002; Benita et al., 2012). Different mechanisms involving K⁺ currents have been proposed, including adenosine triphosphate-dependent K⁺ currents (Cunningham et al., 2006), γ-aminobutyric acid (GABA)_B receptor-mediated responses (Perez-Zabalza et al., 2020), and Ca²⁺ and Na⁺-dependent K⁺ currents (Sanchez-Vives and McCormick, 2000). The up-state-to-down-state transition across distinct cortical regions suggests that subcortical input may synchronously facilitate the onset of the down-state (Volgushev et al., 2011; Sheroziya and Timofeev, 2014), and the thalamus has been considered to play a particularly important role by producing SOs in coordination with the cortex (Steriade et al., 1993b; Crunelli and Hughes, 2010). A recent study indicates the important role of the claustrum in providing global control of slow waves via cortical inhibitory cells (Narikiyo et al., 2020; Timofeev and Chauvette, 2020).

While SO refers to the temporal organization of the cortical activity, the term “slow wave” includes a spatial component because propagation is a fundamental property of waves. A study using source modeling of high-density EEG (hdEEG) recordings has reported that slow waves mostly originate from the center of the lateral sulci in the left hemisphere. Slow waves then mainly migrate from the anterior to the posterior brain regions, and the axis often involves cortical structures, such as the inferior frontal gyrus, anterior cingulate, posterior cingulate, and precuneus (Murphy et al., 2009). In addition, studies with other methodologies have also demonstrated that slow waves do not simply travel (Massimini et al., 2004), but they often do so preferentially through the major connectional backbone of the human cortex (Hagmann et al., 2008) and typically only involve a subset of brain areas (Nir et al., 2011). Thus, certain slow waves may be preferentially triggered by synapses that were recently strengthened during the wake period, priming specific circuits for consolidation (see Sections 2.1.2 and 2.2.1.1). Although these results have challenged several influential topographic observations of slow waves that have persisted since the original EEG recordings of sleep, they further posited the question of whether slow waves are local or global phenomena (see Section 2.1.2).

In the early days of EEG, all waves were considered global phenomena from the viewpoint of macroarchitecture. However, after the aforementioned report that indicated SOs propagate, the local aspects of SO have been well recognized. Herein lies a shift in perspective from macro- to microarchitectures. Although the relative occurrence of local vs. global events remains debatable, their existence has been widely acknowledged (Nir et al., 2011; Vyazovskiy and Harris, 2013; Seok et al., 2022). Evidence has suggested that the SO in humans may be locally modified based on previous experiences. Specifically, the amplitude of the SO may be modulated by recent waking activities (Huber et al., 2004). This finding is closely linked to the synaptic homeostasis hypothesis (SHY) (Tononi and Cirelli, 2014) (see Section 2.2.1.1). This issue is directly related to what drives the SWS and how it is regulated. Although SWS and SWA are closely related to homeostatic regulation, several studies have reported that they are not influenced or regulated by the circadian clock (Dijk et al., 1990). Multiple studies have demonstrated that the amount of SWS and SWA in both humans and rodents is influenced by the time spent awake. Additionally, studies on healthy young individuals have indicated that SWA increases during the deeper stages of sleep and decreases as the night progresses (Achermann et al., 1993; Dijk, 1995, 2009) (Figure 2). Compared to the later part of the sleep, the early phase of sleep exhibits a steeper slope of slow waves, and the slopes were positively correlated with neuronal synchrony, indicating that neuronal firing patterns and slow wave slopes reflect the homeostatic control of SWS (Vyazovskiy et al., 2009) (Figure 3). Stage N3 sleep commonly predominates during the first half of the night. Regarding the characteristics of SWA in sleep stages, a very small amount of high-amplitude SWA can be observed during sleep in NREM stage 1 (N1), with SWA gradually increasing in NREM stage 2 (N2) and N3 and then returning to a very small amount in the rapid-eye movement (REM) sleep stage (Dijk, 2009).

Although the entire mechanism of SWS regulation remains to be elucidated, the quest for a mechanism is directly linked to the question of why SWS is necessary for organisms.

The human brain develops at various stages of life. EEG during sleep has been suggested as a reliable method for monitoring age-related brain changes owing to its minimal susceptibility to confounding variables, such as environmental conditions during the day, thereby facilitating an unbiased evaluation of brain activity (Gorgoni et al., 2020). Exploring SWS throughout one's lifespan is essential for understanding dynamic shifts in sleep architecture and their potential impact on overall health.

SWA may be observed from the age of 2 months and commonly appears by the age of 4–5 months (Fattinger et al., 2014; Castelnovo et al., 2023). During early childhood, SWA tends to originate in more posterior regions of the brain than in adulthood (Timofeev et al., 2020). These changes in SWA topography progress from the posterior regions to the central area, and eventually to the frontal part of the brain in late adolescence (Castelnovo et al., 2023). In this study, the origin of SWA in SWS persisted in the frontal region throughout adulthood, possibly because of frontal cortex maturation. Although the same amplitude and frequency criteria for SWS are used in children and adults, slow waves in children display a larger amplitude (100–400 μV), steeper slope, and less widespread activity as compared to SWA in adults (Grigg-Damberger et al., 2007; Kurth et al., 2010) (Figure 3). Furthermore, SWA observed in children exhibits a greater tendency to originate from the right rather than the left hemisphere, potentially reflecting the incomplete development of interhemispheric white matter connections compared to that in adults (Castelnovo et al., 2023). SWS accounts for ~20%−30% of the total sleep time (TST) in children and 18%−22% of the TST in adults (Ohayon et al., 2004; McLaughlin Crabtree and Williams, 2009; Lee et al., 2022a). A meta-analysis of sleep parameters across the lifespan revealed that SWS comprises the largest percentage of sleep time in young children before puberty, diminishing with age at ~7% per 5 years from 5 to 15 years of age and 2% per 10 years in adulthood (Ohayon et al., 2004). However, a more recent meta-analysis that used only the 2007 AASM manual for scoring has revealed an age-related decline in TST without a significant age-related reduction in SWS (Boulos et al., 2019). Importantly, the former meta-analysis has also reported no significant change in SWS after 60 years of age. Meanwhile, a study based on two large cohort data with late-middle-aged and older adults has reported that delta power (1–4 Hz) decreases while slow power (<1 Hz) increases with age (Djonlagic et al., 2021a). The seemingly inconsistent reports on SWS/SWA changes in later life could be ascribed to different patterns of alterations between delta oscillations and SOs. When discussing changes in slow waves in older adults, being aware of the difference between sleep macro- and microarchitecture is particularly important.

The evolution of SWS characteristics throughout life, including amplitude, morphology, and regional origin, reflects the intricate maturation of the brain. Age-related SWS duration revealed in recent meta-analyses challenges conventional expectations, suggesting the need for further evidence.

Initially, sleep analysis primarily focused on macroarchitecture, encompassing objective parameters assessed in a 30-s EEG epoch, such as sleep stages (N1, N2, SWS, and REM), sleep onset latency, and TST. Currently, the importance of sleep microarchitecture in elucidating sleep and its applications has been increasingly recognized. The analysis of sleep microarchitecture involves the quantification and evaluation of intra-state elements at a finer level by examining factors such as SO, delta power, and SWA. In contrast to macroarchitectural analysis, which evaluates sleep by stage, microarchitectural analysis enables researchers to evaluate sleep across stages or during a short nap. Various analysis methods, such as power spectrum analysis, connectivity analysis (see Section 2.1.5), and phase–amplitude coupling analysis (Cohen, 2008; Tort et al., 2010; Hyafil et al., 2015), have been widely used.

One of the most striking examples of sleep microarchitectural analysis is the memory function of SO–spindle coupling. Communication between different brain networks relies on the interaction and coupling of various EEG signals and local field potential rhythms (Siegel et al., 2012). SWS is dominated by three types of oscillations: SOs, spindles, and hippocampal ripples. The anterior-to-posterior migration of SOs reaches subcortical structures, including the hippocampus, and thalamus-generated spindles spread through thalamocortical fibers to the neocortex and also reach the hippocampus. Several human studies have demonstrated that the timing or strength of SO–spindle coupling is crucial for the overnight consolidation of declarative (Helfrich et al., 2018) and non-declarative (Solano et al., 2021; Hahn et al., 2022) memories (Figure 4). Furthermore, as discussed in Section 3, research linking sleep microarchitecture to various diseases has become increasingly active in recent years. Growing evidence has revealed that SO–spindle coupling is impaired in older adults (Muehlroth et al., 2019) and in individuals with psychiatric disorders such as schizophrenia (Kozhemiako et al., 2022) and neurodevelopmental disorders (Mylonas et al., 2022). Low absolute delta power was significantly associated with incident hypertension, whereas no sleep macrostructural features were associated with it (Berger et al., 2022).

While the gold standard for sleep stage evaluation is visual inspection, automatic detection of microarchitectures such as SO and spindles is increasingly utilized, and various algorithms have been employed to detect SOs (Mölle et al., 2002; Esser et al., 2007; Staresina et al., 2015). Several open-source software/toolboxes are currently available for this purpose (Mensen et al., 2016; Djonlagic et al., 2021a). Nevertheless, no consensus on the criteria for SO has been established, and each study has used different definitions. The absolute number of detected SOs varies by definition, and as discussed later, the function of slow waves may differ depending on their frequency and amplitude, even if lumped into the same category of SWA. Therefore, referring to the definitions applied in each study is crucial to interpreting study results properly.

Sleep microarchitecture provides evidence of important dynamic characteristics of the sleep process that are not reflected in the conventional framework of macroarchitectural sleep evaluation. Although sleep staging is the basis of sleep medicine, it truncates finer details. Analyses of sleep microarchitecture provide potentially significant information regarding sleep.

Functional connectivity in the field of neuroscience refers to the statistical dependence between the time series of brain activity quantified at various spatial scales, ranging from the single-neuron level to the brain region level (Friston, 1994). Although functional connectivity can be evaluated using both invasive and non-invasive methods, non-invasive imaging techniques such as EEG, magnetoencephalography (MEG), and functional magnetic resonance imaging (fMRI) have been increasingly employed. These techniques enable researchers to examine the integration and synchronization of the human brain during sleep, which is difficult to assess using conventional research methods.

An EEG–fMRI study has demonstrated a strong reduction in corticocortical connectivity in SWS, whereas a widespread increase in corticocortical connectivity was observed in N1 and N2 (Spoormaker et al., 2010). Other studies analyzing the blood oxygen level-dependent (BOLD) signal fluctuations measured with fMRI during sleep have reported a breakdown of intra-network connectivity in SWS, specifically in the correlation between the anterior and posterior nodes of the “default mode network” (Horovitz et al., 2009; Sämann et al., 2011; Boly et al., 2012). This breakdown of corticocortical connectivity in SWS is consistent with a finding from an EEG and transcranial magnetic stimulation (TMS) combined study during sleep, which reported that TMS-induced activation spread over the cortex during wakefulness but remained local in SWS (Massimini et al., 2005). These findings indicate that although SWS is characterized by the synchronization of cortical neurons at the microscopic level, it can also be accompanied with a loss of integration among specific cortical areas at the network level. Several studies utilizing a whole-brain model have also suggested that these connectivity changes during SWS can explain the transition from local to global SWS (Deco et al., 2014; Cakan et al., 2021).

Since the analysis of functional connectivity during sleep also provides information on the functional differences of the brains, applications of this technique include the detection of brain changes associated with aging and the development of biomarkers for psychiatric disorders (Leistedt et al., 2009; Hein et al., 2019).

SWS plays a key role in the regulation of various hormonal axes. Its impact can be dual, with both promoting and inhibitory effects, depending on the specific hormones involved (Figure 5).

A robust correlation has been identified between SWS and the secretion of growth hormone (GH), a sleep-stage-dependent hormone that is primarily produced during SWS (Takahashi et al., 1968), irrespective of age (Van Cauter et al., 2000). For individuals with isolated GH deficiency, SWS is diminished compared with healthy controls (Astrom and Jochumsen, 1989). However, the results may vary depending on the type of GH deficiency (pituitary vs. hypothalamic) (Copinschi et al., 2010). Previous studies have reported the conflicting effects of insulin-like growth factor-1 (IGF-1), a neurotrophic mediator of GH, on SWS (Prinz, 1995; Shah et al., 2013). The highest release of prolactin occurs after sleep onset, irrespective of the time of day. Although a close temporal association between SWA and increased PRL secretion is known (Spiegel et al., 1995), no quantitative correlation has been identified between SWA and PRL as seen between SWA and GH release in men. Meanwhile, increased SWS may be associated with higher prolactin levels, as observed in breastfeeding women with higher levels of SWS and delta power (Blyton et al., 2002; Nishihara et al., 2004). SWS exerts an inhibitory effect on the hypothalamic-pituitary-adrenal (HPA) axis, leading to elevated cortisol levels during the latter portion of the night, which coincides with reduced SWS and increased REM sleep (De Feijter et al., 2022). During puberty, the release of luteinizing hormone (LH) following sleep onset is positively associated with SWS (Shaw et al., 2012). LH levels tended to increase 15 min after the initiation of SWS, suggesting the potential influence of hypothalamic NREM-active neurons on gonadotropin-releasing hormone (GnRH) or other neurons responsible for GnRH secretion (Shaw et al., 2012). Thyroid-stimulating hormone secretion, regulated by the circadian rhythm, has been demonstrated to be negatively correlated with SWS (Shekhar et al., 2021). Regarding the renin-angiotensin-aldosteron system, higher plasma renin activity and aldosterone levels during SWS than during REM sleep have been reported (Brandenberger et al., 1994; Charloux et al., 1999).

In addition to the effects on hormones described so far, the impact of SWS on glucose and lipid metabolism has received increasing attention. A shorter SWS is associated with reduced insulin sensitivity and may be correlated with a higher incidence of type 2 diabetes (Kianersi et al., 2023). A decline in the SWS over time may be associated with weight gain (Van Cauter et al., 2008; Reither et al., 2021). These findings may be due to a combination of factors, including alterations in appetite-regulating hormones such as leptin, ghrelin, and GLP-1 via activation of the orexigenic system (for reviews, see Reutrakul and Van Cauter, 2018; Antza et al., 2021).

The physiological role of SWS on the endocrine system is diverse and underscores the multifaceted impact of SWS. Although the relationship between sleep and hormones has been known for a long time, it is still an “old and new” field full of unexplained details.

A bidirectional relationship between SWS and immunity has been known; activation of the immune system enhances SWS and increased SWS benefits the immune system.

Regarding the immune-to-SWS directionality, cytokines induced by the inflammatory response, such as interleukin (IL)-1β and tumor necrosis factor (TNF), are known to be involved in the regulation of SWS (Majde and Krueger, 2005). Administration of granulocyte colony-stimulating factor, which increases endogenous IL-1 antagonists and soluble TNF receptors, reduces SWS and delta power (Schuld et al., 1999). Intranasal administration of IL-6 has also been reported to enhance SWS/SWA in the second half of the night (Benedict et al., 2009). Meanwhile, other human studies have indicated that acute administration of IL-6 or interferon-α, which is reported to induce NREM sleep in animal studies, decreases SWS during the early part of the night (Späth-Schwalbe et al., 1998, 2000). More recent studies have suggested that the administration of anti-inflammatory drugs, such as an IL-1 receptor antagonist and tetracycline that suppresses TNF production, increases SWA (Schmidt et al., 2015; Besedovsky et al., 2017). As most of the knowledge regarding the effect of cytokines on sleep to date is from animal studies, these inconsistent results may be due to more complex pro- and anti-inflammatory effects in humans (Irwin, 2015). A few studies have reported the effect of commonly used nonsteroidal anti-inflammatory drugs on SWS. In healthy adults, acute administration of aspirin at the daily dose decreases SWS (Horne et al., 1980), and ibuprofen delays the onset of SWS (Murphy et al., 1994). These findings may be related to reduced prostaglandin synthesis, including decreases in prostaglandin D2, but evidence is still lacking.

As for the SWS-to-immune system directionality, the interaction between SWS and the endocrine system (Section 2.2.2) has been considered to play a crucial role. The amount of SWS and SWA during the night following hepatitis A vaccination has been reported to correlate with the percentage of antigen-specific CD4⁺ T (Th) cells measured up to 1 year later (Lange et al., 2011). Moreover, the accompanying increases in GH and prolactin and reduction in cortisol levels are also associated with this correlation. GH, prolactin, and aldosterone support the release of cytokines, cell proliferation, lymphocyte responses, and T-cell migration (Petrovsky, 2001; Kelley et al., 2007; Besedovsky et al., 2012). Low cortisol levels during SWS may enhance efficient antigen-presenting cells (APCs)-T cell interactions (Petrovsky, 2001; Glaser and Kiecolt-Glaser, 2005). Taken together, SWS may play a role in facilitating the interaction between APCs and T cells, which leads to a stronger immunological memory, by inducing the optimal pro-inflammatory “endocrine milieu” (Besedovsky et al., 2019).

As with memory and sleep research, the challenge regarding research on immune and sleep interaction is to assess the involvement of each sleep stage separately. However, increasing evidence supports the concept that immunological memory also consolidates during SWS as an analogy to the active systems consolidation hypothesis in memory research (Section 2.2.1.2) (Westermann et al., 2015). The interaction between SWS and immunity is deeply related to hormone secretion, and this can be viewed as a SWS-mediated neuro-immune-endocrine linkage. Although evidence in humans is still limited, this is an area expected to develop in the future.

SWS plays a significant role in reducing sleepiness and aiding restoration, both subjectively and objectively. However, waking from SWS can cause residual sleepiness and sleep inertia. Understanding the connections among feelings of rest, sleepiness, sleep inertia, and SWS is crucial in the field of sleep physiology.

Reports have mainly associated a subjective feeling of being rested in the morning with higher sleep efficiency, longer total sleep time, and duration of SWS (Keklund and Akerstedt, 1997; Kohyama, 2021). SWS is the only sleep stage associated with subjective feelings of rest (Keklund and Akerstedt, 1997). Objectively, a higher SWS is crucial for physiological restorative sleep because of several factors such as a significant reduction in sympathetic activity, energy conservation, and an increase in growth hormone secretion that helps with the repair process. Disruption of the macro- and microarchitecture of the SWS, such as the occurrence of arousals or the presence of alpha intrusions (alpha–delta sleep), has been reported to be associated with unrefreshing sleep in various disorders, such as chronic fatigue syndrome, fibromyalgia, and some psychiatric disorders (Van Hoof et al., 2007; Vijayan et al., 2015; Wu et al., 2017).

Although the extent of SWS is closely linked to the subjective feeling of rest, awakening during SWS produces more sleep inertia than during other sleep stages (Tassi and Muzet, 2000). Sleep inertia is a transitional state characterized by reduced alertness and impaired cognitive performance upon awakening. It is characterized by grogginess, confusion, and a decline in cognitive and sensorimotor abilities (Hilditch and McHill, 2019). Individuals may also experience a decline in mood following awakening from SWS at night compared to their mood before sleep. Although the mechanism of sleep inertia is still unclear, the continuance of low-frequency brain activity in the minutes after awakening, particularly in the posterior regions of the brain, has been observed during sleep inertia (Ferrara et al., 2006; Marzano et al., 2011). Evidence supports a relationship between sleep inertia, intrusions of sleep-specific EEG characteristics, and sleep-specific brain functional connectivity. Compared to N2, waking up from SWS causes greater disruption of brain functional connectivity (Vallat et al., 2019). Global spectral power analysis after waking from the SWS revealed that the recovery of the global power of lower frequencies is more rapid than that of higher frequencies during sleep inertia. Moreover, the reduction in clustering and higher path length in the delta band was associated with increased sleepiness scales and higher lapses in a psychomotor vigilance test (Hilditch et al., 2023).

Several studies have confirmed the benefits of napping on mood and cognitive performance (Hsouna et al., 2019; Dutheil et al., 2021). However, longer naps are more inclined to transition from a lighter sleep stage to deeper sleep with a higher delta power, which is the main component of SWS. SWS is closely associated with sleep inertia, and can negatively affect short-term cognitive performance and alertness (Takahashi et al., 1998; Tietzel and Lack, 2001). Several studies have recommended limiting nap duration to 10–20 min to avoid this transition (Hayashi and Hori, 1998; Hayashi et al., 1999; Tietzel and Lack, 2002). Immediately after a 10-min nap, a notable increase in cognitive performance was observed, and this heightened state was maintained for at least an hour (Tietzel and Lack, 2001). By contrast, a 30-min nap in this study resulted in a decline in cognitive performance immediately after the nap, possibly because of sleep inertia. Interestingly, a study comparing 10-, 30-, and 60-min naps has reported that awakening during SWS was most prevalent in the 30-minute nap, despite an increased percentage of SWS with longer nap durations (Leong et al., 2023). However, this study also demonstrated that the memory encoding benefit is moderately associated with only a 30-min nap.

The link between higher SWS and feelings of rest is unique and has not yet been fully elucidated. Although longer nocturnal sleep may have desirable effects on restfulness, a limited nap duration is essential to maximize the benefits of naps (Tietzel and Lack, 2001; Fushimi and Hayashi, 2008). Exploring alterations in sleep microarchitecture, such as spectral power and functional connectivity, upon awakening from SWS can aid in elucidating the mechanisms underlying sleep inertia.

Sleep restriction, a highly effective treatment for chronic insomnia, typically involves an initial reduction in the amount of time spent in bed (Ong et al., 2016b). Over time, this intervention often leads to enhanced sleep efficiency and a gradual increase in the TST.

The impact of sleep restriction on various sleep stages has been studied, with a particular focus on SWS. Studies have revealed that reducing the time spent in bed to 4 h per night decreases the duration of sleep stages N1, N2, and REM, but not SWS, which is preserved (Brunner et al., 1990, 1993). In cases of sleep restriction lasting ~5 h per night, the N1, N2, and REM sleep stages tend to decrease, although the percentage of SWS often increases (Voderholzer et al., 2011). This change in SWS can persist even during the recovery phase when longer sleep durations are permitted (Xin et al., 2022). An increase in SWS and REM sleep may also occur during sleep recovery after sleep fragmentation or deprivation (Cheng et al., 2021). Delta power tended to increase initially with partial sleep deprivation. A direct relationship has been observed in rats, where shorter sleep durations correspond to greater increases in delta power (Kim et al., 2007). During sleep restriction in humans, the cumulative sum of delta power across each sleep period, referred to as slow wave energy, was lower than the baseline. However, the response during the recovery phase varied with sleep recovery duration (Banks et al., 2010). An increase in delta power was observed during both sleep restriction and recovery night, whereas the total slow-wave count increased only during the recovery night (Åkerstedt et al., 2009; Plante et al., 2016). A higher aggregate slow-wave amplitude in the frontal region occurs owing to a proportional increase in the quantity of high-amplitude slow waves, coupled with a simultaneous decrease in the count of low-amplitude slow waves from baseline to the sleep restriction and recovery periods. During the recovery period, the slopes of low-amplitude waves increase in the frontal region, but not of high-amplitude waves (Plante et al., 2016).

Although sleep restriction is a valuable tool for the management of chronic insomnia, its potential impact on sleep staging and microarchitecture needs to be considered. The adaptability of the SWS to sleep restriction highlights its significance in restorative processes. The dynamic changes in microarchitecture observed during sleep restriction and subsequent recovery contribute to a deeper understanding of sleep homeostasis.

Studies on the glymphatic system have indicated its involvement in the pathogenesis of proteinopathies, including AD (Nedergaard and Goldman, 2020). The glymphatic system is a collaborative neural lymphatic network consisting of the circulation of CSF in the perivascular space through the astrocytic aquaporin-4 channel and lymphatic vessels located in the dura, covering both the brain and spinal cord (Iliff et al., 2012; Nedergaard, 2013). They work together to effectively clear toxic and nonfunctional waste materials from the brain, such as Aβ, tau, α-synuclein, and inflammatory cytokines.

The glymphatic system is considered active during SWS. The volume of CSF entering the brain parenchyma from the subarachnoid space and into the periarterial spaces may increase during SWS (Xie et al., 2013). A study with mice has demonstrated that glymphatic influx correlates positively with cortical delta power (Hablitz et al., 2019). A short SWS duration is correlated with enlargement of the perivascular space, representing impaired perivascular drainage (Baril et al., 2022). Existing research has suggested that the accumulation of Aβ, tau, and α-synuclein can lead to reduced glymphatic activity (Lopes et al., 2022). Several studies have reported a negative correlation between the amount of SWS and Aβ levels in CSF; further, the disruption of SWA during sleep resulted in increased CSF Aβ levels in individuals with normal cognitive function in several studies. However, conflicting data also exist (Ju et al., 2017; Sangalli and Boggero, 2023), and data indicating a causal relationship between SWS enhancement and AD biomarkers are lacking.

Although current data suggest that a higher SWS may contribute to the optimal glymphatic clearance of neurotoxic substances, data regarding the relationship between the glymphatic system and sleep characteristics in healthy adults remain limited (Sangalli and Boggero, 2023). Developments in this field may contribute to future dementia treatment.

The use of acoustic stimulation has been shown to boost SO (<1 Hz) and SWA and improve sleep-dependent memory retention in young adults (Ngo et al., 2015; Ong et al., 2016a, 2018; Leminen et al., 2017) and older adults (Papalambros et al., 2017). This procedure has also been reported to enhance SWA in individuals with MCI and improve performance in word-recall tasks, although data on its long-term benefits remain limited (Papalambros et al., 2019). A more recent pilot study has demonstrated that acoustic stimulation could increase SWS in adults with mild-to-moderate AD; however, its effect on cognitive function has yet to be proven (Van Den Bulcke et al., 2023).

Transcranial direct current stimulation (tDCS) improves SWA and overnight memory retention in young adults (Marshall et al., 2004, 2006). Many studies have demonstrated an association between enhanced SO using tDCS and improved declarative memory in young and older adults, specifically for word-pair learning (Westerberg et al., 2015; Barham et al., 2016) and visuospatial tasks (Ladenbauer et al., 2016). However, there have also been some inconsistent results showing no enhanced SO or improved memory (Eggert et al., 2013; Paßmann et al., 2016; Bueno-Lopez et al., 2019). One meta-analysis has reported that SO augmentation did not enhance procedural memory (Barham et al., 2016). In individuals with MCI, an increase in SO has been observed with the improvement in visual declarative memory, but not with verbal memory improvement (Ladenbauer et al., 2017).

A few studies applying transcranial alternating current stimulation (tACS) have demonstrated an increase in SO and a dose–effect relationship (Jones et al., 2018; Ketz et al., 2018). Moreover, TMS applied during NREM sleep has been reported to augment SWS (Massimini et al., 2007). Although the effect of TMS may be more spatially precise than that of acoustic stimulation or tDCS, limited studies have been conducted to date (Feher et al., 2021).

In sum, many studies have succeeded in augmenting SWS using these techniques; however, there are mixed results concerning their beneficial effects on memory, especially in older adults. Further studies examining the long-term effects of augmenting SWS and its impact on the development of cognitive impairment are warranted.

Several pharmacotherapies have been discovered that extend the duration of SWS through different mechanisms, such as GABA_A agonist, GABA_B/gamma hydroxybutyrate (GHB) agonist, serotonin-2A receptor antagonists (e.g., ritanserin and eplivanserin), and multiple receptor antagonists (e.g., trazodone and mirtazapine) (Walsh, 2009). Medications acting on α2-δ calcium channel ligands (e.g., pregabalin and gabapentin) have also been shown to enhance SWS. A study using Tiagabine, an inhibitor of GABA transporter 1, has demonstrated increased SWS and improved performance on cognitive tasks during sleep restriction (Walsh et al., 2006). However, another study with Tiagabine has reported no benefit of augmented SWS on memory consolidation in healthy adults (Feld et al., 2013). Sodium oxybate has also been associated with increased SWA and better cognition in healthy young adults (Hall, 2009; Walsh et al., 2010). Olanzapine, a serotonin-2A-dopamine-D3 receptor antagonist, similarly increased SWS in adults with schizophrenia (Göder et al., 2008) but decreased SWS latency without increasing SWS in adults with depression (Lazowski et al., 2014).

Meanwhile, several pharmacological interventions have been reported to decrease SWS. While antiepileptics that act on voltage-gated sodium channels, such as valproic acid and phenytoin, exhibit conflicting effects on SWS, phenobarbital, which prolongs the opening time of chloride channels by acting on GABA_A receptor subunits, typically appears to attenuate SWS (Carvalho et al., 2022). Similarly, benzodiazepines, a GABA_A receptor agonist, have also been associated with reduced SWS (De Mendonca et al., 2023). Although sex hormones might impact SWS, and women generally have higher SWS, no significant improvement in SWS was observed after gender-affirming hormone therapy (e.g., estrogen and antiandrogen) in the transfeminine group. However, SWS reduction after testosterone administration has been observed in the transmasculine group (Morssinkhof et al., 2023).

The benefits and disadvantages of drug-induced changes in SWS to human health have not been fully explored to date due to difficulties in the study design.

Exercise enhances SWS in both young and older adults (Cassim et al., 2022). When compared to the absence of exercise, acute evening moderate-intensity exercise had the highest ranking for improving the proportion of SWS, followed by high-intensity exercise and low-intensity exercise. Comparing acute evening moderate- and high-intensity exercise to low-intensity exercise, both interventions led to an increase in the proportion of SWS, and no significant difference was observed (Yue et al., 2022). Listening to the suggestion to “sleep deeper” while falling asleep can promote SWS and SWA (Cordi et al., 2014; Besedovsky et al., 2022). However, no improvement in declarative memory was observed with this intervention, which could be due to the lack of SO-spindle coupling (Beck et al., 2021).