Journal of Sleep Disorders: Treatment and CareISSN: 2325-9639

Reach Us +18507546199

Review Article, J Sleep Disord Treat Care Vol: 6 Issue: 1

Nocturnal Arousals: Stirred, Not Shaken: A consideration of the EEG and sleep

Haigh PM*
Foundation Year 2 Doctor, King’s College Hospital, USA
Corresponding author : Haigh PM
Foundation Year 2 Doctor, King’s College Hospital, London, UK
Tel: 07799554453
E-mail: [email protected]
Received: October 03, 2016 Accepted: October 13, 2016 Published: January 05, 2017
Citation: Haigh PM (2017) Nocturnal Arousals: Stirred, Not Shaken: A consideration of the EEG and sleep. J Sleep Disor: Treat Care 6:1. doi: 10.4172/2325-9639.1000189


Where are ‘we’ when we sleep? Does any part of us remains cognizant of the external environment in which our body rests? The closest we can come to approaching this quasi-psychophilosophical enigma is through monitoring the neuro-electrical activity of our brains during this time. This activity is categorized into high-amplitude and low-amplitude frequency waves visible on the electroencephalogram (EEG). This paper reviews the evolution of the EEG as it is has been utilized to inform our knowledge of the stages of, and transient emergences from, sleep. There is discussion distinguishing sleep from other semi/unconscious states such as anaesthesia or coma while the primary focus is on the detection, aetiology and purpose of ‘arousals’ – both cortical and somatic – and where these feature in the Cyclic Alternating Pattern of normal and disordered sleep. The sequelae of aberrant arousals is also considered in relation to daytime hypersomnolence and apneas with a reiteration of the importance of combining academic tools with clinical practice.

Keywords: Sleep; EEG; Arousals; REM Non-REM sleep; Disordered sleep


Sleep; EEG; Arousals; REM Non-REM sleep; Disordered sleep

Watchman, What of the Night?

From the mid-1930s, sleep and the electroencephalogram (EEG) have been close bedfellows, when Loomis et al. [1] observed the EEG readings of sleeping subjects and dismantled the myth of sleep as an unvarying, homogenous state beginning when we close our eyes and ending when we open them again. A couple of decades later, Aserinsky and Dement, working with Kleitman, pioneered the concept of rapid-eye movement (REM) sleep and gradually our knowledge evolved [2,3]. Without tools such as – and especially – the EEG, objective analysis of the sleeping brain remains a quite philosophical conundrum à la Schrödinger’s cat (or perhaps better: catching the fridge light going out). The EEG is a literal nightwatchman over the various neural events that precede, accompany, or follow pathophysiological occurrences during sleep. It has a watching brief, passively recording the electrical hustle and bustle, and I will consider some of the clinically significant results of its surveillance. Although relevant, I have preferred to side-step dreams, nightmares and somnambulism in order to do justice to a more controversial area [4] where sleep medicine and electroencephalography converge. Although the chief clinical use of EEG is in epilepsy [5], recording brain activity in respect of partial and generalized seizures, for this paper I look elsewhere: at the phenomenon of the EEG arousal.
I do not intend an overview of polysomnographic techniques in general, or an exhaustive analysis of the mechanics underlying the EEG in particular, however, it is necessary to sketch out some principles for interpretation. The EEG detects electrical activity on the surface of the brain discharged by inter-neuronal communication. Collectively, and colloquially, these signals are ‘brain waves’, visible (when considerably amplified) on the EEG as a contemporaneous commentary on the functioning of the brain by representing activity in its specific topographical regions. These waves are no arbitrary, random oscillations; over time their individual idiosyncrasies have been systematized. In 1929, a few years after falling off his horse (and proceeding to change midstream from astronomy to psychiatry), Hans Berger discovered two low amplitude-high frequency (LAHF) waves, alpha (α) and beta (β), by placing electrodes on (among others) the head of his admirably compliant teenage son, Klaus [6]. The α waves, with a frequency of 8-13 waves per second, are seen in all age groups (as the Berger children attested to) but most typically in relaxed adults with their eyes closed. Also known as the ‘posterior dominant rhythm’, these waveforms are clearest in the occipital lobes (round the back; responsible for vision) but occur rhythmically on both sides of the head, possibly with a slightly higher amplitude on the non-dominant side [7]. Generally, β waves are quicker, more than 13 per second and also mostly seen in awake subjects. Principally, they are detected in the frontal and central areas of the brain and may be misread for muscle action potentials.
Theta (θ) and delta (δ) waves are considered together as the high amplitude-low frequency (HALF) waves, significant here as the predominant waveforms of slow wave sleep (SWS). Delta waves especially, with a frequency of 3 Hz or less but the highest amplitude (making them the strongest signal), are normally seen in deep sleep in all ages, especially young infants; but an abnormal finding in the awake adult.
We must also introduce some key morphological variants seen on an EEG. ‘K complexes’ are δ waves, sometimes with a sharp apex, distinctive in the middle stages of Non-REM sleep. They occur symmetrically throughout the brain, usually more prominent in the frontal regions on partial arousal from sleep. A transient complex often follows brief noises, while sustained sounds produce repeated complexes followed by an ‘arousal burst’ – a run of rhythmic θ waves. Sharp ‘V waves’ also occur during sleep, most prominent at the vertex (the ‘bull’s-eye’ on the top of the cranium). Very distinctive, they are most common during stage II sleep (see below) and often after disruption and, like K complexes, during partial arousals. Sleep spindles are groups of waves generated in the thalamus, especially, like V waves, during stage II. They have frequencies in the upper α or lower β levels; lasting a second or less, they increase in amplitude initially and then slowly decrease. So-named from the resemblance to a spindle (‘fusiform’ being a ubiquitous descriptor in medicine though rather passé now, as few students are still able to visualize, unprompted, a spindle of any form).
In sum, in our vigilant hours, the global electrical activity of the brain shows characteristic β waves signifying an internally coherent ‘white noise’ of desynchronised nerve cell activity, independently firing off their charges but permitting effective co-ordinated function. As we settle down to relax and rest our eyes, these brain waves also slow to a gentler α rhythm. The drowsiness that overcomes us after half an hour (or less) is marked by continued decline in wave frequency and the slow HALF θ waves take over, indicating increased levels of synchronized brain activity.

How Being Awake Differs from Being Asleep Differs from Being Comatose

The purpose of sleep notoriously evades any more explicit elucidation than simply to prevent sleepiness; but more of the neurobiology is known, including its origin in the hypothalamus, brainstem and basal forebrain; there is a homeostatic component too, sleep may derive from a protein that accumulates in wakefulness until crossing a threshold registers a ‘need to go to sleep’ that is relieved when we submit [8]. Sleep is a voracious consumer of man (and animal) hours: if deprived of its rightful allotment within a 24 hour cycle, it will ruthlessly snatch it back wherever and whenever it can, leading miserably to the sleep disorders considered below.
To understand the various electrical episodes within the arc of a sleep we must be clear as to structure; the stages of progressively deepening sleep, proposed in the early 20th century (with periodic refinement) more or less apply today. Chronologically, ‘going to sleep’, begins with a quiet wakefulness, a winding-down of physical and conscious mental activity prior to the onset of the first stage of Non- REM sleep. As there can be rapid cycling between wakefulness and Stage I, during the drowsiness, α rhythm dominates. As sleep deepens in later stages, the waves become longer and less frequent, settling into θ. This phase lasts around 5 to 10 minutes before slipping into a slightly deeper Stage II. A minor auditory stimulus might rouse the individual who would think they ‘just nodded off for a minute’. In sleep fragmentation associated with subcortical events there is often an increase in the absolute amount (and proportion) of Stage I [9]. Stage II is a phase of 10-20 minutes, showing θ waves interspersed with spiked K-complexes and sporadic bursts of high frequency spindles. This stage is considered to be fully asleep and relatively sensorially detached from the external world.
The latter stages III and IV are considered SWS. Beginning as a mixture of θ and δ waves, there is increasing diffuse δ activity, until it constitutes half the overall EEG. The θ waves progressively disappear and spindles dwindle. The deepest level of sleep. After around half an hour in Stage IV there is a rapid resurfacing through III and II, but then, rather than emerging at Stage I, we divert into a wholly new phase: the first tranche of REM sleep, occupying a fifth of the adult sleep epoch but in neonates about a half. (An interesting exception is found in narcoleptics, who experience REM onset sleep.) Not long after Kleitman’s laboratory had proposed REM sleep, Michel Jouvet, in Lyon, recognized it as third state distinct from both Non-REM sleep and waking and covered the term ‘paradoxical sleep’[10]: if the reduced cortical activation of sleep is represented by diffuse slowing of the EEG, REM sleep flies in the face of this, demonstrating α and β bursts despite near bodily paralysis (from hyperpolarization of alpha-motor neurones; a putative theory being to prevent the physical enactment of dreams). In addition to its eponymous characteristics, REM sleep is generally the more exciting; it accommodates deep dreams (though dreams may inhabit all stages), breathing and heart rate is more erratic, there are erections of nipples and penis/clitoris, and loss of tone in airway smooth muscle and skeletal-muscle. (Drop in temperature was perhaps one of the first observed physical sequelae of deep sleep: Hippocrates believed sleeping occurred by migration of blood from the limbs to ‘inner regions’ of the body’s core to be warmed [11]) An EEG here will display active LAHF rhythms, whereas, Non-REM sleep shows HALF [12]. Within REM sleep are two further stages: the tonic – with well-delineated and synchronous spindles, and diffuse slowing in θ and δ with the emergence of V-waves and K-complexes [13] – intermittently interrupted by the phasic stage: typified by a slower, fragmented EEG (similar to Non- REM Stage I) until eventually the α wave disappears and the lower voltage background brings the ‘saw-tooth’ β to prominence. This is the twitching hour, as seen in dogs by the fire. REM and Non-REM sleep take it in turns with a turnover rate about the length of a football match – spending progressively longer periods in stage II and REM sleep and towards the end of the sleep epoch less time is spent in deeper stages. On blinking into life at the reveille or a partner’s elbow it is probable we have just been in the REM stage.
It has been suggested [14] that the body at rest (in Non-REM sleep) reveals the default state of the brain, its ‘relaxed’ pattern when vigilance is low. Consciousness maintained but reduced, as on standby. The EEG reflects this with its repose into ‘lazy rhythms’. So how do arousals feature? As the afferent nerve fibres from the body’s outlying sense-organs (eg) in the glottis, arrive at the stem of the brain, the cortex is informed of the detected stimulus. Via the reticular formation, connections rapidly ripple throughout the cortex; the brain cells are kicked out of their gentle firing cadence, ‘arousing’ the brain and desynchronizing the placid regularity of the waves. These arousals may be implicated in the reciprocal interaction between Non-REM sleep and waking; between Non-REM and REM sleep; and also in distinguishing between states of consciousness:
“[The] functional significance of arousal in sleep, and particularly in NREM sleep, is to ensure the reversibility of sleep, without which it would be identical to coma. Arousals may connect the sleeper with the surrounding world maintaining the selection of relevant incoming information and adapting the organism to the dangers and demands of the outer world” [15].
If arousals are such dynamic liaisons, it begs the question whether they serve to warn the conscious cortex of (very) imminent trouble ahead; or more of a post-resolution feedback relay to the sub-conscious, reassuring that further attention is unwarranted (“everything’s all right, go back to sleep.”). The precise sequence is disputed territory, especially with respiratory-related arousals (see below).
To say the presence of arousals prevents a deep sleep from becoming a coma seems not wholly accurate. Though ‘reversibility’ can offer a working distinction (and in distinguishing a transient arousal from an awakening) between the states, how deep sleep differs from general anaesthesia (GA), differs from a coma is better informed by their respective EEG pictures. Lack of arousal certainly does differentiate GA from sleep: it is postulated that anaesthetics suppress cortical function, which blocks arousal through the thalamus to cause unconsciousness [16] but GA also shows a progressive increase in HALF activity as the level of anaesthesia deepens. Although EEG patterns in comatose patients are inevitably contingent upon the extent of the brain damage, they too commonly take the HALF form seen in GA, in addition to sharing a lack of arousals and reduced cortical responsiveness. GA, therefore, is closer to a coma than to a deep sleep, despite patients being ‘put to sleep’ with anaesthetists’ misleading reassurance [11].

“Alarum’d by his sentinel . . .”

A transient arousal from sleep is what we colloquially refer to as ‘stirring’ without fully waking up; when the forces promoting wakening (from the pontine and mid-brain tegments and posterior hypothalamus) temporarily overwhelm the forces of sleep (in the medial brainstem, dorsal reticular medulla and anterior hypothalamus) [15,16]. Electroencephalographically, arousals are disruptions to normal deep sleep depicted as abrupt deviations in frequency from HALF to LAHF on the heels of at least 10 seconds of continuous SWS, possibly with θ and α waveforms but no spindles. Originally, an EEG desynchronization was synonymous with an arousal but this misnomer conflates the phenomenon with its representation (like thinking heat the same as temperature). To facilitate progress ‘arousals’ will serve as an umbrella term for all such disturbances manifest in the sleeping EEG and I propose to adopt the hierarchical categorization of nocturnal arousal suggested by Thomas, 2003:
“absent (no EEG or autonomic marker change), subcortical (no EEG change but autonomic arousal), cortical-I (K-complexes or delta waves), and cortical-II (presence of alpha/beta frequencies)” [3].
This allows us to consider their broader clinical effects rather than becoming bogged down in a quagmire of exact classification. However, we will further pin down arousal as: ‘cortical’ as defined above (incorporating micro-arousals); or ‘somatic’, with an objectively observable change to their sleep – an ‘awakening’. There has been fairly vociferous dissatisfaction with the 1992 criteria for defining and grading arousals, as originally laid down by ASDA [17], (the American Sleep Disorders Association; wisely since renamed the American Academy for Sleep Medicine – AASM). This debate has occupied considerable print-space and is effectively summarized in an evidence-based review by Bonnet et al. [18]. It is worth noting, that one point of issue with the ASDA manual – other than, by its proscriptive attitude to duration it disregarded various other morphologies (such as K complexes and δ bursts) that might potentially be viable arousal contenders – was that it implicitly characterized arousals as detrimental trespassers upon sleep, whereas the alternative position advanced by the Parma group chiefly, was of arousals as an integral element in the fabric of normal sleep, possibly with a role to play in its physiological regulation [19].
Described in 1985 by Terzano et al. [20] the cyclic alternating pattern (CAP) is a feature of the four normal non-REM sleep stages, and incorporates periodic ‘natural’ arousals that occur spontaneously and cyclically between A and B phases (B phase is devoid of arousals). On an EEG the AI subtype is a ‘synchronization arousal’ characterized by a predominance of δ bursts, K-complex sequences, and V-waves (excluded from the original ASDA gradings) [15]. The AII subtype is a muddle of slow and fast rhythms, while the AIII subtype is almost exclusively fast. The theory underlying CAP countered the received view of arousals (that brought the original ASDA classification) as hostile to sleep. Within CAP some arousals may even protect sleep from disruption. This protective function (or the absence of it) was paid greater heed in investigations into the pathophysiology of cot death (‘Sudden Infant Death Syndrome’) [18]. But the notion was not a fresh one; back in 1968, in a key paper in Science [21]. Broughton outlined arousal disorders as abnormalities which prevent normal protective arousals in Non-REM sleep from doing their job of eliciting full alertness, and instead produce a pathologic semi-arousal. Studies since have primarily viewed arousal as an oscillating phenomenon with an adaptive function, although arousals are consistently positively correlated with levels of sleep fragmentation [22,23].
In focussing on definition, thus far, I have somewhat taken for granted the issue of identification, which grossly misrepresents the difficulty in practice of sifting out the transient crests of interest amid the rolling reams of innocuous peaks and troughs. Because of their brevity, EEG arousals are easily mistaken for artefact (distortions from the million-fold amplification) and disregarded. Moreover, whilst the background of deep SWS differs substantially from that of the LAHF of arousal; the background in light and REM sleep however, is also LAHF, making it trickier to distinguish from arousal. All further constraints on accurate interpretation; like trying to discern (and quantify) discrete ripples created by a single pebble plopped into a choppy river.
Arousals, as ‘expected guests’ of physiological sleep, have been characterized as pathology-associated markers without being necessarily harmful in themselves; in that respect rather analogous to DNA polymorphisms in the study of the human genome. Similarly, arousals on an EEG enable temporal correlations with other physical events, for example, an apnoeic episode, and as. Markers of cortical activity they have been used in concert with autonomic response measurements with which they are highly correlated [18,19]. The latter having greater sensitivity at lower levels, being induced by stimuli below the conscious threshold and not invoking a detectable cortical response. Arousal identification is used chiefly in diagnosis of obstructive sleep apnoea (OSA) [24], and also in Periodic Limb Movement Syndrome (aka Willis-Ekbom Disease or ‘restless leg syndrome’), [25,26] both of which may be associated with tens to hundreds of arousals per noctem [9]. Arousals associated with OSA arise predominantly before and during REM sleep, when the cortical and midline structures are more active. However, in A phase of the CAP (a Non-REM event) there are episodic (though not wholly regular) arousals. As early as 1971 some attempt was made [27] to differentiate transient arousals (then ‘phases d’activation transitoire’) in Non-REM sleep (increased frequency, low amplitude and the disappearance of δ waves spindles) from REM sleep (α bursts and decreased ocular activity). These phases occurred most frequently in superficial sleep (Stage I and REM). Whatever the stage, arousals can be generated endogenously by the cortex, or in response to exogenous sensory perturbation (i.e) noise, pain or respiratory distress. There is supportive evidence to suggest that the arousal stimulus in Non-REM sleep is related to the level of inspiratory effort (though Berry et al. [28] found it more likely the arousal stimulus consisted of multiple components, each increasing with inspiratory effort). However, arousals evoked by sensory stimuli are indistinguishable in experience from ‘spontaneous arousals’ [15].

He Sleeps Well Who Knows not that He Sleeps Ill

The primary pathological consequence of nocturnal arousals and awakenings is sleep fragmentation leading to daytime hypersomnolence and concomitant impairment of function, mental status and a host of other miserable sequelae handsomely represented in the literature [24,29]. To consider the connexion (either causal or coincidental) between arousals and sleep disorders, we must first formulate two questions: (i) given that arousals occur in normal physiology, at what point – by either their frequency or their intensity (or both) – do they become abnormal, or inherently pathological? (ii) does the presence (or absence) of arousals consistently indicate the presence (or absence) of sleep disorder? And underpinning both these questions remains the central ‘chicken-and-egg’ issue: do more arousals cause sleep disorders, or vice versa?
To address (i) we must ideally be in possession of a normative baseline of arousals in each slice of the demographic (ie) age, gender, ethnicity etc. Mathur et al. [30] in Edinburgh found that awakenings increased significantly with age and suggested there may be a high level of arousals in the general healthy population. As ever, establishing definitive norms can only work with consensual definition, and those supplied here both comply with the AASM criterion of 3 seconds duration and more studies have reported using an arousal index based upon the manual; however these do ignore CAP and consequently may only provide some indication of a normative level [31].
The response to (ii) is to examine the pathologies associated with nocturnal disorders. The majority of studies of arousals implicate sleep apnoea; predominantly OSA. While little is known of the mechanisms of arousal in central sleep apnoea (CSA), it may be significant here also [28]. Such preoccupation with OSA is reflected now; although brief arousals have been indicated in other disease states: allergic rhinitis; juvenile rheumatoid arthritis and Parkinson’s [9], these are beyond our immediate scope.
OSA justifies its headline status by being one of the chief clinical rationales for referral to polysomnography and by no means an openand- shut case. Previously, the general view of OSA (since Remmers et al. [32]) was of a complete upper airway occlusion occurring during sleep, preventing breathing. This state persisted for a matter of seconds until ‘physiological alarm bells’ sounded to induce a gasp to re-open the airways through an influx of positive pressure. These mini-asphyxiations could occur as many as a hundred times per notcem, not always waking the patient, but, notwithstanding whether they were ‘aroused’ (by our definition), or even ‘awakened’, these episodes form the primary cause of disturbed sleep [15]. However, the physiological reality of this version of events – in which the arousals are helpful mediators from the subconscious, summoning aid from the conscious to ensure the continued supply of oxygen – has been questioned and bears examination here as it brings into focus the fundamental purpose of nocturnal arousals as perceived by EEG.
The old orthodoxy just recounted was really challenged in the last decade when Younes [33] in Canada gave it short shrift in a thorough but highly-involved experiment where he induced apnoea and examined the temporal relationship between the onset of an EEG arousal and restoration of airflow. He concluded that upper airway occlusions resolved with or without an arousal; thus, some apnoeic incidents may be the result of a pathological arousal rather than arousals simply being remedies for respiratory distress; alternatively, arousals and restoration of flow are simply incidental, he conjectured, and arousal responses may promote initiation of further airway obstruction (a vicious cycle). There was some earlier support for the possibility that the received view was erroneous: an 1998 editorial [23] in the European Respiratory Journal reported: “Apnoeas that do not result in a definite arousal have been shown to have the same duration as apnoeas with arousal” and hitherto this phenomenon was attributed mostly to subcortical arousals or even inadequate reading of the EEG, Younes at least offered something other than human error by way of explanation (Appendix 1).
As for the chicken and the egg? Whilst there is undeniable plausibility to Younes’ densely-elaborated hypothesis, the fact that his baton has yet to be decisively taken up, now five years on, suggests we should baulk from jettisoning the grounds beneath twenty years research into EEG arousals and OSA. This notwithstanding, in ‘restless legs syndrome’, Karadeniz et al. [26] found periodic limb movements consistently preceded by arousals, and as such could be considered symptoms of arousal disorder. If the cap fits one, it would be myopic to eschew the possibility of it being worn by others.

“Ce que j'ôte à mes nuits, je l'ajoute à mes jours”

Among its various, perhaps more prominent, clinical functions, the EEG allows us to recognize and characterize the unseen events of sleep. More significantly, it permits analysis of an individual’s nocturnal neural activity; to comb for patterns and markers attendant with pathology, whether it be syncope or limb movement. As such it remains a vital diagnostic tool, and, while conflict still rages over fundamental issues as the role of the arousal in apnoea it shows little sign of being superseded, despite fears of losing the race with computerized tomography and it continues to plough an adjacent furrow to functional neuroimaging. However, the attendant caveat is to bear in mind the result of nocturnal disruptions; the sleep fragmentation and consequently reduced quality of waking life. This is what brings the patient to the clinic and this is what must be kept in sight in order that we do not lose our way amid the enchanted forest of the EEG.


Rowan AJ, Tolunsky E (2003) Primer of EEG: with a mini-atlas. 1st Edtn. Butterworth Heinemann.
• Gilmartin GS, Thomas RJ (2004) Mechanisms of arousal from sleep and their consequences. Curr Opin Pulm Med 10: 468-474.


Track Your Manuscript

Share This Page