The Future of Parasomnias

ABSTRACT

Parasomnias are abnormal behaviours or mental experiences during sleep or the sleep–wake transition. As disorders of arousal (DOA) or REM sleep behaviour disorder (RBD) can be difficult to capture in the sleep laboratory and may need to be diagnosed in large communities, new home diagnostic devices are being developed, including actigraphy, EEG headbands, as well as 2D infrared and 3D time of flight home cameras (often with automatic analysis). Traditional video‐polysomnographic diagnostic criteria for RBD and DOA are becoming more accurate, and deep learning methods are beginning to accurately classify abnormal polysomnographic signals in these disorders. Big data from vast collections of clinical, cognitive, brain imaging, DNA and polysomnography data have provided new information on the factors that are associated with parasomnia and, in the case of RBD, may predict the individual risk of conversion to an overt neurodegenerative disease. Dream engineering, including targeted reactivation of memory during sleep, combined with image repetition therapy and lucid dreaming, is helping to alleviate nightmares in patients. On a political level, RBD has brought together specialists in abnormal movements and sleep neurologists, and research into nightmares and sleep–wake dissociations has brought together sleep and consciousness scientists.

1. Introduction

Parasomnias are abnormal behaviours ‐ such as sleepwalking, night terrors, REM sleep behaviour disorder (RBD), enuresis, and cataplexy ‐ or abnormal mental experiences ‐ such as nightmares, sleep paralysis, and hypnagogic hallucinations) that occur during sleep or at the borderline between sleep and wakefulness. Most parasomnias correspond to hybrid states between wakefulness and sleep, containing both characteristics of wakefulness (such as movements, behaviours, vocalisations and sometimes a consciousness close to wakefulness) and characteristics of sleep (such as muscular atony, blunted reactivity, dreams, altered consciousness and lack of reflection).

The most recently identified parasomnias include RBD, which is being studied in particular as an early biomarker for neurodegenerative diseases, mainly synucleinopathies; sexsomnia, which is responsible for the most common complaints in forensic sleep medicine; sleep‐related eating disorder (SRED) which can be confused with nocturnal eating syndrome, sleep‐related choking syndrome, which can be confused with sleep apnoea and nocturnal dyspnoea (Flamand et al. 2015); and stage N2 behaviour disorder, a marker of autoimmune encephalitis (Arnulf et al. 2005; Ribeiro et al. 2024; Sabater et al. 2014).

NREM and REM parasomnias associated with abnormal behaviour have variable expressions from night to night. They require video and audio recordings, in addition to polysomnography, often over two consecutive nights, meaning that diagnosis is both costly and may be missed in the absence of behavioural events. For the latter reason, simpler indirect markers have been developed to aid diagnosis, such as the amount of REM sleep without atonia in the case of RBD, or the N3 arousal index and N3 associated behaviours in the case of arousal disorders. It is clear that there is scope in the future for extended home video recordings, possibly combined with automatic movement analysis, to improve the diagnostic process. In the case of disagreeable mental experiences that are likely to disrupt patients' nights, the future could open up to the new field of dream engineering, including existing therapies for nightmares such as lucid dreaming or imagery rehearsal therapy, but enhanced by technological methods aimed at including lucid dreaming or reactivating targeted memories.

2. Disorders of Arousal

2.1. Context

Disorders of arousals (DOA) initially included sleepwalking (with usually quiet, amnesic walking), night terrors (with an expression of intense fear, tachycardia and often loud screaming) and confusional arousals (people remain in bed but suddenly raise their head and often their torso with eyes wide open, explore the environment, may talk and look confused or surprised). As previously reviewed (Idir et al. 2022), the category has now been extended to include some recently identified and more specialised forms of NREM parasomnia, including sexsomnia (a form of confusional arousal with amnesic sexual behaviour) and sleep‐related choking syndrome (a form of night terror associated with the terrifying sensation of a coin being stuck in the throat to the point of suffocation). Sleep‐related eating behaviour is a form of sleepwalking associated with compulsive and confused eating). It is considered a NREM parasomnia extending arousal disorders as it mostly emerges from N3 sleep but can also emerge from N2 sleep, and involves less altered consciousness than during classical sleepwalking. Arousal disorders are most common in children but can persist or even newly develop in 4% of adults. Young age and family history predispose individual to DOA, while situational stress and sleep debt the preceding day can promote episodes, which may occur spontaneously or be triggered by arousing events such as sudden noises. The behaviours occur mostly during local arousals in N3 sleep (although severe cases of sleepwalking and sleep‐related eating disorder episodes can occur during local N2 arousals) in a hybrid state involving both wake‐like activation in motor and limbic regions and preserved, or even increased sleep intensity across a frontoparietal network. Current international diagnostic criteria for DOA are clinical. Treatment is based on experience rather than evidence‐based medicine and includes control of priming and precipitating factors, as well as the use of sedatives.

2.2. Finding Accurate Polysomnography Markers

The most recent international diagnostic criteria for DOA are purely clinical, based on age, timing, and characteristics of reported behaviours, and evidence of confusion (American Academy of Sleep Medicine 2023). The absence of polysomnography criteria contrasts with the diagnostic criteria for defined RBD, which require both a history of dream enactment behaviours and REM sleep without atonia and/or behaviours on the accompanying video, or criteria for sleep‐related hypermotor epilepsy, which also require documentation of either behaviours or EEG markers). The authors justify this absence of EEG or behavioural markers by: (i) the high frequency of typical, uncomplicated, and non‐injurious parasomnias for which a clinical diagnosis is sufficient; (ii) the rare occurrence of arousals from N3 sleep accompanied by behaviours typical of confusional arousals and the exceptional out‐of‐bed behaviours in the sleep laboratory; and (iii) the lack of specificity of N3 arousals, which can occur in normal people. However, there are several medical situations in which we believe a more formal diagnosis is required, including harmful or potentially harmful DOA in adults, lack of a bed partner, situations where the differential diagnosis has a similar likelihood to arousal disorders (such as nocturnal behaviours in people over 40 years of age when RBD or sleep‐related hypermotor epilepsy are also plausible), and malingering and forensic medicine (i.e., crimes committed during the night; in our experience the most frequent case is sexual assault during the night). Plus, video‐polysomnography is key for the identification of potential arousal triggers such as noises, hypopnea, epileptic activity, or periodic leg movements.

The use of video‐polysomnographic evidence to support the DOA diagnosis has long been hampered by the absence of sensitive and specific criteria. However, several teams have recently made significant efforts to identify such criteria (Table 1), but these have not yet been considered sufficient to change the international DOA criteria in 2023. Provocative sleep studies have used overnight sleep deprivation before the night's recording (to deepens sleep, which primes episodes) and acoustic stimuli during the night (to induce confusional arousals), either alone or in combination. In 40 patients with DOA, 32 episodes were recorded from 20 (50%) sleepwalkers at baseline, followed by a full night sleep deprivation, with recovery sleep resulting in 92 episodes being recorded from 36 (90%) patients (Zadra et al. 2008). In 10 adults with DOA, the frequency of confusional arousals increased from 1/night in the undisturbed baseline night to 2/night with auditory stimuli (but only 30% of participants with DOA had at least one event), 3.5/night after sleep deprivation, and a further 4/night combing the two (100% of participants had at least one event) (Pilon et al. 2008). However, these measures were carried out for research purposes and seem too complex to be implemented in clinical routine. In fact, it is rare in practice to get a patient to come into the sleep laboratory after voluntarily undergoing a full night's sleep deprivation. In a recent study exploring responsivity during confusional arousals, despite auditory stimuli being calibrated for each individual in order to obtain a tone and intensity that will induce arousal, auditory stimulation had a limited effect in inducing confusional arousals (4%—7/157 trials, and not 30% as before) (Idir et al. 2024).

TABLE 1.

Sensitivity and specificity of indices of N3 fragmentation to support disorders of arousal diagnosis.

Studies	Lopez et al. (2018)	Barros et al. (2020)	Rossi et al. (2023)	Rossi et al. (2023)
Adult patients	Classical DOA N = 100	Classical DOA N = 52	Classical DOA N = 41	Sexsomnia N = 24
Controls	Healthy N = 50	Healthy/clinical N = 40	Clinical N = 40	Clinical N = 40
Arousals > 6.8/h of N3 sleep
Sensitivity	79%	19%	61%	46%
Specificity	82%	96%	95%	95%
Slow/mixed arousals > 2.5/h of N3 sleep
Sensitivity	94%	ND	83%	67%
Specificity	68%	ND	72.5%	72.5%
Slow/mixed arousals > 6/h of N3 sleep
Sensitivity	60%	ND	29%	25%
Specificity	100%	ND	100%	100%

Abbreviations: DOA, disorder of arousal; ND, not done.

Without using sleep deprivation or auditory stimuli, Lopez et al. studied the DOA markers in 100 adults with DOA compared to 50 controls (Lopez et al. 2018). They determined an index of N3 fragmentation (index of arousals/h in N3 sleep and index of slow‐wave sleep arousals/mixed arousals/h of N3 sleep), which showed good sensitivity and specificity for the diagnosis of DOA. The N3 fragmentation index cut‐off value of 6.8/h reached a sensitivity of 79% and a specificity of 82% in this series (Lopez et al. 2018), but this sensitivity decreased to 61% in a smaller series of 41 adults with DOA (Barros et al. 2020) and down to 46% (but reaching a specificity of 95%) in 24 participants with sexsomnia (Rossi et al. 2023). The slow/mixed arousal index had a sensitivity of 94% for the 2.5/h cut‐off and a specificity of 100% for the 6/h cut‐off. A highly specific cut‐off such as this may be useful for future forensic medicine, serving as a supportive criteria to be added to the full case file. The cut‐offs determined by receiving operator curves to optimise the sensitivity and specificity in the parent study, and how they functioned in other series, are shown in Table 1. Visual measurement of these events is quick and, in our experience, improves diagnostic accuracy. Like any new method, it will need to be routinely implemented outside of a few specialised sleep laboratories, but its utility in diagnosis and in the context of forensic medicine may accelerate its adoption in the future. Further, one could imagine using artificial intelligence for this purpose and training a deep learning algorithm to differentiate DOA polysomnography from control polysomnography. Finally, the association of sleepwalking with metrics indicative of consolidated sleep on polysomnography has recently been made possible thanks to de‐identified data from > 370,000 sleep clinic patients (Hanif et al. 2024), underpinning the interest of big data in better understanding the mechanisms of DOA (Table 3).

TABLE 3.

Big data and biomarkers from multicentre studies in the future of parasomnia.

	Field	Examples
Web‐based surveys	Sleep paralysis	Denis and Poerio (2017); Solomonova et al. (2008)
Phone‐based epidemiological surveys	NREM parasomnia, hypnagogic hallucinations, sleep paralysis	Ohayon et al. (1996); Ohayon, Guilleminault, and Priest (1999); Ohayon et al. (2012); Ohayon, Guilleminault, and Priest (1999)
Clinical data from large transversal series and cohorts	RBD (e.g., international RBD study group, national cohorts)	Elliott et al. (2023); Postuma et al. (2019)
	NREM parasomnia	Battiato et al. (2025); Hanif et al. (2024)
	Sleep paralysis, hypnagogic hallucinations	Hanif et al. (2024)
Cognitive tests	RBD	Joza et al. (2024) On line tests: To be determined
Large polysomnography databases	RBD	Vargas Gonzalez et al. (2025); Vargas Gonzalez et al. (2024)
	NREM parasomnia	Battiato et al. (2025)
	Sleep paralysis, hypnagogic hallucinations	Hanif et al. (2024)
Large brain imaging database	RBD (MRI, DAT scan)	Arnaldi et al. (2024); Rahayel et al. (2023)
DNA bank	RBD	Krohn et al. (2022)
	NREM parasomnia	To be determined

Abbreviation: RBD, REM sleep behaviour disorder.

2.3. Behavioural Markers for Improving the Diagnosis of Arousal Disorders

The common assertion that parasomnias rarely occur in sleep laboratories is probably inaccurate, as 59% of 100 adults with DOA had a confusional arousal during the first night of video‐polysomnography (Lopez et al. 2018), and 41.7% of 41 patients with sexsomnia displayed an apparently sexual behaviour during N3 arousal during the first night (Rossi et al. 2023). However, defining what constitutes apparently abnormal behaviour on the accompanying video during N3 arousals can be difficult, as immediate questioning (to check level of consciousness) is unusual in a routine sleep clinic, although it has recently been done for scientific purposes (Cataldi et al. 2024). Recently, efforts have been made to characterise all behaviours resulting from N3 arousals in 104 individuals with and without DOA, and to determine how sensitive and specific some common behaviours are, such as opening eyes or speaking (Barros et al. 2020). In the DOA group, the onset of behaviours emerging from N3 sleep was more abrupt, an observation that led to excellent inter‐rater agreement. Behaviours such as eye opening (69 vs. 16%), head raising (41 vs. 9%), visually exploring the environment (27 vs. 1%), expressing fear/surprise (21 vs. 0%), talking (18 vs. 0.3%), raising the trunk (13 vs. 0.3%), and interacting with the environment (13 vs. 0.5%) were more frequent than in controls, in contrast to quiet comfort behaviours. A cut‐off of 2 or more arousals in N3 sleep with eye opening was 94% sensitive and 77% specific for the diagnosis of DOA. This accuracy was confirmed on the second night of monitoring. Behaviours including expression of fear/surprise, sitting, screaming, and standing up were specific to patients with DOA and were never observed in clinical and healthy controls. In 41 patients with sexsomnia and 40 controls, an N3 arousal with trunk raising, sitting, speaking, showing an expression of fear/surprise, shouting, or exhibiting sexual behaviour was 100% specific for a diagnosis of sexsomnia (Rossi et al. 2023). This first approach based on behavioural observation upon N3 arousal needs to be validated in other and larger groups.

2.4. Using Home Test and Automated Techniques to Improve Diagnosis

Changes between home and sleep lab environments, timing, habits, and other factors can reduce the likelihood of sleepwalking in the lab. Inexpensive home infrared cameras, commonly used for home monitoring, are readily available and have been used by patients with DOA on their own initiative, capturing far more frequent and complex events than a single night in the sleep lab (Mwenge et al. 2013). Similarly, motion‐activated cameras are commonly used by hunters in the forest to detect wild animals (Table 2). They have recently been used at home in adults with DOA and were able to capture many more behavioural episodes suggestive of DOA than two nights in a sleep laboratory (Lopez et al. 2023). Improvements in the diagnostic processes can also be obtained with the implementation of automatic algorithms to detect patients with NREM parasomnias. Moro et al. developed an algorithm based on a convolutional neural network able to distinguish patients with DOA from patients with sleep‐related hypermotor epilepsy (Moro et al. 2023). In this field, automatic analysis of overnight 3D videos could also be a promising approach (Cesari, Ruzicka, et al., 2023). Compared to other research areas in sleep medicine, no study has so far investigated actigraphy or more recent wrist watches as supporting tools for the diagnosis of NREM parasomnias. Future research is therefore needed in this area. Similarly, no study has investigated EEG headbands (de Gans et al. 2024), a field that is worth exploring further.

TABLE 2.

Digital devices in the future of parasomnia.

Type	Field and technics	References
Diagnostic support
Home surveillance camera (infrared)	Sleepwalking, sleep terrors: Hunter camera	Lopez et al. (2023)
	REM sleep behaviour disorder: 3D camera with automatic gestures analysis	Cesari, Kohn, et al. (2021)
Connected mattress	Not yet tested for parasomnia
Wrist actigraphy	REM sleep behaviour disorder	Brink‐Kjaer, Gupta, et al. (2023); Brink‐Kjaer, Winer, et al. (2023); Raschella et al. (2023).
Treatment support
Anti‐theft doormat	Sleepwalking
Bed alarm	Sleepwalking, enuresis (with machine learning to prevent episodes)	Lee et al. (2024); Wang et al. (2024)
Targeted memory reactivation	Nightmare disorder; enhancing image rehearsal therapy	Schwartz et al. (2022)
Targeted lucid dreaming	Lucid dreaming therapy of nightmares	Carr et al. (2023)

2.5. Future DNA Markers of Arousal Disorders, and Animal Models?

As the factors that prime and precipitate parasomnia episodes are better understood, the reasons why some people develop these behaviours while others do not are unclear. Despite the familial nature of DOA, the clinical and polysomnographic characteristics of familial vs. sporadic DOA have been just recently studied in a large sample (Table 3), showing that familial DOA starts earlier in life and is more severe than sporadic cases (Battiato et al. 2025). In addition, DNA studies are scarce even though DOA is one of the most commonly inherited sleep disorders. Progress may be made in the future with extensive DNA collection in well‐studied patients. Unlike RBD, where an animal model was developed 20 years before human RBD was identified, an animal model of sleepwalking or sleep terrors is still lacking and needs to be developed to shed light on DOA mechanisms.

2.6. Future in Therapies for Disorders of Arousal

Treatment of DOA is usually graded according to severity. Behavioural and psychological treatment may be suggested for all patients, especially children (Mundt et al. 2023). These include control of priming factors such as psychological stress (the most common priming factor according to patients, which could be reduced by relaxation, mindfulness, education/reassurance, hypnosis and psychotherapy), sleep debt (sleep hygiene, sleep extension, scheduled naps), and increased body temperature (avoidance of evening exercise, reduction of high temperature and fever in the evening). Precipitating factors that trigger arousal, such as noise (reduced through hearing protection), physical contact, alcohol consumption, oesophageal reflux, upper airway obstruction, or periodic leg movements, can be eliminated or treated. Drugs or conditions that deepen sleep and increase confusion upon awakening from N3 sleep, such as antipsychotics, may worsen parasomnia episodes. The role of alcohol use, which has been described as protective against DOA in one association study (e.g., Hanif et al. 2024) or as facilitating episodes in other studies, should be formally evaluated. The epidemiologic association between DOA and psychopathology (anxiety, depression, obsessive‐compulsive disorder, attention‐deficit/hyperactivity disorder, and autism spectrum disorder) has been reviewed (Tomic et al. 2025). Why and how psychopathology, especially depression, contributes to DOA (Hanif et al. 2024) and whether reduction of depressive symptoms improves DOA should also be investigated in the future. Early large‐scaled epidemiological studies have already noticed this association (Ohayon et al. 1999; Ohayon et al. 2012).

Medications that reduce arousal from N3 sleep are commonly prescribed in severe cases and include clonazepam (and other long‐acting benzodiazepines), gabapentin, carbamazepine, and sometimes paroxetine (e.g., for sexsomnia). Of interest, a recent large association study in clinical samples of 370,000 people with sleep disorders found a decreased risk of reporting somnambulism in people taking antihistamines, but no beneficial effect of melatonin, benzodiazepines, and trazodone (Hanif et al. 2024). It illustrates the future potential of using large databases of deidentified clinical and polysomnographic measures to validate some of these associations (Table 3). Notably, to our knowledge, none of these therapeutic approaches have been tested against placebo or control, as required by evidence‐based medicine. In the future, more interest is expected from clinical groups and the pharmaceutical industry to conduct these trials in these neglected disorders. Ultimately, the future of DOA will begin with more interest from clinicians, scientists, and pharmaceutical companies in these disorders that are common and potentially harmful to patients and their loved ones.

3. N2 Behaviour Disorder: A New Parasomnia With Important Consequences for Neurological Diagnosis

3.1. Brief Outline

The presence of prolonged, finalistic, quiet behaviours during N2 sleep was noticed early on in a patient with an unknown fatal, rapidly progressive neurological disorder, named N2 behaviour disorder (Arnulf et al. 2005). Although extremely rare, similar behaviour has been described 10 years later during a poorly defined N2 stage in patients with a novel autoimmune encephalitis associated with anti‐IgLON‐5 antibodies (Sabater et al. 2014). More recently, such observations were extended to another autoimmune encephalitis associated with anti‐NMDA receptor antibodies (Ribeiro et al. 2024).

3.2. Importance for the Future

Autoimmune encephalitis is a growing field in neurology and oncology, with severe clinical symptoms and consequences, as well as the potential to cure serious diseases if detected and treated early. Sleep disorders, including severe insomnia, hypersomnia, narcolepsy, parasomnia, and sleep‐related hypoventilation are common in autoimmune encephalitis (Munoz‐Lopetegi et al. 2020). Some of these disorders represent life‐threatening conditions (e.g., sleep‐related hypoventilation, nocturnal stridor) that must be treated rapidly, or lead to disabling symptoms (e.g., insomnia or narcolepsy) that can be alleviated. Symptoms of nocturnal agitation should be distinguished from seizures. However some of these symptoms may serve as markers that can prompt the diagnosis and treatment of autoimmune encephalitis. Their importance in the context of seronegative autoimmune encephalitis may be critical in the future. Indeed, the presence of N2 behavioural disturbances appears to be so specific in signalling autoimmune encephalitis that a simple videotape of these behaviours in a given clinical context would prompt the diagnosis ‐ and the rapid immune treatment ‐ of anti‐Iglon‐5 encephalitis, for example, in a patient with recurrent episodes of hypercapnic coma of unknown cause. Autoimmune encephalitis is rare, serious, but curable disorders that benefit from any input. Video recording in the intensive care unit, possibly combined with polysomnography, may help to identify these rare parasomnias in the future.

4. Enuresis

4.1. Context

Sleep enuresis is the recurrent, involuntary urination during sleep, at least once a month (frequent if > 4/week) in a person over the age of 5, more often a male than a female. Enuresis may occur during any sleep stage. The mechanisms are complex and include a mismatch between excessive urine production during sleep (either familial, including decreased nocturnal ADH production, or secondary, for example, to sleep‐disordered breathing), difficulty waking from sleep in response to an urge to urinate (high arousal threshold), and storage issues (reduced bladder capacity, underactive or overactive bladder). Simultaneous cystography and polysomnography have shown that bladder contractions can occur before micturition and before waking. Nocturnal enuresis affects quality of life and self‐esteem in children. In the elderly, the need to wear pampers or a urine‐guiding system at night has been developed for adults with nocturnal enuresis and for adults with normal nocturnal continence, but with severe motor disability and risk of falling when going to the toilet.

4.2. Future in Sleep Enuresis

Some connected pyjamas and wearables currently in development contain sensors (ultrasound or gyroscopic) that monitor bladder fullness, as well as heart rate and other measurements, which can trigger an alarm when a certain volume is reached, leading to machine learning‐trained pre‐void (rather than post‐void) alarms (Lee et al. 2024; Wang et al. 2024). As nocturnal enuresis runs in families, some genetic linkage studies have identified associations in some genes, but these efforts should be continued in the future, including larger groups and well‐characterised participants.

The evidence‐based first‐line treatments for monosymptomatic nocturnal enuresis include an enuresis conditioning alarm and antidiuretic treatment with desmopressin. A recent prospective large‐scale randomised controlled trial provided evidence that tailored characterisation including voiding diaries increases the likelihood of treatment response to desmopressin when nocturnal polyuria was present, and to alarms if the maximum voided volume was reduced (Jorgensen et al. 2024). Models as shown in this study could be developed in the future to predict response to treatment and guide clinician's choices.

5. The Future of Rem Sleep Behaviour Disorder

5.1. Brief Outline

REM sleep behaviour disorder is characterised by dream enactment behaviours and by the absence of physiological muscle atonia during REM sleep (American Academy of Sleep Medicine 2023). It is classified as isolated RBD (iRBD), when no other neurological disorders are present, or secondary RBD when associated with other neurological or neurodegenerative disorders, autoimmune diseases, or brainstem lesions (Hogl et al. 2018). Secondary RBD is most commonly associated with synucleinopathies including Parkinson's disease (PD), dementia with Lewy bodies (DLB) or multiple system atrophy (MSA). In these disorders, RBD is equally observed in men and women, while iRBD predominates in men (American Academy of Sleep Medicine 2023). It has been shown to represent a prodromal alpha‐synucleinopathy in most cases, as > 90% of patients with iRBD develop PD or DLB over time, and a minority develop MSA (Galbiati et al. 2019). The prevalence of RBD confirmed by polysomnography is estimated to be around 1% (Haba‐Rubio et al. 2018; Kang et al. 2013). However, there is a need for large epidemiological studies to better estimate the prevalence of RBD and to investigate possible geographical and ethnic variations.

Patients with RBD typically describe violent and realistic movements, associated with action‐packed dreams in which they are attacked by people or animals and have to defend themselves or their loved ones. These behaviours pose a risk of injury to the patients themselves and their bed partners, and injuries are frequently reported (Fernández‐Arcos et al. 2016). RBD may be underdiagnosed in women, as sex differences have been reported with less aggressive dream‐enacting behaviours and dream content more related to daily activities in women. This aspect needs to be addressed in future studies in order to improve diagnosis and treatment of RBD in women and reduce current sex differences.

Diagnostic criteria for RBD are provided in the International Classification of Sleep Disorders (American Academy of Sleep Medicine 2023). The diagnosis is based on the presence of vocalisations, jerks, and motor behaviours during REM sleep (as demonstrated by video‐polysomnography or presumed to occur during REM sleep based on clinical history), which are often associated with REM‐related dream content (dream enactment) and are accompanied by the absence of physiological muscle atonia during REM sleep (REM sleep without atonia, RWA) (American Academy of Sleep Medicine 2023). The neurophysiology taskforce of the International RBD study group recently published video‐polysomnography procedures guidelines for the diagnosis of RBD, which require the capture of at least one RBD episode with video/audio recording during REM sleep (Cesari et al. 2022). Current challenges and recent developments in the diagnosis of RBD are discussed below.

5.2. REM Sleep Without Atonia

As previously stated, the demonstration of RWA is mandatory for diagnosis of RBD (American Academy of Sleep Medicine 2023) and several different methods for manual quantification of RWA have been proposed. While the International RBD study group recommends quantification of RWA in 3‐s mini‐epochs according to the SINBAR criteria and defines clearly cut‐offs to be employed (Cesari et al. 2022), the AASM recommends quantification with similar criteria to the SINBAR ones in 30‐s epochs, but does not clearly specify the cut‐offs to be used to identify abnormally elevated RWA (American Academy of Sleep Medicine 2023). In line with the International RBD study group guidelines, the last AASM manual version 3.0 recommends the recording of electromyography at the flexor digitorum superficialis muscles and the quantification of RWA at the upper limbs (Cesari et al. 2022). To ensure inter‐center reproducibility and reliability, it is crucial that common rules for RWA quantification and cut‐offs are clearly defined and universally accepted.

In parallel to the definition of rules for manual quantification of RWA, the last years have seen a significant increase of automatic methods for RWA scoring (Cesari and Rechichi 2024). One of the first proposed algorithms is the so‐called REM atonia index (Ferri et al. 2008). Several comparison studies and recent reviews have shown its high sensitivity and specificity in identifying patients with RBD (Byun et al. 2022; Cesari et al. 2019; Puligheddu et al. 2023). While the REM atonia index has been originally proposed as a measure of atonia for the chin muscle, recent evidence suggests that this index is very accurate for identification of patients with RBD also when applied to the flexor digitorum superficialis muscles (Cesari, Heidbreder, et al. 2023; Leclair‐Visonneau et al. 2024), though clear cut‐offs are still missing. Automatic scoring according to SINBAR criteria is also currently available as open‐access and open‐source though the ‘RBDtector’ software (Rothenbacher et al. 2022), which has also been recently externally validated (Joza et al. 2025). The main limitation of ‘RBDtector’ is that it requires annotation files in a specific format, which is generated by one specific polysomnography software, thus limiting its applicability.

As manual quantification of RWA is a time‐consuming task, requires highly trained personnel, and is prone to inter‐rater variability (Bliwise et al. 2018), it is reasonable to think that future guidelines will recommend automatic RWA quantification methods. In this regard, two important steps are needed. First, to move the field forward and to integrate automatic quantification of RWA into clinical routine, automatic RWA quantification methods that are respecting the FAIR4S principles (i.e., Findable, Accessible, Interoperable and Reusable principle for software) are needed (Barker et al. 2022). Second, future research should better clarify the influence of artefacts on automatic RWA quantification. Previous studies indicated that the chin is more affected by artefacts than the flexor digitorum superficialis muscles (Cesari, Heidbreder, et al. 2021) and that artefact correction is mandatory to have reliable automatic RWA quantification according to the SINBAR method (Frauscher et al. 2014). Other automatic RWA quantification methods seem to be more robust to artefacts, but further research in larger cohorts is needed (Cesari et al. 2018, 2019).

5.3. Video Analysis of RBD: Thinking About the Future

In addition to the demonstration of RWA, a clinical history of dream‐enactment behaviour is enough to diagnose RBD according to the international criteria (American Academy of Sleep Medicine 2023). On the contrary, the guidelines from the International RBD Study Group require that, for a definite RBD diagnosis, an RBD episode is documented during video‐polysomnography (Cesari et al. 2022). These discrepancies led to a partial endorsement of these guidelines by the World Sleep Society (Schenck et al. 2023). However, this aspect warrant further research. In fact, evidence indicates that patients with RBD mainly show simple minor events and vocalisation of mild intensity, while complex movements are a minority (Frauscher et al. 2007; Marino et al. 2025; Nepozitek et al. 2021; Oudiette et al. 2012). The question of whether an RBD episode, even in the form of a simple minor event, can be detected every night needs to be clarified in careful audiovisual analyses in large cohorts of patients with RBD.

In this regard, recent advancements in automatic analyses of videos could be potentially helpful. The Innsbruck group has investigated in several studies the automatic analysis of depth videos as an innovative method to identify patients with RBD (Cesari, Kohn, et al. 2021; Cesari, Ruzicka, et al. 2023; Waser et al. 2020). Inspired by these studies, automatic analyses of infrared videos recorded during video‐polysomnography have been performed (Abdelfattah et al. 2025), showing very high sensitivity and specificity in identifying patients with RBD. Interestingly, all these studies substantiate that short movements (i.e., up to 2 s) are very frequent in REM sleep in patients with RBD and they can best discriminate patients with RBD from healthy and clinical controls. In the future, it can be foreseen that automatic video analysis could be combined with automatic RWA quantification to improve sensitivity and specificity in automatically identifying patients with RBD in clinic settings.

5.4. Neurodegenerative Sleep or “Polysomnogramome”

In addition to RWA, which is both a diagnostic criterion for RBD and a biomarker of neurodegeneration on EMG, polysomnography contains a wealth of additional information about sleep stability and transitions (Christensen et al. 2016). Measures provided by EEG, EOG, ECG, and respiratory channels over 8 h and across different sleep stages can indicate neurodegeneration in brain structures that control sleep and arousal. These include changes in EEG rhythms, figures (O'Reilly et al. 2015; Sunwoo et al. 2021), spectrum, coherence, connectivity, and hypnodensity (the probability that an epoch corresponds to each sleep stage), reduced heart rate variability in the ECG as an indication of prodromal autonomic dysfunction (Postuma et al. 2010), abnormal rapid eye movements during NREM sleep (Vargas Gonzalez et al. 2025) or excessive sighing during wakefulness and sleep indicating multiple system atrophy (Vargas Gonzalez et al. 2024). Such peculiarities could be detected by automated techniques to identify patients with neurodegenerative disorders among the large number of polysomnograms obtained for other purposes in the elderly. Deep learning methods, which automatically integrate all these changes, have started to achieve the same goal (Feuerstein et al. 2024).

5.5. Isolated RBD as Early Stage Alpha‐Synucleinopathy

As mentioned above, iRBD is recognised as prodromal alpha‐synucleinopathy based on several long‐term longitudinal data showing that the vast majority of these patients develop an overt alpha‐synucleinopathy over time (Galbiati et al. 2019). This is also supported by studies investigating patients with longstanding iRBD, demonstrating that also those patients present markers of synuclein‐related neurodegeneration (Iranzo et al. 2017; Yao et al. 2018). Additionally, a subtantial number of studies have shown α‐synuclein pathology in several tissues and biofluids, including the cerebrospinal fluid, olfactory mucosa, skin, salivary glands, and blood (Miglis et al. 2021; Okuzumi et al. 2023). Among them, detection of pathological alpha‐synuclein in the cerebrospinal fluid and in the skin using seeding amplification assays seems currently the most promising based on diagnostic accuracy (Stefani et al. 2025). Of note, p‐tau/Aβ42, neurofilaments light chain and α‐synuclein ‐measured by immunocapture in neuronally derived extracellular vesicles in the blood‐ might be useful in predicting disease progression and phenoconversion, pending future data (Stefani et al. 2025).

The good characterisation of iRBD as early stage alpha‐synucleinopathy makes these patients an ideal population for synucleinopathy neuroprotective trials. However, this also poses some challenges. Recent developments in this field, in particular the detection of pathological synuclein aggregates by seeding amplification assays, have led to the proposal of a biological definition/classification of neuronal alpha‐synuclein diseases. According to the SynNeurGe (Synuclein, Neurodegeneration, Genetic) research diagnostic criteria (Hoglinger et al. 2024), most iRBD patients with pathological alpha‐synuclein on seeding amplification assays would be classified as sporadic PD or sporadic PD‐type synucleinopathy, based on the presence or absence of neurodegeneration on imaging. According to the Neuronal α‐Synuclein Disease Integrated Staging sSystem (NSD‐ISS) (Simuni et al. 2024), subtle clinical manifestations without functional impairment (including iRBD) mark stage 2, with sub‐classification based on dopaminergic neuronal dysfunction. Relevant implications for patients with iRBD include: (i) the potential exclusion of patients without pathological alpha‐synuclein from neuroprotective trials (not necessarily evaluating drugs modulating alpha‐synuclein); (ii) the neglect of iRBD patients who will convert to MSA; (iii) the neglect of molecular (e.g., lysosomal, mitochondrial) and neuropathological (comorbid proteinopathies) data with the risk of oversimplifying a complex disease (Maya et al. 2024).

5.6. Prodromal REM Sleep Behaviour Disorder

REM sleep behaviour disorder does not begin abruptly but is instead preceded by “longstanding prodromes consisting of accentuated twitching, simple behaviours, and vocalizations”, as already reported by Schenck and colleagues in the first description of RBD in 1986 (Schenck et al. 1986). Prodromal RBD has been described as a stage of the disease in which symptoms and signs of evolving RBD are present but do not yet meet established diagnostic criteria for RBD (Hogl et al. 2018), based on the presence of excessive muscle activity and/or dream enactment behaviours/excessive movements during REM sleep that do not meet the criteria for RBD diagnosis. The implications go beyond the early identification of a REM parasomnia, as current literature supports the concept of isolated RWA, i.e., RWA in the absence of dream enactment behaviours, as a prodrome of RBD, and therefore as prodromal α‐synuclein related neurodegeneration (Cesari et al. 2022). Available data on behavioural events also support this concept (Huang et al. 2023; Sixel‐Doring et al. 2023, 2014), suggesting a potentially critical role for prodromal RBD in defining at‐risk cohorts for α‐synuclein‐related neurodegeneration and in selecting subjects for future clinical trials of neuroprotective treatments at very early stages of the neurodegenerative process, likely decades before the development of motor or cognitive symptoms.

The neurophysiology taskforce of the International RBD study group published the first video‐PSG procedures guidelines for the identification of prodromal RBD, for research purposes only (Cesari et al. 2022), to facilitate the much‐needed longitudinal studies with large sample sizes and long follow‐up periods that will allow a better understanding and characterisation of prodromal RBD and its evolution over time.

5.7. Current and Future Developments in REM Sleep Behaviour Disorder

Accurate, objective, and fast diagnosis of RBD is still a topic of major interest in research. Data‐driven methods that go beyond the diagnostic criteria are of central focus. As an example, recent research investigated deep learning algorithms that can accurately identify patients with RBD from whole polysomnography recordings (Brink‐Kjaer et al. 2022; Feuerstein et al. 2024; Gunter et al. 2023). Related to this (Table 2), research is also focusing on enabling accurate sleep recording and RBD diagnosis employing only EEG bands (Levendowski et al. 2025; Possti et al. 2024) or recording in home environments (Brink‐Kjaer, Gupta, et al. 2023; Brink‐Kjaer, Winer, et al. 2023; Raschella et al. 2023). Future research should further explore this field in order to facilitate the work of clinicians and reduce the burden of sleep laboratories worldwide.

Another focus of research involves the development of accurate methods to predict the risk of conversion to an overt alpha‐synucleinopathy in patients with iRBD (de Natale et al. 2022). Such research field is of high importance for the design of clinical trials of future neuro‐protective and/or neuro‐modulatory treatments (Postuma 2022). As not all possible biomarkers can be covered in this work, we refer the reader to the most recent reviews on the topic of iRBD (de Natale et al. 2022; Miglis et al. 2021). While significant advances in the field have been made, so far no single biomarker is considered adequate for predicting the risk of conversion. Future studies should evaluate the combination of different biomarkers and the use of machine learning for this purpose (Arnaldi et al. 2024; Cesari et al. 2024; Jeong et al. 2024).

Another important and emerging topic of research, which will need further attention in the future, concerns the ethical aspects of diagnosing iRBD and the issue of prognostic counselling (Stefani et al. 2023).

5.8. Political Aspects in the Future of RBD

Over the years, evidence has accumulated showing that RBD is a prodromal synucleinopathy, gradually convincing PD specialists of the importance of studying these patients. The International RBD Study Group, composed mainly of sleep neurologists (often with an additional specialty in movement disorders), has played a major role in collecting clinical, cognitive, radiological and DNA information in large groups of iRBD patients who are rarely seen outside specialised centres (although iRBD cannot be considered an orphan disease). This effort, not supported by the European Union or other national institutions, has produced more than 20 important original articles and should continue to determine trajectories in a given patient, potentially using artificial intelligence based on a combination of clinical, cognitive, polysomnographic, brain imaging, biological and DNA measures (Table 3). This information will be critical in designing future neuroprotection trials and convincing pharmaceutical companies to conduct these trials in iRBD patients.

In parallel, cohorts of iRBD patients were progressively built in centres (e.g., Montreal, Barcelona, Oxford, Paris‐Iceberg study, Seoul) and later in countries such as the USA (the North American prodromal Synucleinopathy cohort (Elliott et al. 2023)), France (France‐RBD) and Italy (Farpresto) (Puligheddu et al. 2022). Sleep specialists, who are the first to see patients with iRBD and make the diagnosis, neurologists specialising in abnormal movements, who manage them once they convert to PD, and geriatricians or neuro‐geriatricians specialising in dementia, who manage patients once they convert to DLB, have moved closer together (Table 4). This includes joint cohorts, joint symposia at sleep or PD conferences, and joint consensuses.

TABLE 4.

Transversal policy in the future of parasomnias: What links with scientific societies?.

Parasomnia	Link with societies	Specialised societies/field	Example of collaboration
RBD	Neurology	Movement Disorders Societies Dementia societies—Neurodegeneration	Prodromal cohorts for PD and DLB Sleep in PD Trajectories in PD with versus without RBD
Stage N2 behaviour disorder	Neurology	Neuro‐oncology—Autoimmune encephalitis	N2 behaviour disorder as a marker of autoimmune encephalitis
Nightmare disorder	Psychiatry Psychology	Mood disorders	Nightmares as markers of suicidality—state markers of nightmares
Lucid dreaming	Neuroscience	Consciousness	Insight into the perception of inner and outer worlds during sleep

6. The Future of Sleep Paralysis and Hypnagogic Hallucinations

6.1. Recurrent Isolated Sleep Paralysis

Recurrent isolated sleep paralysis (RISP) is a REM sleep‐related parasomnia (Table 5). It is characterised by a recurrent inability to move the trunk and limbs at sleep onset or awakening from sleep, lasting seconds to minutes, causing distress (bedtime anxiety or fear of sleep) in the absence of narcolepsy (American Academy of Sleep Medicine 2023). There is a co‐existence of subjective wakefulness (preserved awareness and full recall) and the REM‐related active inhibition of striated muscles (including the accessory respiratory muscles, which may contribute to the feeling of a weight on the chest or oppression of the thoracic cage). The first episode of sleep paralysis is generally perceived as a frightening experience because of the subjective wakefulness and complete inability to move or to vocalise. On spectral EEG analysis, sleep paralysis is different from false awakenings (sleep‐related experiences in which the subjects erroneously believe that they have woken up, only to discover subsequently that the apparent awakening was part of a dream), normal REM sleep, and lucid REM sleep (Mainieri et al. 2021). Theta EEG rhythm predominates during sleep paralysis and false awakenings, suggesting that the brain during sleep paralysis is in a sleeping and dreaming state rather than awake. Factors usually promoting RISP include sleep debt, multiphasic sleeping schedule, and sleeping supine. In a meta‐analysis, associations were found with substance use, stress and trauma, genetic influence, physical illness, personality, intelligence, demonic beliefs, sleep problems, and disorders (Denis and Poerio 2017), although some of these points were not confirmed in an online survey of 380 subjects with RISP (Mayer and Fuhrmann 2022).

TABLE 5.

Phenomenological aspects of various normal and hybrid REM sleep/wake states.

	Quiet wake	Cataplexy	Sleep paralysis	False awakening	Lucid dreaming	REM sleep
Clinical features
Feel paralysed	No	Yes	Yes	No	No	No
Feel the environment as familiar	Yes	Yes	Yes	Yes	No	No
Accurate awareness of the state	Yes	Yes	Yes	No	Yes	No
Associated anxiety	None	Yes	Frequent	Frequent	None	None
Mental experience	Awake thoughts	Awake thoughts (hallucinations if prolonged)	Feels awake but hallucination	Feels awake but dreaming	Dreaming	Dreaming
Postural muscle tone	High	Low	Low	Low	Low	Low
Eyes
Closed eyes	Yes	Yes if total	Unclear	Yes	Yes	Yes
Spontaneous REMs	Present	Present	Present	Present	Present	Present
Voluntary REMs (codes)	Present	?	Present	?	Present	Absent

Abbreviation: REMs, rapid eye movements.

The Prague group examined the EEG and MRI trait factors in RISP sufferers. Their sleep and REM sleep macrostructure did not differ compared to healthy subjects without RISP, but there was a higher bifrontal beta activity during REM sleep (Klikova et al. 2021), and abnormal microstate temporal dynamics and stage decision beyond REM‐related changes (Cerny et al. 2024). In morphological 2‐D MRI, RISP participants had increased cerebellum height and diameter of the midbrain‐pontine junction, compared to individuals without RISP, potentially reflecting compensation attempts to dysfunctional sleep‐wake regulatory pathways (Miletinova et al. 2024).

In terms of treatment, an increased frequency of sleep paralysis leads to habituation and de‐dramatisation in some people (Mayer and Fuhrmann 2022). Avoiding lying in a supine position and preventing sleep debt, sleeping with a bed sharer or being pinched by someone in the household, are classic suggestions to stop sleep paralysis. Relaxation, rather than trying to move when paralysed, which increases stress, has been mentioned as effective in case reports (Jalal 2016). A small proof‐of‐concept study of meditation‐relaxation showed benefits for recurrent sleep paralysis in patients with narcolepsy (Jalal et al. 2020). In this population, drugs that improve cataplexy reduce sleep paralysis in controlled trials, including oxybate at doses greater than 6 g (as lower doses may actually increase sleep paralysis) and pitolisant (Dauvilliers et al. 2019; Xyrem International Study 2005), while evening doses of paroxetine and venlafaxine are also beneficial, according to expert opinion.

6.2. Sleep Related Hallucinations

Hypnagogic and hypnopompic hallucinations are also seen as dissociated states. According to ICSD's criteria, there must be a complaint of recurrent hallucinations that are experienced just near sleep onset or upon awakening during the night or in the morning. These hallucinations are experienced as predominantly visual and are not better explained by another current sleep disorder, in particular narcolepsy, where those hallucinations can also occur.

When they occur in isolation, they do not need to be coded. However, when the frequency is high or causes distress requiring clinical attention, the diagnosis should be applied. The hallucinations can be experienced as very real and frightening. Among the differential diagnoses are nightmares, optical illusions, complex hallucinations, RBD, sleep terrors, peduncular hallucinosis, oneiric stupor, Charles Bonnet Syndrome, delirium, sleep‐related epilepsy, migraine, and narcolepsy. Exploding head syndrome is classified differently from sleep‐related hallucinations, although one may consider that this feeling of exploding head, sudden sound, or light at sleep onset or offset, without any headache, is a sensory hallucination.

Recent studies noted an association between self‐reported parasomnia and psychiatric illness in a very large sample of 370,000 patients with sleep disorders. They may also be considered as indicators of depression and anxiety (Hanif et al. 2024). Whether the mental experience during sleep terrors is dream scenes during which patients felt awake (“awake dreaming”) or hypnopompic hallucinations projected in the bedroom is still a matter of debate (Siclari 2025).

The future of RISP and sleep related hallucinations depends on the need to better understand the differences and similarities between sleep paralysis, sleep‐related hallucinations, false awakenings, out‐of‐body experiences, dreaming and lucid dreaming (Table 5), as lucid dreamers can oscillate between these different states (Denis and Poerio 2017). This debate focuses on the distinctions between consciousness and wakefulness, as well as all the hybrid states between wakefulness and sleep.

7. The Future of Nightmare Disorders

7.1. The Context of Nightmare Disorder

Nightmares are defined as prolonged, extremely dysphoric, and well‐remembered dreams that usually involve efforts to avoid threats to survival, safety, or physical integrity (American Academy of Sleep Medicine 2023). Nightmares can occasionally occur in healthy people. A nightmare disorder is defined when the nightmares are recurrent and associated with significant impairment of daytime functioning, such as mood disturbance, insomnia, cognitive impairment, fatigue, drowsiness, and behavioural or social problems. Classifications distinguish between nightmares without an identified cause, called “idiopathic”, and nightmares associated with a medical condition, including mental illness or the use of treatments. The occurrence of weekly nightmares is found in 4% of the general population (all causes combined) and is overrepresented in psychiatric disorders such as anxiety (16%) and depression (37%), but especially in post‐traumatic stress disorder, where the prevalence is as high as 67% (van Schagen et al. 2017). Post‐traumatic nightmares can occur in all stages of sleep (Phelps et al. 2018; Richards et al. 2023; Saguin et al. 2023). In contrast, a single study of patients suffering from idiopathic nightmares showed that they were reported exclusively after periods of REM sleep (Paul et al. 2019). The mechanism by which nightmares are generated is still unknown. One theoretical model proposes that the generation of nightmares is related to a dysfunction of the fear extinction processes and emotional regulation during REM sleep, wherein an overactivation of the amygdala and related limbic structures is insufficiently modulated by prefrontal control systems, leading to an increase in negative emotions and eventually an awakening (Nielsen and Levin 2007).

7.2. First Challenge: “State” Markers of Nightmares

One way of understanding how nightmares are generated is to look at the structure of the sleep structure directly preceding a patient's report of a nightmare, known as the ‘state’ approach. The sleep period preceding reports of traumatic nightmares showed no spectacular changes, for example, no difference in heart rate in one study (Phelps et al. 2018), and a limited change of 3 heartbeats per minute in the 10 min before waking in another (Richards et al. 2023), in contrast to the intense anxiety often associated with nightmare reports. However, a limitation of these studies is the lack of contrast between sleep preceding a nightmare and sleep preceding an awakening without a nightmare in the same patients. Paul et al. conducted ambulatory polysomnography in nightmare sufferers to contrast within the same individual the 5 min of sleep preceding a report of an idiopathic nightmare with the sleep preceding a simple dream report. They were able to demonstrate a mild acceleration in heart rate (6 beats per minute) and respiration, and a doubling in the frequency of rapid eye movements preceding reports of idiopathic nightmares (Paul et al. 2019). Such an approach seems promising for a better understanding of the mechanisms associated with nightmares and could be used to replicate the results of Paul et al. in a larger number of idiopathic nightmares and in post‐traumatic nightmares.

7.3. Second Challenge: “Trait” Markers of Nightmares

Another approach to understanding nightmares is the “trait” approach, which focuses on the entire sleep (or sleep stages) of patients with nightmares, regardless of the occurrence of the nightmare itself. This approach is particularly suited to the study of nightmares in sleep laboratories where the occurrence of nightmares is rare, leaving little room for a “state” approach. The structure of sleep in patients with idiopathic nightmares has so far shown little difference from the sleep of healthy people, with studies yielding mixed results. Only a few studies found differences in sleep architecture (e.g., reduced slow wave sleep, increased wakefulness within sleep, sleep fragmentation). Furthermore, sleep fragmentation was only observed in NREM sleep, whereas, as we have seen, idiopathic nightmares seem to occur only in REM sleep (Nielsen et al. 2010; Perogamvros et al. 2019). A limited difference of 5 beats per minute in both NREM and REM sleep between patients with nightmares and healthy controls was also found in a third study (Tomacsek et al. 2024). Finally, the analysis of EEG markers by frequency band has not yet shown reproducible results between different studies. Thus, while there is some evidence for altered sleep and autonomic activity in nightmare sufferers, there is to date no robust marker that specifically identifies nightmares as a “trait” that is reproducible between studies. Finding such markers may currently require the study of sleep characteristics that have never been studied at the trait level, such as sleep breathing rate, or more complex markers based on a combination of different signals. Finding robust “trait” markers of nightmares could (i) help understand the mechanisms underlying their occurence during sleep and (ii) serve as biomarkers ‐ if they show good discriminatory value (sensitivity and specificity) ‐ helping clinicians in detecting nightmares from sleep recordings or confirming clinical diagnosis.

7.4. Third Challenge: Nightmares Related to Psychiatric Diseases

Apart from post‐traumatic nightmares, nightmares related to psychiatric disease are rarely studied with polysomnographic recordings, either as “state” or “trait”. Nightmares associated with depression would be especially interesting to investigate with a “trait” approach, comparing patients with both depression and nightmares to those with depression but no nightmares, as well as healthy controls. A “state” approach of depression related nightmares using at‐home recordings could help determine whether they are occurring exclusively in REM sleep, like idiopathic nightmares, or aboth in REM and NREM sleep, as seen in post‐traumatic nightmares and help identify markers associated with such nightmares. The presence of nightmares has recently been highlighted as a strong proximal predictor of suicidality in large prospective epidemiologic studies in adolescents (Guo et al. 2024; Liu et al. 2021) and in retrospective clinical series of depressed patients (Geoffroy et al. 2022). These findings should focus the attention of psychiatrists and psychologists on gathering information about nightmares in people with depression.

7.5. Fourth Challenge: Epidemiologic Studies of the Interplay Between Nightmares and Psychiatric Disease

While psychiatric illnesses are commonly associated with nightmares, little is known about how the occurrence of nightmares might impact the development of psychiatric illness. Answering this question would require studies that retrospectively question patients with a psychiatric illness (such as depression) about the presence of idiopathic nightmares preceding the disorder. Another approach would be prospective monitoring of patients with idiopathic nightmares regarding the development of psychiatric illness over their lifetime.

7.6. Fifth Challenge: High‐Density Home Recording Methods

To our knowledge, nightmares have never been studied at the “state” level using an EEG approach in one study, the 20 s of sleep preceding fearful dream reports (non‐nightmarish) was contrasted with fearless dreams obtained by induced awakenings using high‐density EEG recording (256 channels) (Sterpenich et al. 2020). This approach showed, using source reconstruction, an increase in activity in the insula (in REM and NREM sleep) and the anterior cingulate cortex (in REM sleep only) during fearful dreams. This type of high‐density EEG approach could be used to understand the brain regions involved in nightmare production. However the main limitation is the rarity of nightmares in the sleep laboratory, eir in contrast with greater frequency in home recordings. One possible solution is to develop an approach that allows for high‐density recordings in the homes of patients with nightmares.

8. Lucid Dreaming and Dream Engineering as Therapy for Nightmares and Insight Into the Dream Process

8.1. Lucid Dreaming

Lucid dreaming is a hybrid state of consciousness in which the dreamer becomes aware of being in a dream while remaining asleep. In some cases, the dreamer also gains the ability to exert control over dream content (Lemyre et al. 2020). Estimates suggest that around 55% of people experience at least one lucid dream in their lifetime, with 23% experiencing them monthly or more frequently (Saunders et al. 2016). Lucid dreams can occur spontaneously, particularly in children, but their frequency tends to decline with age (Voss et al. 2012). They can also be induced and enhanced through training at any stage of life (Tan and Fan 2023).

While lucid dreaming is considered a non‐pathological variant of REM dreaming, it has been associated with other parasomnias and hybrid states between wakefulness and sleep (Table 5), including nightmares (Drinkwater et al. 2020), out‐of‐body experiences (Campillo‐Ferrer et al. 2024), false awakenings (Buzzi 2019), and sleep paralysis (Ableidinger and Holzinger 2023; Mainieri et al. 2021). It is also prevalent in individuals with narcolepsy (Dodet et al. 2015; Rak et al. 2015). Given these associations, lucid dreaming is increasingly being explored as a tool for understanding altered states of consciousness in both healthy and pathological sleep, and as a potential therapeutic intervention for various sleep disorders.

8.2. New Insights Into Lucid Dreaming: Consciousness in Sleep

Lucid dreaming provides a unique window into the neural mechanisms underlying conscious experience during sleep. Because voluntary eye movements are unaffected by the muscle atonia of REM sleep, lucid dreamers can use predefined eye signals that can be recorded via electrooculography, serving as an objective marker of both dreaming and lucidity. These markers have allowed investigations into core aspects of dreams, including time perception (Erlacher et al. 2014), visual processing (LaBerge, Baird et al. 2018, LaBerge, LaMarca et al. 2018); voluntary control over breathing (Oudiette et al. 2018) and motor movements (Dresler et al. 2012). Furthermore, recent advances in lucid dreaming research demonstrated that two‐way communication with lucid dreamers is possible, enabling real‐time dialogue and insights into the dreaming mind (Konkoly et al. 2021). While REM sleep is traditionally thought to disconnect dreamers from the external world, some sensory channels remain accessible. In one study, expert lucid dreamers were able to perceive external stimuli—such as spoken questions or Morse code via light flashes and tactile stimulation—and respond through eye signals or brief frowning and smiling. Communication was successful in 18% of trials, indicating that dreamers can process and react to stimuli under certain conditions. The same “dialogue” technique showed that lucid dreamers were able to perform a complex semantic task (distinguishing words from pseudo‐words) during REM sleep as well as N1 and N2 NREM sleep (Turker et al. 2023).

Efforts into studying the neural correlates of lucidity itself also provide valuable insights into how meta‐awareness can emerge and be maintained during sleep. Lucid dreaming has been described as a hybrid state between wakefulness and sleep, with increased cortical and physiological activation during REM sleep (Voss et al. 2009), although these results have not been reproduced in a larger sample (Dodet et al. 2015). Studies have reported heightened REM density, autonomic nervous system arousal ‐including elevated heart rate and respiration‐, and increased parietal beta and fronto‐lateral gamma activity, alongside reduced fronto‐central delta power during lucid REM sleep (review in Baird et al. (2019)). Functional imaging studies suggest that lucid dreaming involves increased activation and connectivity within fronto‐parietal cortical networks, particularly in regions linked to insight and agency (Dresler et al. 2012). However, small sample sizes and methodological limitations, including potential ocular artefacts, complicate the interpretation of some of these findings, and it was argued that lucid dreams may occur in a deeper REM sleep phase rather than a state closer to wakefulness (Baird et al. 2019). Lucid dreams predominantly occur in REM sleep, but they have also been observed in NREM stages, including N1 and N2 (Dresler et al. 2012). Further investigating the neurophysiological mechanisms underlying lucidity in both REM and NREM sleep may help understand other related altered states of consciousness in sleep disorders and parasomnias.

8.3. Lucid Dreaming: From Personal Development to Medical Treatment

Lucid dreaming has long been pursued for personal curiosity, creativity, and well‐being (Blagrove and Hartnell 2000). Research suggests that lucid dreaming can foster positive dream content, increase the feeling of being refreshed in the morning and improve mood upon waking (Schredl et al. 2020; Stocks et al. 2020). As a result, there is growing clinical interest in exploring lucid dreaming as a therapeutic tool.

One of the most promising clinical applications of lucid dreaming is nightmare therapy. Imagery Rehearsal Therapy, the standard treatment for nightmares, involves rescripting distressing dreams while awake. Similarly, lucid dreaming allows the dreamer to modify nightmares, albeit directly from within the dream. Even when the dreamer does not actively alter the dream scenario, simply recognising that the nightmare is not real can help reduce its emotional intensity (de Macedo et al. 2019). Lucidity may also help alleviate sleep paralysis distress by allowing individuals to recognise the episode as harmless or even transform the experience into something more neutral or enjoyable (Ableidinger and Holzinger 2023). Reviews of studies on lucid dream therapy for nightmares, including nightmares related to post‐traumatic stress disorder, suggest generally positive outcomes, although research is still limited by small sample sizes (de Macedo et al. 2019; Ouchene et al. 2023). In insomnia disorder, lucid dreaming therapy reduced insomnia symptoms in a preliminary study, as if becoming aware of one's mental experience during sleep would reduce sleep misperception (Ellis et al. 2021). Whether this effect involves a reduction in sleep misperception through increased awareness of mental experiences during sleep remains to be explored.

8.4. Methods for Inducing Lucid Dreams: Science‐Fiction in the Bedroom?

A key challenge in using lucid dreaming for therapy is its relative rarity in the general population, as traditional induction techniques require significant time and effort, limiting their feasibility for clinical applications. Common cognitive and behavioural methods include keeping a dream journal and performing reality checks throughout the day, using intention and prospective memory techniques, engaging in sensory attention training, and sleep interruption techniques, such as the wake‐back‐to‐bed method (Stumbrys et al. 2012). While these approaches have shown varying success rates, novel interventions like virtual reality‐based lucid dream training may improve their effectiveness. In addition to cognitive techniques, pharmacological and neurostimulation methods have gained interest not only as potential tools for enhancing lucidity but also for shedding light on its underlying neural mechanisms. In particular, acetylcholinesterase inhibitors, such as donepezil and galantamine (LaBerge, LaMarca et al. 2018), have been shown to increase the likelihood of lucid dreaming in healthy participants. Non‐invasive brain stimulation methods, including transcranial direct current stimulation (Stumbrys et al. 2013) and gamma‐frequency transcranial alternating current stimulation (Voss et al. 2014) have produced mixed results—enhancing some lucid‐related aspects of dream content but not inducing full lucid dream experiences. Sensory stimulation during REM sleep has also been investigated, including light cues, acoustic stimulation, and tactile stimulation (Erlacher et al. 2014; Stumbrys et al. 2012). While some studies report success, results remain inconsistent. The combination of cognitive lucidity training with sensory stimulation during REM sleep, a method called Targeted Lucidity Reactivation, recently demonstrated a 50% success rate, even in individuals with limited prior experience and in patients suffering from high nightmare distress (Carr et al. 2023) this technic was alos shown to reduce nightmares, sleep paralysis, and RBD in patients with narcolepsy (Mundt et al. 2025). Wearable technology that delivers automated sensory stimulation at targeted sleep stages is a promising area of research, although minimising risks of sleep disruption should remain a priority, especially for application in populations with pre‐existing sleep impairment.

8.5. Potential Risks and Considerations

Although lucid dreaming is generally associated with improved well‐being, concerns have been raised about potential negative effects of inducing lucid dreaming (Aviram and Soffer‐Dudek 2018; Mallett et al. 2022). Self‐reported adverse experiences associated with lucid dreaming include sleep paralysis, lucid dysphoria, poor sleep, and reality confusion (Aviram and Soffer‐Dudek 2018; Mallett et al. 2022). Being lucid while unable to change or escape a nightmare can also lead to high distress levels. This highlights the importance of considering levels of control and agency when developing lucid dream induction techniques for therapeutic applications. A large‐scale survey found that any association between lucid dreaming and heightened stress, anxiety, and depressive symptoms was primarily attributable to the co‐occurrence of nightmares (Carr et al. 2025).

8.6. The Future Promises of Lucid Dreaming

In sum, lucid dreaming and other dream engineering techniques (Carr et al. 2020) aimed at enhancing more positive, insightful, and volitional dream content hold potential for treating nightmares, improving sleep quality, and alleviating mood disorders. They may also offer benefits in treating parasomnias involving negative dream experiences, such as sleep paralysis, sleep terrors, and RBD. Yet, challenges remain in developing reliable and accessible induction techniques, ensuring clinical feasibility and safety. Although still in its early stages, the ability to engage with dreamers in real time opens new avenues for exploring both normal and pathological dream states, as well as the neural mechanisms of sensory disconnection during sleep. Future efforts should focus on refining communication methods, for example, with more complex eye movement sequences, facial contractions, and sniffing patterns to enhance the effectiveness and reliability of these interactions.

Overall, lucid dreaming continues to serve as a powerful tool for both clinical and fundamental research on sleep, consciousness and mental health. However, challenges remain due to small sample sizes and the inherent difficulties of studying lucid dreaming in laboratory and home settings. Research should also prioritize standardised procedures, large‐scale open science collaborations, advancements in wearable technology, and validation in clinical populations. Additionally, incorporating citizen science initiatives, working with expert lucid dreamers—including individuals with narcolepsy—and further developing two‐way communication methods are key areas for future progress (Esfahani et al. 2024; Zerr et al. 2024).

Author Contributions

Claudia Picard‐Deland: writing – original draft. Matteo Cesari: writing – original draft. Ambra Stefani: writing – original draft. Jean‐Baptiste Maranci: writing – original draft. Birgit Hogl: writing – original draft. Isabelle Arnulf: writing – original draft.

Conflicts of Interest

The authors declare no conflicts of interest.

Picard‐Deland, C. , Cesari M., Stefani A., Maranci J.‐B., Hogl B., and Arnulf I.. 2025. “The Future of Parasomnias.” Journal of Sleep Research 34, no. 5: e70090. 10.1111/jsr.70090.

Data Availability Statement

Data sharing not applicable to this article as no datasets were generated or analysed during the current study.