Acetylcholine and metacognition during sleep

Acetylcholine is a neurotransmitter and neuromodulator involved in a variety of cognitive functions. Additionally, acetylcholine is involved in the regulation of REM sleep: cholinergic neurons in the brainstem and basal forebrain project to and innervate wide areas of the cerebral cortex, and reciprocally interact with other neuromodulatory systems, to produce the sleep-wake cycle and different sleep stages. Consciousness and cognition vary considerably across and within sleep stages, with metacognitive capacity being strikingly reduced even during aesthetically and emotionally rich dream experiences. A notable exception is the phenomenon of lucid dreaming — a rare state whereby waking levels of metacognitive awareness are restored during sleep — resulting in individuals becoming aware of the fact that they are dreaming. The role of neurotransmitters in these fluctuations of consciousness and cognition during sleep is still poorly understood. While recent studies using acetylcholinesterase inhibitors suggest a potential role of acetylcholine in the occurrence of lucid dreaming, the underlying mechanisms by which this effect is produced remains un-modelled and unknown; with the causal link between cholinergic mechanisms and upstream psychological states being complex and elusive. Several theories and approaches targeting the association between acetylcholine and metacognition during wakefulness and sleep are highlighted in this review, moving through microscopic, mesoscopic and macroscopic levels of analysis to detail this phenomenon at several organisational scales. Several exploratory hypotheses will be developed to guide future research towards fully articulating how metacognition is affected by activity at the acetylcholine receptor.


Introduction
Sleep is a universal physiological phenomenon in animals and humans that is essential for normal brain function and overall health (Dresler et al., 2014).Since the discovery of REM sleep (Dement and Kleitman, 1957), mammalian sleep has been categorized into two classes of actively generated sleep stages based on distinct electrophysiological signatures: REM and non-REM (NREM) sleep.REM sleep characteristically shows desynchronized EEG activity that is similar to wakefulness, combined with rapid eye movements and muscle atonia.NREM sleep is characterized by slow wave activity and sleep spindles and can be further divided into multiple substages.Regulation of the sleep-wake cycle relies on a complex interplay between multiple neurochemical systems, that are subject to ongoing research and investigation (Holst andLandolt, 2018, Jones, 2019).
A central role in the neurochemistry of sleep plays the neurotransmitter Acetylcholine, reciprocally interacting with other neuromodulatory systems to generate the dynamics of sleep cycles.Initially discovered at the beginning of the twentieth century, acetylcholine (ACh) is one of the most well-studied neurotransmitters (Horiuchi et al., 2003).While ACh acts as an excitatory neurotransmitter in the peripheral nervous system, its function in the central nervous system is largely modulatory.Instead of directly exciting or inhibiting neuronal populations, central cholinergic neuromodulation alters the threshold potential of individual neurons, effectively altering the excitability, firing rate and presynaptic neurotransmitter release (Picciotto et al., 2012) and thereby producing cascading network-level changes in affected neuronal populations.Cholinergic neurons are clustered in various nuclei in the brain, including the basal forebrain (BF) and the pedunculopontine and lateral dorsal tegmental area (PPT-LDT) of the brainstem.These nuclei exhibit widespread projections to the entire neocortex, hippocampus and thalamus.Other cholinergic nuclei include the striatum, the vestibular nuclei and the spinal-cord pre-ganglion (Perry et al., 1999).
As a neuromodulator, acetylcholine is involved in a variety of different cognitive functions, resulting from activity from multiple receptors with differing kinetics, distributions and effects.Due to the complexity of theseintricate pharmacodynamics, attempts to understand Acetylcholine in the context of its psychological and behavioral effects-or alterations to thought, memory and emotion at the macroscopic level-are not easily realized, and as such, a range of computational and theoretical models have arisen to describe the effects of this neuromodulator, and how these interact with other systems to affect cognition at large.Several domains of aberrant cognitive process that arise during sleep (for example, the impoverishment of insight and critical awareness) have been suggested to be cholinergically driven (Hobson and Pace-Schott, 2002).As such, there is likely to be considerable overlap between the systems that suppress metacognition during sleep and those that facilitate and maintain it during wakefulness.Studying this overlap might further our understanding of the role of acetylcholine in metacognition, and could shed new light on the role that this ligand and its dynamics play in the synthesis of cognitive activity as it is experienced.This review will summarise and explore the tentative connections between acetylcholine and metacognition during sleep, and attempt to model how activity of this ligand and its receptors at the microscopic and mesoscopic level can result in cognitive and behavioral changes at the domain of human psychological process.Studying this overlap might further our understanding of the role of acetylcholine in metacognition, and could shed new light on the role that this ligand and its dynamics play in the synthesis of cognitive activity as it is experienced.

Acetylcholine receptors and subtypes
Cholinergic neurotransmission starts with ACh production in the presynapse.Enzymatic synthesis of acetyl-coA is catalysed by transformation of a functional acetyl-group from citrate onto co-enzyme A by the ATP-citrate-lyase.Co-enzyme A is the educt for acetyltransferase, which catalyses acetylation of choline to ultimately form ACh. ACh is then transported into synaptic vesicles, that will later release into the synaptic cleft by exocytosis, binding to two different types of receptors (muscarinic and nicotinic) that are expressed at the pre and post synapse, among other locations (Soreq, 2015).Termination of cholinergic neurotransmission takes place via the enzyme acetylcholinesterase (AChE), which hydrolyses ACh in the synaptic cleft into its components acetate and choline.Choline is then transported back into the presynapse via choline transporters (Soreq, 2015).ACh can work as a neuromodulator, altering the release of other neurotransmitters such as glutamate, GABA or dopamine.This functional diversity links ACh to a variety of different brain states and functions.These functions include attention, memory, arousal and most importantly wakefulness and sleep (Soreq, 2015), which may play an intricate role in how the different features of cognition are differentially activated and deactivated between these states.
Muscarinic acetylcholine receptors (mAChRs) are G-coupled receptors that encompass five subtypes (M1-M5) and are expressed throughout the neocortex, hippocampus, striatum and other brain structures.Upon activation, mAChRs are metabotropic, and initiate intracellular signalling cascades that result in opening and closing of ion channels, which ultimately lead to depolarization or hyperpolarization of the neuron.By coupling to different types of G-proteins, mAChRs can have different effects.M1, M3 and M5 (collectively called 'M1-like' receptors) are coupled to G q proteins, leading to depolarization and increase in synaptic transmission.They are predominantly located at the postsynapse and can be found in the cortex, striatum and other brain structures.'M2-like' receptors, including M2 and M4, are coupled to G i/o proteins and inhibit synaptic transmission at the presynapse or hyperpolarize the postsynaptic membrane.M2-like receptors can be found in the basal forebrain where they are presumed to regulate synthesis and release of ACh (Colangelo et al., 2019;Eglen, 2006).
The Muscarinic receptor subtype is named after the selectively binding Acetylcholine analogue Muscarine, which was isolated and extracted from the deliriant/dissociative psychoactive mushroom Amanita muscaria-and erroneously thought to be the source of its intoxicating effects (Schmiedeberg, 1869).The most prominent cognitive effects produced through this receptor subtype can instead be seen during administration of ligands that block receptor activity, such as Scopolamine, which have a long history of use in shamanic practices (Rudgley, 1995) and frequently induce vivid and notably unpleasant hallucinations and memory disruptions.These are described as being similar in nature to having severe Lewy-body dementia or Alzheimer's disease (Perry & Perry, 1995;Ebert & Kirch, 1998) though the precise mechanism of action for these effects remains poorly understood.
Nicotinic acetylcholine receptors (nAChRs) are ionotropic receptors that are formed from five common combinations of twelve theoretical subunits (α2-α10; β2-β4), the most common of which being α4β2 and α7 (Hernandez & dineley, 2012).Nicotinic AChRs are located at both the presynaptic terminal, where they modulate neurotransmitter release via influx of sodium and calcium, depolarizing the presynaptic membrane; as well as the postsynaptic terminal of the dendritic spine, where they either excite or inhibit the postsynaptic membrane (Colangelo et al., 2019).Nicotinic acetylcholine receptors are named after the substance Nicotine, an alkaloid frequently found in the nightshade plant family and most prominently known as the active ingredient in tobacco, where it broadly mimics the role of acetylcholine, binding at high affinity for the α4β2 receptor subtype, while acting as a selective antagonist for the α9 and α10 subtypes (Wonnacott, 1997;Gotti, Marks, Millar, & Wonnacott, 2019).Nicotine has a complex role in human behaviour, acting as both a mild stimulant and relaxant (Ksir, Hakan, Hall, & Kellar, 1985), memory enhancer (Heishman et al., 2010) and potentially a form of self-medication for intrusive thoughts; particularly for ADHD and Schizophrenia symptomology (Adler et al., 1998;Levin, 2002).Despite being highly addictive and sought after as a recreational substance, the effects of Nicotine are comparatively mild by the standards of other recreational narcotics.

Acetylcholine and sleep regulation
Most evidence of neurochemical regulation of sleep has been gathered by measuring and manipulating sleep in animals.Despite varying sleep patterns and complexity of sleep architecture, many sleep features are conserved across species.Rodents and other mammals such as cats are frequently used as animal models in sleep research, providing the possibility for genetic and pharmacological interventions to investigate neurochemical sleep regulation, neural pathways and sleep disorders (Paterson et al., 2011).
Early in vivo pharmacological and lesion studies pioneered by Michael Jouvet, and followed by many others, have suggested an important role for cholinergic signaling in REM sleep generation (Jouvet, 1972, Szymusiak andMcGinty, 1986).Subsequent research has started to elucidate the underlying pathways of neurochemical sleep regulation.For instance, brainstem injections of ACh receptor agonists induced REM sleep and muscle atonia (Baghdoyan et al., 1984).Furthermore, selective cholinergic cell lesions in the brainstem reduced REM sleep time, phasic and tonic REM events, implicating pontine cholinergic neurons as a key component for REM sleep generation (Webster andJones, 1988, Shouse andSiegel, 1992).
Early on, Robert McCarley and Allan Hobson proposed a reciprocal-interaction model, suggesting that sleep is regulated by the interaction of brainstem neurons that are either active (REM-ON) or silent (REM-OFF) during REM sleep (McCarley and Hobson, 1975).Over the years, this model was modified and updated, with a 'flip-flop' switching circuitry between REM-ON and REM-OFF areas and other synaptic details added (Fig. 1B) (Lu et al., 2006;Fuller et al., 2007;Pace-Schott and Hobson, 2002).It has now been established that there are two major cholinergic pathways involved in REM sleep generation.Cholinergic neurons in the LDT-PPT have ascending projections to the basal forebrain and reticular and relay cells of the thalamus (Fig. 1A).These thalamic cells in turn have glutamatergic projections to the cortex; while other cholinergic neurons in the basal forebrain directly project to the cortex (Holst Fig. 1. A. Regulation of REM sleep in the brain.During REM sleep, cholinergic REM-ON cells in the pedunculopontine and lateral dorsal tegmental area (PPT-LDT) project to the cortex via the thalamus and cholinergic neurons in the basal forebrain (BF), activating the cortex.GABAergic REM-ON cells in the sublaterodorsal nucleus (SLD) produce muscle atonia by inhibitory projections to the medulla and spinal cord.These SLD GABAergic REM-ON neurons share mutual inhibitory connections with GABAergic REM-OFF cells in the lateral pontine tegmentum (LPT) and periaqueductal gray (PAG), creating a "flip-flop" switch between REM/NREM sleep (adapted from Holst andLandolt, 2018, Saper et al., 2010).B. Updated reciprocal interaction model with REM/NREM "flip-flop" switch.REM sleep initiation takes place via stimulation of glutamatergic REM-ON cells in the pontine reticular formation (PRF) by cholinergic REM-ON cells in LDT/PPT.Additionally, cholinergic REM-ON cells excite SLD REM-ON cells of the "flip-flop" switch that in turn inhibit monoaminergic REM-OFF cells.With progression of REM sleep, cholinergic REM-ON cells excite aminergic REM-OFF cells.These REM-OFF cells then provide inhibitory feedback to REM-ON neurons, leading to REM cessation.Self-inhibition of aminergic REM-OFF neurons lifts suppression of cholinergic activity, leading to REM initiation (adapted from Brown et al., 2012). andLandolt, 2018).These cholinergic neurons in the LDT-PPT and basal forebrain are active during wakefulness and peak during REM sleep, therefore being considered REM-ON cells.Via their direct and indirect projections to the cortex and subcortical structures, they contribute to the typical REM features including cortical activation and hippocampal theta waves.Additionally, ACh prevents slow wave activity via muscarinic receptors in the cortex (Peever and Fuller, 2016, Scammell et al., 2017, Jones, 2019).Rodent studies with injections of ACh receptor antagonists and receptor subtype knockouts have identified muscarinic receptor subtypes M1 and M3 as key players in REM sleep generation (Niwa et al., 2018), while feline studies have indicated the M2 receptor to be responsible in this species (Baghdoyan and Lydic, 1999).
REM-ON cells in the sublaterodorsal nucleus (SLD) have descending projections to the ventral medulla and the spinal cord and-where they activate inhibitory interneurons-produce REM-characteristic muscle atonia (Peever and Fuller, 2016).During NREM sleep, REM-ON cells show minimal firing.This is supported by microdialysis studies in cats, measuring highest ACh release during REM sleep and lowest release during NREM sleep (Marrosu et al., 1995, Vazquez andBaghdoyan, 2001).Optogenetic activation of cholinergic activity increased fast activity in the cortex while suppressing slow waves, suggesting the decrease in cholinergic activity as a prerequisite for NREM sleep (Jones, 2019).
Cholinergic REM-ON neurons in PPT-LDT and basal forebrain do not act in isolation: They are interspersed with glutamatergic and GABAergic neurons with heterogeneous functions that respond differently to monoaminergic arousal-promoting agonists.For instance, basal forebrain cholinergic neurons depolarize in response to noradrenaline, whereas basal forebrain GABAergic neurons respond with hyperpolarization, suggesting a complex interplay between the different neurochemical systems (Jones, 2019).During wakefulness, cholinergic REM-ON neurons are tonically inhibited by aminergic REM-OFF neurons (Fig. 1).These include adrenergic neurons in the locus coeruleus (LC), serotonergic neurons in the dorsal raphe nucleus (DRN) and histaminergic neurons in the hypothalamus.These REM-OFF neurons possess widespread ascending projections that split dorsally to the thalamus and ventrally to the cortex, brain stem and basal forebrain, promoting and maintaining wakefulness (Holst and Landolt, 2018;Scammell et al., 2017;Fuller et al., 2007).GABAergic REM-OFF neurons in the periaqueductal gray (PAG) and lateral pontine tegmentum (LPT) are reciprocally connected with GABAergic-REM-ON neurons in the SLD, providing a mutually inhibitory "flip flop" switch (Fig. 1B).During NREM sleep, GABAergic REM-OFF cells are active, inhibiting GABAergic REM-ON cells.Monoaminergic REM-OFF neurons facilitate this REM inhibition by acting on both sides of the switch, exciting REM-OFF neurons and inhibiting REM-ON neurons (Saper et al., 2010).As the name suggests, REM-OFF neurons are shut off during REM sleep.They are maximally active during wakefulness and exhibit decreased activation during NREM sleep, gradually lifting their inhibition on cholinergic REM-ON cells.Reciprocally, REM-ON activation has the opposite effect on the "flip-flop" switch system, exciting REM-ON and inhibiting REM-OFF cells.Additionally, aminergic self-inhibition from REM-OFF cells facilitates REM-ON activation.Subsequently, cholinergic REM-ON activation increases until REM sleep is initiated, and REM-OFF cells are shut off.With REM sleep progression, cholinergic activity excites aminergic REM-OFF cells.These REM-OFF cells in turn provide inhibitory feedback, leading to REM cessation (reviewed in Fuller et al., 2007, Pace-Schott and Hobson, 2002, Saper et al., 2010, Brown et al., 2012).

Consciousness during sleep
The human conscious experience has long been considered in terms of two overlapping, but qualitatively distinct aspects, described as dual processes by William James (James, 1890), primary vs. secondary process thinking by Sigmund Freud (Freud, 1966) and primary vs. secondary consciousness by Gerald Edelman (Edelman, 2003).This distinction has also been used within the context of lucid dreams where it arguably produces its most striking contrast: only in this context can secondary processing transition between such extreme quantities in such a short period (seconds) without major shifts in physiological vigilance or surgical or pharmacological intervention and without risk (Dresler et al., 2009;Hobson, 2009;Hobson & Voss, 2010).
Primary consciousness describes the basic experience of emotion and perception, and constitutes the highly reactive aspect of human behavior, whereas secondary consciousness describes the experience of analytical, self-referential and metacognitive processes.This capability for metacognitive processes including self-awareness and critical self-reflection has been associated with prefrontal cortical structures (Fleming & Dolan, 2012), which also play a key role in executive functioning and other higher cognitive tasks such as working memory and decision making (Qiu et al., 2018).
The availability and quality of types of conscious experience vary heavily during the sleep-wake cycle.With access to rich sensory stimuli from the external world and the capacity to reflect on their processing, the waking state is characterized by availability of both primary and secondary consciousness.In contrast, during dreams, we omit external sensory input and experience a purely internally generated simulation of reality, which we take at face value without further reflection.In other words, typical dreams consist of primary but not secondary consciousness.This loss of metacognitive capacity is one of the most apparent changes between waking and sleep: no matter how unusual the contents of experience, under normal circumstances, we rarely are able to question whether the contents of our dreams make sense, due to explicit inhibition of this cognitive process.
Selectively diminished form of metacognition seen during sleep can be attributed to the global reduction of prefrontal cortex activation, in the transition from wakefulness to NREM sleep.PET studies have shown an association between prefrontal deactivation and depth of NREM sleep.With the transition to REM sleep and activation of the cholinergic system, parts of the prefrontal cortex, including anterior cingulate (ACC) and ventromedial prefrontal cortex, and higher visual areas are reactivated.Consistent with this reactivation, REM sleep is the sleep stage that is most associated with full-blown dream experiences, indicating a resumption of primary conscious experience, however absent of the self-critical secondary aspects that are ubiquitous with waking.Dreaming also occurs during NREM, however typically lacks the vivid imagery and detailed environments of REM dreams-instead being generally more thought-like (Nielsen et al., 2000;Pace-Schott et al., 2003).Furthermore, REM dreams are reportedly more bizarre and emotional than NREM dreams (Muzur et al., 2002, Hobson andFriston, 2012) rendering this incapability to spot incongruities in the dream narrative even more outstanding.During REM sleep, the dorsolateral prefrontal cortex (DLPFC) becomes relatively deactivated, in contrast to ACC and ventromedial PFC.This deactivation is proposedly instigated via cholinergic modulation (Muzur et al., 2002).Consistent with this proposal, cholinergic M2 receptors in ACC axons and inhibitory neurons projecting to the DLPFC have been found, suggesting explicit cholinergic inhibition of the DLPFC activation during REM sleep (Medalla & Barbas, 2012).
The three-dimensional AIM-model of conscious state control proposes that the different states of consciousness are determined by cortical activation, input-output gating and neuromodulation (Fig. 2) (Hobson, 2009;Hobson, Pace-Schott, & Stickgold, 2000).Activation (A) describes the intensity of consciousness, indicated by EEG activity.Many cortical and subcortical areas including the regions mentioned above are active during waking, deactivate during NREM and reactive during REM sleep.Input-output gating (I) classifies the information source that determines the state of consciousness indicated by the degree of sensory input and motor output.During REM sleep, external sensory input and brainstem mediated motor output is suppressed, relying on internal information processing.The modulation level (M) is determined by the activity of cholinergic and aminergic neurons as proposed by McCarley and Hobson's reciprocal interaction model.Waking is characterized by both aminergic and cholinergic activity, that decreases during NREM sleep.During REM sleep, cholinergic neurotransmission peaks, shutting off aminergic neurons (Hobson, 2009).

Lucid dreaming
During dreams, we typically experience primary consciousness alone.Wemostly implicitlyaccept this as 'reality' uncritically, only to recognise the nature of the dream after waking up (Laberge, 2014).Absent of secondary consciousness, one may even explicitly come to question the nature of their dream-and, finding themselves cognitively incapable of doing so proficiently-falsely conclude that they are actually awake.Lucid dreams present an exception to this, and can be defined as a reactivated state of secondary consciousness and metacognition during sleep.In order to experience a lucid dream, an individual essentially has to succeed in recognizing that their experience of reality is a simulation within their own mind.Accordingly, they may further regain awareness of their waking reality 'at large'; being a place which they will inevitably return to once the dream itself has ended.Lucid dreaming has been described as a 'hybrid' state, as it contains features of both waking and sleep (Fig. 2).For instance, lucid dreamers have full access to their cognitive functions-they are able to consciously execute specific actions such as pre-agreed eye movements, while maintaining a REM-characteristic polysomnogram (LaBerge et al., 1981).These pre-agreed eye movements that are clearly distinguishable in an electro-oculogram, are used as an objective marker to verify lucid dreams and can further be used to temporally mark the execution of pre-agreed task performance within a dream (Dresler et al., 2011;Baird 2019).Within the AIM model, lucid dreaming has been located at a medium position in the (I) axis, as in contrast to non-lucid sleep the dreamer is aware of the existence and nature of the external world, albeit directly experiencing only the internally generated dream world.Furthermore, subjects during lucid dreams are able to understand communications from external sources, such as mathematical problems, and communicate back to people in the waking world using these pre-agreed eye movements, effectively communicating from the dreaming and waking world in both directions (Konkoly et al., 2021).
Despite sharing many characteristic REM features (Baird, Tononi, & LaBerge, 2022), lucid dreaming shows distinctive changes in brain activity compared to non-lucid REM sleep (Baird et al., 2019).Per definition, lucid dreams are strongly connected to metacognition.Neurobiologically, this is reflected by increased structure and function in multiple regions associated with metacognition such as the frontopolar, parietal and dorsolateral prefrontal cortices (Shimamura, 2000;Flemming & Dolan, 2012).For instance, Filevich and colleagues showed increased gray matter volume in the frontopolar cortex, in frequent lucid dreamers compared to participants with low lucid dreaming frequency (Filevich et al., 2015).This is a region that has previously been associated with metacognition and self-reflection (Flemming & Dolan, 2012).They also measured increased BOLD signal in the frontopolar cortex during metacognitive thought processing, demonstrating similar neural correlates for lucid dreaming and metacognitive function.Other regions that are linked to metacognition and are reactivated in lucid compared to non-lucid REM sleep include parietal regions such as the precuneus, and the dorsolateral prefrontal cortex (DLPFC) (Dresler et al., 2012;Baird et al., 2019).Deactivation of those regions is thought to underlie reduced self-awareness and restricted cognitive function during non-lucid sleep.Overall, lucid brain activity is seen in multiple regions of the fronto-parietal control network that includes the DLPFC and parietal regions.This frontoparietal network has been associated with integrating information from anti-correlated internal and external awareness networks and shift between those, suggesting a mediating role for metacognitive processes during lucid dreaming (Demertzi, Soddu, & Laureys, 2013;Dresler et al., 2015;Spoormaker, Czisch, & Dresler, 2010).Furthermore, frequent lucid dreamers exhibit increased functional connectivity between anterior PFC and tempoparietal association areas that are deactivated during non-lucid sleep (Baird et al., 2018).
Lucid dreams offer a unique possibility to study loss and recovery of self-awareness within the same vigilance state.This holds unique potential to study consciousness during sleep.Volitional control over the dream content enables lucid dreamers to execute specific tasks during dreaming that can later be correlated to physiological sleep measurements (Dresler et al., 2011(Dresler et al., , 2014)).Lucid dreaming also has clinical implications, for instance in the treatment of nightmares (Spoormaker & Van Den Bout, 2006;Rak et al., 2015;Macedo et al., 2019).Moreover, lucid dreaming research may contribute to establishing neuroimaging markers of self-awareness and consciousness that could help with diagnosis and treatment of patients in unresponsive states (Baird et al., 2019).Since lucid dreaming is a rare state, a considerable fraction of contemporary lucid dreaming research explores new strategies to make it more accessible for a broader range of volunteers in order to efficiently investigate and exploit its benefits.Numerous methods for the induction of lucid dreaming have been developed and tested, ranging from cognitive training and behavioral sleep manipulations to electrical brain stimulation, sensory cueing and virtual reality (Stumbrys et al., 2012;Baird et al., 2019, Erlacher et al., 2020;Carr et al., 2020;Gott et al, 2020).Anecdotal and semi-systematic evidence (Yuschak, 2006), and more recently controlled studies have further explored pharmacological strategies of lucid dream induction.

Acetylcholine in lucid dreaming
Recent research has demonstrated a promising role of acetylcholine for pharmacological lucid dream induction.Recent studies by LaBerge (LaBerge et al., 2018), Sparrow (Sparrow et al., 2016;Sparrow et al., 2018) have found increases in lucid dreaming frequency by enhancing cholinergic neurotransmission via administration of acetylcholinesterase inhibitors (AChEIs).This makes the proposed medium position of lucid dreaming on the (M) axis questionable, and an adapted version with lucid dreaming moved towards the ACh end of the axis more reasonable (Fig. 2).
AChEIs are a diverse category of drugs that prevent the enzyme acetylcholinesterase from degrading ACh.Galantamine in particular has been used in multiple studies due to its mild side-effects at low doses, and has shown promising results for lucid dream induction.Galantamine is unique in terms of its dual mode of action: in addition to selectively and competitively inhibiting acetylcholinesterase, it is also working as an allosteric modulator of nAChRs, thereby increasing nAChR agonist affinity and producing synergistic effects with elevated ACh levels (Lilienfeld, 2002).
LaBerge and colleagues tested an integrated lucid dreaming protocol, combining cholinergic stimulation with the practice of mnemonic induction of lucid dreams (MILD), a technique that induces lucidity by recognizing a dream sign, an anomaly within the dream (LaBerge et al., 2018).129 participants received either 0, 4 or 8 mg of galantamine orally over three consecutive nights.Each night, participants were woken up after approximately 4.5 h of sleep, took the assigned dose of the drug, went back to bed after 30 min and practiced MILD while going back to sleep.The results showed a significant dose-related increase in lucid dreaming frequency for both doses of galantamine compared to the placebo procedure.Galantamine also increased recall, vividness and complexity of dreams.In a previous study using the AChEI donepezil, LaBerge and colleagues also found an increase in sleep paralysis by cholinergic enhancement (Laberge, 2004).Sparrow et al. tested galantamine individually and combined with other lucid dream induction techniques (dream reliving and meditation) on 35 participants (Sparrow et al., 2018).They found a significant increase in lucidity for galantamine versus placebo treatment.Galantamine treatment combined with dream reliving and meditation also increased lucidity compared to placebo treatment with dream reliving and meditation.However, dream reliving and meditation did not significantly enhance lucidity compared to galantamine treatment only.The underlying mechanism by which AChEIs (and specifically, galantamine) induce lucid dreaming and metacognitive awareness during sleep remains unknown; however here are several mechanisms proposed, that are not mutually exclusive (Baird et al., 2019).Given that the dynamics of nAChRs and their specific effects on waking and sleeping cognition are so complex, in the following we describe several explorative hypotheses on the association between ACh and metacognition during sleep in order to facilitate experimental setups and contextualize empirical findings for future research.

Acetylcholine and vivid/aberrant dreams
The link between Acetylcholine and disordered, aberrant and unusual dream experiences is well established-so much so, that "usually vivid dreams" are standardly listed side effects on nicotine patches and replacement alternatives (Centers for Disease Control and Prevention, 2021).Meanwhile, clinical trials with nicotine patches have reported 'vivid dream' incidences in as much as 30 % of participants, constituting a 5-fold increase over placebo (Richmond et al., 1994;Rose & Davis, 2020) while acetylcholinesterase inhibitors such as Donepezil, Rivastigmine and Tacrine also have 'disordered dreaming' as listed possible side effects as well (Pagel & Helfter, 2003).A recent investigation into the Acetylcholinesterase inhibitor Galantamine demonstrated induction of both nightmares and "inception-style, layered dreams" in 25 % of participants respectively (Biard et al, 2015).
Perhaps the strongest indication that acetylcholine plays a potent role in the genesis of aberrant altered conscious states during REM can be derived from the literature into the dissociative anaesthetic Ketamine.The connection between Ketamine, nightmares and parasomnias is so well established that nightmares and dysphoric hallucinations, as a potential side-effect even predate the formal naming of the compound itself (King & Stephen, 1967).In one study, Ketamine was found to increase the nightmare incidence rate by a odds-ratio of 3, while significantly altering the negative-to-positive dream emotion ratio in healthy population at sub-anaesthetic doses, in the days following administration (Blagrove et al., 2009).Another study, consisting of a clinical (post-surgical) population, showed a nightmare incidence rate of 50 % in the days following Ketamine administration, through a physiologically verified increase in REM duration, REM density and sleep fragmentation (Knill et al., 1990).
While the most common medically and recreationally available form of this compound, (R,S)-Ketamine, acts as a potent NMDA antagonist-Keramine rapidly breaks down into multiple metabolites following ingestion or infusion, such as norKetamine, hydrox-ynorKetamine and dehydroxynorKetamine-the majority of which loose most or all of their functional NMDA antagonism, instead becoming potent antagonists of the α7 Nicotinic Acetylcholine receptor (Moaddel et al., 2013).The most interesting of these, (2R-6R) hydroxynorKetamine-a remarkably potent α7 antagonist-reaches maximum blood serum concentration 24 h after administration; and lingers at substantial levels for as long as 3 days (Zhao et al., 2012) and is therefore thought to play a role in the proportionately delayed nature of Ketamine's rapid-acting antidepressant properties.Whether this 'delayed' effect (consisting of a shift from NMDA to α7 Nicotinic antagonism) also plays a role in the genesis of 'Ketamine nightmares' remains to be investigated, but also appears particularly plausible, given that the preponderance of nightmares occur on days 2 and 3 following administration of the racemic compound.
Ketamine nightmares bear a striking formal resemblance with sleep paralysis and out-of-body experiences (Fine & Firestone, 1973;Willkins et al, 2011) the latter of which is independently associated with recreational Ketamine use, both on and off the drug (Wilkins et al., 2011).A formal neurobiological mechanism explaining these aberrant, dysphoric and particularly vivid dream experiences is yet to be established, but may involve the α7 Nicotinic Acetylcholine receptor, and its unique role in sensory gating (Adler et al., 1998;Levin, 2012;Martin et al., 2007).This mechanism of action would seem especially plausible, given the accumulation of theory and evidence that out-of-body-experience and sleep paralysis are associated with disorders in sensory processing and integration in their own right (Blank et al, 2004;Braithwaite et al., 2011) in addition to the purported role that sensory integration is now thought to play in the genesis of both nightmares (Carr et al., 2022) and lucid dreams (Schredl et al., 2022).
Taken together, there would appear to be a strong association between nicotinic acetylcholine activation (specifically, antagonism of the receptor α7 subtype) and large-scale alterations of sensory processing and sensory integration, with pronounced follow-on effects on dream behaviour; particularly with regard to dysphoric, surreal and 'lucid like' dreams.Since lucid dreams, out-of-bodyexperiences, sleep paralysis and false awakenings independently and seemingly interdependently correlate (Blackmore, 1988;Levitan et al, 1999;Denis & Poerio, 2017;Buzzi, 2019) there appears to be ambiguous and potentially multi-directional causation between these phenomenon: aberrant dream experiences may either serve as potential lucidity triggers, or downstream epiphenomenon from failed lucidity attempts and episodes.It would therefore appear plausible that cholinergically inhibited sensory gating may in certain contexts contribute to aberrant, vivid and parasomnic dream experience, which in turn may lead to lucid dream induction as a secondary psychogenic phenomenon.This may present a promising line of research in future, but is outside the scope of this review to model to describe in greater depth.

Stabilization of REM sleep and increase of phasic REM
One proposed explanation for the role of acetylcholinesterase inhibitors in lucid dream induction involves the stabilization of REM sleep: AChEIs inhibit the ACh-degrading enzyme acetylcholinesterase, thereby causing an accumulation of ACh at the synapse.It is widely accepted that ACh, its synthetic derivatives, and AChEIs are involved in REM sleep generation and maintenance.For instance, cholinergic REM-ON neurons in the brainstem and basal forebrain suppress REM-OFF activity (Fig. 1) and cholinergic agonist injection into the brainstem produce REM sleep.Accordingly, in particular considering lucid dreaming as a hybrid state half way to wakefulness, AChEIs such as galantamine might stabilize REM sleep and thus keep the sleeper in a dream even in cases when they are on a trajectory towards wakefulness: where under normal conditions a rapid state shift to wakefulness would occur, under the influence of AChEIs the dreamer stays asleep and might thus achieve full-blown dream lucidity rather than wake up.
Phasic REM has been associated with increased dream vividness, dream emotion, PGO wave and REM density (Simor et al., 2020) all of which are indirect precursors for lucid dreaming.Furthermore, lucid dreaming has been found to be associated with increased phasic REM activity in its own right, indicated by increased eye movements and physiological activation and suppression of the Hreflex, a muscle reaction to electrical stimulation of innervating fibres (Brylowski et al., 1989).Interestingly, galantamine administration specifically has been shown to decrease REM latency and increase REM duration at the expense of NREM sleep.Galantamine furthermore increases phasic REM activity, including REM density (Riemann et al., 1994).Administration of muscarinic agonists had similar effects on REM sleep duration and latency as galantamine (Riemann et al., 1994).Since nicotinic receptors are thought to be involved in generating phasic REM components, such as PGO waves and rapid eye movements which are likely to represent generation and experience of specific visual epochs (Gott et al., 2017) AChEI may increase phasic REM activity via acting on nAChRs, thereby creating a favourable state for lucid dreaming.

Indirect stimulation of lucidity via other neuromodulators
Acetylcholine strongly interacts with other neuromodulators in the brain, AChEIs may enhance lucidity in an indirect manner via acting on other neuromodulatory systems such as norepinephrine, serotonin and dopamine (LaBerge et al., 2018).There is evidence that AChEI administration not only increases ACh levels but also increases dopamine levels (Schilström et al., 2007).What is more, previous studies have demonstrated that ACh administration increased dopamine release and turnover in vivo (Xu et al., 1989), while mAChR antagonists reduced frontal and hippocampal dopamine turnover and impaired cognitive performance (Memo et al., 1988).Dopamine is involved in arousal, memory and higher order cognition (Clos et al., 2019, Hauser et al., 2017).The cholinergic system also interacts with the serotonergic system.For instance, systemic administration of the cholinergic agonist nicotine increased serotonin levels in the frontal and cingulate cortex.This cholinergic-serotonergic interaction also plays an important role in mediating cognition (Steckler and Sahgal, 1995).Overall, high levels of ACh may lead to excessive activation of nAChRs that in turn enhance the release of other neuromodulators that affect consciousness and cognition (Hasselmo and Sarter, 2011).Since NA/5-HT and ACh have been shown to interact antagonistically during sleep-and since Galantamine, unlike other AChEIs, has been shown to have no effect on NA release, unless co-administered with Nicotine (Sharp et al., 2004)-it would appear likely that Galantamine's purported effects on metacognition may own themselves in part to the effects of Dopamine.Given the established link between Dopamine, dream recall and dream vividness (Solms., 2000;De Gennaro et al., 2016) this may point towards a more direct link between the intensity of conscious experience during REM and its predilection to result in spontaneous metacognitive epochs.Alternatively, indirect effects via Dopamine may have their specific mechanisms grounded within 'local' sleep framework (Vyazovskiy et al., 2011;Siclari & Tononi, 2017).However, such purported effects would be difficult to model for explicit hypothesis generation at this stage, with even unihemispheric sleep in animals resulting from somewhat enigmatic and counterintuitive neuromodulatory dynamics (Konadhode et al., 2016).As such, parallel investigations into the effects of nAChRs on both lucidity, dream vividness, dream recall and mind wandering could be useful, since these may co-occur within a single continuum of local sleep phenomena (Andrillon et al., 2019).

Direct facilitation of metacognition
While lucid dreams have sporadically been reported to occur also in NREM sleep, lucid dreaming largely considered a phenomenon bound REM sleep (Baird et al., 2019).This sparseness of lucid dreaming coincides with low levels of ACh during NREM sleep, which in turn might point to a directly facilitating role of ACh for lucid dreaming rather than a merely indirect role via REM sleep stabilization.However, on the specific mechanisms of such a direct lucidity-stimulating role of ACh can only be speculated.
AChEIs including galantamine are commonly prescribed in Alzheimer's Disease to treat cognitive impairments.According to the 'cholinergic' hypothesis of this disease, decreased cholinergic neurotransmission is involved in many of the cognitive symptoms observed in AD (Lleó, 2007).Besides memory, acetylcholine plays an important role in attention as shown by selective lesion studies of cholinergic input to the cortex, causing attentional impairments (Hasselmo and Sarter, 2011).Clinical trials of galantamine and other AChEIs, have shown small but significant improvements in cognitive performance, especially episodic memory, attentional functions and several behavioral symptoms of AD, including hallucinations and motor coordination (Lleó, 2007).In addition, Alzheimer's disease is characterized by unique and idiosyncratic failures in metacognition; resulting in the characteristic 'denial of symptoms' or anosognosia (McGlynn & Kaszniak, 1991) which increases in extremity as the disease itself progresses (Edmonds et al., 2018).Long thought to be driven by degeneration of the prefrontal cortex (Bertrand et al., 2018) and degeneration of nicotinic acetylcholine receptors brain-wide (Nykas et al, 2011) recent theoretical models into this disease have suggested a specific causal pathway between memory disruption and the emergence of metacognition (Morris & Hannesdottir, 2004;Morris & Mograbi, 2013;Sunderaraman & Cosentino, 2017) with a particular regard to the role of self-modelling in this process (Hallam et al., 2020).This presents a somewhat nuanced explanation for the role of Galantamine in metacognition.Metacognition failure in Alzheimer's has been shown to predominantly impact subjective (or consciously accessible) metacognition judgments, with implicit or 'unconscious' metacognitive processes largely in line with those of healthy subjects (Geurten, Salmon & Bastin, 2021).In other words, prefrontal and acetylcholinergic receptor destruction in this disease potentially contribute to a rather distinctive category of systems disorder: one which affects the emergence of metacognition as a distinct, reportable mental state; and not at the mechanistic or functional level.Given that lucid dreaming is the result of metacognitive judgement becoming available within conscious awareness per definition: the destruction of this particular subset of metacognition in Alzheimer's disease would seem remarkably similar to that which is restored in lucid dreams.It is therefore unsurprising that metacognition deficits in AD empirically correlate with lowered metabolic activity in several of the key regions associated with Lucid Dreaming (Mondragón et al., 2019).Furthermore, accuracy in meta-memory tasks has been shown to correlate with cortical thickness in midline structures, including the Dorsomedial Prefrontal Cortex, Insula Cortex and Precuneus (Sunderaraman & Cosentino, 2017; Bertrand et al., 2018)-overlapping with regions implicit in lucid dreaming (Dresler et al., 2012;Dresler et al., 2015;Baird et al., 2019).Taken together, it can be speculated that part of Galantamine's positive effects in Alzheimer's disease, but not in other memory-related diseases (Loy & Schneider, 2006;Birks & Craig, 2006;Auchus et al., 2007) could be due to this compound's explicit capability to restore metacognition and related processes such as metamemory.To which degree the suppression of these processes in both the disease and in REM sleep share biological overlap however remains to be investigated.While Galantamine does addresses the psychological symptoms of this disease, such as delusions and hallucinations (Kavanagh et al., 2011), there have been no studies conducted that explicitly investigate its relationship with denial of symptoms or related metacognitive measures.As such, this remains an open question in the field of Alzheimer's research, and a possible translational area between mechanisms of lucid dreaming induction and research into this disease.

Indirect stimulation of lucidity via prospective memory facilitation
The most prominent symptom of Alzheimer's Disease, for which cholingeric drugs including galantamine are prescribed, are memory impairments.In particular, multiple studies have reported that administration of the nAChR agonist nicotine improves prospective memory (Marchant et al., 2008, Rusted et al., 2009, Rusted et al., 2011).
The MILD technique for lucid dream induction crucially depends on prospective memory recall: it elicits lucidity by forming the strong intention to recognize any anomalies within the upcoming dream (LaBerge et al., 2018).Any enhancement of prospective memory abilities could thus be expected to increase the efficacy of the MILD technique.
However, in Alzheimer's Disease, galantamine is administered repeatedly over a longer period of time and at a much higher dose than in the lucid dreaming studies mentioned above.Moreover, studies of single dose AChEI effects on memory and attention have found inconsistent results, some of them reporting no memory enhancing effects at all (Repantis et al., 2010).In addition, it has been hypothesized that AChEIs memory enhancing effects may simply be based on increased arousal (Husain and Mehta, 2011).Therefore, the possibility of lucid dream induction via simple pro-cholinergic memory enhancing effects appears rather unlikely, and remains to be proven.

Conclusions and future directions
As a neuromodulator, acetylcholine is involved in a variety of cognitive functions.In the service of these functions, acetylcholine binds to multiple receptors with different kinetics, distributions, and effects-however, at present, the theoretical models that can bridge these fundamental mechanics to psychological and cognitive phenomena have much room for improvement.Administration of acetylcholinesterase inhibitors have shown promising results in producing specific cognitive changes, through induction of metacognitive states during dreaming.While these underlying mechanisms are still unknown, several hypotheses are available, and specific research into the effects of acetylcholine on dream lucidity could bridge this gap; and bring the well-established body of knowledge of these fundamental mechanism into contact with a rich body of theory which underpins the state of knowledge with regard to the neural correlates of lucid dreaming.Since these hypotheses remain to be tested, significant promise could be realized by further developing these theoretical models, and better integrating them with the state of knowledge of dream lucidity, with the aim of conducting an experimental research program into the effect of acetylcholine on metacognition during wakefulness and sleep.

Fig. 2 .
Fig. 2. Left: Hobson's AIM model of consciousness determined by the level of activation (A), input-output gating (I) and modulation (M) that form the axes.Waking (red circle) is characterized by external sensory input and motor output, high cortical activation and high aminergic activity.During NREM sleep (green circle) aminergic activity and external sensory input is decreased.Moreover, multiple cortical and subcortical areas are relatively deactivated in NREM sleep compared to waking.During REM sleep (blue circle) many of these areas are reactivated.External input remains suppressed and muscle atonia inhibits muscle output.Further, aminergic activity is shut off and cholinergic activity takes over.Lucid dreaming (purple circle) is considered a hybrid state between waking and non-lucid REM sleep, placed between both states of consciousness (adapted fromHobson, 2009).Right: An adapted version of the AIM model that takes the induction of lucid dreaming through cholinergic agents into account.In this model, lucid dreams are produced through downward (cholinergic) pressure upon the 'M' Axis, driving to its lowest possible position.Correspondingly, REM is lifted on the 'M' axis such that it no longer holds the lowest-possible position on this axis.After such adaptation, the proposed hybrid nature of lucid dreaming as a state between non-lucid REM sleep and wakefulness is less obvious.Considering recent findings on the possibility of two-way communication with the real world during lucid dreaming(Konkoly et al., 2021), lucid REM sleep arguably inhibits a more medium position on the 'I' axis of the model, though.(For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)