Combined facsimile of scans in nine cases with global cessation of dreaming caused by deep frontal lesions, illustrating the strong involvement of the white matter surrounding the frontal horns of the lateral ventricles 
Combined facsimile of scans in 14 cases with preserved dreaming with bifrontal lesions, illustrating the relative preponderance of cortical convexity involvement 
Global cessation of dreaming is associated with subcortical lesions located in the deep frontal white matter (areas F09 and F14 in the classification of Damasio and Damasio , shown here)
It is noteworthy that the psychotropic medications that replaced prefrontal leucotomy as the treatment of choice for psychotic disorders, block dopamine (DA) transmission in a mesial forebrain pathway that projects primarily to the nucleus accumbens. Probably related to this is the observation that both prefrontal leucotomy in general and cessation of dreaming in particular, due to lesions in this general area, are associated with reduced motivational incentive  as indeed are most antipsychotic medications . Also of interest in this connection is the observation by Piehler  and Schindler  to the effect that early recovery of dreaming after prefrontal leucotomy typically coincided with psychiatric relapse, suggesting that absence of dreaming could serve as an index of the clinical success of the operation. Dreaming is, after all, a psychotic state.
Effects of Pontine Brainstem Lesions
Cessation of dreaming following circumscribed pontine lesions—with or without cessation of REM sleep—has never been demonstrated (see Solms [6, 12] for reviews), despite the longstanding assumption that dreaming is caused by—if not identical with—the cyclical, spontaneous activation of cholinergic (ACh) cells in the mesopontine tegmentum during the REM state, together with reciprocal inhibition of serotonergic (5HT) and noradrenergic (NA) cells in the dorsal raphe and locus coeruleus complex [66, 67]. Consciousness in general is of course frequently compromised by pontine lesions, but at least eight cases with cessation or near-cessation of REM sleep have been reported in which patients were capable of communicating meaningfully about their dreams [68–71]. Indeed, one such patient did actually report loss of dreaming , but the lesion—caused by ruptured traumatic aneurysm of the basilar artery—almost certainly extended beyond the pontine brainstem and included the visual–spatial cortical areas discussed above. Even this isolated case therefore does not support the old equation of pontine brainstem mechanisms with dream generation. (The relationship between dreaming and REM sleep is discussed further below.)
Neuroimaging and Transcranial Magnetic Stimulation Studies Related to Dreaming
Neuroimaging studies have determined patterns of regional brain activation and deactivation during REM sleep, the stage of sleep during which dream reports are most frequently obtained . These patterns of activity are highly consistent with those areas of the brain linked with dreaming by the clinical lesions studies reviewed above. Significant increases in regional brain activity have been observed in the basal forebrain and other limbic and paralimbic structures, including the hippocampal complex, the anterior cingulate cortex and the pontine tegmentum, during REM sleep [73–75].
Furthermore, Braun et al.  reported a dissociated pattern of activity between visual association areas (extrastriate cortices—fusiform, inferotemporal, and ventral lateral occipital) and primary visual areas (striate cortices) during REM sleep compared with slow-wave sleep. Activation within the visual association cortices was also shown to correlate positively with activity within parahippocampal gyri and contiguous portions of the hippocampus, and with deactivation of dorsolateral and orbital prefrontal association areas. Based on these findings, the authors concluded that ‘during REM sleep, the extrastriate cortices and paralimbic areas to which they project may be operating as a closed system, functionally disconnected from frontal regions in which the highest order integration of visual information takes place. Such a dissociation could explain many of the experiential features of dreams’ (p. 94). This notion of a closed loop between certain medial forebrain and limbic regions and higher order visual association areas is consistent with Solms’s [6, 12] suggestion that dreaming is a product of deep frontal structures activating higher visual association areas during sleep, instead of the executive and motor areas they activate during waking, thereby generating imaginary (versus real) action.
A recent study, using [15O]H2O PET in healthy subjects with habitually high and low dream recall frequencies, also showed that high dream recallers—compared with low dream recallers—had greater regional cerebral blood flow in the temporal–parietal junction during slow wave sleep, REM sleep, and waking, as well as greater regional cerebral blood flow in the medial prefrontal cortex during REM sleep and waking . These brain regions are highly consistent with those identified by lesion studies as being critically related to the dream process. As the temporal–parietal junction may facilitate the orientation of attention during sleep to external stimuli, it has been argued that the increase in activation of this region may be responsible for the observed increase in intrasleep wakefulness in high dream recallers. According to the arousal–retrieval model of dreaming , intrasleep wakefulness may facilitate the encoding of the dream content into long-term memory, consequently facilitating dream recall upon awakening in high dream recallers.
Consistent with this explanation, high dream recallers are more reactive to their external environments during all stages of sleep, as well as during wakefulness, when compared with low dream recallers . (Using a novelty oddball paradigm, high dream recallers were shown to have enhanced P3a and late latency event-related potentials (ERPs) to novel and unexpected auditory stimuli. The P3a and late latency potentials are associated with complex cognitive processes such as familiarity, episodic memory and emotional processing [80–82]. Eichenlaub et al.  have argued that these robust differences in brain responsivity show that the cerebral organisation of high recallers is intrinsically different to that of low recallers and that this difference may potentially facilitate either production or encoding of the dream.
This propensity to be more reactive to the external environment is consistent with enhanced activation of the posterior attentional networks in high dream recallers. Whether increased dream recall in habitually high dream recallers is due to more efficient encoding of the memory trace of the dream into long-term memory, or whether activity within these regions indicates a genuine increase in dream activity, remains inconclusive.
However, the latter possibility is supported by a recent set of transcranial magnetic stimulation (TMS) studies which found that visual dream imagery could be enhanced by inhibiting certain frontal regions while stimulating the right posterior parietal cortex, during stage 2 sleep but not during slow-wave sleep [83, 84]. This finding must be understood in relation to the fact that TMS is unable to propagate through connected networks during slow-wave sleep as efficiently as during lighter NREM sleep (Massimini et al. 2005) . It nevertheless corroborates the notion that activity in posterior association cortex is responsible for the perceptual construction of dreams and moreover that activity in these regions during sleep represents an increase in dream activity.
Despite strong evidence that higher visual association cortices are responsible for the perceptual construction of dreams, it has recently been shown that primary visual areas may also be involved. In an innovative combined EEG/fMRI study, participants’ dream imagery (as verbally reported) was decoded from neural activity measured during sleep onset, by software trained to correlate discrete visual stimuli (i.e. pictures) with brain activity during waking . Horikawa et al.  concluded that the ‘principle of perceptual equivalence, which postulates a common neural substrate for perception and imagery, generalizes to spontaneously generated visual experience during sleep’ (p. 642). As lesions to primary visual regions do not result in dream loss, or any visual disturbances in dreams , the exact contribution of the primary visual regions to dream imagery remains unclear.
Excesses of Dreaming
Solms  loosely grouped together 12 case reports in the literature and 10 of his own cases under the heading of dream/reality confusion (or ‘anoneirognosis’). These patients reported excesses of dreaming, ranging from increased frequency and/or vivacity of dreams to intrusions of dreaming and dream-like thinking into waking cognition. The principle justification for collecting these cases under a unitary nosological heading was that the focal lesions (representing a wide variety of pathologies) that cause ‘anoneirognosis’ were typically located in the transitional zone between the anterior diencephalon and basal forebrain. Kindred phenomena are, however, also observed with visual de-afferentation, peduncular hallucinosis, delirium, parkinsonian syndromes, Guillain–Barre syndrome and a variety of toxic and metabolic conditions. The common denominator in these cases may therefore simply be degradation of constraints on consciousness. Certainly, any suggestion at this stage that dream–reality confusion may be considered to be a focal symptom is unjustified.
Dream/reality confusion in parkinsonian syndromes is difficult to interpret. Increased dreaming and hallucinations are frequently seen with Parkinson’s disease (PD) but this may be iatrogenic. It is well established that hallucinations and excessive dreaming can be provoked by the administration of l-dopa, both in PD  (and in normal subjects, independently of any concomitant changes in REM sleep . Accordingly, it has been shown that reduction of dopaminergic medication, and administration of dopamine blockers, reduces hallucinations and excessive dreaming in PD . However, visual hallucinations in PD may also be an indication of the presence of Lewy body pathology, with involvement of parieto-occipital and limbic regions [90–92]. Excessive dreaming in parkinsonian syndromes may, therefore, have a different mechanism in cases with and without cortical Lewy bodies. In PD, hallucinations occur late in the course of the disorder, whereas they are an early feature of dementia with Lewy bodies (DLB).
Hallucinations and dream/reality confusion are also common in narcolepsy . In these cases, hallucinations may accompany or follow attacks of cataplexy and sleep paralysis. Hallucinations of a presence of someone nearby (‘sensed presence’) or a pressure on the chest with breathing difficulties (‘incubus/succubus’), and floating/flying and ‘out of body’ experiences, are typical in these cases. Dreams can occur at sleep onset (at night or during daytime naps) as well as on awakening (Rosenthal’s syndrome). The retention of elements of normal waking mentation, such as volitional control or environmental awareness, is characteristic of narcoleptic dreams.
Various other rare disorders are associated with dream/reality confusion. Idiopathic hypersomnia manifests in excessive daytime sleepiness, prolonged unrefreshing sleep and ‘sleep drunkenness’ on attempting to wake up. Habitual dreaming, hypnagogic hallucinations and sleep paralysis are common in these cases . Kleine–Levin syndrome is a rare disorder characterised by recurrent episodes of hypersomnia, compulsive eating behaviour and various psychopathological changes like hypersexuality, irritability or apathy. Hallucinations, delusions and ‘dreamy states’ are reported in 14–24 % of patients with KLS . In fatal familial insomnia, a variant of Creutzfeldt–Jakob disease (CJD), progressive insomnia is coupled with an oneiric stuporous state in which patients perform complex, jerky movements that correspond to dream content which patients are later able to report . Dream/reality confusion with hallucinations also occurs in sporadic CJD .
In populations without neurological disorders, excessive dreaming has been reported as a primary complaint in certain sleep clinics . A study comparing these patients to controls revealed that complaints of excessive dreaming were related to significant microstructural changes, including increased arousals, intrasleep awakenings, period leg movements, alpha–delta sleep and REM density; however, no macrostructural changes were noted, and no differences in REM sleep and sleep onset latencies were apparent. Excessive dreamers were also found to be significantly more stressed, fatigued and anxious than controls, and to have more headaches.
Nocturnal seizures (and complex partial seizures in particular) sometimes present as recurring nightmares [97, 98]. Solms  identified 24 cases of this type in the literature and nine in his own series. Of theoretical interest is the fact that such nightmares typically occur during non-REM sleep. The content of the nightmares frequently coincides with that of the patient’s typical aura or ‘dreamy state’ seizures [12, 98, 99]. Penfield was able to artificially generate a waking aura resembling the recurring nightmare in one case by stimulating exposed cortex in the region of the epileptogenic focus [100–102]. Successful pharmacological or surgical treatment of the seizure disorder invariably results in disappearance of the recurring nightmares. These facts further support the interpretation of the nightmares in these cases as seizure equivalents (and indeed as non-REM phenomena).
As with dream/reality confusion (which frequently co-occurs with nightmares), increased frequency of nightmares is associated with a wide range of toxic and withdrawal states and metabolic abnormalities. The grounds for detaching these two ‘excesses of dreaming’ from each other are not entirely clear. The common denominator here may therefore, once again, simply be general degradation of constraints on consciousness.
REM Behaviour Disorder (RBD)
In this disorder, dreamed behaviours are physically acted out. This is due to disruption of pontomedullary mechanisms that induce REM atonia . The enacted behaviours may be dramatic or even violent, and usually relate to vivid, frightening dreams. A fair proportion of cases injure their bed partners . The disorder is most common in males, and onset is often in the sixth or seventh decade. RBD manifests mainly in the second-half of the sleep cycle (where REM is predominant). Increased slow-wave sleep, and increased periodic limb movements across all sleep stages are also seen [107–111].
Of special interest is the association of RBD with the parkinsonian syndromes. The presence of RBD in PD patients is associated with cognitive deficits and appears to predict dementia [112, 113]. Disorders with Lewy body pathology often involve RBD. The incidence of RBD in PD is 25–50 % and more than 50 % in DLB and MSA. In contrast, disorders without Lewy bodies rarely involve RBD. Notably, idiopathic RBD may present many years prior to the other symptoms of an incipient parkinsonian syndrome [107, 113–118]. The prognostic significance of RBD as a precursor to PD, DLB and MSA is now well established, resulting in the suggestion that the term ‘cryptogenic’ RBD should replace ‘idiopathic’ RBD [109, 111, 119, 120].
The chemical and pharmacological evidence is extremely difficult to interpret. This is due partly to the dynamic interactions that characterise neurotransmitter systems, and the paucity of rigorous pharmacological studies . Mention will only be made here of recent findings which seem particularly relevant to understanding dream generation, and the distinction between dreaming and REM sleep.
The neurochemical signature of the REM state is well established: namely, autochthonous activation of ascending pontine ACh cells—which is thought to produce characteristic pontine–geniculate–occipital (PGO) waves—and reciprocal inhibition of pontine aminergic (5HT and NA) cells—which is thought to demodulate the dreaming forebrain . Equally well established is the fact that non-REM sleep has the opposite pattern. Less widely known is the fact that, unlike other aminergic brainstem cells, the source cells in the ventral tegmental area (VTA) of the mesocortical DA pathway described above in connection with prefrontal leucotomy continue to fire at equal rates during sleeping and waking [121, 122]. These cells also fire with greater interspike variability during REM than non-REM sleep . This has recently been shown to indicate prominent burst activity in the REM state , resulting in greater terminal DA release. DA delivery to the nucleus accumbens is in fact maximal during REM sleep when compared with NREM sleep and waking .
The REM state is also characterised by minimal prefrontal glutamate release , which presumably coincides with the observation reported above to the effect that dorsolateral prefrontal lesions have no obvious effect on dream content (and with the observation that this region is strongly deactivated in PET imaging studies of REM sleep ). The chemical signature of the REM state, as regards the neurotransmitter interactions underlying the observed regional patterns of forebrain activation and deactivation, is certainly more complex than was previously assumed .
This complexity is underscored by the impenetrable thicket of psychopharmacological evidence. Of particular value is any evidence that could clarify the pathophysiology of dream cessation following deep ventromesial frontal lesions. Since the sleep cycle is unaffected by such lesions , it is reasonable to assume that they impair a mechanism which is specific to dream generation (as opposed to REM generation). Two competing hypotheses have been advanced to account for dream cessation following deep ventromesial frontal lesions (and the commensurate hyperactivation of this region in fMRI and PET imaging of dreaming sleep and schizophrenic hallucinations [72, 125, 126]). The first hypothesis is that it reflects the activation of ACh cells in the basal forebrain; the second is that it reflects the activation of DA cells in the VTA.
Against the former hypothesis is the observation that ACh antagonists (like scopolamine), rather than suppressing dreaming and dream-like thinking, have the opposite effect: they produce dream/reality confusion [127, 128]. In fact, in this respect, anticholinergic drugs mirror the effects of lesions in cholinergic basal forebrain nuclei . These and other considerations led Braun  to observe that activation of these nuclei during REM sleep may actually reflect inhibition of forebrain ACh in dreaming sleep.
In favour of the latter hypothesis is the observation that DA agonists (like l-dopa) increase dream bizarreness, vivacity, complexity and emotionality without having any commensurate effects on REM sleep . DA agonists, of course, also provoke other symptoms of psychotic cognition. Systematic studies of the effects on dreaming of DA antagonists have not yet been performed. However, a preliminary study by Yu (unpublished observation) of the effects on dreaming of antipsychotic medications recently found significant dream-suppressing effects.
Particularly incompatible with the view that dreaming and REM sleep are generated by the same pontine mechanisms is the accumulating evidence to the effect that 5HT agonists (SSRIs), like anticholinergics, have the opposite effect to what the REM-dreaming hypothesis would have predicted. SSRIs suppress REM sleep but produce excesses of dreaming, of both types described above [131–137].
The available pharmacological evidence therefore supports the view that dreaming—like other forms of psychosis—is primarily generated by (demodulated) DA mechanisms rather than ACh ones . It is likely that interactions between the DA system and other neurotransmitter systems also affect dreaming, as enhanced dreaming has been shown in populations receiving noradrenergic beta blockers. However, the neurochemical basis of dreaming is likely to be far more complex than this, the only conclusion that the limited current evidence reasonably allows. Nevertheless, the roles of reward processing and addiction, as they relate to DA activity in the brain, have more recently been used to try and elucidate the dream process.
Reward and Motivational Processes During Sleep and Dreaming
DA circuitry is also central to reward processing. The mesocortical–mesolimbic (MC-ML) DA system is defined as the ‘system [that] is formed by dopamine neurons located in the ventral tegmental area … which project to the nucleus accumbens, prefrontal cortex, septum, amygdala, and hippocampus’ . This system has been termed the SEEKING system by Panksepp  and it is thought to ‘drive and energize many mental complexities that humans experience as persistent feelings of interest, curiosity, sensation seeking’ (p. 145). It is also involved in reward processing, which refers to ‘an instinctual affective and exploratory drive to seek biologically-important stimuli in the external or internal (‘intrapsychic’) environment’ (Perogamvros and Schwartz, p. 1936) .
Perogamvros and Schwartz  have proposed the reward activation model (RAM) of dreaming, which postulates that reward processing during sleep may contribute to the consolidation of memories with a high motivational/emotional relevance, as well as aid in the modulation of REM sleep through projections to REM generating brainstem structures. Due to the strong interconnections between the hippocampus and the ventral tegmental area—which drives DA activity in the MC-ML circuits—RAM proposes that activation of the hippocampus during sleep may stimulate the VTA and lead to reward activation during sleep; in turn, VTA activity can lead to the reactivation of certain memories in the hippocampus. It is thought that during SWS in particular, the reactivation of the ventral striatum and the hippocampal complex allows for the consolidation of ‘memory-reward associations’ . Therefore—and as originally proposed by Freud —‘the fabric of the dream-thoughts is respolved into its raw material’ (p. 543).
In support of motivational theories of dreaming are studies of addiction, which has been associated with activity in the MC-ML dopamine system . It is common for abstinent drug addicts to have increased dream content related to finding and taking drugs, a phenomenon that Johnson  first termed ‘drug dreams’. As many as 80 % of acutely abstinent drug addicts experience drug dreams , which is related to drug craving in addicts and previous drug users [146–149]. Drug craving can be triggered by cues related to drug taking, both conscious and unconscious . These types of drug-related cues have also been shown to be associated with the subsequent occurrence of drug dreams [151, 152]. In the absence of these drug-related cues, drug dreams tend to dissipate . Drug dreams persist in drug addicts undergoing pharmacological treatment during abstinence, such as methadone treatment in heroin addicts and nicotine gum in smokers , (indicating that drug dreams do not merely result from physical withdrawal, but rather are a type of psychological withdrawal . Furthermore, up-regulation of the MC-ML dopamine system—as measured indirectly by the Limbic System Activity Scale—has been associated with drug dreams . The association between this up-regulation of the MC-ML dopamine system and the presence of drug dreams provides further evidence for this system’s role in the motivational aspects of dream genesis . This line of research has been used to argue that not only is the DA system involved in dream genesis, but that the content of drug dreams favours a motivational impetus for dream genesis [144, 155].