OUP user menu

Can we still dream when the mind is blank? Sleep and dream mentations in auto-activation deficit

Smaranda Leu-Semenescu, Ginevra Uguccioni, Jean-Louis Golmard, Virginie Czernecki, Jerome Yelnik, Bruno Dubois, Baudouin Forgeot d’Arc, David Grabli, Richard Levy, Isabelle Arnulf
DOI: http://dx.doi.org/10.1093/brain/awt229 awt229 First published online: 11 September 2013


Bilateral damage to the basal ganglia causes auto-activation deficit, a neuropsychological syndrome characterized by striking apathy, with a loss of self-driven behaviour that is partially reversible with external stimulation. Some patients with auto-activation deficit also experience a mental emptiness, which is defined as an absence of any self-reported thoughts. We asked whether this deficit in spontaneous activation of mental processing may be reversed during REM sleep, when dreaming activity is potentially elicited by bottom-up brainstem stimulation on the cortex. Sleep and video monitoring over two nights and cognitive tests were performed on 13 patients with auto-activation deficit secondary to bilateral striato-pallidal lesions and 13 healthy subjects. Dream mentations were collected from home diaries and after forced awakenings in non-REM and REM sleep. The home diaries were blindly analysed for length, complexity and bizarreness. A mental blank during wakefulness was complete in six patients and partial in one patient. Four (31%) patients with auto-activation deficit (versus 92% of control subjects) reported mentations when awakened from REM sleep, even when they demonstrated a mental blank during the daytime (n = 2). However, the patients’ dream reports were infrequent, short, devoid of any bizarre or emotional elements and tended to be less complex than the dream mentations of control subjects. The sleep duration, continuity and stages were similar between the groups, except for a striking absence of sleep spindles in 6 of 13 patients with auto-activation deficit, despite an intact thalamus. The presence of spontaneous dreams in REM sleep in the absence of thoughts during wakefulness in patients with auto-activation deficit supports the idea that simple dream imagery is generated by brainstem stimulation and is sent to the sensory cortex. However, the lack of complexity in these dream mentations suggests that the full dreaming process (scenario, emotions, etc.) require these sensations to be interpreted by higher-order cortical areas. The absence of sleep spindles in localized lesions in the basal ganglia highlights the role of the pallidum and striatum in spindling activity during non-REM sleep.

  • apathy
  • spindles
  • dream
  • basal ganglia, anoxic lesions


Bilateral damage to the basal ganglia (caused by anoxia, ischaemia, encephalitis or surgical procedures) may result in a major form of apathy characterized by a complete lack of self-initiated behaviour, whereas motor and cognitive abilities are preserved when externally driven (Laplane et al., 1984, 1989). This neuropsychological syndrome has been called auto-activation deficit (AAD) (Laplane and Dubois, 2001), although further cases have been termed ‘athymhormia’ ‘psychic akinesia’ and ‘reversible inertia’ (Luaute and Saladini, 2001; Habib, 2004). There is a sharp contrast between the dramatic quantitative reduction of self-generated actions and the normal production of behaviours in response to external solicitation. This striking syndrome has shed light on the role of the basal ganglia (specifically, the associative and limbic component of the globus pallidus) in driving motivated behaviours (Schmidt et al., 2008). When patients with AAD were asked about their thoughts, several patients reported that their mind was empty or blank, i.e. that they have no spontaneous activation of thoughts. It appears likely that the ‘auto-activation’ deficit results in a failure to reach a threshold of initiation/activation of thoughts or actions when subjects should behave on an internal basis and not in automatic response to perception (Levy and Dubois, 2006).

During sleep, the brain is operating on an exclusively internal basis. When awakened, most normal subjects remember some mentations that were associated with their previous sleep state. These sleep-associated mentations are defined as dream activity (Pagel et al., 2001). There are qualitative and quantitative differences between dreams obtained during rapid eye movement (REM) versus non-REM sleep, with more visual, emotional, complex and longer scenarios occurring in REM sleep dreams (Nir and Tononi, 2010). The physiology of dream activity during REM sleep has been characterized on the basis of its functional correlates in imaging and deep brain recordings (Pace-Schott, 2011). Several motor components of the frontal cortex are active, whereas the prefrontal and cingular areas are less active during REM sleep (Braun et al., 1998). Executive systems in the brainstem (pontine tegmentum) stimulate the limbic, paralimbic and sensory cortex during REM sleep (Maquet et al., 1996). It is speculated that these bottom-up stimulations result in emotional, visual, sensory and auditory internal activity, through ascending cholinergic projections from the pedunculopontine nucleus to the thalamus and then cortex in specific ponto-geniculo-occipital and ponto-geniculo-temporal waves. In contrast, alternative theories speculate that dreaming begins in the higher cortex structures (based on psychic motive for Freud and based on abstract knowledge and figurative thought for neurocognitive theories) and then proceeds backwards (translated into perceptions) as imagination proceeds during wakefulness (Nir and Tononi, 2010). These two theories (bottom-up and top-down) of brain activation during dreaming are not mutually exclusive and may represent different times in the dreaming processes. To test these theories, we asked whether the self-stimulation of the cortex by the brainstem is sufficient in patients with AAD to stimulate spontaneous mental content during sleep, although this characteristic is lost during wakefulness. In addition, the role of the basal ganglia in sleep mechanisms is unclear, with the exception of the thalamus (thalamic reticular nuclei generate sleep spindles in non-REM sleep and thalamic intralaminar nuclei relay stimulation from the brainstem to the cortex during REM sleep). We thus examined dream activity and sleep structure in patients with AAD.

Materials and methods


Thirteen consecutive patients with AAD with bilateral striato-pallidal lesions were recruited from the registries of the neurology department at a university hospital nationwide. All of the participants agreed to participate in the study. This series of patients is the same group of patients used in a previously published study investigating incentive motivation (Schmidt et al., 2008). The patients were compared with 13 non-medicated, paid healthy control subjects and were screened for neurological or psychiatric conditions and matched in age and gender. This study was approved by the local ethics committee (CCP Ile de France 06). All of the participants signed a written consent. Before their inclusion, a clinical examination, brain MRI and neuropsychological evaluation were performed on the patients. The inclusion criteria were: (i) presence of AAD; and (ii) exhibited bilateral lesions of the basal ganglia. Auto-activation deficit was defined according to the symptoms previously described by Laplane and Dubois (2001), including: (i) a deficit in the spontaneous activation of mental processing, as observed in behavioural, cognitive and affective domains; and (ii) the deficit can be partially reversed by external stimulation that activates normal patterns of response. In addition, patients with AAD may typically express the feeling that their mind is empty when they are not stimulated, but this symptom was optional during the inclusion process.


The subjects had a medical and psychological interview followed by a neurological examination. The patient, family and psychologist completed the motivation and action disorders rating scale (Habib, 1995), which contained subscores evaluating the severity of apragmatism (sum of the scores by the three scorers, 0–24), of affective indifference (0–24), loss of drive (0–24) and mental blank (0–3). The mental blank was scored as: 0 (normal, almost continuous mental activity); 1 (able to remain thoughtless for several seconds, more frequently than before); 2 (intense mental blank, the subject remains thoughtless for long moments, but still exhibits spontaneous activity even in the absence of external stimulation); and 3 (total mental blank, no thoughts except when an external stimulus or an interrogator provokes them). The apathy inventory was also performed (Starkstein et al., 1992). The presence of a depressive mood was assessed using the Montgomery and Asberg Depression Rating Scale (Montgomery and Asberg, 1979) and daytime sleepiness was self-evaluated using the Epworth sleepiness scale (Johns, 1994). The neuropsychological tests that could affect the dream report were the Mini-Mental State Examination (Folstein et al., 1975), Frontal Assessment Battery (Dubois et al., 2000), phonemic verbal fluency (Cardebat et al., 1990), and Free and Cued Selective Reminding Test which were used to evaluate free immediate recall and total delayed recall (performed only in patients) (Grober et al., 1988).

Sleep monitoring

Video and sleep monitoring were performed on the participants for two consecutive nights. The first night was used to evaluate the patient’s sleep duration, structure and microstructure. The monitoring included Fp1-Cz, O2-Cz, C3-A2 EEG, right and left electrooculogram, electromyography of the levator menti and tibialis anterior muscles, nasal pressure through a cannula, tracheal sounds through a microphone, thoracic and abdominal belts to assess the respiratory efforts, electrocardiography, pulse oximetry, and ambiance microphone. The sleep stages, arousals, alpha rhythms on EEG, respiratory events, periodic leg movements and muscle activities were scored by visual inspection according to standard criteria (Iber et al., 2007) by two experienced neurologists (S.L.S. and I.A.) with high inter-rater agreement. Sleep spindles were assessed by visual scoring and defined as a figure lasting 0.3 to 1 s, with a frequency between 12 and 15 Hz (De Gennaro and Ferrara, 2003). Sleep spindles were examined during the entire sleep monitoring and defined as being present or consistently absent. Next, the spindle index (number of spindles within 10 min of sleep stage 2) was manually quantified in artefact-free epochs within the first 10-min (equivalent to 20 epochs of 30 s) period of stage N2 in the second sleep cycle.

Dream collection and analysis

A collection of home dreams was obtained in both groups using written dream diaries during the first week. During the second night of sleep monitoring, the investigators collected the subjects’ mental contents after forced awakenings in various sleep stages. The subjects were systematically awakened once in non-REM sleep stage N2 (during the first sleep cycle of the night) and twice during REM sleep (during the second cycle after 5 min of REM sleep and during the third cycle after 10 min of REM sleep). The investigator would enter the room, call out the subject’s name until they positively responded and then asked, ‘Tell me everything that was going through your mind just before I called you.’ When the subjects finished their report, the investigator added, ‘Can you remember any more details?’ The response verbatim was recorded and videotaped.

The dream reports were independently analysed by two scorers (S.L.S. and G.U.), who were blinded to the patient/control status and to the sleep stage. The length of the reports was measured with the total recall count (Antrobus, 1983), which is the sum of all of the words that described the mentation before the awakening. It excludes commentary, associations with the mentation content, hesitations, and redundancies in the report. The inter-scorer agreement for the number of words was very high (intraclass correlation coefficient = 98.2%). All of the narrations were scored according to Orlinsky’s classification (Supplementary material) from 0 to 7 depending on the degree of elaboration and complexity (Oudiette et al., 2009), and to the bizarreness scale (Revonsuo and Salmivalli, 1995). This latter method consists of counting all of the expressions that describe 1 of 14 contents (self, place, time, persons, animals, body parts, plants, objects, events, actions, language, cognition, emotions and sensory experiences). Each element was then categorized as non-bizarre or bizarre. Three types of bizarreness were possible: (i) incongruous elements (which included internally distorted or contextually incongruous elements, exotic elements and impossible elements); (ii) vague elements; and (iii) discontinuous elements. The same two scorers independently assigned a compound score (content/bizarreness) to each separate identified element. Next, the scores were crosschecked and any inconsistencies were resolved by discussion between the judges. Using this method, 54 (4.2%) of the 1285 elements were excluded because no agreement on the scores could be reached between the judges. The final data therefore consisted of 1285 scored elements. The judges had independently scored 1231/1285 elements as belonging to the same content category and 169 as belonging to the same bizarreness category, thus the content and bizarreness agreements were 95.8% and 95.5%, respectively.

Locations of the brain lesions

Visual examination of the MRI (12 patients) and CT scans (one patient who could not undergo MRI because of a cardiac pacemaker) confirmed that all of the patients with AAD demonstrated bilateral basal ganglia damage. T1-weighted structural scans were acquired using a MRI scanner (GE Medical Systems) of 1.5 T for 10 patients with AAD and 3 T for the remaining three patients with AAD. Lesions of the 13 patients were manually segmented using MRIcro (Rorden, 1999–2005, Columbia, USA, http://www.sph.sc.edu/comd/rorden/mricro.html). Regions of interest corresponding to the segmented lesions were normalized to the Montreal Neurological Institute neuroanatomical space using Statistical Parametric Mapping (SPM5) software (Wellcome Trust Centre for NeuroImaging, London, UK), as previously described (Schmidt et al., 2008). An example (Patient 9) is shown in Fig. 1.

Figure 1

Axial 1.5 T MRI slice in Patient 9 showing the lesion centred on the external globus pallidus. The caudate nucleus, putamen and external globus pallidus are delineated in blue, the thalamus and reticular perithalamic nucleus in green, after registration of a 3D deformable atlas of the basal ganglia.

Statistical analysis

Because the sample size was small, the statistics described the median, upper and lower quartiles for quantitative variables, and numbers and percentages for qualitative variables. Relationships between the qualitative variables were tested using Fisher Exact tests, while the quantitative variables were compared using the unpaired Wilcoxon test. The inter-scorer agreement was estimated by the weighted Kappa coefficient for the Orlinsky scale and bizarreness scale and by the intraclass correlation coefficient for the number of words. All of the tests were two-sided and a P-value <0.05 considered statistically significant. The computations were performed using the SAS V9 statistical software programme.


Characteristics of the subjects

The demographic characteristics of the sample, findings on the neurological examination, presence of a mental blank at the time of the study, cause and location of the basal ganglia lesions on MRI of the patients with AAD are shown in Table 1. Registration with the canonical T1 template showed that the main maxima were located bilaterally in the caudate nucleus, putamen and pallidum. The lesions were located in the caudate nucleus (n = 7 patients), putamen (n = 6) and pallidum (n = 8), as well as in the thalamus (n = 2). The disease course ranged from 6 to 246 months at the time of study. Six patients had a mental blank score of 2 (almost complete), one patient had a partial mental blank (score of 1) and six patients had no mental blanks. The individual neuropsychological data are detailed in Supplementary Table 1. Compared with the matched controls (Table 2), the patients with AAD had lower scores on the Mini-Mental State Examination and Frontal Assessment Battery. The patients with AAD had higher scores than controls at the Montgomery and Asberg Depression Rating Scale but the scores were not pathological (except for two patients), in contrast to the moderate or major depressive features found during medical interview. The sleepiness scores were similar in the two groups. However, their free immediate recall memory was normal in two patients, moderately impaired in five patients and markedly impaired in six patients. The total delayed recall was normal in eight patients, mildly impaired in one patient, moderately impaired in two patients, and markedly impaired in two patients. The letter verbal fluency was normal in three patients, mildly impaired in two patients, moderately impaired in four patients, and markedly impaired in four patients.

View this table:
Table 1

Demographic, clinical and MRI measures of 13 patients with AAD

PatientAge (years)SexDiagnosticMotor signsMental blank* 0–3Basal ganglia lesion in MRI
P 127FViral encephalitisGeneralized dystonia0NoneR, LNoneNone
P 244MVasculitisNo focal deficit2NoneR, LRR, L
P 329FIschemiaMild right paresis0NoneNoneNoneR, L
P 436MCardiac arrestMild parkinsonism, generalized dystonia0NoneR, LR, LNone
P 557MCardiac arrestMild parkinsonism2NoneNoneR, LR, L
P 622MVasculitisTransient diplopia0NoneNoneR, LNone
P 725MCerebral anoxiaMild parkinsonism1NoneNoneNoneNone
P 844MCardiac arrestNo focal deficit2NoneNoneR, LNone
P 930FCardiac arrestPostural tremor2NoneNoneR, LR, L
P 1029MCardiac arrestGeneralized dystonia2No MRINo MRINo MRINo MRI
P 1157FStrokeRight upper limb dystonia2RR, LNoneR, L
P 1277MStrokeNo focal deficit0R, LR, LR, LR, L
P 1376MStrokeNo focal deficit0NoneR, LR, LR, L
  • *0 = normal, almost continuous mental activity; 1 = able to remain thoughtless for several seconds, more frequently than before; 2 = intense mental blank, remains thoughtless for long moments, but still exhibits spontaneous activity even in the absence of external stimulation; 3 = total mental blank, no thoughts except when an external stimulus or an interrogator provokes them. L = left; R = right.

View this table:
Table 2

Demographic and clinical characteristics in patients with AAD and healthy subjects (controls)

Patients with AADControlsP-value
Age36 (29; 57)34 (26; 57)0.53
Sex, number of males99
Mini-Mental State Examination, 0–3024 (21; 28)30 (30; 30)0.0007
Frontal Assessment Battery, 0–1813 (12; 14)17 (16; 18)0.0009
Starkstein Apathy Scale, 0–4222 (14.2; 29)10 (8; 12)0.0008
Montgomery and Asberg Depression Rating Scale, 0–607 (4; 13)0 (0; 2)0.005
Epworth Sleepiness Scale, 0–2411 (4; 15)7 (6; 9)0.45
  • Data are the median (lower quartile; upper quartile).

Sleep architecture

There was no difference between patients with AAD and control subjects regarding sleep measures (Table 3). However, despite all of the participants having normal alpha rhythms when awake, K complexes of usual shapes, duration and frequency during non-REM sleep stage N2 and preserved muscle atonia during REM sleep, 6 of 13 (46%) patients with AAD had a complete absence of sleep spindles (which is in sharp contrast with control subjects, in whom the spindles were always present) during non-REM sleep (Patients 2, 3, 4, 10, 12 and 13). These six patients had bilateral lesions in the striatum and pallidum, but lacked a common specific location (pallidum, putamen, caudate) or side. Interestingly, five of these six patients were free of thalamic lesions. When present, the spindle index was similar in patients with AAD and in controls. One patient who lacked thalamic lesions had a spindle frequency of 10.6 Hz, a frequency that was lower than the normal range (12–15 Hz) and the lowest spindle index of the group (seven spindles per 10 min of stage N2).

View this table:
Table 3

Sleep measures in AAD patients and healthy control subjects

Patients with AADControlsP-value
Total sleep period, min507 (467; 562)473 (459; 514)0.39
Total sleep time, min444 (379; 481)442 (403; 476)0.82
Sleep efficiency, %86.8 (77; 90.2)91.9 (87.4; 93)0.15
Latency to, min
    Sleep onset25 (14; 29)18 (9; 36)0.61
    REM sleep onset188 (98; 229)83 (74; 137)0.06
Sleep duration, % of total sleep time
    Stage N14.9 (4.3; 8.3)3.2 (2.9; 4.7)0.10
    Stage N253.3 (48.4; 61.2)56.2 (51.8; 57)0.98
    Stage N322.6 (20.8; 26.1)20.2 (17.1; 21.9)0.25
    REM sleep17.4 (11.6; 20.3)20.2 (15.7; 22)0.39
Sleep fragmentation, number of events/hour
    Arousals14.3 (8.8; 14.9)12.6 (8.8; 15.8)0.98
    Apnoea and hypopnoea2.1 (0.5; 10)1.9 (0.5; 2.4)0.40
    Periodic leg movements00 (0; 0.7)0.74
Sleep figures
    Occipital alpha waking rhythm, Hz9 (9; 10.6)10 (9.5; 10.6)0.25
    Presence of spindles, % of patients53.8100<0.0001
    Spindles index, no of spindles/10 min*51 (12; 60)54 (46; 76)0.33
    Presence of K complex, % of patients1001001
    Complete REM sleep atonia, % of patients1001001
  • Data are the median (lower quartile; upper quartile), *excluding the patients without any spindles.

Dream collection and analysis

During the general interview, two patients (one with and one without a mental blank) reported that they had stopped dreaming since the accident. When their dreams were collected at home, there was a sharp contrast between patients with AAD and control subjects, with a quasi-absence of any dream report written in the diaries of patients with AAD (Table 4). Only one patient with AAD (Patient 4, no mental blank; moderately impaired immediate memory) had a full dream diary, in which he reported poor and repetitive dream content (he observed the same person, an unknown girl) for five nights in a week. In contrast, all of the control subjects returned completed dream diaries with numerous dream reports. In dreams collected in the laboratory, there were 38 forced awakenings during sleep in patients with AAD and 30 in control subjects. Only one patient with AAD (Patient 5) reported a mental content when awakened in non-REM sleep stage N2 versus 7 of 13 of the control subjects. This patient said that he dreamt of ‘the millionth of a movie.’ Four (30%) patients with AAD and 12 (92%) control subjects reported at least one mental content when awakened in REM sleep. Among these four patients, two patients had a total mental blank during wakefulness. These four patients had moderately impaired (n = 2) and markedly impaired (n = 2) immediate memory, whereas the nine patients with AAD, who had no dream recall, exhibited normal (n = 2), moderately impaired (n = 3) and markedly impaired (n = 4) immediate memory. As for the four patients who reported dream recall, the verbal fluency was normal in one patient (Patient 3), moderately affected in one patient (Patient 6) and markedly affected in two patients (Patients 5 and 10). The remaining nine patients with AAD, who had no dream recall, had normal (n = 2), mildly impaired (n = 2), moderately impaired (n = 3) and markedly impaired (n = 2) verbal fluency.

View this table:
Table 4

Mental content obtained after awakening from during sleep in patients with AAD and healthy subjects

AAD patients n = 13Controls n = 13P-value
Home dream collection
    Total number of dreams0.57 (5; 9)0.0009
    Number of subject with dreams113
    Total recall count per subjecta80715 (210; 752)0.23
    Sleep mentation complexity 0–7b1.665 (3.28; 5.42)0.23
Sleep mentation bizarrenessc
    Number of elements (per subject)21129 (60; 178)0.0007
    Non-bizarre elements (Type 1)7683 (80; 91)0.41
    Bizarre elements2417 (9; 20)0.41
Laboratory-based dream collection
 Non-REM sleep stage N2
        Forced awakenings, n1313
        Patients with mental content, %7.753.80.01
        Total recall count, per subject102.5 (0; 9)0.44
        Sleep mentation complexity, 0–721 (1; 2)0.42
        Sleep mentation bizarrenessc
        Total number of elements, per subject33 (1; 5)1
        Non-bizarre elements, %100100 (75; 100)0.76
        Bizarre elements00 (0; 25)0.76
 REM sleep
        Forced awakenings, n2627
        Patients with mental content, %1957.70.02
        Total recall count, all subjectsa0 (0; 10)47 (9; 108)0.007
        Sleep mentation complexity, 0-7b24 (3; 4)0.09
Sleep mentation bizarrenessc
    Total number of elements per subject4 (2; 7)15 (10; 24)0.07
    Non-bizarre elements, % of total number100 (100; 100)87.5 (80; 91)0.15
    Bizarre elements, % of total number0 (0; 0)12.5 (9; 20)0.15
  • Data are the median (lower quartile; upper quartile).

  • aAntrobus (1983); bOudiette et al. (2009), cRevonsuo and Salmivalli (1995).

The mental contents obtained in REM sleep were rarer, shorter, tended to be less complex, and did not contain any bizarre elements in patients with AAD (no incongruity, no discontinuity). The response verbatim was as follows: (i) ‘I saw my niece, she didn’t said a word, she was sitting, she couldn’t speak. It took place in Ivory Coast, in the lounge, with her cousin. It was a bank holiday because I was seeing my nieces and nephews well dressed’ (Patient 3, no mental blank; moderate impaired memory); (ii) ‘I was writing’ (Patient 5, complete mental blank; markedly impaired memory); (iii) ‘I was going for a walk in Créteil’ (a town close to Paris), and ‘I was talking to somebody about my mother’ (Patient 6, no mental blank; moderately impaired memory); and (iv) ‘I was shaving’ (Patient 10, complete mental blank; markedly impaired memory). Importantly, Patient 10 dreamt of shaving while he never initiated this activity during the daytime without being motivated by his caregivers (because of major apathy and a mental blank), and would not shave by himself because of severe hand dystonia. Similarly, Patient 5 dreamt of writing, although he would never initiate any writing without being invited by his caregivers to do so (because of major apathy and a mental blank). The Orlinsky total score for dream complexity correlated mildly (r = 0.55; P = 0.052) with the Frontal Assessment Battery score. There were no significant differences between AAD dreamers and non-dreamers in terms of Mini-Mental State Examination [20 (18–24.5F [lower quartile-upper quartile]) versus 25 (24–29), P = 0.11], verbal fluency [7 (4–13) versus 9 (6–9), P = 0.76], immediate free recall [22.5 (15–27) versus 25 (15–26), P = 1], delayed free recall [7.5 (4.5–10.5) versus 8 (7–10), P = 0.94], scores at the Frontal Assessment Battery [11 (6–13) versus 13 (12–15), P = 0.14], at the Motivation and Action Disorders Rating Scale [16.5 (8–18) versus 11 (4–16), P = 0.33] and at the Motivation and Action Disorders Rating Scale [5.5 (2–10) versus 11 (7–16), P = 0.27]. Within the AAD subgroup, there were a similar proportion of dreamers with (33%) versus without (28%) mental blank. Within the subgroup of six patients with mental blank, the two dreamers had worse Mini-Mental State Examination, Frontal Assessment Battery and apathy scores, a reduced immediate and delayed memory, a worse verbal fluency, but were less sleepy and less depressed than the four non-dreamers (no valid statistics on small samples). There were no obvious differences in brain lesions, comorbidities and other neurological symptoms. In contrast, the control subjects reported frequent, long and complex dreams when awakened during REM sleep (Table 4).


One third of the patients with AAD reported dreams when awakened in REM sleep, even when they had an almost complete mental blank during the daytime. However, their dream reports were infrequent, short, devoid of any bizarre elements and tended to be less complex than the dream reports of control subjects. The sleep duration, continuity and stages were similar between the groups, with the exception of a striking absence of sleep spindles in 6 of 13 patients with AAD.

The generation of dreams by some patients with AAD was the most unexpected result in this study. This finding contrasts with the complete cessation of dreaming during REM sleep reported by patients after a prefrontal lobotomy (Jus et al., 1972). Importantly, the first patient described by Laplane et al. (1982) assured that he had no dream activity. It is all the more remarkable that the two patients with AAD who had a complete absence of self-initiated thoughts were able to recall mentations from REM sleep. Although limited to two patients, this unexpected result supports one influential hypothesis on dreams in REM sleep, which stipulates that dreams are generated through bottom-up processes (Hobson and McCarley, 1977). In this hypothesis, the dream scenario is initiated from activity in the lower-level sensory areas when stimulated by the brainstem REM sleep executive systems (mostly the ponto-geniculo-occipital waves) and are later interpreted in higher-order visual areas (Nir and Tononi, 2010). The brainstem simultaneously sends corollary discharges to the entire cortex (i.e. the visual, oculomotor and motor cortices), which ensures the timely coordination between the visual images, eye movements and body movements within the dream. One may therefore imagine that this bottom-up activation from the brainstem to the sensory cortices is sufficient to elicit at least some internal images and sounds, even when the self-generation of thoughts by the basal ganglia-prefrontal loop is deficient. The reverse, top-down theory on dreaming (dreaming originating from high-order cortex as imagination does) is not supported here, as patients with AAD who have a mental emptiness and no imagination during wakefulness do report some dream mentations upon emerging from sleep.

However, the mentations elicited from REM sleep in patients with AAD are rare and poor (in terms of words, complexity of the scenario and bizarreness) as well as devoid of any apparent emotional content. The paucity of dreams in AAD contrasts with the numerous vivid dreams (often associated with REM sleep behaviour disorder, a phenomenon not seen in the patients with AAD) reported by patients with Parkinson’s disease (Bugalho and Paiva, 2011) who have more limited damage within the basal ganglia but can suffer from apathy. There may be several reasons for this aspect in AAD. The patients may produce shorter dream reports because their memory and verbal fluency are impaired. The patients may also exhibit lack of motivation to provide an extensive rather than a brief report of their dreams. Thus, the lack of motivation and interest towards dreaming discriminates the dreamers from the non-dreamers (Schredl et al., 2003). However, we found no difference in the verbal fluency, immediate and delayed memory, cognitive and frontal scores between non-dreamers and dreamers with AAD. Similarly, lower frontal scores and Mini-Mental State Examination correlate with more aggressive dreams in Parkinson’s disease, but do not attenuate dreaming (Bugalho and Paiva, 2011). Plus, apathy in AAD by definition disappears when patients are stimulated, which was the case here as they were questioned immediately at sleep offset by the investigators. The brevity of the dream reports may also directly relate to the AAD process during REM sleep. A specific hypothesis regarding REM sleep dreams suggested by Roffwarg et al. (1966) and further developed (Nir and Tononi, 2010), states that the eventual development of dream imagery may involve a process by which the cortex ‘fits’ sensory images into discharge patterns of brainstem origin, which is established before the accumulation of sensory experience. The cortex may internally develop some modulating influence over these pontine discharges (whereas the basic discharge rhythm most likely has a brainstem genesis). Thus, the dream would truly appear to be born in the brainstem but clothed in the cortex (Roffwarg et al., 1966). If true, this means that in this study, in patients with AAD, some cortex activity is elicited by the brainstem but the cortex-to-cortex interactions, which are later necessary to produce complexity and synthesis leading to a ‘story’ in the dream, could be missing. This also suggests that damage to the basal ganglia affects this cortex-to-cortex activation, whereas it does not affect REM sleep (desynchronized EEG, saw-tooth waves, rapid eye movement density, muscle atonia). Although this is speculative, there is an example that cortico-cortical integration in patients with AAD may be impaired, as revealed during a motor task based on a money reward. Patients had a normal grip force and evaluated adequately the amount of money that they could win with their force, but they failed to increase their force in order to increase reward, showing that bilateral striato-pallidal damage specifically disconnects motor output from affective evaluation of potential rewards (Schmidt et al., 2008).

Interestingly, these rare dreams were obtained only after a forced awakening during REM sleep (and in one case during the non-REM sleep stage N2) in the sleep laboratory setting, but almost never occurred when a diary was collected from home. Because the opposite is usually observed (1-week home dream collections are larger than one night laboratory-based dream collections, as observed in controls; Weisz and Foulkes, 1970), we suspected that patients with AAD were too apathetic to spontaneously recall dreams at the awakening and even more when writing the dreams down without being strongly requested to do so. Thus, the patients were solicited by an investigator in the sleep laboratory to elicit a dream report, and this was obtained in only four of the patients. Because AAD is acquired after an accident, all of the patients knew what a dream was and experienced dreaming before the accident. Two of the patients were aware that they had lost their ability to dream since the accident, which was verified during the home and laboratory dream collections. Thus, the high proportion of non-dreamers in the AAD group, including those during forced awakenings, was not a consequence of an inability to self-represent the dream.

The question arises whether this major difference in dream quantity and quality between patients with AAD and controls is linked to different types of sleep. Interestingly, there were no differences observed between groups of general sleep architecture and REM sleep duration as well as the usual REM sleep markers. This normal sleep structure in subjects with a major, bilateral lesion of the basal ganglia (mostly the caudate nucleus, but sparing the thalami, except in two patients) suggests that the basal ganglia do not play a major role in the generation of most sleep features, except for spindles. Data on the role of the caudate nucleus in sleep are scarce. Neurons in the globus pallidus internal have a regular discharge pattern across the sleep-wake stage (including REM sleep) in rat, suggesting that they are not affected by sleep stages (Urbain et al., 2000). However, spindles were lacking in 6 of 13 patients with AAD. This absence of spindles cannot explain the reduction in dream reports during REM sleep, as sleep spindles occur only during non-REM sleep. However, it may affect N2-associated mentations, as almost nothing was obtained in patients with AAD (only one patient with AAD was reported to have dreamt about ‘the millionth of a movie’ without any further development), whereas more than half of the control subjects recalled some short, non-complex and non-bizarre mentations, which is very similar to the description of N2-associated mentations collected in a normal series (Foulkes, 1962). Sleep spindles are a distinctive EEG phasic feature of non-REM sleep. They originate from the thalamic reticular nuclei and project to multiple neocortical regions and less frequently to the parahippocampal gyrus and hippocampus (De Gennaro and Ferrara, 2003; Andrillon et al., 2011). An isolated suppression of spindles was found in patients with degenerative, vascular or surgical lesions of the thalamus. Spindles may be reduced but not totally suppressed in degenerative diseases affecting the basal ganglia including Parkinson’s disease (Arnulf et al., 2000) and progressive supranuclear palsy (Arnulf et al., 2005). Interestingly, in our study, parametric mapping failed to show thalamic involvement, except in one patient who had a bilateral thalamic lesion and no spindles, while another had a right thalamic lesion and normal spindling activity. This result suggests that the thalamocortical networks involved in the genesis of the spindles require intact striatal and pallidal function. When present in AAD, the spindle density was similar to the density observed in the control subjects, suggesting that lesions in the basal ganglia operate as an on/off mechanism on sleep spindles, but does not influence their frequency.

Taken together, this collection of dream and sleep characteristics in a limited sample of patients with a rare, but fascinating neuropsychological syndrome, supports the hypothesis that dream sensations are generated by bottom-up brainstem stimulation to the sensory cortex, whereas the full dreaming process (scenario and complexity) requires higher-order cortical areas. This study also newly highlights the role of the pallidum and striatum in spindling activity during non-REM sleep.


Study funded by the Pitié-Salpêtrière University-Hospital Institute of Neurosciences (IHU de neurosciences).

Supplementary material

Supplementary material is available at Brain online.

auto-activation deficit


View Abstract