go to content


P. INDURSKY AND V.S. ROTENBERG vadir@post.tau.ac.il
Abarbanel Mental Health Center, Tel-Aviv University, Israel
Correspondence Address
Dr Vadim S Rotenberg, MD, PhD, DSc. 15 Keren Kayemet Street, Bat-Yam, Israel Fax +972 3 6589952

International Journal of Psychiatry in Clinical Practice 1998 Volume 2 Pages 47-51
Received 13 June 1997, accepted for publication 9 September 1997

Polysomnography was performed during two consecutive nights in 23 patients with major depression. After every final awakening patients estimated the change of their mood from evening to morning: I. Mood worse in the morning than in the evening; 2. Mood does not change; 3. Mood better in the morning. When mood was estimated as being better in the morning (20% of all nights), eye movement density in REM sleep increased from the first to the fourth cycle. In all other nights eye movement density was slightly higher in the first than in subsequent cycles. Mood improvement correlated positively with eye movement density in the fourth cycle and negatively with REM sleep duration in the first cycle. Eye movement density in the first cycle correlated positively with the subjective estimation of sleep latency. A possible functional difference between initial and final REM sleep periods is proposed. (International J. Psychiatry in Clinical Practice, 1998, 2: 47-51)

Keywords: mood, eye movements, REM sleep, depression, sleep cycles


The systematic investigations performed by Kramer and his co-workers(1) have shown that dreaming serves as a selective affective regulator, changing the subject's mood in a positive direction. The dream mechanism is especially effective in the relief of a state of subjective unhappiness. According to Kramer, this integrative capacity of dreaming is not solely related to the content of dreams. For instance, manifest dream content may differ both during the same night and from night to night, while the dream's positive influence on mood in healthy subjects is still apparent. Although dream content differs in men and women, the change in mood is independent of gender(1).

Kramer did not investigate the possible role of the objective sleep variables in mood restoration Although it is well known that dreams are reported different sleep stages, not only in REM sleep(2), the physiological variables of REM sleep correspond with the typical dream experience better than any other objective sleep variables(3; 4). In a previous investigation (5) we have shown that the active participation of healthy subjects in their own dreams correlates positively with eye movement (EM) density in REM sleep. At the same time, in a patient with cyclic mood disorder, the shifts to the elevated mood were accompanied by increased EM density the last cycles, while shifts to the depressed mood were associated with decreased EM density in the second half of the night (6). According to these data, it would be possible to suggest that the elevation of EM density might relate to shifts of mood in a positive direction in depressed patients. However, it is well known that in depressed subjects, EM density is increased, in comparison to normal controls, especially in the first cycle(7; 8). This contradiction can be solved if there is a difference between the functional meaning of the first and the last REM sleep periods in mood regulation. In the present report we describe our investigation of this hypothesis.


Twenty-three patients (13 men and 10 women) with major depression were investigated clinically and in the sleep laboratory. The mean age of patients was 51. 5 13. 2 years The clinical diagnosis of major depression was based on a structured clinical interview and was established according to DSM-IV criteria. The 21-item Hamilton Rating Scale for depression (9) was used on the day before the first polysomnography, to score the severity of the present depressive episode. The mean score of the Hamilton Rating Scale was 32. 5 9 4 for the group of patients. The mean number of previous episodes of depression was 4. According to the results of the routine clinical investigation (including routine blood test and ECG) all patients were physically healthy. By clinical interview, we also excluded other diagnoses associated with altered sleep (alcoholism, apnea, myoclonus etc). Patients were free from psycho-tropic medication for at least 10 days preceding the investigation. Data were collected from all patients on two consecutive nights in the sleep laboratory. Lights were usually turned off between 10. 00 and 10. 30 pm and turned on between 6. 00 and 6. 30 am (according to the ordinary sleep/wake habits in the ward). During the day, patients were under the supervision of the staff in order to exclude napping. Polysomnography was performed using the electroencephalograph Neurofax EEG-4400, EEG and EOG were recorded at a low filter setting of 0. 3 Hz. Eye movement recording was performed by using two channels. Poly-somnograms were analyzed according to international criteria (10). Sleep latency was defined as the first minute of stage 2 followed by at least 10 minutes of stage 2, or slow-wave sleep not interrupted by awakenings or stage 1 for more than 2 minutes. REM sleep latency was defined as the time from the onset of sleep until the onset of the first REM period, including intervening time awake. EM in REM sleep were counted visually, the criterion for EM detection was a minimum of 25 mV,(11) and we counted all simultaneous pen excursions on both channels. Artifacts caused by body movements and blinks have been excluded from the analysis of EM. Slow semicircular eye movements have also been excluded from this analysis. EM density was counted per minute of stage REM, separately in every cycle.

Every patient, 10 minutes after the final (induced or spontaneous) awakening in the laboratory, was requested to answer questions concerning subjective feelings about sleep duration, sleep latency, number of awakenings, sleep depth and the state after sleep. The question about mood after sleep had three possible answers: 1. mood in the morning is worse than in the evening, 2. mood does not change from the evening to morning, 3. mood in the morning is better than m the evening. The task of the present pilot investigation was to compare a generalized subjective estimation of mood before and after sleep, and to investigate possible relationships of such estimation of mood shift with the objective sleep variables. We were not interested in the precise evaluation of mood separately in the evening and in the morning, but only in the subjective feeling as to whether mood, on the whole, had changed after sleep and in what direction.

Sleep structure and EM density were compared for nights which were followed by different types of reports about mood. In addition, correlations between objective sleep variables and subjective sleep estimation and estimation of mood have been performed for all nights and for the several groups of nights selected according to the peculiar features of sleep structure (% REM high (> 28) vs. low (<15), predomination of the slow-wave sleep in the first vs. the second part of the night, long vs. short first REM sleep period, etc), or according to the subjective estimation of sleep (sleep is deep or superficial in the first part of the night, subject felt refreshed or not refreshed after sleep, etc). We have used subjective-objective correlations in different selected subgroups in order to check the stability of such correlations.


One-way analysis of variance (ANOVA) was used for sleep variables for three groups of nights (according to mood estimation). If the ANOVA was found to be significant at the 0. 05 level, it was followed by the paired Student's t-test in order to compare sleep variables in different groups. Product - moment correlation was used for the estimation of relationships between subjective sleep/mood estimation and EM density. In order to perform such correlations categorical sleep ratings have been ranged.


Forty-five polysomnograms of nights accompanied by the definite reports of mood state were analyzed in this investigation. The difference in reports could not be attributed to differences in clinical state between subjects or between the first and the second night, as the same patients presented different reports for different nights, and there were no relationships between reports and the night order.

Table 1. represents sleep structure for three groups of nights according to the subjective reports about mood. Figure 1 represents the change of EM density from cycle to cycle in these groups of nights. On nights which correspond to stable negative mood or to deterioration of mood (answer 1 or answer 2), EM density was highest in the first cycle and slowly decreased in the subsequent cycles. On nights accompanied by mood improvement, EM was lowest in the first cycle and increased dramatically in the following cycles. Mood was estimated as better in the morning in 20% of all nights.

Figure 1.Dynamic of REM density in night sleep cycles

The morning mood in depression: groups: A - better, B - without change, C- worse.

Table 1.Sleep variables in depressed groups with different mood dynamics (all nights)
Mood groups Worse No change Better
Nights 18 18 9
Total sleep time, min 326 (121) 374 (80) 306 (95)
Sleep efficiency, % 89 (10) 94 (8)** 84 (10)**
Sleep latency, min 35 (67) 10 (13) 30 (37)
Slow wave sleep, % 11.5 (11.7) 8.6 (9.1) 6.3 (6.1)
SWS, 1st cycle, min 16.1 (26.7) 9.9 (11.4)* 2.9 (6.1)*
SWS, 2nd cycle, min 14.1 (16.0)* 3.5 (5. 8)** 12.6 (19.2)
SWS, 3rd cycle, min 3.9 (4.3) 8.8 (21.1) 7.2 (10.1)
SWS, 4th cycle, min 3.1 (5.4) 4.6 (10.4) 1.1(1.9)
REM sleep latency, min 73.8 (79.9) 100.1 (68.4) 44.0 (47.0)
REM sleep, % 25.2 (13.8) 23.8 (7.0) 19.4 (13.5)
REM, 1st cycle, min 23.2 (18.3) 24.5 (19.8) 13.4 (18.4)
REM, 2nd cycle, min 22.1 (16.9) 25.3 (18.2) 22.8 (14.8)
REM, 3rd cycle, min 20. 8 (16. 6) 20.1 (19.7) 13.6 (13.7)
REM, 4th cycle, min 21.7 (10.9) 21.8 (18.1) 20.1 (13.2)
EM density, 1st cycle 6.3 (4.4)** 7.2 (5.4)* 2.9 (2.0)**,*
EM density, 2nd cycle 5.2 (2.8) 6.1 (3.6) 6.0 (4.9)
EM density, 3rd cycle 5.6 (4.5) 5.0 (4.5) 6.6 (6.7)
EM density, 4th cycle 4.3(2.3) 4.9 (4.0) 9.8 (6.7)
EM density, total 17.4 (8.8) 20. 7 (10.6) 21.7 (13.2)
Number of awakenings 3.9 (2.7) 2.9 (5.0) 3.1 (2.8)

Figures in brackets denote standard deviation

Sleep efficiency and slow-wave sleep (SWS) in the first cycle were significantly less in Group III than in Group II (see Table 1). In Group III, EM density in the first cycle was lower than in both other groups (the difference was significant for Group II), while m the fourth cycle EM density increased more than in other groups.

In two subgroups (women, and patients without the first-night effect) the improvement of mood correlated negatively with sleep efficiency ( - 0.54; -0.53), in three other subgroups it correlated negatively with the duration of the first REM sleep period. Such correlation was relevant for the nights when sleep was estimated as deep (- 0.59) and refreshing ( - 0.47), and in nights with REM sleep latency shorter than 40 min (- 0.47). In nights when sleep was estimated as deep, there was also a negative correlation between mood improvement and the duration of wakefulness during sleep ( - 0.61). In one subgroup of nights (with REM sleep latency longer than 110 mm) the improvement of mood correlated negatively with SWS (- 0.74).

The most stable, strong and positive correlations have been found between subjective estimation of mood improvement and EM density in the fourth cycle. Such correlations have been found in five subgroups in nights with short first REM sleep periods (0.65), with a high amount of SWS (0.94), with less than 15% REM sleep (0.97), with predomination of SWS in the last cycles (0.93), and with high EM density (>6/min on average for the whole night, 0.84). At the same time, EM density in the first cycle correlated positively with the subjective estimation of wakefulness before sleep (sleep delay, 0.51), and EM density in the second and third cycles showed a positive correlation with the subjective estimation of the number of awakenings (0.33). These significant correlations were relevant for the whole group of patients. According to correlations, there was no difference between the first and the second night of the investigation.


In depressed patients, mood in the morning is usually worse than in the evening. In the present investigation, in 20% of all nights, mood was estimated as more positive in the morning than in the previous evening.

The most prominent results may be summarized as follows:
1. the increase of EM density from cycle to cycle on nights which precede the improvement of mood, as compared with flattened EM distribution with a tendency of EM density to be maximal in the first cycle on all other nights for depressed patients,
2. the positive correlation between mood improvement after sleep and EM density in the fourth cycle,
3. the negative correlation between mood improvement and REM sleep duration in the first cycle,
4. the positive correlation between EM density in other cycles and subjective estimation of wakefulness before sleep or of the number of awakenings.

Of course, correlations cannot establish causal relationships between variables, and there are three possible interpretations of these results. First, the increase of EM density from cycle to cycle, being typical for healthy subjects, corresponds to mental activity with increasing participation of the dreamer in dream events [5], and such mental activity is thought by some [1] to be responsible for mood restoration. Second, mood improvement, brought about during sleep by unknown causes, is accompanied by the restoration of normal EM distribution from cycle to cycle. Third, another unknown factor causes both mood restoration and increase of EM activity from cycle to cycle.

From the present data it is difficult to choose any one of these explanations without reservation. However, some indirect reasons support the first one. It has been shown in many investigations that, in healthy subjects as well as in patients, REM sleep is important for psychological adaptation [12-17]. Secondly, deprivation of REM sleep in healthy subjects caused emotional disturbances[18], while similar REM sleep deprivation in depression caused mood improvement [19]. Finally, if dreaming is related to the restoration of mood [1], and EM density is related to dream content (5), it is reasonable to suggest that the intensification of the REM sleep mental activity which is manifested in the increased EM density, is responsible for the episodic mood restoration after sleep.

According to the results of the present investigation, different REM sleep periods may have different functional meanings. While EM density in the fourth cycle correlated positively with mood restoration, REM sleep in the first cycle correlated negatively with the same variable. EM activity in the first cycle correlated in our patients with the subjective estimation of wakefulness, and EM activity in the subsequent cycles correlated with the estimation of awakenings. This might be the reason why Reynolds et al. [20] did not find any difference in REM activity and REM density (measured for the whole night) between bereaved subjects who were depressed and those who were not. Depression is characterized by the increased EM density in the first REM sleep period while healthy subjects are characterized by increased EM density in later cycles. Although an increase in EM density in the first cycle may represent an attempt by the brain to compensate for depressive disorder [21], such an attempt is not efficient enough-mood is not restored. Moreover, such increase of EM density in the initial cycles is typical of untreated depression. According to our hypothesis [22], the nature of the first REM sleep episode is not such as to contribute successfully to the adaptive process. In depression, REM sleep is usually functionally inefficient in the subsequent cycle [23], which is confirmed by the above-mentioned correlation between EM density and the subjective feeling on awakening. Only rarely, and predominantly in the last cycles, after long-lasting previous attempts, did REM sleep in depression become functionally sufficient to perform a compensatory function and to restore mood. Our general conclusion is that even in depressed patients, mood can be periodically restored after sleep, and in such cases EM density is increased in the last cycles in comparison to the initial cycles. The initial and the last REM sleep periods are functionally unequal, and it is important to take this into consideration in investigations as well as in the treatment of depression by using REM deprivation.


  • Sleep architecture is correlated with mood
  • There is a functional difference between initial and final REM periods


    1. Kramer M (1993) The selective mood regulatory function of dreaming. An update and revision. In: The functions of dreaming (eds A. Moffitt, M. Kramer, R. Hoffmann) 139-196, State University of New York Press New York

    2. Foulkes D (1996) Sleep and dreams. Dream research 1953-1993. Sleep 19: 609-24.

    3. Herman JH, Erman M, Boys R et al (1984) Evidence for a directional correspondence between eye movements and dream imagery in REM sleep. Sleep, 7: 52-63.

    4. Rotenberg VS (1982) The adaptive function of sleep (in Russian). Moscow. Nauka

    5. Rotenberg VS (1988) Functional deficiency of REM sleep and its role in the pathogenesis of neurotic and psychosomatic disturbances. Pavlov. J. Biol. Sci. 23: 1-3

    6. Kupfer D, Heninger GR (1972) REM activity as a correlate of mood changes throughout the night. Arch Gen Psychiatry 27: 368-73.

    7. Reynolds C.F. III, Kupfer D (1988) Sleep in depression. In: Sleep disorders, diagnosis and treatment (Eds. R. Williams, I. KaraCan, C.A. Moore) 147-164 Wiley New York

    8. Benca RM, Obermeyer WH, Thisted RA, Gillin C (1992) Sleep and psychiatric disorders. A meta-analysis. Arch. Gen. Psychiatry 49: 651-68.

    9. Hamilton M (1960) A rating scale for depression. J. Neurol, Neurosurg. Psychiatry 23: 56-62.

    10. Rechtschaffen A, Kales A (1968) A manual of standardized terminology, teChniques and sCoring system for sleep stages of human 18 subjects. National Institutes of Health PubliCations 204 US Government Printing Office, Washington, DC

    11. Benson E, ZarCone VP (1993) Rapid eye movement sleep eye movements in schizophrenia and depression. Arch. Gen. Psychiatry 50: 474-82.

    12. Greenberg R (1987) Self-psychology and dreams: The merging of different perspectives. PsyChiatr. J. Univ. Ottawa 12: 98-102.

    13. Greenberg R, Pearlman C (1974) Cutting the REM sleep nerve: An approaCh to the adaptive role of REM sleep. PerspeCt. Biol. Med. 17: 513-21.

    14. Greenberg R, Pearlman C, Campel D (1972) War neuroses and the adaptive funCtion of REM sleep. Br. J. Med. Psychol. 45: 27-33.

    15. Rotenberg VS (1993) REM sleep and dreams as meChanisms of the recovery of search activity. In The functions of dreaming (eds. A Moffit, M Kramer R Hoffmann) 261-92, State University of New York Press Albany,i NY

    16.Rotenberg VS (1996) The psychobiological dream functions. A new solution for old Contradictions. In: Sense and nonsense. Philosophical, Clinical and ethical perspectives (ed J Rosenberg) 187-97, Magnes, Jerusalem.

    17. Rotenberg VS, Arshavsky V.V. (1979) REM sleep, stress and search activity. Waking & Sleeping 3: 235-44.

    18. Greenberg R, Pillard R, Pearlman C (1972) The effeCt of dream deprivation on adaptation to stress. PsyChosomatiC Med 34: 257-62.

    19. Vogel GW, Vogel F, MCAbur RS, Thurmond AJ (1980) Improvement of depression by REM sleep deprivation. Arch. Gen. Psychiatry 37: 247-53.

    20.Reynolds CF, III, HoCh CC, Buysse DJ et al (1993) Sleep after spousal bereavement: A study of recovery from stress. Biol. Psychiatry 34: 791-7.

    21. Cartwright RD, Lloid SR (1994) Early REM sleep: A Compensatory Change in depression? Psychiatry Res. 51: 245-52.

    22. Rotenberg VS, Kayumov L, Indursky P et al (1997) REM sleep in depressed patients: Different attempts to aChieve adaptation. J. Psychosom. Res. 42; 565 - 75.

    23.Rotenberg VS (1994) The revised monoamine hypothesis. MeChanism of antidepressant treatment in the Context of behavior. Integrative Physiolog. Behav. SCi. 29: 182-8.