go to content


Vadim S. Rotenberg vadir@post.tau.ac.il L. Kayumov(2), P. Indursky(1), R. Kimhi(1), A. Venger(1), Y. Melamed(1).

Abarbanel Mental Health Center, Tel-Aviv University, Israel(1), Sleep and Alertness Clinic, Toronto Hospital, Western Division, Canada(2).

Homeostasis.VOL 39 APRIL 1999 No 3-4

In depressed patients EM density in REM sleep and Slow Wave Sleep (SWS) are often redistributed. The goal of the present investigation was to analyze the relationships between these variables. - Twenty-four depressed patients underwent two or three consecutive polysomnography studies before treatment. Polysomnograms (PSG) were divided into two groups. In Group I SWS concentrated in the 3rd or 4th cycles, while group II included all other PSGs. In group I Eye Movement Density (EMD) in REM sleep periods which preceded the enhancement of slow wave sleep was found to be higher in all cases compared to the EMD in the previous REM sleep periods, REM sleep latency was shorter than in group II. In group II EMD was not regularly increased from cycle to cycle. In group II the first REM sleep period was the longest one while REM sleep periods in group I followed the normal tendency to increase from cycle to cycle. Only in group I the improvement of mood from evening to morning displayed a strong positive correlation with EM density in the 4th cycle. The increased phasic activity of REM sleep in depression may reflect an adaptive activity of REM sleep, which creates a condition for slow wave sleep to appear.

Key words: EM density, REM sleep, slow wave sleep.


In spite of the long history of investigations of the phenomenology and functions of Rapid Eye Movement (REM) density in REM sleep there are still many contradictions and unsolved problems. One important question is how the distribution of EM from cycle to cycle in healthy subjects differs from that of patients suffering from mental disorders. A second unsolved question is what is the functional meaning of EM activity in REM sleep.

Aserinsky (1969) was the first to describe that EM density in healthy subjects increases during the night. A recent investigation has confirmed this finding and has shown that the same regularity characterizes schizophrenic patients, while depressive patients display a flattened distribution of EM density during the night. Although the range of EM densities in psychiatric patients overlap the normal range, EM density in the first, second and third REM periods are almost equal in depressed patients with a very small and nonsignificant tendency to increase from REM 1 to REM 3 (Benson & Zarcone, 1993). Patients with primary depression in comparison with patients with severe somatic medical illnesses, displayed significantly more EM activity in the first REM sleep period, while in the second REM period the difference was not significant (Foster et al, 1976). The relationship between phasic REM sleep activity and the seventy of depression is unclear. On the one hand, healthy veterans displayed a positive correlation between measures of self-rated depression (in Zung scale) and measures of EM activity (Zarcone & Benson, 1983). On the other hand, m depressive patients, the correlation between EM density and Hamilton Depression Rating Scale was absent (Benson & Zarcone, 1993, Foster et al, 1976). At the same time, there was a significant positive correlation between affect intensity (the sum of all positive and negative affects) before sleep, and EM density during sleep (Reynolds et al, 1993).

It was also shown that in a patient with Bipolar Mood Disorder, the shifts to an elevated mood were accompanied by increased REM activity during the second half of the night while the shifts to a depressed mood were associated with decreased REM activity during the second half of the night (Kupfer & Heninger,1972). We have confirmed these data on patients with major depression (Indursky & Rotenberg, 1998). This suggests that the increased EM density in the last cycles may relate to mood improvement. EM density correlates also positively with some other mental functions: 1) The number of phasic eye movements per epoch of sleep among normal and senile persons correlated significantly and are positively related to Wechsler Adult Nonverbal Intelligence Subscale (Reynolds et al, 1988); 2) In patients with mental retardation, the objective index of cognitive competence was the density of EM activity during REM sleep (Feinberg et al, 1969); 3) In healthy subjects, EM density m REM sleep correlated with active participation of the subject in his own dream scenario (Rotenberg, 1988); 4) In depression, active participation of a dreamer in his/her own dream is rare, and a dreamer is often in a state of helplessness (Greenberg & Pearlman,1979); 5) Elevated EM density is a characteristic feature of late-life bereavement in the absence of depression (Reynolds et al, 1993). All of the preceding data suggest that the high phasic activity in REM sleep may have a positive outcome on mental functions and may relate to the adaptive function of REM sleep and dreams (Greenberg & Pearlman, 1974, Rotenberg, 1984, Rotenberg & Boucsem, 1993, Rotenberg et al, 1997).

In the present study, we searched for an additional independent index (sleep variable) in order to confirm the adaptive function of the phasic REM activity. Slow wave sleep (SWS) seemed to be a good candidate for this role. The total amount of SWS is an important index of the successful adjustment of normal subjects to the night shift work (Rotenberg, 1992). SWS is decreased in depression and is often restored after successful treatment (Kupfer et al., 1989; Reynolds et al., 1991). According to the S-deficiency hypothesis (Borbely & Wiz-Justice, 1982), a depressed mood is related to low levels of S (the pressure for nonREM sleep, which is built up during wakefulness and is satisfied during sleep). In depression, SWS is often redistributed compared to healthy subjects. It predominates in the second and sometimes even in the third cycle. However, it is not known whether such unusual increase of SWS during the second part of the night is related to the duration or quality of the previous REM sleep. Thus the aim of the present study was to compare the distribution of SWS in different cycles with the objective variables of REM sleep periods (EM density, REM sleep duration) which preceded the increase of SWS in the subsequent sleep cycles. If the prominent increase of SWS in the second part of the night is preceded by increased EM density, then it would be possible to speculate that phasic activity in REM sleep contributes to the temporary improvement of brain function which allows SWS to appear (for instance, due to the reduction of depressed mood). Another aim of the present investigation was to compare sleep structure on nights characterized by SWS predomination in the first and second part of the night in order to check whether it is an alteration of sleep structure related to SWS redistribution. As far as we know, such an investigation has not been previously performed, although the delayed SWS in depression is not an unusual finding.



We investigated 24 subjects (12 males and 12 females), age ranged from 32 to 70. All of them were suffering from major depression (diagnosis based on DSM-IY criteria). The seventy of depression was evaluated by using the Hamilton Rating Scale for Depression (Hamilton, 1960). The mean number of past admissions in this group was 4.1. Most of these patients (20) displayed classical diurnal variations of mood, the latter being worse in the morning than in the evening. Nine patients suffered from major depression with psychotic features, only one patient had a past history of alcohol abuse. The initial score of the Hamilton Scale prior to investigation was on average 36.7. All patients have been investigated before initiating of any treatment. In eleven patients investigations were performed also after 2 months of treatment (6 received - 12 sessions and 5 received antidepressants- Amitriptyline - 50 mg per day or Fluvoxamine - 150 mg per day).

The study was approved by the Helsinki committee of the Abarbanel Mental Health Center and all patients gave informed consent to participate in the study. As a control group we have investigated ten healthy subjects without any health complaints or sleep disorders.


Sleep data were collected from all patients and control subjects at least on two consecutive nights, including the first night for adaptation. In 15 patients, it was available to record sleep during 3 consecutive nights. The total number of the night polysomnograms before treatment was 62, after treatment was 24, and 20 in the control group.

Lights were usually turned off between 10 00 and 10 30 pm and turned on at 6 30 a m according to the habitual schedule the ward patients were adapted to. Electroencephalogram, submental electromyogram, electrooculograms and pneumogram were recorded on an electroencephalograph Neurofax EEG-4400. EEC and EOG were recorded at a low filter setting at 0. 3 Hz. Polysomnograms were analyzed according to the international criteria (Rechtschaffen & Kales, 1968) on 62 nights before treatment and on 24 nights after two months of treatment. We defined sleep onset as the first 30 second epoch of stage 2 sleep, SWS or REM sleep which was followed by no more than 1 minute of wakefulness during the first 10 minutes of sleep. REM sleep latency was defined as the time from the onset of sleep until the onset of the first REM sleep episode. REM sleep percentage was the percentage of total sleep time spent in REM sleep. Eye movements in REM sleep were counted visually. The criteria for rapid EM detection were a minimum of 25 microV excursion (Benson & Zarcone, 1993). The EM density was counted as EM frequency per 1 minute at stage REM in every cycle. In addition, we also counted in REM sleep a frequency of 5-second intervals which contained at least one eye movement. Fragments of wakefulness or non-REM sleep occurring within a REM episode were not analyzed for EM activity and were not included in the calculation of REM period length. A minimum of REM duration was 1 minute, in order to calculate EM ratio. However, only one REM sleep period was excluded according to this criteria. The first sleep cycle was estimated as sleep duration from sleep onset until the end of the 1st REM sleep period. The subsequent sleep cycles were estimated as sleep duration from the end of the previous REM sleep period until the end of the subsequent REM sleep period. After every morning awakening, every patient received a questionnaire (Rotenberg et al, 1997) allowing subjectively assess his sleep quality, feeling refreshed after sleep and change of mood from evening to morning (patient was asked whether his/her mood after sleep became better, worse or unchanged in comparison to mood before sleep). In the present investigation, we were interested in the subjective estimation of mood change after sleep with SWS redistribution vs sleep without SWS redistribution. Subjective variables were ranged and correlated with the objective sleep variables (Rotenberg et al, 1997).

Compared groups.

All polysomnograms performed before treatment were divided into two groups according to the SWS distribution during the night: group I included 11 polysomnograms in which SWS was predominantly m the 3rd and the 4th cycle. We used the following criteria: SWS in the 3rd (or 4th) cycle had to be at least three times longer than in any other previous cycles. Thus, this group consisted of nights with a real explosion" of SWS in the 3rd or 4th sleep cycles. Group II collected all other polysomnograms (n=51) and contained polysomnograms without SWS, or with SWS concentrated in the first two cycles, or with SWS equally distributed through all cycles. Previous studies have shown that in depression, SWS is redistributed and often concentrates in the second, rather than in the first sleep cycle (Benca et al, 1992). However, in healthy

subjects and especially in young testees, SWS duration in the second cycle is also long enough and sometimes comparable to SWS duration in the first cycle. For this reason, we decided in our analysis to select only nights with an obvious redistribution of SWS that is usually absent in normal subjects with normal work-rest schedule. In our control group of 10 healthy subjects (20 nights) such redistribution of SWS was absent. Thus, SWS redistribution was found only in depressed patients, and for further analysis we have used only nights of depressed patients in order not to form mixed groups.

Only one patient demonstrated such redistribution of SWS in all nights.

In other patients, some nights belonged to group I and other nights belonged to group II. Group I included some nights of 4 patients with psychotic features and of 4 patients without psychotic features, however, other nights of the same patients belonged to group II. Therefore, the two groups could not be differentiated clinically or by age. The same patient on one night demonstrated a normal distribution of SWS (group II) and on another night - redistribution of SWS (group I). The difference between nights was not attributed to the 1st night effect. Group I included only 3 polysomnograms of the 1st night. We were unable to find any obvious differences in the prior sleep/wake schedule or circadian rhythm variations between nights of Group I and Group II. Thus, it was possible to speculate that the redistribution of SWS is related to the precise mental state of the patient at the particular time of sleep investigation.

Statistical analysis.

We used paired t-test and nonparametric test U according to Mann and Witney (Himmelblau, 1970). We have also analyzed the correlation between objective and subjective sleep variables.


Table 1 represents data comparing the sleep structure of the two groups. There are some significant differences between those groups. In group I, latencies of stage 2 and REM sleep are shorter, and there is a tendency for REM sleep to increase from cycle 1 to cycle 2 (a normal distribution of REM sleep). REM sleep in the first two cycles is significantly longer in Group II.







Sleep Duration (mm)




Stage 2 Latency (mm)




Stage 3 Latency (mm)




REM Sleep Latency (mm)




hi REM Duration (mm)




2nd REM Duration (mm)




3d REM Duration (mm)




4th REM Duration (mm)




REM Sleep%




EM Density in the 1st REM (No/mm)




EM Density in the 2nd REM (No/mm)




EM Density in the 3ri REM (No/mm)




EM Density m the 4th REM (No/mm)




SWS m the 1st cycle (mm)




SWS in the 2nd cycle (mm)




SWS m the 3rd cycle (mm)




SWS in the 4th cycle (mm)








EM= eye movement, SWS=slow wave sleep

SWS percentage is higher in group I. EM density in group I is higher than in group II in all cycles except the first one.

In group I, EM density in the REM sleep episode which preceded the explosion" of SWS (REM 2 or REM 3) was, in all cases with only one exception, higher than EM density in the previous REM sleep episodes.



SWS3>SWS1 + SWS2 or SWS4 > SWS1 + SWS2 + SWS3

SWS3<SWS1+SWS2 or SWS4 < SWS1 + SWS2 +SWS3

EM2 > EM1 or EM3 > EM1 and EM2

11 nights

17 nights

EM2 < EM1 or EM3<EM1 and EM2

3 nights

19 nights

The relationship between EM density and SWS in the subsequent cycles is displayed in Table 2. In most cases when SWS obviously predominates in the 3rd or in the 4th sleep cycles and is higher than the sum of SWS of all previous cycles, EM density in the preceding cycle is higher than in all previous cycles. When SWS is not dominant in the last cycles, EM density can be either increased or decreased in the 2nd/3rd cycle in comparison to the previous cycles. Although it was a general tendency of EM density to increase in group I from cycle to cycle, the coincidence between the increased EM density and the increased SWS in the subsequent cycles was not a result of this general tendency. In group I EM density in REM 2 exceeded EM density in REM 1 in 9 cases out of 10, and exactly in these cases, SWS displayed a dramatic increase in the 3rd cycle. In one case, EM density was increased initially only in REM 3, in comparison to the first two cycles, and precisely in this case SWS was increased in the 4th cycle. However, this regularity was shown only in group I, where SWS increased at least after REM 2. In group II, EM density in REM 2 exceeded EM density in REM 1 in 38% of all cases, however even in these cases SWS in the 3rd cycle was not increased. In group II, EM density in REM 1 was higher than in REM 2 in 46% of all cases. However, SWS in the 2nd cycle was longer than in the 1st one in only 16% of all cases. Thus, while the increase of SWS in the second part of the night is regularly preceded by increased EM density, the increased EM density by itself is not regularly followed by SWS explosion: it is an essential but insufficient condition for such explosion.

The same relationship has been discovered between SWS and the density of the 5 second intervals that contain EM. In 83% of all nights with SWS, being concentrated in the second part of the night the density of these 5 second intervals was higher in REM 2 or REM 3 than in REM 1 or REM 2 respectively. When SWS was distributed in the first part of the night, or equally between all cycles, REM 2 exceeded REM 1 in the density of these 5 second intervals only in 47% of all nights (p<0.01).

In order to check our main finding, in addition we have selected nights with similar distribution of SWS from all polygraphical investigations performed after treatment. Although treatment by itself can change the sleep structure, it was not a topic of ouranalysis. The only aspect we have been interested in was whether the abovementioned regularity (increased EM density before SWS explosion) is present also after treatment. It was possible to find 5 nights with SWS increase in the late cycles. In 4 of them, regularity was the same, while for one night it was difficult to draw any conclusions because of the total absence of EM during REM sleep. However, when SWS after treatment was not concentrated in the last cycles, EM density in the second cycle was higher than in the first one in 52% of all cases.

The distribution of REM sleep duration according to the SWS distribution was very similar in both groups before treatment. In group I, REM sleep, which preceded the increase of SWS, was longer than the previous REM sleep period in 50% of all cases; in group II REM 2 was longer than REM 1 in 54% of all cases. Thus, it was only EM density but not REM sleep duration, which predicted the increase of SWS in the second part of the night.

The correlations between subjective and objective sleep variables are also different in both groups. Only in group I the subjective estimation of mood improvement correlated positively with EM density in the 4th cycle (r=0.93; p<0.01) and negatively with REM sleep latency (r=-0.73; p<0.05). REM sleep latency in this group in general is shorter than in group II. The subjective estimation of the number of awakenings during the night correlates positively (r=0.79; p<0.02) with EM density in the first cycle of group I. In group II, the subjective estimation of mood dynamics does not correlate with REM sleep variables, while the subjective estimation of the number of awakenings correlates positively with EM density in cycles 2 and 3. The subjective estimation of the duration of wakefulness after awakenings correlates positively with REM sleep duration in the 2nd cycle (r=0.45; p<0.01) and with EM density in the 1st cycle (r=0.46; p<0.01), however, it correlates negatively with EM density in the 4th cycle (r=-0.47; p<0.01). It is worth taking into consideration that EM density is less in the 4th cycle in group II than in group I.

In group I, the feeling of total sleep duration correlates positively (r=0.62; p<0.05) with SWS and in group II with SWS duration in the 2nd cycle (r=0.38; p<0.05). Group I is characterized, in addition, by negative relationships between REM sleep duration and EM density. REM% correlates negatively with EM density in the 3rd cycle ( r= -0.71; p<0.01). REM sleep in the 2nd cycle correlates negatively with EM density in the 3rd and 4th cycles (r=-0.87; p<0.01). Group II shows a positive correlation between REM duration and EM density in the first two cycles.


The most important finding of the present investigation was the intensification of the eye movement density in REM sleep exactly before the prominent increase of SWS in the second part of the night. Although such SWS increase is a relatively rare finding it may elucidate some mechanisms of sleep disorders in depression. Before discussing the nature of these relationships between EM density and the subsequent SWS, it is necessary to exclude the possibility that EM density on all nights had a regular tendency to increase from cycle to cycle and as a result a combination of the increased EM density with the subsequent increase of SWS (caused by some unknown additional factor) was only coincidental. According to our findings, this was not the case. In group II, which was larger than group I and displayed the sleep structure which is more typical for depression, EM density in the second cycle was usually lower than in the first cycle. This finding confirms the well-known data that in depression, EM density in the first cycle has a tendency to be increased. In addition, in group I EM density in REM 3 exceeded EM density m REM 2 only in 50% of all cases. Thus, the increase of EM density before SWS explosion" in the second part of the night was not a by-product of the general tendency of EM density to increase from cycle to cycle.

Secondly, in group I the average duration of REM sleep in the 2nd cycle was not significantly longer than the duration of REM sleep in the first cycle. In individual cases, REM 2 was longer than REM 1 only in 50% of all cases. In group II, regularity was almost the same. Thus, REM sleep duration is not increased before SWS explosion". According to this data, it is possible to speculate that in some particular cases REM sleep characterized by enhanced EM density, contributes to the final improvement of the mental state of patient and determines the condition, which allows SWS to appear. Our suggestion about the role of the increased EM density in the subsequent increase of SWS is in line with the above-mentioned data that EM density characterizes a quality of REM sleep (Kupfer & Heninger, 1972, Rotenberg, 1988). EM density also correlated with absence of depression and the presence of normal SWS in late life bereavement (Reynolds et al, 1993). In our investigation, EM density in group I in the 4th cycle displays a strong positive correlation with subjective mood improvement, and it is an additional argument for restorative capacity of REM with high EM density (Indursky & Rotenberg, 1998). If enhanced EM density reflects in these particular cases the active position of a dreamer in his dream scenario, as it was shown in healthy subjects (Rotenberg, 1988), and activation of image thinking (Hong et al, 1997), it may relate to the restoration of search activity, brain monoamine functions and to improvement of adaptation.

REM sleep latency in group I was decreased. Thus, it is possible to suggest the presence of high REM sleep pressure in group I. This high REM sleep pressure notwithstanding, REM sleep was not increased in Group I, thus it is unlikely that the delay of SWS is caused by its competition with REM sleep in the first cycles. REM sleep performs its adaptive function only in the second or in the third cycle, but not in the first cycle, although REM sleep pressure is high. Why SWS is not increased during the second cycle in group II after the first REM sleep period, which is often characterized by increased EM density? By considering both groups together, we can see the following regularity: while any significant increase of SWS during the second part of the night was preceded by enhanced EM density, the increased EM density in the 1st cycle did not predict or determine the increase of SWS. How is it possible to explain these contradictions? It is reasonable to suggest that the second and all subsequent REM sleep periods have some advantages according to their adaptive capability compared with the first REM sleep period. In our previous investigation (Rotenberg & Arshavsky, 1979), we have shown that the second REM sleep period was increased in healthy students due to the process of adaptation to stress. At the same time, it is well known that in depressive patients the second REM sleep period is usually not longer and often shorter than the first one and EM density shows on average a flattened within-night distribution (Benson & Zarcone, 1993). In the bipolar patients, the shifts to an elevated mood were accompanied by increased EM activity only during the second half of the night (Kupfer & Henmger, 1972). This corresponds to our present data. At the same time, high EM density in the first cycle correlates positively in group I with the subjective estimation of awakenings and in group II with the subjective duration of wakefulness (while EM density in the 4th cycle correlates negatively with this variable). Thus, it is possible to suggest that the 1st REM sleep period is functionally different in comparison with the subsequent REM sleep periods, at least in depressive patients. If in group I, REM sleep in most cycles, except the first one, is functionally sufficient and restorative (decreases the level of depression, helps to restore SWS), and this function correlates with high EM density, then this can possibly explain, why in this group REM duration correlates negatively with EM density. The requirement in REM sleep is decreased if REM sleep quality is sufficient. If REM sleep is functionally sufficient and protects healthy subjects from emotional disturbances then REM sleep deprivation can cause emotional disturbances in healthy subjects and can destroy SWS. It can explain the data of Beersma and Van den Hoofdakker (1992) that REM deprivation for one or two nights can suppress non-REM intensity in healthy subjects. From our point of view, it happens only if REM sleep by itself is functionally sufficient.

In depression, however, REM sleep is more often functionally insufficient, like in group II. This insufficiency is manifested in contentless dreams (Riemann et al, 1990) or in dream content which represents feelings of helplessness and hopelessness (Greenberg & Pearlman, 1979). It is possible to speculate that in depression EM density, even being increased in comparison to healthy subjects as happens in the 1st cycle, is usually not increased enough and not in the appropriate cycles to restore REM sleep functions, a normal emotional state and SWS. A broad fluctuation of REM sleep functional activity from night to night and from cycle to cycle can explain a wide range of REM sleep EM densities in mentally ill patients (Benson & Zarcone, 1993). If the report in REM sleep is contentless (Koulack, 1993), it can be a sign of REM sleep insufficiency, even if EM density is high. Thus, EM density by itself is not an unequivocal sign of REM sleep efficiency and it may be a reason why increased EM density is not always accompanied by SWS restoration, and why in group II EM density correlates with the subjective estimation of the number of awakenings. If REM sleep is functionally insufficient, it becomes deteriorative in its nature (Rotenberg, 1994). In such conditions, REM sleep deprivation may have a therapeutic effect, while sleep by itself may cause a relapse in depression (Vogel et al, 1980).

In conclusion, it is worth stressing once more the point that nights in group II represent a typical architecture of sleep in depression. REM sleep was shifted to the first part of the night and sleep latency was increased. There was also flattened distribution of eye movement between different REM sleep periods and SWS was reduced. According to our data, this typical sleep structure in depression is characteristic for the functionally insufficient REM sleep. Even a normal REM sleep latency in this group corresponds to the data (Cartwnght et al, 1991) that normal but not reduced REM sleep latency corresponds to the worse outcome of depression. However, because group I in most cases contains the same patients as group II, it is possible to suggest a fluctuation of REM sleep efficiency, at least in some depressive patients. When the adaptive REM sleep function becomes episodically improved, all relationships in sleep are changed, as it was shown in the present investigation

Acknowledgments. The authors are very thankful to Dr. Fisch for the revision of the manuscript. The study was supported by the Israeli Ministry of Health


Aserinsky E. The maximal capacity for sleep rapid eye movement density as an index of sleep satiety. Biol.Psychiatry 1969, 1: 147-159.

Beersma D.G. and Van den Hoofdakker R.H. Can non-REM sleep be depressogenic'? J. Affect. Disease, 1992, 24: 101-108.

Benca R.M., Obermeyer W.H., Thisted R.A. and Gillin J.C. Sleep and psychiatric disorders: A meta-analysis. Arch Gener Psychiatry, 1992,49: 651-668.

Benson E. Z. and Zarcone V. P Rapid eye movement sleep in schizophrenia and depression. Arch. Gen. Psychiatry, 1993, 50: 474-482.

Borbely A.A. and Wiz-Justice A. Sleep, sleep deprivation and depression: A hypothesis derived froma model of sleep regulation. Hum Neurobiolog. 1982,1: 205-210.

Cartwright R.D., Krawitz H.M., Eastman C.J. and Wood M.A. REM latency and the recovery from depression. Getting over divorce. Am. J. Psychiatry, 1991, 140: 1530-1535.

Feinberg J., Braun M. and Shulman E. Sleep patterns in mental retardation EEG Clin. Neurophysiol. 1969, 27: 368-373.

Greenberg R. and Pearlman C. The private language of the dream. In Natterson, (Ed.) The dream in clinical practice. New-York, J Aronson, 1979, 85-96.

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

Himmelblau D.M. Process analysis by statistical methods NY, London John Willey Inc 1970.

Hong Ch. Ch.-H., Potkin S.G., Antrobus J.S., Callaghan G.M. and Gillin J.Ch. REM sleep eye movement counts correlate with visual imagery in dreaming. A pilot study. Psychophysiology, 1997, 34:377-381.

Indursky P. and Rotenberg V.S. Change of mood during sleep and REM sleep variables. International J. of Psychiatry in Clinical Practice, 1998,2: 47-51.

Koulack D. Dreams and adaptation to contemporary stress. In A. Moffitt, M. Kramer & R Hoffman, Eds. The function of dreaming. State University of New York Press, 1993, 321-340.

Kupfer D.J. and Heninger G.R. REM activity as a correlate of mood changes throughout the night. Arch. Gener. Psychiatry, 1972, 27: 368-373.

Kupfer D.J., Ehlers C.Z., Pollock B.C., Nathan R.S. and Perel J.M. Clomipramme and EEG sleep in depression. Psychiatric Res. 1989,30: 165-180.

Rechtschaffen A. and Kales A. Eds. A manual of standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. Washington, D National Institute of Health, 1968, No.204.

Reynolds C.F., Hoch C.C., Buysse D.J., George C.J., Houck P.R., Mazumdar S., Miller M., Pollock B.G., Rifai H. and Frank E. Sleep in late-life recurrent depression. Change during early continuation therapy with nortriptyline. Neuropsychopharmacol 1991,5: 85-96.

Reynolds C.F., Kupfer D.F., Houck P.R., Hoch C.C., Stack J.A., Herman S.R., and Zimmer B. Reliable discrimination of elderly depressed and demented patients by electro-encephalographic sleep data. Arch.Gener. Psychiatry, 1988, 45: 258-264.

Reynolds C.F., Hoch C.C., Buysse D.J., Houck P.R., Scherlnitzauer M., Pasternak R.E., Frank E.,Mazumdar S., and Kupfer D.J. Sleep after spousal bereavement. A study of recovery from stress. Biol. Psychiatry, 1993, 34: 791- 797.

Riemann D., Low H., Schredl M., Wiegand M., Dippel B. and Berger M. Investigation of morning and laboratory dream recall and content in depressive treatment with trimipramine. Psychiat. J. Univ. Ottawa, 1990, 15: 93- 99. Rotenberg V.S. Search activity in the context of psychosomatic disturbances, of brain monoamines and REM sleep function. Pavlov. J. Biolog. Sci, 1984, 19: 1-15.

Rotenberg V.S Functional deficiency of REM sleep and its role in the pathogenesis of neurotic and psychosomatic disturbances. Pavlov. J. Biol. Sci, 1988, 23: 1-3.

Rotenberg V.S The competition between SWS and REM sleep as index of maladaptation to shift work.Homeostasis, 1992, 33: 235-238.

Rotenberg V.S. REM sleep and dreams as mechanisms of the recovery of search activity. In Moffitt A , Kramer M. and Hoffman R.( Eds.) The function of dreaming. State University of New-York Press, 1993, 261-292.

Rotenberg V.S. The revised monoamine hypothesis. Mechanism of antidepressant treatment in the context of behavior. Integrative Physiological and Behavioral Science, 1994, 29: 182-188.

Rotenberg V.S. and Arshavsky V.V. REM sleep, stress and search activity. Waking and Sleeping, 1979, 3: 235-244.

Rotenberg V.S. and Boucsein W. Adaptive vs maladaptive emotional tension. Genetic, Social and General Psychology Monographs, 1993, 119: 207-232. Rotenberg V.S., Kayumov L., Indursky P., Hadjez J., Kimhi R., Sirota P., Bichucher A. REM sleep in depressed patients: different attempts to achieve adaptation. J. Psychosom. Res. 1997, 42:565-575.

Vogel G.M., Vogel F., McAbce R.S., and Thurmoud A.J. Improvement of depression by REM sleep deprivation. New findings and a theory. Arch. Gener. Psychiatry, 1980, 37: 247-253.

Zarcone V. P. and Benson K. L. Increased REM eye movement density in self -rated depression. Psychiatric. Res. 1983,8: 65-71.