REM Sleep = Dreaming: Only a Dream

Nicholas Rosenlicht

Department of Psychiatry, University of Medicine and Dentistry of New Jersey,
671 Hoes Lane, Piscataway, NJ 08855


Irwin Feinberg

University of California, Davis and VA Northern California Health Care System

There is an ironic contradiction in current literature on sleep and dreaming. Those who have conducted controlled research, or who have evaluated this research objectively, know that equating REM sleep with dreaming is no longer a supportable proposition - if it ever was. On the other hand, many neurophysiologists ignore the accumulated body of rigorous psychophysiological data and describe their research on REM sleep as studies of the "neurophysiology" or "biochemistry" of dreams.

Given the peculiar hierarchy of scientific authority, and the allure of linking psychology to physiology, it is these latter scientists that tend to dominate current opinion. This produces several unfortunate consequences. First, it suggests to our students that it is OK is ignore scientific data that does not fit one's prejudices. It also misleads the public on a topic of enormous lay interest. But its most destructive effect is to impede research progress: the daunting complexity of one of the most interesting problems in human brain-behavior relations is being sacrificed in favor of an outmoded (and false) simplification.

In the remainder of this brief essay, we outline the evidence that compellingly refutes a REM=dream isomorphism. A more detailed review will be published shortly.

Goodenough et al. (1959) appear to have been the first to seriously question the REM=dreaming equivalence. They reported that a group of subjects (Ss) who reported frequent spontaneous recall of dreams at home had a high (53%) rate of dream recall in experimental awakenings from NREM sleep. Three years later, Foulkes (1962) reported a similarly high (54%) rate of dream reports from NREM awakenings in Ss unselected with respect to dream recall. With reports of "thinking" included as dreams, this rate jumped to 74%, close to the ~80% usually obtained from REM awakenings. Foulkes did, however, find qualitative differences between REM and NREM dreams. Narratives from NREM awakenings tended to have a lower incidence of affective events and visual images than those from REM. This observation, which remains influential in textbooks and other literature, turned out to be misleading, as we discuss below.

Foulkes and Vogel's early study (1965) of mental activity at sleep onset further challenged the REM=dream equivalence. They found that dreamlike reports, often similar to those obtained from REM, could be obtained from awakenings at sleep onset. These reports were common during descending stage 1 and 2, and could occur in the presence of alpha. Although the hypnagogic reports as a group tended to be shorter and to lack the visual continuity of many REM reports, they described sleep onset narratives that seemed indistinguishable from the dreamlike reports Pg. 10 obtained from REM.

Early evidence cited in support of a REM=dream isomorphism was an apparent correlation between the pattern of rapid eye movements during REM sleep and that which would be expected if the subject had been scanning the dream images he later reported (Dement and Kleitman, 1957; Roffwarg et al., 1962). However, initial positive findings were not confirmed in later, better-controlled studies. Moskowitz and Berger (1969) found that only 18 of 56 blind matchings were correct, little above the fourteen expected by chance. Studies by Firth and Oswald (1975) and Jacobs et al. (1972) also failed to find a consistent relationship between eye movements and dream imagery. Koulack (1972) reviewed the evidence on this question and concluded that "the notion of a constant isomorphic relationship [between eye movements and dream imagery] is untenable." Nevertheless, one still finds contemporary texts that describe the scanning hypothesis as confirmed. We would also like to reiterate here our hypothesis that the eye movements of REM hold no special significance for the psychology of this state. The effector response to neuronal firing in the oculomotor nuclei - eye movement - does not cause arousal. In contrast, the REM discharges of motor neurons that control large muscles would awaken the sleeper were they not inhibited. Nature, being frugal, did not develop an unneeded inhibitory pathway to block eye movements. The biological significance of eye movement activity during REM probably lies elsewhere. Eye movement density appears to be roughly proportional to the within-sleep level of brain arousal. It could therefor provide a valuable non-invasive index of that important sleep variable (for details and some relevant data see Feinberg et al., 1987).

A REM=dream isomorphism also predicts that dream reports should be longer with longer periods of time in REM. A relation between narrative length and time in REM was reported by Dement and Kleitman in their pioneering study (1957). However, this early work did not mention control for time of night. Since REM periods are longer as sleep progresses it is probable that length of time in REM was confounded by time of night. A recent study (Rosenlicht et al., 1994) that included control for time of night did not find that narrative length was proportional to time spent in REM prior to the awakening. Mean total word count (TWC - a measure of dream length) in reports from awakenings after 10 minutes of REM sleep was not longer than after 5 min. However, mean TWC from REMP4 was almost twice that from REMP2, indicating that prior sleep duration is a much more potent determinant of narrative length than time in REM.

As it became evident that mental activity occurs in NREM sleep, some argued that REM dreams were qualitatively different because they were often more bizarre and affect laden. However, REM reports are usually longer than NREM reports. Therefore, when comparing the incidence of "dreamlike" events (visual images, discontinuities, etc.) in narratives from the two states, the length of the report must be taken into account. Antrobus (1983) demonstrated that when length of dream report, as measured by TRC (total recall count) is partialed out, the apparent qualitative differences between REM and NREM reports disappeared. A number of investigators including Foulkes and Schmidt (1983), Fein et al. (1985), Cavallero and Foulkes (1990), and Cavallero et al. (1992) have also shown that when length of dream report is controlled, there is little or no difference between REM and NREM reports. Thus, if narratives elicited by awakenings from REM have distinguishing features, these have yet to be demonstrated

More recently, the mnemonic sources of dream content have been the subject of a number of interesting studies (Cavallero et al., 1990; Cavallero, 1987; Cipolli et al., 1988; Cicogna et al., 1986, 1991; Foulkes, 1982; Antrobus, 1987; Ehrlichman et al., 1985). Stage differences in dream recall appear more closely related to level of mnemonic activation and to access to memory traces than to any special dream production mechanism unique to one stage of sleep (Cicogna et al., 1986, 1991). Thus, Cipolli et. al. (1988) found that thematic units from different stages on the same night were more closely related than those from the same sleep stage on different nights. When Ss were asked to make associations to dream material, Cavallero and his colleagues (1987) found that mnemonic traces were similar for sleep onset and REM dreams. Finally, dream sources, as well as content, appear similar in REM and NREM mentation (Cavallero et al., 1990). In general, these results suggest that the same cognitive systems produce mental activity irrespective of EEG sleep stage, as Foulkes proposed in 1982. Moreover when Cicogna et al. (1986) compared memory traces from day dreaming and sleep onset dreaming, they found a similarity suggesting that žcognitive processes involved in the creation of original narrative sequences may be similar in sleep and waking.

We must, at last, explicitly reject the notion, as appealing as it may be, that REM sleep is the brain correlate of the dream. Sleep researchers, including physiologists, should move on to study the true situation - messier, more complex, but no less interesting. Testable alternative hypotheses to explain some of the variance in sleep mentation already exist. Of these, we believe that the simplest and most plausible are those that propose a positive relation between memory accessibility and brain "activation" or the within-sleep arousal level.

Rejecting the false REM=dreaming equivalence also can advance research on the purely physiological level. If the wild activity in sensory and motor neurons does not affect mentation, it suggests that this state involves an žinternalÓ deafferentation that prevents these discharges from affecting consciousness (Feinberg and March, 1995). Thus, a remarkable attribute of the REM state, and what may be an essential clue to its function, is the existence of a gross Pg. 11 psychophysiological mismatch rather than congruence. However, to pursue this possibility requires that we accept the falsification of the REM=dreaming isomorphism.

One final imponderable must be acknowledged. The discussion above is based on the assumption that mental activity reported on waking is recall of mentation that was ongoing during sleep. This inference has never been proved, and it remains possible that Goblot (cited originally by Calvin Hall, 1981) was correct 100 years ago: the dreams we report on awakening may be constructed entirely during the waking process itself. While we consider the extreme version of this hypothesis unlikely, we do believe that a considerable portion of the dream report is constructed during and immediately after awakening.


  1. Goodenough,D; Shapiro,A; Holden,M; Steinschriber,L (1959): A comparison of "dreamers" and "nondreamers": eye movements, electroencephalograms, and the recall of dreams. Journal of Abnormal and Social Psychology 59, 295-302.

  2. Foulkes,WD (1962): Dream reports from different stages of sleep. Journal of Abnormal and Social Psychology 65, 14-25.

  3. Foulkes,D; Vogel,G (1965): Mental activity at sleep onset. Journal of Abnormal Psychology 70, 231-243.

  4. Dement,WC; Kleitman,N (1957): The relation of eye movements during sleep to dream activity: An objective method for the study of dreaming. J. Exp. Psychol. 53, 339-346.

  5. Roffwarg,HP; Dement,WC; Muzio,JN; Fisher,C (1962): Dream imagery: relationship to rapid eye movements of sleep. Arch. Gen. Psychiatr. 7, 235-258.

  6. Moskowitz,E; Berger,R (1969): Rapid eye movements and dream imagery: are they related? Nature 224, 613-614.

  7. Firth,H; Oswald,I (1975): Eye movements and visually active dreams. Psychophysiology, 12, 602-606.

  8. Jacobs,L; Feldman,M; Bender,MB (1972): Are the eye movements of dreaming sleep related to the visual images of the dreams? Psychophysiology 9, 393-401.

  9. Koulack,D (1972): Rapid eye movements and visual imagery during sleep. Psychol. Bull. 78, 155-158.

  10. Feinberg,I; Floyd,TC; March,JD (1987): Effects of sleep loss on delta (0.3-3 Hz) EEG and eye movement density: new observations and hypotheses. Electroenceph. clin. Neurophysiol. 67(3), 217-221.

  11. Rosenlicht,N; Maloney,T; Feinberg,I (1994): Dream report length is more dependent on arousal level than prior REM duration. Brain. Res. Bull. 34, 99-101.

  12. Antrobus,J (1983): REM and NREM sleep reports: comparison of word frequencies by cognitive classes. Psychophysiology 20, 562-568.

  13. Foulkes,D; Schmidt,M (1983): Temporal sequence and unit composition in dream reports from different stages of sleep. Sleep 6, 265-280.

  14. Fein,G; Feinberg,I; Insel,TR; Antrobus,JS; Price,LJ; Floyd,TC; Nelson,MA (1985): Sleep mentation in the elderly. Psychophysiol. 22, 218-225.

  15. Cavallero,C; Foulkes,D; Hollifield,M; Terry,R (1990): Memory sources of REM and NREM dreams. Sleep 13, 449-455.

  16. Cavallero,C; Cicogna,P; Natale,V; Occhionero,M; Zito,A (1992): Slow wave sleep dreaming. Sleep 15, 562-566.

  17. Cavallero,C (1987): Dream sources, associative mechanisms, and temporal dimension. Sleep 10, 78-83.

  18. Cipolli,C; Fagioli,I; Baroncini,P; Fumai,A; Marchio,B; Sancini,M (1988): The thematic continuity of mental experiences in REM and NREM sleep. International Journal of Psychophysiology 6, 307-313.

  19. Cicogna,P; Cavallero,C; Bosinelli,M (1986): Differential access to memory traces in the production of mental experience. International Journal of Psychophysiology 4, 209-216.

  20. Cicogna,P; Cavallero,C; Bosinelli,M (1991): Cognitive aspects of mental activity during sleep. American Journal of Psychology 104, 413-425.

  21. Foulkes,D (1982): A cognitive-psychological model of REM dream production. Sleep 5, 169-187.Antrobus,J (1987): Cortical asymmetry and sleep mentation. Psych.Rev. 94, 359-368.

  22. Ehrlichman,H; Antrobus,J; Weiner,M (1985): EEG asymmetry and sleep mentation during REM and NREM sleep. Brain and Cognition 4, 477-485.

  23. Feinberg,I; March,JD (1995): Observations on delta homeostasis, the one-stimulus model of NREM-REM alternation, and the neurobiologic implications of experimental dream studies. Behavioural Brain Research 69, 97-108.

  24. Hall,CS (1981): Do we dream during sleep? Evidence for the Goblot Hypothesis. Perceptual and Motor Skills 53, 239-246.

Pg. 12