Relation of Eye Movements During Sleep to Dream Activity: An Objective Method for the Study of Dreaming

Introduction

  • William Dement and Nathaniel Kleitman's 1957 study in the Journal of Experimental Psychology investigates the relationship between eye movements during sleep and dream activity.
  • The study aims to establish an objective method for determining when dreaming occurs by correlating subjective dream reports with physiological phenomena.
  • Previous research by Aserinsky and Kleitman (1) found a high incidence of dream recall during periods of rapid eye movements (REM) and a low incidence at other times.
  • Further studies confirmed the relationship between REM and dreaming in both normal subjects and schizophrenics (4), and showed that REM periods occur at regular intervals related to cyclic changes in sleep depth as measured by EEG (5).

Methods

  • Three approaches were used to rigorously test the relation between eye movements and dreaming:
    • Dream recall during REM or quiescent periods was elicited without direct contact between experimenter (E) and subject (S) to eliminate unintentional cuing.
    • Subjective estimates of dream duration were compared with the length of REM periods, hypothesizing a positive correlation if dreaming and eye movements are concurrent.
    • The pattern of eye movements was related to dream content to test whether they represent a specific expression of the visual experience of dreaming or merely a random motor discharge of a more active central nervous system.
  • Participants included seven adult males and two adult females. Five were studied intensively, while data from the other four were used to confirm the results of the first five.
  • In a typical experiment, the subject (S) reported to the laboratory before their usual bedtime, having abstained from alcohol or caffeine on the day of the experiment.
  • Electrodes were attached near the eyes to register changes in the corneoretinal potential fields as the eyes moved, and to the scalp for recording brain waves as a criterion of sleep depth.
  • Electrode lead wires were attached to the top of the head and then to a lead box at the head of the bed to minimize entanglement and allow free movement.
  • Potentials were amplified by a Model III Grass Electroencephalograph in an adjoining room, run continuously throughout the sleep period at a speed of 3 or 6 mm per second.
  • A faster speed (3 cm/sec) was used for detailed examination of brain waves, although the slower speed permitted an approximate estimation of the gross pattern.

Procedure

  • Subjects were awakened at various times during the night to test their dream recall.
  • Return to sleep after awakening invariably took less than 5 minutes.
  • Table 1 summarizes the experiments, showing the number of nights each subject slept and the number of awakenings.
  • 21% of awakenings occurred in the first 2 hours of sleep, 29% in the second two, 28% in the third two, and 22% in the fourth two.

Results

  • Discrete periods of rapid eye movements (REM) were observed in all nine subjects every night they slept.
  • These periods were characterized by a low-voltage, relatively fast pattern in the EEG.
  • Interspersed periods without rapid eye movements (NREM) showed EEG patterns indicative of deeper sleep, either a predominance of high-voltage, slow activity, or frequent, well-defined sleep spindles with a low-voltage background.
  • No REMs were observed during the initial onset of sleep, although the EEG always passed through a stage similar to that accompanying REM periods later in the night.
  • The mean duration of REM periods varied between 3 and 50 minutes, with a mean of about 20 minutes, tending to be longer later in the night.
  • Eye movements occurred in bursts of one or two, up to fifty or a hundred movements.
  • A single movement generally lasted 0.1-0.2 seconds and was followed by a fixational pause of varying duration.
  • The amount, pattern, and size of the movements varied irregularly from period to period.
  • REM periods occurred at fairly regular intervals throughout the night, with frequency characteristic for each individual.
    • DM and WD averaged one eye-movement period every 70 and 75 minutes, respectively.
    • KC averaged one eye-movement period every 104 minutes.
    • The average for the whole group was one REM period every 92 minutes.
  • Awakening during a NREM period did not trigger REMs upon return to sleep, nor did it change the onset of the next REM period.
  • Awakening during an REM period generally terminated the REMs until the next period, with the EEG sequence resembling that following a spontaneous end to an REM period.
  • Exceptions occurred during the final hours of sleep, where REMs sometimes restarted after awakening.

Dream recall

  • Subjects were awakened by a doorbell and spoke into a recording device, stating whether they had been dreaming and relating the dream content.
  • There was no communication between subject and experimenter until the subject committed themselves regarding dream recall.
  • Subjects were considered to have been dreaming only if they could relate a coherent, fairly detailed description of dream content.
  • Awakenings occurred during REM periods or at varying times after the cessation of eye movements during NREM periods.
  • Subjects were not informed whether their eyes had been moving.
  • Table 2 shows the results of dream recall attempts after various awakenings.
  • PM and KC's awakenings were chosen randomly, DN followed a pattern (three REM, then three NREM), and WD was told he would only be awakened when dreaming but was randomized. IR was chosen according to the whim of E.
  • Subjects showed a high incidence of dream recall following REM awakenings and a low incidence following NREM awakenings, regardless of how the awakenings were chosen.
  • DN was not more accurate despite the pattern, and WD was not less accurate despite being misled.
  • Practice was not a significant factor, with only one subject showing improvement in recall on later nights.
  • Dream recall dropped rapidly after cessation of REMs.
    • 5 dreams were recalled in 17 NREM awakenings within 8 minutes after a REM period.
    • Only 6 dreams were recalled in 132 awakenings when NREM awakenings followed REM periods by more than 8 minutes.
  • Subjects were most emphatic about not dreaming when NREM awakenings occurred during an intermediate stage of sleep, characterized by a brain-wave pattern of spindling with a low-voltage background.
  • When aroused during deep sleep (high-voltage, slow waves in the EEG), subjects often felt bewildered, unsure if they had been dreaming or asleep.
  • Most instances of inability to recall dreaming after REM awakenings occurred early in the night.
    • 19 negative reports occurred after REM awakenings in the first 2 hours of sleep, 11 in the second 2 hours, 5 in the third 2 hours, and 4 in the last 2 hours.
    • There was no such variation for NREM awakenings.

Dream Duration

  • Assessing if REM period lengths were proportional to subjective dream duration estimates helps establish the relatedness of REM and dreaming.
  • Subjects were initially awakened at various times after REM onset and asked to estimate dreaming time, but this was too difficult.
  • A series was then done in which subjects were awakened either 5 or 15 minutes after REM onset and had to decide which duration was correct based on their dream recall.
  • Table 4 shows the results of these awakenings.
  • All subjects except DN accurately chose the correct dream duration.
  • DN's errors were consistent with recalling only the latter part of a longer dream.
  • Dream narrative lengths were also influenced by factors like the subject's loquacity.
  • However, dream narrative lengths still showed a significant relationship to REM period duration.
  • Table 5 shows the correlations between minutes of REMs and lengths of dream narratives for each subject, measured by the number of words.
  • r values ranged from 0.40 to 0.71, with p values ranging from <0.001 to <0.05
  • Dream narratives after 30 or 50 minutes of REMs were not much longer than those after 15 minutes, possibly due to inability to remember all details of very long dreams.

Eye-Movement

  • The quality and quantity of REMs showed endless variation.
  • The movements occurred in bursts of activity separated by periods of relative inactivity.
  • The brain-wave stage remained the same whether there was much or little movement at any given moment.
  • The hypothesis was that the movements represented the visual imagery of the dream, i.e., corresponded to where and at what the dreamer was looking.
  • An attempt to account for every movement by having subjects state chronologically what directions they gazed in the dream proved futile.

Eye Movement Patterns

  • Subjects were awakened as soon as one of four predominant patterns of movement had persisted for at least 1 minute and were asked to describe the dream content just before awakening.
    • The four patterns were:
    • Mainly vertical eye movements
    • Mainly horizontal movements
    • Both vertical and horizontal movements
    • Very little or no movement.
  • The prevalence of horizontal or vertical components was determined by placing leads both vertically and horizontally around the eyes.
  • A total of 35 awakenings was accumulated from the nine subjects.
  • Periods of either pure vertical or horizontal movements were extremely rare.
  • Three periods of vertical movements were seen. The dream content involved action in the vertical plane.
    • One subject dreamed of standing at the bottom of a tall cliff, operating a hoist, and looking up at climbers and down at the machinery.
    • Another subject dreamed of climbing ladders, looking up and down.
    • A third subject dreamed of throwing basketballs at a net, looking up at the net and down to pick up another ball.
  • One instance of pure horizontal movement was seen. The subject was watching two people throwing tomatoes at each other.
  • On 10 occasions, subjects were awakened after 1 minute of little or no eye movement. The dreams all had the common property that the dreamer was watching something at a distance or staring fixedly at some object.
  • In two of these awakenings, the patterns were the same:
    • About a minute of ocular inactivity followed by several large movements to the left just a second or two before awakening.
    • In one case, the subject was driving a car, staring at the road, approached an intersection, and was startled by a car speeding from the left as the bell rang.
    • In the other case, the subject was also driving, staring at the road, and saw a man standing on the left side of the road and hailed him as he drove past.
  • In the 21 awakenings after a mixture of movements, subjects were always looking at things close to them.
    • Typical reports were of talking to a group of people, looking for something, fighting with someone, etc.
    • There was no recall of distant or vertical activity.
  • In order to confirm the meaningfulness of these relationships, 20 naive subjects as well as 5 of the experimental subjects were asked to observe distant and close-up activity while awake.
  • Horizontal and vertical electrodes were attached. The eye-movement potentials were comparable in both amplitude and pattern to those occurring during dreaming.
  • There was virtually no movement when viewing distant activity, and much movement while viewing close-up activity.
  • Vertical eye-movement potentials were always at a minimum except for upward movements accompanying blinking, and in a few cases when the experimenter tossed a ball in the air for them to watch.

Discussion

  • The results indicate that dreaming accompanied by REMs and a low-voltage electroencephalogram occurred periodically in discrete episodes during the course of a night's sleep.
  • It cannot be stated with complete certainty that some sort of dream activity did not occur at other times.
  • The lack of recall and the fact that the brain waves were at the lightest level of sleep only during REM periods and at deeper levels at all other times, makes this unlikely.
  • The few instances of dream recall during NREM periods are best accounted for by assuming that the memory of the preceding dream persisted for an unusually long time.
  • This is borne out by the fact that most of these instances occurred very close, within 8 minutes, after the end of REM periods.
  • Other workers have attempted to relate dreaming to physiological phenomena during sleep.
    • Wada (12) felt that dreaming and gastric contractions occurred simultaneously, but this conclusion was based on only seven awakenings in two subjects.
    • Scantlebury, Frick, and Patterson (11) also studied gastric activity and dreaming and felt, on the basis of three instances of dream recall out of seven awakenings, that the two were probably related.
    • The occurrence of dreaming during a series of foot twitches occurring immediately after the onset of sleep was postulated by McGlade (9), but this conclusion was based mainly on dreams recalled on the morning after the experiments, which is highly unreliable.
  • Incidental observations have been made on the occurrence of dreaming by investigators studying brain waves during sleep (2, 3, 6, 7, 8).
  • All stages of brain waves were related to dreaming in these five papers, but no mention was made of whether or not actual dream content was recalled, and the number of reports by sleepers was generally very small.
  • In other studies of dreaming, excellently reviewed by Ramsey (10), attempts were made to localize dream activity by simply awakening subjects at various times during the night.
  • In general, it was found that dreams might be recalled at any time during the night, but that most were recalled in the later hours of sleep.
  • This would correspond to the statistical incidence of REMs as previously reported (1, 4), and is also consistent with the finding in this study that, even when the awakenings occurred during REM periods, recall was still more difficult earlier in the night.
  • It was stated herein that all subjects showed periods of REMs every night they slept. This was also the case in another briefly reported series of experiments involving 16 subjects who were observed a total of 43 nights (5).
  • It is felt on the basis of these and other studies that periods of REMs and dreaming and the regularity with which they occur are an intrinsic part of normal sleep.
  • In view of this, the failure to observe REMs in occasional subjects reported in earlier work (1, 4) deserves some consideration.
    • One explanation is that the recording was done by sampling rather than continuously.
    • Another explanation is that a lower amplification of the REM potentials was employed which, although usually adequate, did not clearly record very small movements.
    • A third possibility is that the dreams of these subjects happened to be the sort, such as watching distant activity, in which eye movement was at a minimum.
  • There was nothing in the experiments reported in this paper to indicate that the dreams occurred instantaneously, or with great rapidity, as some have supposed.
  • Rather, they seemed to progress at a rate comparable to a real experience of the same sort.
  • An increment in the length of REM periods was almost invariably associated with a proportional increase in the length of the dream.
  • It seems reasonable to conclude that an objective measurement of dreaming may be accomplished by recording REMs during sleep.
  • This stands in marked contrast to the forgetting, distortion, and other factors that are involved in the reliance on the subjective recall of dreams.
  • It thus becomes possible to objectively study the effect on dreaming of environmental changes, psychological stress, drug administration, and a variety of other factors and influences.

Summary

  • Regularly occurring periods of REMs were observed during every night of experimental sleep in nine adult subjects.
  • A high incidence of dream recall was obtained from subjects when awakened during REM periods and a very low incidence when they were awakened at other times.
  • A series of awakenings was done either 5 or 15 minutes after the REMs (dreaming) had begun and subjects judged the correct dream duration with high accuracy.
  • The pattern of the REMs was related to the visual imagery of the dream, and the eye movements recorded in analogous situations while awake corresponded closely in amplitude and pattern to those observed during dreaming.