Eye Movements During Sleep


Have you ever noticed that a sleeping person’s eyes sometimes flicker around, up and down, left and right? Unless you have some sort of hobby that involves watching people sleep for long periods of time, you probably haven’t noticed this…

Anyway, this rapid eye movement is known as… rapid eye movement (REM). Dement and Kleitman, two very well-known researchers on sleep, hypothesised a connection between REM sleep and dreaming.

Dreams, especially during the time that this experiment took place, have fascinated many psychologists and psychiatrists. Freud argued that dreams preserve sleep by unconsciously fulfilling our wishes that would otherwise upset or disturb the sleeping dreamer. He called them the “royal road to the unconscious” and spent a whole lot of time analysing them during therapy sessions.

While Freud had his own perspective of dreams, more recent research focused on the cognitive and physiological explanations for dreaming. A cognitive viewpoint may see dreams as a way that humans deal with their personal or life problems; a physiological viewpoint may explain dreams as the result of neurons creating an image that we ourselves give meaning to.

The average sleeper will pass through different levels of non-REM (N-REM) sleep in a cycle before entering the REM period, and this occurs five to seven times.

  • Level one and Level two – light sleep characterised by irregular wave patterns on an EEG (electroencephalograph).
  • Level three and Level four – deeper sleep characterised by regular wave patterns on an EEG (electroencephalograph).

Level four is called “slow wave” or “deep” sleep. After this level, the sleeper goes back to Level two and a period of REM sleep begins, lasting for about 15 to 20 minutes. These sleep states alternate during the night: they start with a rapid fall into deep sleep, followed by increasing episodes of lighter sleep and REM sleep.

This can therefore be looked at in terms of five stages:

The researchers created three hypotheses:

  1. There will be a significant association between REM sleep and dreaming.
  2. There will be a significant positive correlation between the estimate of dream duration and eye-movement length.
  3. There will be a significant association between the pattern of eye movement and the context of the dream.

The sample consisted of nine people (seven males and two females). Five of the participants provided the most detailed information; the other four were used to support and compare the findings from the main five participants.

A participant would report to the laboratory a while before their usual bedtime. They had been instructed to eat normally but to avoid beverages containing alcohol or caffeine.  Electrodes were attached to their face and scalp; these were connected to an EEG (electroencephalograph) that would record electrical changes caused by eye movement and to measure brain activity. They then went to sleep in this dark and quiet room.

To test the first hypothesis, the participants were awakened during the night by a loud doorbell next to their bed and their dream recall was tested. The participant would then speak into a tape recorder near the bed. They had to state whether or not they had been dreaming and then, if possible, to report the dream content. Their dream was only recorded if they could relate a coherent and somewhat detailed description of the dream content. Different participants were woken according to different schedules: two were woken up randomly, one was woken at the experimenter’s whim, one was randomly woken but told he would be woken only during REM, and one was woken three times during REM followed by three in N-REM, and so on.

To avoid experimenter effects, the experimenter did not communicate with participants at night. To avoid bias, the participants were not told about their eye movement after they were woken up.

To test the second hypothesis, participants were woken up either five or 15 minutes into an REM period and asked whether they thought they had been dreaming for five or 15 minutes.

To test the third hypothesis, participants were woken up as soon as one of four patterns of eye movement lasted for at least one minute. When woken, the participant had to describe their dream content in detail. The four patterns that prompting an awakening were:

  1. Mainly vertical eye movements
  2. Mainly horizontal eye movements
  3. Both vertical and horizontal eye movements
  4. Very little or no eye movement

Nathaniel Kleitman was a Russian physiologist and sleep researcher; he also taught Dement in university

All participants showed periods of REM sleep at every night.

  • The REM EEG was characterised by a low-voltage, relatively fast pattern.
  • In between REM periods, the EEG patterns were high-voltage, slow activity or spindles with a low-voltage background; both of these are characteristics of deeper sleep.

REM never occurred at the beginning of a sleep cycle.

  • REM periods that were not terminated by an awakening varied between three minutes to 50 minutes (mean: 20 minutes) and tended to increase as the night progressed.

REM periods occurred at regular intervals but each participant had their own pattern.

  • The periods between REM sleep lasted between 70 to 104 minutes.

The results of the procedure that tested the first hypothesis were as follows:

  • REM sleep is predominantly (but not exclusively) associated with dreaming.
  • N-REM sleep is associated with periods of sleep without dreams.
  • All dream recall in N-REM awakenings occurred within eight minutes of an REM; this suggests that the dream content might just be remembered from the previous REM.

The results of the procedure that tested the second hypothesis were as follows:

  • All participants were able to choose the correct dream duration fairly accurately, except for one participant who underestimated the duration.

The results of the procedure that tested the third hypothesis were as follows:

  • Periods of only vertical or horizontal eye movements were rare.
  • When participants were woken up after a series of virtual eye movements, they reported dreams involving things like: standing at the bottom of a cliff to operate a hoist and looking up at the chambers, then down at the machinery; climbing up a series of ladder while looking up and then down; throwing basketballs at a net, first shooting while looking up at the net and then looking down to pick another ball off the floor.
  • In the only incident of horizontal eye movements, the dreamer was watching two people throwing tomatoes at each other.
  • In the 21 awakenings that occurred after a mixture of eye movements, participants were always looking at people or objects close to them (e.g. talking to a group of people or looking for something) and there was no recall of distant or vertical activity.
  • In the 10 incidents where participants showed little or no eye movements, the dreamers recalled watching something at a distance or just staring fixedly at an object.

In order to check the meaningfulness of the above relationships, 20 new naïve participants and five of the experimental participants were asked to observe distant and close-up activity while awake. Measurements from all these cases were comparable to those that occurred during dreaming.

William C. Dement is an American sleep researcher and founder of the Sleep Research Centre, the world’s first laboratory specifically catering to research on sleep

Type of research method
This was a laboratory study that took place with scientific equipment. No independent variable was manipulated (the participants slept themselves) so this can also be seen as a natural experiment. Self-reporting methodology was also applied: participants recalled and described their dreams and approximated their dream durations.


  1. Good control:There was a high level of controls over many extraneous variables, for example: none of the participants had caffeine or alcohol before the experiment, they slept in the same location, they were woken up with the same doorbell sound, the duration of sleep was controlled, and so on. This means the validity of the results is good.
  2. Replicable: The procedures were standardised and scientific equipment was used throughout the experiment. This means that the study can be replicated multiple times and the reliability of the results can be checked.


  1. Unrepresentative sample:The results of this experiment cannot be generalised onto a larger population because the sample consisted of only nine people, the majority of whom were men. The size was too small and there was no diversity in sex, age, ethnicity, etc.
  2. Low ecological validity: The entire environment was artificial; participants slept in a laboratory with electrodes plastered to their scalp and face and were woken up by the sound of a doorbell. This may have affected the standard of realism and distorted the results.
  3. Cause and effect identification: It is not clear, from this experiment alone, whether dreaming caused REM or if REM was causing dreams. A relationship was found, but the nature of that relationship is left unclear.

Ethical issues

  1. Informed consent: Yes, this was obtained.
  2. Deception: At a few points, some participants were deceived, such as during awakenings.
  3. Confidentiality: All the identities of participants were kept private.
  4. Emotional or physical harm: There was no real harm inflicted on participants.
  5. The right to withdraw: Participants could leave the study.
  6. Debriefing: No mention of debriefing.

Reference: Dement, W. and Kleitman, N. (1957). The Relation of Eye Movements During Sleep to Dream Activity. Journal of Experimental Psychology.