Dement and Kleitman (1957)

Aim General: To investigate the relationship between eye movements and dreaming. Specific: 1. Does dream recall differ between REM and NREM stages of sleep?

General: To investigate the relationship between eye movements and dreaming.

  1. Does dream recall differ between REM and NREM stages of sleep?
  2. Is there a positive correlation between estimated dream duration and REM period length?
  3. Are eye movements related to the dream content?

Research Hypothesis

  1. There will be a significant association between REM sleep and dreaming.
  2. There will be a positive correlation between estimated dream duration and REM period length.
  3. There will be a significant association between eye movement patterns and dream content.

Psychometrics used in the study:

  1. EEG (electroencephalogram) = traces cyclical changes that occur in brain activity during sleep. Electrodes are placed around the skull to analyse brain waves.
  2. EOG (electrooculogram) = traces eye movements during sleep. Uses electrodes placed around the eye region.

REM ➜ Rapid Eye Movement is when dreaming occurs.
NREM ➜ Non-Rapid Eye Movement.

Aserinsky (1955) (a student of Kleitman) revealed that participants who were woken from REM sleep were more likely to report a vivid, visual dream than participants that woke up from NREM. They also showed that we have several sleep stages alternating between REM and NREM.

Research Method: A laboratory experiment, but different methods to test each aim.

Research Design
Approach 1: natural experiment; repeated measures design; IV - whether they woke up from REM or NREM; DV - whether they recalled a dream or not.

Approach 2: True experiment using correlational study; repeated measures design; IV - waking participants after 5 or 15 minutes into REM sleep; DV - participants guess on dream duration (5 or 15 minutes); correlational analysis used to cross-check participants' estimate dream duration and the word count of their respective dream narrative.

Approach 3: natural experiment; repeated measures design; IV - eye movement patterns; DV - dream content.


  • 7 males and 2 females were recruited through opportunity sampling. 5 studied in detail and 4 used to confirm the results of the first 5.
  • 5 main participants spent between 6 - 17 nights in the lab. Approximately 50 - 77 times awakening.
  • 4 spent only 1 - 2 nights. 4 - 10 times awakening.
  • Participants were identified by their initials.

Participants reported to the lab before their personal bedtime. They ate their normal diet but were asked to avoid caffeine (alertness) and alcohol (drowsiness) on the day of the study. They slept in a dark, quiet room. They had 2 EOG electrodes near their eye and 2/3 EEG electrodes to the scalp. A doorbell (for standardisation) had been used to wake participants up at random from REM or NREM. All participants were woken up when an eye movement pattern lasted for at least a minute. Everyone returned to sleep in less than 5 minutes.

Procedure 1 - They were woken up at various times to test their dream recall (during REM and NREM). Dream narrative recorded on a tape recorder (to prevent researcher bias). They were asked if they had a dream or not, and if they did, then they recorded it. Dream only counted if the recall was clear.

Procedure 2 - Participants were woken up after either 5 or 15 minutes into their REM sleep. Participants guessed the duration they had dreamt for. The number of words in the dream narrative was counted after the participants reported their dream.

Procedure 3 - Participants’ eye movement direction was detected with the EOG. Participants were woken up and they reported their dream.


  • All subjects showed REM every night. REM patterns varied per individual but each individual had a regular REM pattern.
  • 92 minutes was the average time gap between different dreams. The range was 70 - 104 minutes.
  • The average REM length was 20 minutes and the range was 3 - 50 minutes. It was longer later in the night. Bursts of 2 - 100 rapid eye movements.
  • Those woken in NREM returned to NREM. Those woken in REM went to NREM (but sometimes went to REM to complete the final phase).

Results 1:
Awakenings from REM produced a dream recall of 79.6%, and from NREM produced a dream recall of 7%.
Waking participants under 8 minutes of completing their REM period resulted in 5/17 dreams being recalled. However, waking participants after 8 minutes resulted in only 6/132 dreams being recalled.

Results 2:
The estimation of REM duration was accurate and very high. 88% for 5 mins; 78% for 15 minutes.
There was a positive correlation between REM duration and words in the recall. The narratives of 152 dreams were collected, but 26 were omitted due to poor recording thus, there were 126.


Results 3:
Vertical eye movement - standing at the bottom of a tall cliff and operating a hoist.
Horizontal - 2 people throwing tomatoes at each other.
Vertical and horizontal - talking to people standing close to them.
Little or no movement - watching something in the distance or staring at an object.


  1. Dreams occur during REM sleep only. Dreams reported from NREM sleep are from previous REM episodes.
  2. Estimated dream duration and REM period lengths are very similar, thus it shows that dreams are not instantaneous events but rather experienced in real-time.
  3. Eye movements correspond to where, and what the dreamer is looking at in the dream, hence, it explains that they are not random events.

Strengths and Weaknesses

  • Reliability is high as it was a lab experiment with many controls. For example, the doorbell made people instantly wake up so that dreams wouldn't be forgotten by slow-woken people.
  • Demand characteristics were avoided as participants were not told whether they were in REM or not as otherwise, they would try recalling harder.
  • Exhibits validity as the details recorded focused on dreaming; the definition of a 'dream' had been operationalised; and asking participants to choose between 5 or 15 mins helped reduce participant variables such as the ability to guess.
  • Quantitative data such as brain waves, eye movement patterns, and REM sleep duration was collected through the EEG and EOG.
  • Qualitative data such as the dream content was collected, but it's subjective and can affect the validity since the narrative length is not only dependent on the REM phase length, but also on the participant's expressiveness.
  • Both genders were included thus, there's generalisability. However, the sample size is too small, therefore limiting generalisability.
  • Deception of participant WD being woken up in the wrong sleep stage can cause distress as they'd try recalling dreams harder. Context: Participant WD had been told they'd be woken up in their REM stage of sleep, but was actually woken up randomly during their REM or NREM stages of sleep.
  • Lacks ecological validity as people who are used to taking alcohol & caffeine may experience atypical dreams. Sleeping in a lab connected to electrodes would be unusual, and this may tamper with their sleeping behaviour.

Nature vs. Nurture
Dream content relates to our experience, so it is a product of nurture. However, the ability to dream is a product of nature.