The 1992 study by Williams and colleagues tested whether dreams are more bizarre than daytime fantasies. Most dreaming happens in REM sleep, when the brain is very active. The Activation Synthesis Theory says the brain’s random signals then become a story. Judges who did not know which reports were which found dreams far more bizarre, which supports the theory.
Key Takeaways
Aim
- To compare how bizarre (odd or impossible) dream reports and waking fantasy reports are. The theory says brain signals in REM sleep (when most dreaming happens) are turned into a story.
- Hypothesis: dreams will contain more bizarreness than waking fantasies.
Method
- A natural experiment (the researchers did not control whether a report was a dream or a fantasy) using self-report (people write about their own experiences).
- Twelve American students, two male and ten female, aged 23 to 45, kept a journal for a term. They wrote down dreams on waking and fantasies (daydreams) while awake. Each gave at least 10 of each.
- Sixty dream reports and 60 fantasy reports were split into one-sentence units. Three judges, who did not know which was which, scored bizarreness on a coding scale. Then 12 dreams and 12 fantasies went into a blind test.
- The independent variable (what is compared) was dream or fantasy. The dependent variable (what is measured) was the bizarreness score.
Results
- Dreams were more bizarre. Dreams scored 0.223 bizarre items per sentence and fantasies only 0.089, less than half as much.
- Ninety per cent of dream reports contained something bizarre, against about 27% of fantasy reports.
- Judges who did not know which was which told dreams from fantasies 88.7% of the time.
Conclusion
- Dreams contain more bizarreness than fantasies. This supports the Activation Synthesis Theory, because the brain is in a different state in REM sleep.
- Dreaming and fantasising are different kinds of thinking, because judges could tell them apart.
Strengths
- Reliability: three judges scored every report on their own and did not know the type. This reduces observer bias (judges seeing what they expect), and inter-rater reliability (how far judges agree) could be checked.
- Quantitative data: a coding scale turned dreams into scores that could be compared.
- Same people: the same 12 students gave both kinds of report, so differences between people could not explain the results.
Limitations
- Representativeness: only two men, all university students, aged 23 to 45. This shows gender bias (too many of one sex) and age bias (a narrow age range), so the sample is not typical of everyone.
- Self-report: dreams were written down at home, possibly after a delay. Reports may have been edited, so bizarreness could be lost.
- REM not recorded: nobody checked that the dreams came from REM sleep, so the link to brain activity is an assumption.
Ethical Issues
- Informed consent: the paper does not say how agreement to use the journals was obtained. A way of dealing with it is to explain the research and ask if people are happy to take part.
- Confidentiality: some full reports were printed and could hold personal details. A way of dealing with it is to use numbers, not names, and to leave out private details.
Exam Use
- In the exam you get marks for the method (natural experiment and self-report), the sample (12 students, two male and ten female) and features of the procedure.
- Do not confuse Williams with Freud (who said dreams are hidden wishes). Williams used journals, judges and scores. Freud studied one man.
Background
The study tests the Activation Synthesis Theory of dreaming, put forward by Hobson and McCarley (1977). The theory says dreams are the mind’s attempt to make sense of brain activity during sleep.
The theory has these steps.
- In REM sleep (rapid eye movement sleep, when most dreaming happens), the pons in the brainstem sends out bursts of signals.
- The signals stimulate the limbic system, which controls emotion.
- Random brain waves travel up to the cerebral cortex.
- The cortex treats the waves as if they were real information from the senses. It uses stored memories to make sense of them. This is called synthesis.
The activity is random, so the dream is often bizarre. Bizarre means odd, impossible or full of sudden changes.
If the theory is right, dreams should be more bizarre than thinking when awake. The brain is in a different state. Not everyone accepted this. Some researchers did not think dreams are very bizarre.
The authors thought it unlikely that waking and REM sleep, two very different brain states, would give the same kind of thinking. So they measured bizarreness. They did this in dreams and in waking fantasies.
The other idea you meet in this topic is Freud’s dream theory. Freud said dreams are hidden wishes in disguise, so a dream needs interpreting to find its meaning. Williams needs no hidden meaning.
Aim
The aim was to compare how bizarre dream reports and waking fantasy reports are. The study put the theory to the test.
The hypothesis was simple. Dreams will contain more bizarreness than waking fantasies. The authors also noticed, in a preliminary check, that dreams had more characters and more remote times and places.
A fantasy here is a mental story or picture that comes into your head when you are awake. It has no clear link to something you can see or hear, and you did not plan it.
Method
Design and Variables
Williams, Merritt, Rittenhouse and Hobson (1992) used a natural experiment with self-report.
In a natural experiment, the researcher does not change the thing being compared.
It happens by itself. Here, whether a report came from sleep or waking life was not controlled by the researchers. Self-report means the people wrote down their own experiences.
- The independent variable is what is compared. It was whether the report described a dream or a fantasy.
- The dependent variable is what is measured. It was how bizarre the report was, as a bizarreness score. The paper does not use these two terms. In an exam answer you should.
The same 12 people supplied both kinds of report. This is like a repeated measures design, where the same people take part in every condition.
Reports were collected at home, not in a sleep laboratory. The authors call the study a pilot study, which means a small first test.
Sample
The sample was 12 American students, two male and ten female, aged 23 to 45. They studied biopsychology at Harvard University Extension School in Massachusetts.
All 40 students on the course were asked to keep journals. The numbers narrowed like this.
- 12 students: They reported at least 10 dreams and 10 fantasies. Their 12 journals made up the data.
- 60 dreams and 60 fantasies: These were chosen from the journals for their length and visual images. All 120 reports were scored.
- 12 dreams and 12 fantasies: These were picked from the 120 for the blind test.
So the sample was chosen by how many reports the students produced.
Materials
- A journal: Students wrote in it on waking in the night and in the morning. They recorded any dreams they remembered.
- A definition of fantasy: Waking mental activity was recorded if it was spontaneous, a story or a picture, and had no clear link to anything outside or to a plan.
- A bizarreness scale: Each report was scored for what was bizarre and what kind of bizarreness it was.
The scale asked two questions about each odd item. What is it? Where is it: in the plot, in a person’s thoughts, or in a feeling?
So a sudden jump to a new scene is an odd item in the plot, of the first type below.
- Discontinuity: A sudden change, such as a switch to an unrelated scene.
- Incongruity: Something that does not make sense in waking life, such as travellers stuffing a sheep into a suitcase.
- Uncertainty: Doubt about what a person, place or object really is.
A unit with nothing bizarre scored 0. A report’s bizarreness score is its number of bizarre items divided by its number of one-sentence units.
Procedure
- During a term, students wrote their dreams and fantasies in a journal. They recorded dreams on waking in the night and in the morning.
- Early in the course, students handed in 10 dreams and 10 fantasies. This was before any lesson on how waking thought and dreaming differ.
- Twelve students had written enough. Their journals were the data.
- Sixty dream reports and 60 fantasy reports were chosen. They were chosen for their length (more than five lines) and for describing formed visual images.
- Each report was split into one-sentence units. Each unit was scored for bizarreness on the scale.
- Three judges scored all 120 reports on their own. The reports were typed up and given random codes, so the judges could not tell a dream from a fantasy.
- In a blind test, 12 dreams and 12 fantasies were put in random order. Judges said which was which, gave reasons and rated how sure they were, from 1 to 3.
- The authors counted features in the judges’ reasons, such as remote times and places and the number of characters.
Results
Bizarreness in Dreams and Fantasies
Dreams were more bizarre than fantasies on every main measure. The table shows the results for the 60 dream reports and the 60 fantasy reports.
| Measure | Dreams | Fantasies |
|---|---|---|
| Bizarre items per sentence (average) | 0.223 | 0.089 |
| Reports with something bizarre | 90% (54 of 60) | about 27% (16 of 60) |
| Reports with a sudden change (discontinuity) | 30% | about 3% |
| Reports with something that made no sense (incongruity) | 85% | about 27% |
The fantasy score was less than half the dream score. The gap was too big to put down to chance, which means it was unlikely to be just luck.
Measured by score per report, the paper says the most clearly different feature was sudden changes of plot. Of the dream reports, 30% had one. Only about 3% of fantasy reports did.
Dreams were longer than fantasies. The authors checked whether length explained the result. Bizarreness was only weakly linked to length. The authors say the result was not just due to length.
All but one of the 12 students had more bizarreness in their dreams than in their fantasies. The one exception scored higher for her fantasies than for her dreams.
How Well Did the Judges Agree?
The three judges worked alone. They gave the same score on about eight items in ten.
This is a measure of inter-rater reliability, which is how far judges give the same scores.
They agreed less often on the exact kind of bizarreness. The authors said the meaning of bizarreness was ambiguous. So they analysed only items on which two or more judges agreed.
Telling Dreams from Fantasies
In the blind test, the judges decided correctly whether a report was a dream or a fantasy 88.7% of the time.
That is almost nine times in ten.
The test also showed two other differences.
- Remote times and places: All 12 dream reports were set in a remote time or place. Only 6 of the 12 fantasy reports were.
- Characters: Dreams averaged 3.25 characters and fantasies 1.67. About a third of the fantasies (4 of 12) had only the dreamer in them.
The judges’ mistakes were all among the reports that were less clear-cut.
These were not the most dream-like dreams or the most fantasy-like fantasies. About 27% of fantasy reports still had something bizarre in them, so the difference is one of amount, not of kind.
Conclusion
The authors reached these conclusions.
- Dreams contain more bizarreness than fantasies: Dreams scored more than twice as high as fantasies. They also had more characters and more remote times and places.
- REM brain activity is linked to bizarre dreams: The brain is in a different state in REM sleep than when awake. This fits the Activation Synthesis Theory.
- Dreaming and fantasising are different kinds of thinking: The judges could tell them apart, so the two are different mental abilities (different ways of thinking).
- The two overlap: The difference is one of amount, not of kind. Some fantasies were bizarre and some dreams were ordinary.
Strengths
The study has these strengths.
- Blind, independent judges: Three judges scored every report on their own. They could not tell dreams from fantasies. This reduces observer bias, which is when judges see what they expect to see.
- Inter-rater reliability: Judges worked separately, so how far they agreed could be checked. Reliability is high when different judges give similar scores.
- A standardised scale: One coding scale was used for every report. It gave quantitative data (numbers) that could be compared.
- Same people in both conditions: The same 12 students gave dreams and fantasies. This removes differences between people, such as how creative they are.
- Close to real dreaming: Reports came from normal sleep at home, not a laboratory. This gives the study higher ecological validity, which means the results fit real life.
- Fewer demand characteristics: Demand characteristics are clues about what a study is for. Early reports came before any lesson on dreams, which may have limited them. Later reports could still be affected.
The study also contrasts with the Freud study. Freud’s evidence is one man and one interpreter.
This study uses 120 reports and judges who did not know the type, so it is more objective. OCR names subjectivity as a criticism of the Freudian theory.
In the exam you get marks for strengths that are tied to the study. A general point such as “it was reliable” earns little. Say who did what, as in “the three judges worked on their own”.
Limitations
The study has these limitations.
- Representativeness: Only two men took part, they were university students, and they were aged 23 to 45. This shows gender bias and age bias, so the sample is not typical of everyone.
- Population validity: Population validity means how far the results apply to other groups. They may not apply to men, to people under 23 or over 45, or to people who are not students.
- Self-report from home: Nobody could check when reports were written. A delay could tidy a dream into a more sensible story and lose bizarreness. The authors suggest checking with laboratory reports.
- Social desirability: Some reports may have been left out or edited to avoid embarrassment. Social desirability means giving answers that look good. The authors do not think this was likely to affect the results.
- Dreams may not be REM dreams: Nobody recorded REM sleep, so it is only assumed that the dreams came from it. The study measures written reports and then infers what the brain was doing.
- Construct validity of bizarreness: Construct validity asks if a scale measures what it claims to. Bizarreness is hard to define, and judges agreed less on its exact kind.
- Length: Dreams were longer than fantasies. Length is a confound, a variable that could explain the results. The authors found bizarreness only weakly linked to length.
- Quantitative, not absolute: Some fantasies were bizarre and some dreams were not. The study cannot show that dreams are bizarre and fantasies are not.
- No control in a natural experiment: The researchers did not decide who dreamed or fantasised. Dreams and fantasies differ in more than brain state, so cause and effect cannot be shown.
Describing the Weakness of the Sample
In the exam you get marks for saying the sample is not representative.
This means it is not typical of the wider population. You must also say who it does not represent.
Examples are men (only two took part), people who are not students, and people aged under 23 or over 45. Saying “you cannot generalise” on its own does not earn the mark.
Ethical Issues
The paper does not describe consent, debriefing or ethics approval. Here is how the usual ethical issues apply to what it does describe.
- Informed consent: Students kept journals for a course. The paper does not say how agreement to use them was obtained. To deal with it, explain the research and ask if people agree.
- Confidentiality: The paper prints several full reports. One holds details such as a named person. A way of dealing with it is to use numbers or letters, not names, and leave out personal details.
- Protection from harm: Writing down dreams is low risk. But some reports may have been left out to avoid embarrassment. To deal with it, tell people they can stop and remove their results.
- Right to withdraw and debriefing: The paper describes neither. A way of dealing with them is to say people can withdraw, and to explain at the end what the study was for.
Exam Questions
What Examiners Look For
These are the question types the exam sets on this study, and what earns marks.
- Identify (1 mark each, multiple choice): The method (natural experiment and self-report), the sample (two male and ten female American students) and a true statement (an equal number of dreams and fantasies were analysed).
- Outline the procedure (4 marks): One mark for each feature you outline. Naming a feature is not enough. Features of the sample do not count as procedure.
- Outline a conclusion (2 marks): One mark for a brief conclusion and one for linking it to the findings. If you give two conclusions, only the first is marked.
- Weakness of the sample (2 marks): One mark for the idea of representativeness. One mark for showing who the sample does not represent.
- 13-mark essay: A judgement question on a theory, with a second area of psychology. See the end of this section.
The features of the procedure that earn marks are these.
- A written journal of dreams, with whether the student woke in the night or morning.
- Waking mental activity recorded if it met the definition of a fantasy.
- 60 dream and 60 fantasy reports chosen from the 12 students’ journals.
- Reports chosen for their length (over five lines) and for describing formed visual images.
- One-sentence units, each scored for bizarreness on a coding scale.
- Judges who did not know whether a report was a dream or a fantasy.
- Three judges scoring all 120 reports on their own, so inter-rater reliability could be worked out.
- The study lasting one term.
In 2023 some candidates confused Williams with Freud. Some also described the study when the question asked about the theory.
The questions and model answers below are examples written for revision. They are not the exam board’s own wording.
Williams or Freud?
Williams and Freud are the two core studies in Sleep and Dreaming. They sit on opposite sides of the topic.
| Feature | Williams study | Freud study (Wolf Man) |
|---|---|---|
| Theory it supports | Activation Synthesis Theory, a biological theory (it explains dreams by brain activity) | Freudian Theory of Dreaming, a psychodynamic theory (it explains dreams by hidden wishes) |
| Method and sample | Natural experiment using self-report: 12 students, two male and ten female, aged 23 to 45 | Case study of one patient: a young Russian man, rich, with depression |
| What was analysed | 120 written reports, scored by three judges as numbers (quantitative) | One man’s dreams and symptoms, centred on a childhood dream of wolves in a tree, interpreted by Freud (qualitative) |
| Idea about dreams | Bizarre dreams reflect REM brain activity, so no hidden meaning is needed | A dream disguises a hidden wish: the dream you remember hides its real meaning |
| Named criticism of its theory | Reductionism (explaining something complex through one simple cause) | Subjectivity |
Two quick tests help. If an answer mentions judges, scores, journals or REM brain activity, it is Williams. If it mentions one man, wolves, the unconscious or latent content, it is Freud.
In the exam, write “unconscious” for Freud. “Subconscious” is not the word to use.
Practice Question 1: Outline the Procedure (4 marks)
Question: Outline the procedure of the Williams study. [4 marks]
Model answer: Students kept a written journal for a term. They wrote down their dreams when they woke in the night and in the morning. They also wrote down waking fantasies.
Sixty dream reports and 60 fantasy reports were chosen from the journals of 12 students. Each report was split into one-sentence units. Each unit was scored for bizarreness on a coding scale.
Three judges scored all 120 reports on their own. They did not know which were dreams.
- Mark scheme: 1 mark for each feature of the procedure that is outlined, up to 4. Any four of the features in the list above are credited, in any order.
- Not credited: Features of the sample, such as “12 students”, do not earn marks here.
Practice Question 2: Outline a Conclusion (2 marks)
Question: Outline one conclusion of the Williams study. [2 marks]
Model answer: Dreams are more bizarre than fantasies. This is because the dream reports scored more than twice as high for bizarreness as the fantasy reports.
- Mark scheme: 1 mark for a brief conclusion. 1 mark for linking it to the findings.
- Other credited conclusions: Dreaming and fantasising are different cognitive abilities (ways of thinking), because the judges could tell them apart. Brain activity in sleep differs from waking, for the same reason.
Practice Question 3: A Weakness of the Sample (2 marks)
Question: Explain one weakness of the sample used in the Williams study. [2 marks]
Model answer: The sample was not representative of the wider population. It had only two men and ten women, so it does not represent men.
- Mark scheme: 1 mark for showing the idea of representativeness, that the sample is not typical of the wider population. 1 mark for showing who it does not represent.
- Credited examples: Two male and ten female students. University students, so not representative of non-students. Aged 23 to 45, so not representative of people under 23 or over 45.
- Not credited: The word “generalise” alone.
Practice Question 4: Identify the Method (1 mark)
Question: Which one of these is the method used in the Williams study? A case study. B natural experiment. C questionnaire. D interview. [1 mark]
Model answer: B, natural experiment.
- Mark scheme: 1 mark for the correct letter. The researchers did not control whether a report was a dream or a fantasy, so it is a natural experiment.
The 13-Mark Question
The 13-mark question is a judgement question. It is worth 6 marks for describing and 7 for evaluating. There are no marks for applying.
To reach the middle levels, you must refer to the named theory or study and to one from a different area. Williams cannot be paired with the Freud study. Both are in Sleep and Dreaming.
The examiners say good partners include social learning theory, Piaget’s theory, theories of schizophrenia or depression, and Eysenck’s personality theory. The mark scheme also names the multi-store model as a possible partner for reductionism or the value of biological explanations.
Check whether the question names a theory or a study. If it names the Activation Synthesis Theory, describing the Williams study earns nothing. Use the Williams study as evidence for the theory, and use its limits as your evaluation.
Reductionism is a good point to use. The theory explains dreams only through brain activity. Williams measures reports and infers the brain state, while the Freud theory looks at meaning.
Later Research
Recent studies still ask whether dreams really are more bizarre than waking thought. None repeats the 1992 method.
Kirberg and Windt (2026) compared dreams with mind wandering. This means thoughts drifting when you are awake. Each had its own pattern of bizarreness. So dreams are not just stronger mind wandering.
Kirberg (2022) argued that odd, jumbled thoughts are normal in any spontaneous thinking, asleep or awake. Yu and Shen (2020) found that lucid dreams, where the dreamer knows they are dreaming, were less bizarre than ordinary dreams.
References
Hobson, J. A., & McCarley, R. W. (1977). The brain as a dream state generator: An activation-synthesis hypothesis of the dream process. American Journal of Psychiatry, 134(12), 1335–1348. https://doi.org/10.1176/ajp.134.12.1335
Kirberg, M. (2022). Neurocognitive dynamics of spontaneous offline simulations: Re-conceptualizing (dream)bizarreness. Philosophical Psychology, 35(7), 1072–1101. https://doi.org/10.1080/09515089.2022.2042231
Kirberg, M., & Windt, J. (2026). The kaleidoscope of bizarreness: The analysis of first-person-reports shows the relationship between dreaming and mind wandering to be complex. Consciousness and Cognition, 137, 103965. https://doi.org/10.1016/j.concog.2025.103965
Williams, J., Merritt, J., Rittenhouse, C., & Hobson, J. A. (1992). Bizarreness in dreams and fantasies: Implications for the activation-synthesis hypothesis. Consciousness and Cognition, 1(2), 172–185. https://doi.org/10.1016/1053-8100(92)90059-J
Yu, C., & Shen, H. (2020). Bizarreness of lucid and non-lucid dream: Effects of metacognition. Frontiers in Psychology, 10, 2946. https://doi.org/10.3389/fpsyg.2019.02946