Williams et al. (1992): Bizarreness in Dreams and Fantasies

Why compare dreams with waking fantasies?

Dreaming and imagining while awake both involve experiences that are not simply descriptions of what is happening around us. Yet dreams often contain sudden changes of scene, impossible events and strange combinations of people or places. Williams et al. (1992) investigated whether dreams were more bizarre than waking fantasies, and whether this supported the activation-synthesis hypothesis.

Activation synthesis explains dreaming as the brain’s attempt to interpret internally generated neural activity. During REM sleep, circuits in the brainstem become active, along with areas of the limbic system involved in emotion and memory, including the amygdala and hippocampus. The brain combines and interprets this activity into a dream. Strange patterns of activation may therefore produce strange experiences.

In ordinary wakefulness, the brain processes information about the surrounding world. A waking fantasy, however, is spontaneous imagined mental activity rather than a direct response to an external stimulus or a deliberately planned thought. This makes it a useful comparison with dreaming: both involve internally experienced stories or images, but occur in different physiological states.

Williams and colleagues reasoned that the different brain states of waking and REM sleep were unlikely to produce identical mental experiences. Their aim was to assess bizarreness in dreams and fantasies as evidence for activation synthesis. A suitable prediction was that dreams would contain more bizarreness than waking fantasies reported by the same participants; this should be distinguished from a formally stated hypothesis.

Who took part and what did they record?

The sample consisted of 12 students: two male and ten female, aged 23–45. They were enrolled on a biopsychology course at Harvard University Extension School in Cambridge, Massachusetts, USA. These were the 12 out of 40 students who supplied sufficient dream and fantasy reports.

The study used home-based self-reports in written journals. The same participants supplied both kinds of report, allowing a comparison of dreaming and waking fantasy within the same group. Researchers did not experimentally assign participants to different brain states or directly measure their neural activity.

The materials were writing equipment, paper and Hobson et al.’s (1987) Bizarreness Scale. During term time, students were asked to record all remembered dreams when they woke during the night or in the morning. They also recorded waking experiences that met the definition of fantasy: spontaneous narrative or perceptual mental activity without a clear link to external stimuli or conscious intention.

Students were asked to submit ten dreams and ten fantasies early in the course, before discussion of specific differences between waking and dreaming cognition. This helped reduce the chance that such teaching would shape their reports.

Turning personal experiences into comparable scores

Researchers selected 60 dream reports and 60 fantasy reports. Each selected report had to be longer than five lines and describe formed visual perceptions: recognisable visual experiences rather than no visual content. The 120 reports ranged from 30 to 600 words.

Each report was divided into sentence units and scored using two aspects of the Bizarreness Scale:

For example, suppose an imagined story suddenly moves from a classroom to a beach without explanation. This would illustrate plot discontinuity: a break in the sequence or setting. An ordinary object behaving impossibly would illustrate plot incongruity, while an unclear character identity could illustrate uncertainty. These examples show how a broad judgement of ‘strange’ can be broken down into more specific categories.

Reports were transcribed and randomly coded before three judges scored all 120 reports. The judges worked independently. Random coding helped reduce biased scoring, while independent judgements allowed researchers to assess inter-rater reliability.

What did the researchers find?

The judges agreed about 80% of the time across bizarre and non-bizarre items. However, agreement on the precise category of bizarre items alone was lower, at 50–60%, showing that the definition was not completely clear. The analysis therefore focused on items where at least two judges agreed. Judges also identified a random sample of reports as dreams or fantasies with 88.7% accuracy, without being told which they were.

The central finding was that bizarreness was more than twice as prevalent in dreams as in fantasies. Mean bizarreness density was 0.223 for dreams and 0.089 for fantasies. Density measures how concentrated bizarre content is within a report, rather than simply counting how much text was written. Dream reports were significantly longer, so this distinction matters when comparing them.

Dreams scored higher for plot discontinuity, plot incongruity, uncertainty and thought incongruity. Plot discontinuity showed the greatest difference. Neither dreams nor fantasies contained items scored as thought discontinuity, emotion discontinuity or emotion uncertainty.

Bizarre dream content came from nearly all participants, whereas bizarre fantasy content came from only a few. Nevertheless, the pattern was not universal: seven of the 12 participants showed a statistically significant difference between their dreams and fantasies, and one participant’s fantasies scored higher for bizarreness than her dreams, although that difference was not significant.

There were also differences beyond bizarreness:

• Dreams were set in remote times or places across 12/12 participants; fantasies were more often current in both time and place, recorded for 6/12 participants.
• Dreams involved more than one character across 12/12 participants. Fantasies involving only the self were recorded for 4/12 participants.

What do the findings mean for activation synthesis?

The researchers concluded that dreams and waking fantasies are different modes of information processing. Dreams contained more bizarreness and other ‘dreamy’ features, including remoteness of time and place and more characters.

This supports activation synthesis because the theory links different neural activity during REM sleep and wakefulness to different mental experiences. The observed differences are consistent with that explanation.

However, the distinction is not absolute. Waking fantasies can also be bizarre, and both states share cognitive features. The conclusion is therefore that dreaming and waking fantasy differ in degree and in typical characteristics, not that one is always strange and the other never is.

How convincing is the study?

Independent scoring was a strength. Three judges used a common coding system, and researchers checked their agreement rather than relying on one person’s impression. Restricting analysis to items agreed by at least two judges reduced the influence of individual interpretations. However, the lower agreement on specific bizarre categories shows that scoring still involved subjective judgement.

Recording experiences at home captured everyday dreaming and fantasy. This avoided requiring all experiences to occur in an unfamiliar laboratory. Comparing reports from the same people also reduced the influence of differences between separate groups of participants.

Home recording lacked control. Participants’ environments and recording techniques could differ. Researchers could not be certain that reports were written immediately. During a delay, a person might forget details or organise the experience into a more coherent story, removing discontinuities or uncertainties. Reports might also be omitted or edited to avoid embarrassment. These possibilities weaken confidence that the journals accurately represented the original experiences.

The physiological states were not verified. Some dreams could have occurred in non-REM sleep, and some fantasies near sleep onset. Researchers tried to reduce this problem by excluding explicitly reported sleep-onset fantasies and selecting reports with visual content, but they did not directly establish that every dream occurred during REM sleep. This limits the strength of the link between the results and REM brain activity.

The sample limits generalisation. Twelve adult biopsychology students from one institution, mostly female and selected because they supplied enough reports, may not represent people of other ages, backgrounds or reporting habits.

Overall, the study provides useful evidence that dreams tend to be more bizarre than waking fantasies. Its systematic comparison supports activation synthesis, but the small sample, subjective scoring and unverified brain states mean it is not a direct demonstration of the theory’s biological mechanism.

Aim and theory link

Williams et al. (1992): compare bizarreness in dreams and waking fantasies to assess support for activation synthesis.

Different neural activity during REM sleep and wakefulness should produce different mental experiences.

Method essentials

• Home-based written self-reports; the same participants reported dreams and fantasies.