Basic Patterns in Dreaming

Basic Patterns in Dreaming by Kelly BulkeleyThe basic patterns of dream content are coming into sharper focus, thanks to new technologies of digital analysis. By using these tools to study large and diverse collections of high-quality dream data, and then making those tools and data publicly available, we can illuminate recurrent frequencies of dream content that others can easily review, replicate, and verify for themselves. The more we know about these basic patterns, the more we can gain helpful insights from people’s dreams regarding their mental and physical health, social relations, cultural interests, and even spiritual beliefs.

When I began this line of research in the mid-2000’s, I used the resources of the Dreambank.net, a site managed by G. William Domhoff and Adam Schneider. In a paper from 2009, “Seeking patterns in dream content: A systematic approach to word searches,” drawing on the resources of the Dreambank, I included this passage in the conclusion:

“Until researchers have gathered many more high-quality reports from a wide variety of people (ideally accompanied by multiple sources of biographical data), we cannot make any definitive declarations about the universal features of human dreaming. But the results of this study suggest several testable hypotheses:

  1. Dreaming perception is primarily visual, with less hearing and touch and almost no smell or taste.

  2. All emotions are represented in dreams, with fear the most frequent.

  3. Many types of cognitive activity occur in dreaming, especially those associated with awareness and social intelligence.

  4. Aggression is more frequent than sexuality, and both are more frequent for men than for women.”

Today, these same hypotheses can easily be tested with the resources of the Sleep and Dream Database (SDDb). The simplest method is to use the SDDb’s built-in word search template of keywords. The word search function has a template of forty categories of dream content, including categories for specific types of perception, emotion, cognitive activity, and social interaction. Starting on the “Advanced Search” page, I would define the data set for this purpose by setting a word limit of 25 words, and then select a category from the keywords menu. Looking at perceptions first, the following results can be generated in a few moments:

Out of a total of 20,510 dream reports of at least 25 words in length, reported by a total of 7,335 people, a word relating to visual perception appeared at least once in 34.6% of the reports. For hearing, the figure was 10.7%, for touch, 13%, and smell and taste combined only 2.7%. Eleven years later, I would still stand by that first hypothesis.

Turning to emotions, the results of the same simple search process (define the data set as having a minimum of 25 words, and selecting a category from the keywords menu) are just as predicted. A word relating to fear appears at least once in 18.2% of the dreams. Anger appears in 7.1%, sadness in 3.7%, happiness 6.5%, and wonder/confusion 14.4%. This hypothesis seems pretty solid, too.

Cognition in dreaming is harder to study for various reasons, but the word search method can still offer some interesting results. A word relating to thinking appears at least once in 41.9% of the dreams. Some kind of speech or verbal communication appears in 37.6%, and a reference to reading or writing in 7.6%. These findings support the basic idea that dreaming has a fair amount of cognitive activity, with plenty of social communication, though more detailed studies are needed to tease out the variations between dreaming and waking cognition. The third hypothesis is worth keeping.

Social interactions in dreaming are also difficult to study, so the results here should be regarded with extra caution. Indeed, the hypothesis from 2009 may not bear contemporary scrutiny, particularly around gender differences. (When defining the data set, gender can be selected as a search variable from the constraints menu.) The SDDb word search approach yields a finding of at least one reference to physical aggression in 20.8% of the male dreams and 17.2% of the female dreams. That’s a difference, but not a huge one. With the category of sexuality, the male dreams had at least one reference in 5.8% of the reports, versus 6.6% for the female dreams. This is the reverse of the predicted difference. The results of this quick analysis confirm that overall references to physical aggression occur much more frequently than references to sexuality, but the results do not support the 2009 hypothesis about higher frequencies of both kinds of content in men’s dreams.

There are other ways to study these questions with the tools of the SDDb. For example, the “baselines” function provides the frequencies on all 40 categories for a specially curated subset of 2,094 male dreams and 3,227 female dreams. These baseline frequencies provide a kind of measuring stick for dream researchers—a more precise way of determining the average frequencies of particular types of dream content and comparing them to other sets of dreams, which might have content features that vary from the baseline patterns in interesting ways. That shall be a topic for another post.

Note: This post first appeared in Psychology Today on September 4, 2020.

The Study of a 32-Year Long Dream Journal

The Study of a 32-Year Long Dream Journal by Kelly BulkeleyThe latest series to be uploaded into the Sleep and Dream Database (SDDb) is the biggest yet: the “Brianna Journal 1984-2016,” 2,448 dream reports from a woman who kept a journal fairly consistently for 32 years.  This series offers an amazing opportunity to observe in unusually close detail the emotional contours of an individual’s life as she makes her way through a challenging and often dangerous world.

Brianna (not her real name) shared these dreams with me and Deirdre Barrett last year, which we initially studied for a presentation at the 2016 conference of the International Association for the Study of Dreams.  Using the word search functions of the SDDb, I performed a “blind analysis” on three subsets of Brianna’s dream journals, meaning I 1) tabulated the frequencies of word usage for several categories of dream content, 2) compared her frequencies with baseline averages for each category, and 3) made inferences, based on nothing other than her dream patterns, about her concerns and activities in waking life.  For instance, I inferred that Brianna is closer to her mother than her father, is interested in books and writing, is not interested in sports, and has significant involvement with issues of death and dying.  Brianna herself, who attended the conference presentation, confirmed these and other inferences, which helped demonstrate the general idea that patterns in dreaming can accurately reflect people’s waking life concerns.

Now I have finally uploaded the complete collection of dreams Brianna shared with me, which provides a broader overview of her dreaming experiences over the span of more than three decades.  I will share more details from my analysis at the upcoming 2017 IASD conference (held in Anaheim, California, June 16-20).  For now, here are some of the initial findings of my study of this remarkable series.

Length: This is a long series in at least three ways: total number of dreams (2,448), time span covered by the journals (32 years), and average number of words per report (292).  The median word length is 168 words, meaning half the reports are shorter than that, and half the reports are longer.  Looking at the distribution of word lengths in the series as a whole, 851 of the dreams have between 1 and 99 words, 794 of the dreams have between 100 and 299 words, and 803 of the dreams have 300 or more words.  A series with this many dreams at both the short and long ends of the spectrum poses special challenges for analysis.  For now, I will study the series as a whole, but at some point I will look at subsets of varying lengths (e.g., the dreams of 50-300 words in length, of which there are 1,192).

Cognition: The series as a whole has a remarkably high frequency of dreams with at least one word relating to thinking (71%), speaking (56%), and reading/writing (19%).  The dreams have lots of strange, irrational material, too, but much of the content is oriented around normal cognitive activities that are also important in her waking life (Brianna is, in fact, a literate, well-educated, and sociable person).  The high proportion of cognition references could be a result of the unusual length of her dreams, and/or it could be an accurate reflection of her waking personality.  Either way, this is a topic worth further investigation.

Death: One out of every seven (15%) of Brianna’s dreams has a reference to death.  That is quite high compared to other dream series I have studied, and it strongly suggests that death and dying are major concerns in Brianna’s waking life.  I know enough about her to confirm the general accuracy of this inference, and now I am curious to look more closely at how this theme weaves its way through her series as a whole.

Religion: The frequency of references to religion is also unusually high in this series, and the list of specific words used in the dreams makes it fairly easy to accurately infer that Brianna is Jewish.  In previous studies I have found that patterns in dreaming offer good clues to a person’s beliefs and attitudes towards religion.  The Brianna series seems to be another illustration of that premise, and through deeper analysis I hope to understand better how religious and spiritual themes in the dreams track with Brianna’s waking life interests, concerns, and experiences.

Note: this post was originally published in Psychology Today, March 10, 2017.