Abstract

Considerable progress has been made in the systematic study of dream content. The most commonly used methods for collecting dream reports—laboratory awakenings, home dream logs, questionnaires, and most recent dreams collected in group settings—all have their uses and inherent advantages and disadvantages. Reliable, comprehensive, and validated instruments for the actual analysis of dream content reports have been developed, and complementary tools are now available to all researchers on the Internet. Quantitative data on dream content from laboratory and nonlaboratory settings generally converge in depicting a reliable picture about the nature of dream content in the general adult population. Few differences emerge between laboratory and home dream reports. Both data sets indicate that for the most part, dreams are a reasonable simulation of waking life characters, social interactions, activities, and settings and that dreams show systematic relationships to various dimensions of the dreamer’s waking life but not to day-to-day events. Developmental changes also occur in dream content until late adolescence, when dream content becomes surprisingly stable and consistent throughout adulthood and old age. More clinically oriented studies suggest that affect and social interactions are two key dream content variables that are most strongly related to measures of psychological well-being. Taken together, these findings have several implications for theories of dreaming and provide convincing evidence that dreams are a unique and meaningful psychological product of the mind.

Researchers and clinicians have long been fascinated by the content of dreams. Although many contemporary dream researchers suggest that dreaming is functionally significant and may subserve a biologically important function, some argue that dreams are epiphenomenal to neurophysiologic activity during rapid eye movement (REM) sleep and have no value in and of themselves even though evidence suggests they have psychological meaning. This chapter reviews methodologic issues in dream research and systematic findings on the content of people’s dreams, and it presents the implications of key findings on normative dream content. The implications for theories of dreaming are briefly considered in the final section.

There is no consensus on what distinguishes dreaming from other cognitive processes, such as thinking or daydreaming, nor on what constitutes dream content. Interdisciplinary groups from the International Association for the Study of Dreams and the American Academy of Sleep Medicine concluded that “a single definition for dreaming is most likely impossible given the wide spectrum of fields engaged in the study of dreaming, and the diversity in currently applied definitions.”¹ Thus, depending on one’s perspective, dreaming can be synonymous with the term “sleep mentation,” which refers to the experience of any mental activity (e.g., perceptions, bodily feelings, thoughts) during sleep, or it may be restricted to more elaborate, vivid, and story-like experiences recalled upon awakening. As highlighted by others,² using a broadly inclusive versus more restrictive definition of dreaming has a direct and significant impact on the nature and sense of empirical data and theoretical modeling in the field.

In this chapter, the term dream is conceptualized as having four interrelated meanings. First, a dream is a form of thinking during sleep that occurs when there is a certain, as yet undetermined, minimal level of brain activation in a context in which external stimuli are typically occluded and the cognitive system that keeps us aware of our surroundings is shut down. Second, a dream is something people experience as a series of actual events (e.g., a sequence of perceptions, thoughts, and emotions) because the thought patterns simulate waking reality. Third, a dream is what people remember upon awakening, so it is a memory of the dreaming experience. Finally, a dream is the spoken or written report provided to investigators based on the memory of the dreaming experience. The empirical studies discussed in this chapter reveal that the events of a dream always include the dreamer as an observer or participant, and that they almost always include at least one other character besides the dreamer (either a person or an animal). In addition, the dreamer or the other characters in the dreams are invariably engaged in one or another activity (e.g., looking, walking, running) or a social interaction. Thus, the sense of participation in an event, along with characters, activities, and social interactions, is what distinguishes dreams from the more fleeting, fragmented, and thoughtlike forms of sleep mentation.

Methods for Collecting Dream Reports

Researchers never study dream experiences directly. Instead, they collect and have access to descriptions of the experience, the dream report. The nature and content of the verbal or written report obtained can be influenced by a number of factors. These include the setting (e.g., home, laboratory, classroom, psychotherapy), method of awakening (e.g., spontaneous, induced), time of awakening (e.g., early, middle, or late in the sleep period), sleep stage prior to awakening (e.g., REM, non-REM [NREM] sleep), type of collection instrument (e.g., questionnaire, dream journal), reporting method (e.g., written by the subject, written by the experimenter, audio recording), instructions provided (e.g., report anything that was going through your mind before your awakening, note only your dreams), probes on reported content (no, fixed, or semistructured questions), interpersonal situation (e.g., reporting directly to an experimenter, clinician), time delay between when the dream was experienced and when it is reported, study duration, and subject characteristics (e.g., gender, personality, habitual level of dream recall).

The degree to which the content of dream reports is influenced by these various factors either individually or in combination varies as a function of the collection method used. The principal sources of dream reports are the sleep laboratory, home dream journals, questionnaires, psychotherapy sessions, and classroom or other group settings where a most recent dream can be collected from everyone willing to participate. Although there is convincing evidence that working with patients’ dreams can be clinically useful,³ dream reports from the psychotherapy relationship are rarely used in systematic studies, and thus this source is not covered here.

Sleep Laboratory

Sleep laboratories are an excellent source of dream reports because they provide the opportunity for collecting a representative sample of a subject’s dream life, both within and across nights, under controlled conditions. Awakening subjects from several REM or NREM periods results in the collection of dream reports that may have been otherwise forgotten by the subjects upon normal awakening in the morning. Awakenings during REM, or from stage II NREM late in the sleep period, maximize the probability of recall and make it possible to collect as many as four or five dreams in a single night. On the other hand, frequent awakenings can be difficult for participants, and factors such as sleep inertia and one’s desire to return to sleep may interfere with the quality of the dream reports. However, a complimentary cued morning report of dreams recalled during the night can yield new and reliable information as to the dreams’ original contents.⁴

The main problem with the laboratory collection of dream reports is that it is a very costly and time-consuming process, and even though several dreams can be collected each night, it still can take many months to obtain 10 or more dreams from each of a dozen participants. Furthermore, some types of dreams, including nightmares and sexual dreams, rarely occur in the sleep laboratory, presumably due to sociocognitive factors. In addition, approximately 20% of laboratory REM dream reports reflect direct incorporations of the laboratory environment,⁵ even when collected over several consecutive nights.⁶ For our purposes, the most important outcome of detailed laboratory studies is that they provide a baseline for assessing the quality of dream reports collected by other methods.

Dream Logs

Prospective daily logs are used by an increasing number of dream researchers even though they require a greater investment of time and resources than do questionnaires. In fields like nightmare research, home journals are considered the gold standard for the measurement of nightmare frequency.⁷ Although limitations associated with longer-term retrospective assessments of dream recall and dream content are increasingly recognized, variations in home logs have received little attention.

Prospective logs can take two different forms. The first is the checklist format, in which participants indicate if there was dream recall and if so, the number and type of dreams recalled (e.g., nightmare). The second is the narrative log, in which participants are requested to provide a complete written transcript of each dream recalled. Findings from one comparison ⁸ of these two methods of data collection suggest that narrative-log participants, having a more time-consuming task, do not take the required time to provide a complete narrative of all of their recalled dreams, as Strauch⁹ found with teenage boys. Instead, they may choose to focus on their more memorable, exciting or salient dreams, which would typically include bad dreams and nightmares. By comparison, people completing checklist logs would be more likely to record all of their dreams (including relatively banal or poorly recalled ones) as each entry is just as quickly completed regardless of dream type.

Although writing down one’s dreams remains the most frequently used method to collect dream content, participants may also use tape recorders to dictate their reports. This approach may be particularly useful with children and younger adolescents.¹⁰ It also proved highly useful in a study of blind participants.¹¹

Questionnaires

In questionnaire studies, participants’ retrospective self-reported information concerning their dream experiences is viewed as a modest but valid way of assessing different aspects of the dream experiences themselves. Three types of information are generally collected.

First, subjects can be queried about the frequency with which they experience certain kinds of dreams (e.g., everyday dreams, nightmares) over a determined period of time. There is increasing evidence, however, that data obtained with retrospective estimates differ considerably from daily prospective home logs. For instance, when compared to results from daily home logs, retrospective self-reports significantly underestimate current nightmare frequency ¹²^,¹³ and this rate of underestimation is not attributable to an increase in recalled dreams caused by keeping a dream log.¹³ Similarly, one study¹² found that the magnitude of the association between trait anxiety and nightmare frequency decreased significantly when daily logs were used to measure nightmare frequency instead of retrospective self-reports. This led the authors to suggest that anxious persons do not necessarily have more nightmares, but rather that they are more likely to remember and report nightmares retrospectively. Finally, a meta-analysis¹⁴ of studies having examined the relationship between dream recall frequency and various personality dimensions found that scores on personality measures were not related to dream recall frequency per se, but rather to people’s tendency to retrospectively underestimate or overestimate their dream recall. Taken together, these findings indicate that correlates of retrospective measures of dream recall should not be assumed to be correlates of log measures of dream recall. Contrary to prospective log measures, retrospective indices of dream recall are best viewed as measures of peoples’ cognitive representations of their dream life.

A second kind of information sometimes elicited via questionnaires focuses on specific dimensions of people’s dreams or their beliefs about their general dream life. This approach assumes that there exists a valid relationship between self-reported information on the content of one’s everyday dreams and the dream experiences themselves. However, comparisons of self-report measures and log-based data indicate that this assumption may be unwarranted. For instance, one ¹⁵ comparison of participants’ questionnaires and 2-week logs found no relationship between estimated frequency for the appearance of aggressive, friendly, and sexual elements and their frequency in the dream reports. Similarly, a subsequent study¹⁶ showed that when people’s level of dream recall is poor, their beliefs about the level of anxiety in their dreams is not related to the actual affective content of their everyday dreams as recorded prospectively in home logs. These findings suggest that the relation between beliefs people hold about the content of their dreams and their actual dream experiences is mediated by autobiographical memory and that these beliefs are particularly inaccurate when dream recall is low (i.e., when memories of one’s dreams are not readily available).

Lastly, questionnaires are used to investigate if participants ever experienced a specific type of dream, and if so to report the most recent occurrence as best recalled. This approach allows the investigation of certain types of dreams that, due to their infrequency, are difficult to capture in laboratory settings or with home dream logs (e.g., recurrent dreams, existential dreams) or dreams that stand out in the person’s past (e.g., earliest dream recalled, most terrifying nightmare). Although useful in some research settings, the resulting dream content findings must be treated cautiously due to possible memory distortions and biases.

In sum, although some dream questionnaires have good internal consistency and test-retest reliability,¹⁷ studies of their relationship to dream content and frequency findings obtained from dream journals reveal important discrepancies and raise questions as to their validity.

Classroom and Other Group Settings

Settings such as classrooms provide an objective and structured context for the efficient and inexpensive collection of dream reports. Anonymous participants are instructed to write down the most recent dream that they can recall on a standardized form while revealing only basic background information such as age and gender. The most recent dream method has been used with children as young as ages 10 to 11 years in different countries with surprisingly similar cross-national results.¹⁸^,¹⁹ However, there is reason to believe that young children up to ages 10 to 11 years are using their waking imaginations to provide a report that fits cultural stereotypes about the nature of dreams.²⁰ The main drawback with this method is that there is not usually time to collect any personality or cognitive measures on the people providing the reports.

Analyzing Dream Content: Instruments and Issues

Most past dream research used either rating scales at the ordinal level of measurement (“more” or “less” of a characteristic) or discrete categories at the nominal level of measurement (an element is “present” or “absent”). Rating scales are most useful for characteristics of dream reports that have degrees of intensity in waking life, such as activity level, emotionality, clarity of visual imagery, or vividness. Cohen ²¹ reports that four dimensions of dream salience can be rated by participants in dream studies: emotionality, bizarreness, activity, and vividness. A factor analysis of the ratings of 100 REM dream reports suggests that rating scales boil down to five basic dimensions: degree of vividness and distortion, degree of hostility and anxiety, degree of initiative and striving, level of activity, and amount of sexuality.²² However, it is often difficult to establish reliability with some scales, and much of the specific information in dream reports is lost or unused with general rating scales.

Of the 150 dream rating and content analysis scales reviewed by Winget and Kramer,²³ the Hall and Van de Castle coding system²⁴ is the best validated and remains the most widely used system for analyzing dream content. The Hall/Van de Castle system, which provides many of the findings presented in the rest of this chapter, rests on the nominal level of measurement and uses percentages and ratios as content indicators that can correct for the varying length of dream reports from sample to sample. The dream reports used in the original normative sample, as well as the codings for them, are available to researchers through www.dreambank.net.²⁵ The normative findings reveal a pattern of gender differences that needs to be taken into account when doing studies of individuals. The Hall/Van de Castle coding system employs nonparametric statistics for determining P values and effect sizes, which can be obtained instantly after entering codings into the DreamSAT spreadsheet available to all researchers on www.dreamresearch.net.²⁶ The general Hall/Van de Castle norms can be used with confidence for a variety of purposes because they have been replicated in several studies.²⁷^,²⁸

As documented by Winger and Kramer, there exist numerous other coding systems, and many new ones have been created since their comprehensive review. However, unlike the Hall/Van de Castle system, most of these instruments have only been used by the original investigators (limiting potential for comparisons across laboratories), many use weighting systems of questionable validity, and few are based on clearly defined and objective scoring criteria that yield good interrater reliability. Moreover, as detailed elsewhere, many of these scoring systems can be duplicated by combining two or more elements of the Hall/Van de Castle system.²⁷

Some research questions (e.g., self-reflectiveness in dreams, contextualizing images in dreams), have necessitated the creation of new instruments.²⁹^,³⁰ The DreamThreat rating scale was developed to test an evolutionary theory of dreams that stipulates that the function of dreaming is to simulate threatening events with the intent of improving the subject’s ability to recognize and avoid diverse threats in real life.³¹ Although this rating scale has been criticized, it is noteworthy in that it has been used by different groups to assess various kinds of dreams and that it yields good to excellent inter-rater agreement.³¹^–³³ Taken together, findings indicate that a significant proportion of dreams contain a wide range of threats, but few of these dreams present realistic life-threatening events, and the dreamer rarely succeeds in escaping the threat.

Finally, given the time-consuming nature of traditional scoring of dream reports, some groups have tried to develop computerized systems that can carry out such tasks both reliably and accurately. Because emotions are viewed by some dream theorists as playing a key role in structuring dream content, and given that dream affect is one of the most frequently investigated dream content variables, some of the more innovative work in this field has focused on the classification of emotions. Online search tools available to researchers at www.dreambank.net allow for rapid and accurate searches for specific words, phrases, and long word strings, and one study shows that the use of word strings for emotions yields results comparable to when standard Hall/Van de Castle codings are scored on the same dream reports.³⁴ Another promising project is based on an algorithm that seeks to accurately categorize dream emotions both at fixed times and dynamically as the dream narrative progresses.³⁵ Moreover, the algorithm has the potential of improving its performance as a function of training (machine learning). Although this research is still in its infancy, such innovations might allow efficient and accurate scoring of large data banks across laboratories.

Problems in Studying Emotions and Bizarreness in Dreams

Although both rating scales and the Hall/Van de Castle nominal coding categories have proved useful for most dimensions and elements of dream content, there are methodologic problems relating to the study of emotions and bizarreness in dreaming. Several different studies using blind coders find that negative emotions outnumber positive ones.²⁴^,³⁶ However, very different results emerge when the dreamers themselves make a global rating of each of their dream reports on a pleasant–unpleasant dimension. Such studies regularly find that the dreamers rate the emotions in their dreams as at least equally pleasant and unpleasant, and sometimes as more pleasant.³⁷^–⁴⁰ There is no ready explanation for these contrasting results with the two different methods.

Dreamers also tend to attribute many more emotions to their home dreams than do blind judges when they are asked to recall the emotions that accompanied the report they have written down. However, it is an open question in need of further study as to whether or not this greater amount of emotions in self-ratings of home dream reports is the result of two extrinsic factors: the demand characteristics of such a rating task and the waking-life assumption that certain emotions would logically be present in many of the situations experienced in the dream.

There is also a lack of agreement on how to assess unusual or bizarre elements in dreams, which leads to widely varying prevalence and frequent estimates. In studies that focus on clearly impossible events, the figure is 10% or less for large samples of both REM and home dreams.⁴¹^,⁴² When sudden scene changes, uncertainties, and small distortions are included, the figure rises to between 30% and 60%.⁴³^–⁴⁵ Using a rating scale based on the degree to which any dimension of the dream differs from waking experience and behavior, it was found that 75% of 500 REM reports from men and women had at least one bizarre aspect, as compared with 7% to 8% that were bizarre in three or more ways.^{46, p. 95-103} In addition, studies of bizarreness in dreams have been handicapped by the fact that there have been no studies comparing the nature and frequency of bizarre elements in dreams and waking thought samples from the same participants, which seems to be an essential step given the evidence that waking thought often contains unexpected and anomalous elements.⁴⁷