Dreaming, Vol. 13 No. 2, June 2003.

Dream Imagery Becomes More Intense After 9/11/01

Ernest Hartmann1,3 and Robert Basile2

We examined a series of twenty dreams - the last ten dreams recorded before 9/11/01 and the first ten dreams recorded after 9/11/01 - from each of sixteen individuals in the United States who regularly record all their dreams. Blind scoring using established scales demonstrated that dreams after 9/11/01 were characterized by more intense imagery, but were not longer nor more “dreamlike,” compared to data before 9/11/01. The dreams after 9/11/01 did not contain significantly more content related to the attacks. The results show that traumatic events such as the attacks of 9/11/01 have a detectable effect on dreams - specifically an increase in dream image intensity - in a population of dream recorders. Whether this finding can be generalized to the entire population is not clear from this preliminary study. The results of this study are consistent with previous findings that dream image intensity is related to emotional arousal.

KEY WORDS: dreams; trauma; 9/11/01; contextualizing image; central image; dream image.

1Department of Psychiatry, Tufts University School of Medicine, Boston, MA.

2Ferkauf Graduate School of Psychology, Yeshiva University, NY.

3Correspondence should be directed to Ernest Hartmann, M.D., 27 Clark St., Newton, MA 02459; e-mail: ehdream@aol.com.



Evidence from a number of sources suggests that emotion has an important influence on dreaming. Brain imaging studies demonstrate clear activation in REM sleep of the amygdala and related structures that deal with emotion (Braun et al., 1997; Maquet et al., 1996; Sutton et al., 1996). Neuropsychological studies examining the details of brain lesions which result in reports of complete cessation of dreaming implicate two forebrain areas, one of them in the medial portion of the frontal lobes, including the mesiofrontal dopamine systems considered important in emotional arousal (Panksepp, 1985; Solms, 1997). Studies of dream content demonstrate the frequent presence of emotion in reported dreams, the exact level depending on how the questions are asked (Merritt, Stickgold, Pace-Schott, Williams, & Hobson, 1994; Nielsen, Deslauriers, & Baylor, 1991).

Recent studies of dreams after traumatic events have led to the suggestion that dream imagery can often be understood as picturing the dominant emotion or emotional concern of the dreamer (Hartmann, 1996, 2001). A powerful image such as "I was swept away by a huge tidal wave," has been repeatedly reported in clinical/anecdotal studies of persons who have experienced a variety of traumas ( fires, attacks, attempted rapes). The image is not related to the specific trauma experienced, but appears to be picturing the person’s emotional state - "I feel terrified," or "I feel overwhelmed." There is evidence based on blind scoring of dream reports that such images are indeed more frequent and intense after acute trauma (Hartmann, Zborowski, McNamara, Rosen, & Grace, 1999; Hartmann, Zborowski, Rosen, & Grace, 2001). Even distant trauma appears to affect dreams: the "most recent dreams" of students who report having been physically or sexually abused at any time, score higher on intensity of the dream image than the "most recent dreams" of other students (Hartmann, Zborowski, NcNamara, Rosen, & Grace, 1999; Hartmann, Zborowski, Rosen, & Grace, 2001). These studies used an established rating scale measuring the intensity of the "Central Image" or "Contextualizing Image" of the dream, which demonstrates high interrater reliability.

It has been difficult to collect adequate data for studies of trauma, since the most useful contrasts would involve within-subject comparisons of routinely recorded dreams before and after trauma. Unfortunately there are few persons who routinely record all their dreams, are willing to share their dream records with researchers, and also happen to have experienced an acute personal trauma during the period of dream recording. However, we believe that the terrorist attacks of 9/11/01 produced some degree of trauma or at least serious stress in everyone living in the United States, which might facilitate such a study. There is in fact evidence that the events of 9/11/01 produced an increase in stress measures and in some symptoms of Post-traumatic Stress Disorder even in persons living far from the sites of the attacks (Schuster, 2001). Under this assumption, we attempted to collect dream series before and after 9/11/01 from anyone in the United States who had been regularly recording his or her dreams and was willing to share these dreams without any selection or alteration.


We report here results on sixteen adults living in the United States, the first sixteen from whom we obtained complete data sets. Participants were obtained through web sites listing research projects maintained by the Sleep Research Society and the Association for the Study of Dreams. Participants were not asked for personal information but some did provide data. There were six males and nine females and one whose sex is unknown. For the eight participants who provided their age, their mean age was 53. Geographically they were scattered throughout the United States, though none lived in New York City.

Each supplied a series of 20 dreams from their dream journals, almost always recorded on a computer. Each series consisted of the last ten dreams recorded before 9/11/01 and the first ten dreams after 9/11/01. A total of 324 dreams were obtained and analyzed. We expected to score 20 dreams from each participant for a total of 320 dreams. In two cases, however, only 19 dreams were obtained, and in several cases what had been submitted as one long dream appeared, in the opinion of two independent judges, actually to consist of two separate dreams. Thus the overall total scored was 324 dreams.

The dreams were assigned random numbers and scored on a blind basis for the intensity of the central image, using a standard score sheet for central or contextualizing images which has been used in a number of previous studies cited (see Fig. 1). Blind scorers are asked to look at a written dream and decide whether there is a "central image" or "contextualizing image" (CI) - "an image which stands out by virtue of being especially powerful, vivid, bizarre, or detailed." If there is no such image, the intensity (CI) score is zero. If such an image is found, the scorer is asked to identify it and give it an intensity score from 0.5 to 3.0, with 3.0 indicating "about as intense an image as you have found in dreams." This results in a seven point intensity scale running from 0 to 3 by half points. Independent scorers have shown interrater correlations of r=.7 to r=.85 on this scale in a number of dream series. This is the CI score or dream image intensity score used here. The scorer is then also asked if possible to choose an emotion which the image might be picturing, from a list of eighteen basic emotions.

All dreams were scored on CI intensity by one experienced scorer. In addition, fifty dreams, selected at random from the 324 dreams, were scored by another experienced scorer. Interrater reliability was r = .76. Each dream was also scored on a blind basis for what emotion might be pictured by the central image, for dream length, and on a well-established scale for "dreamlikeness" (dream-like vs. thought-like) (Foulkes, 1966). Fifty of the dreams, selected at random, were scored on "dreamlikeness" by two experienced scorers. Interrater reliability was r = .72. In addition, each dream was scored from 0 to 3 on three ad-hoc scales measuring dream content related to the events of 9/11/01: a) Content dealing with attacks of any kind; b) Content dealing with buildings resembling either the World Trade Center towers or the Pentagon; 3) Content involving airplanes. The code was then broken, and dreams were regrouped by participant on each scale. For each participant a mean value pre-9/11 and a mean value post-9/11 were obtained, and these means were compared using a t-test for correlated means.


The 10 dreams before and the 10 dreams after 9/11/01 covered a range of time periods. Most commonly the dreams occurred during a period of two to four weeks before and two to four weeks after 9/11/01. However, one participant had 10 "before" dreams between 9/7/01 and 9/11/01 and 10 "after" dreams between 9/11/01 and 9/15/01. Before 9/11 the shortest time span was from 9/7 to 9/11, and the longest time span was from 6/9 to 9/11. In the "after" 9/11 dreams the shortest time span ended on 9/15. The longest ended on 10/8. There was no significant difference between the time spans covered by the "before 9/11" series as opposed to the "after 9/11" series.

The results (Fig. 2, Table I) demonstrated a highly significant increase in intensity of dream imagery (CI score) after 9/11/01. Fourteen of the 16 participants showed a change in this direction. However, dreams after 9/11/01 were not longer than pre-9/11/01, nor were they scored as more dreamlike. And there was no significant before vs. after difference on the three content scales, though there was a trend towards more imagery involving attacks after 9/11/01 compared to before. The data on "emotion pictured" showed an apparent increase for the emotion "fear/terror". This emotion was scored in 27 instances (17 percent of all instances scored) before 9/11/01, and in 37 instances (23 percent) after 9/11/01; however this did not reach statistical significance, and must be considered a trend. No other emotion showed any definite change.


Overall these results indicate that at least in persons who record their dreams, the main difference produced by 9/11/01 was in intensity of the dream image, rather than a change in dream length, "dreamlikeness," or specific content.

One might question whether persons who record all their dreams are typical of the US population. We have limited knowledge of the specific participants who took part in this study. However, detailed interviews in the past with a number of regular dream recorders have convinced us that these are generally normal persons with no diagnosable psychopathology, distinguished chiefly by their interest in their dreams, daydreams, artistic productions, and other aspects of their inner lives. They are persons characterized by having "thin boundaries" in a number of senses (Hartmann, 1991). Such persons often unusually sensitive in various ways, and thus it is possible that they would react more to stress and trauma than the average person. It would be of interest to study persons with thick boundaries, but since these people recall few dreams, it would be impossible to use them in a study such as this.

One might also question whether factors other than the terrorist attacks of 9/11/01 might have an influence on these dreams. In other words, are dreams in September and October systematically different than dreams in July and August for other reasons - for instance the beginning of the school year in September. We are in fact in the process of collecting data on students that may help answer this question, but no data are available as yet. The fact that the average age of the present sample (among those whose age was known) was 53 and none of these, to the best of our knowledge, were students, makes it somewhat unlikely that the beginning of the school year had much effect.

It is of interest that the intensity of dream imagery differentiates the before vs. after time periods more clearly than do other measures, including the emotion rated as pictured by the dream. (There was a strong trend - not reaching significance - towards more "fear/terror" after 9/11/01). This is consistent with previous studies in which intensity of dream imagery differentiated groups such as trauma vs. no trauma or abuse vs. no abuse better than did the specific emotion pictured (see Hartmann, Zborowski, & Kunzendorf, 2001 for a summary). In all studies the emotion "fear/terror" is the one most frequently scored (even in normal students). In fact, previous studies find this emotion to be scored in about the same proportion of all emotions as was found in the pre-9/11/01 data of the present study (Hartmann, Zborowski, & Kunzendorf, 2001).

The fact that none of the three content measures show a significant before-vs.-after change is perhaps surprising. The data suggest no specific explanation for this, except that there was considerable variability between participants, with a few having a number of dreams involving attacks and towers while the majority showed little or no such content. And the mean scores for each of the three categories were quite low. In each of the three content categories (attacks, towers, airplanes), over half the participants had no dreams whatsoever involving that content.

These results suggest that a traumatic event such as the terrorist attacks of 9/11/01 had a measurable effect on dreams, at least in persons who regularly record their dreams. Of course the number of participants studied was small. We hope to report eventually on a larger sample.

The present results support the view that emotional arousal affects dreams, and more specifically that emotional arousal increases the intensity of dream imagery, rather than affecting the length or specific content of the dream. Dream image intensity can be considered a measure of emotional arousal.


 Braun, A., Balkin, T., Wesenten, N., Carson, R., Varga, M., Baldwin, P., Selbie, S., Belenky, G., & Herscovitch, P. Regional cerebral blood flow throughout the sleep-wake cycle. Brain 120, 7 (1997).

Foulkes, D. The Psychology of Sleep (Charles Scribner’s Sons, New York, 1966).

Hartmann, E. Boundaries in the Mind (Basic Books, New York, 1991).

Hartmann. E. Outline for a Theory on the Nature and Functions of Dreaming. Dreaming 6, 147 (1996).

Hartmann, E. Dreams and Nightmares (Plenum, New York, 1998, Perseus, New York, 2001).

Hartmann, E., Zborowski, M., Kunzendorf, R. The Emotion Pictured by a Dream: An Examination of Emotions Contextualized in Dreams. Sleep and Hypnosis 3, 33 (2001).

Hartmann, E., Zborowski, M., McNamara, P., Rosen, R., Grace, N. Contextualizing Images in Dreams: Relationship to the Emotional State of the Dreamer. Sleep 22S, 131 (1999).

Hartmann, E., Zborowski, M., Rosen, R., Grace, N. Contextualizing Images in Dreams: More Intense After Abuse and Trauma. Dreaming 11, 115 (2001).

Maquet, P., Peters, J., Aerts, J., Delfiore, G., Degueldre, C., Luxen, A., Franck, G. Functional neuroanatomy of human rapid-eye-movement sleep and dreaming. Nature 383, 163 (1996).

Merritt, J., Stickgold, R., Pace-Schott, E., Williams, J., Hobson, J.. Emotion profiles in the dreams of men and women. Consciousness and Cognition 3, 44 (1994).

 Nielsen, T., Deslauriers, D., Baylor, G. W. Emotions in Dream and Waking Event Reports. Dreaming 1, 287 (1991).

Panksepp J. in Handbook of clinical neurology, P. Vinken, G. Bruyn, H. Klawans, J. Frederiks, Eds. (Elsevier, Amsterdam, 1985), pp. 271-285.

Schuster, M.  A National Survey of Stress Reactions After the September 11, 2001, Terrorist Attacks. et al., N. Engl. J. Med. 345, 20 (2001).

Solms, M. The Neuropsychology of Dreams (Lawrence Erlbaum Associates, Mahwah, NJ, 1997).

Sutton, J., Breiter, H., Caplan, J., Huang-Hellinger, K., Kwong, J., Hobson, A., & Rosen, B.  Human brain activation during REM sleep detected by fMRI. Neurosci. Abstracts 22, 690 (1996).


Fig. 1


Definition: A contextualizing image is a striking, arresting, or compelling image — not simply a story — but an image which stands out by virtue of being especially powerful, vivid, bizarre, or detailed.


1.        fear, terror 11.      power, mastery supremacy
2.        helplessness, vulnerability, being trapped, being immobilized  12.      awe, wonder, mystery
3.        anxiety, vigilance 13.      happiness, joy, excitement
4.        guilt014.      hope
5.        grief, loss, sadness, abandonment, disappointment  15.      peace, restfulness
6.        despair, hopelessness (giving up)  16.      longing
7.        anger, frustration  17.     relief, safety
8.        disturbing—cognitive dissonance, disorientation, weirdness  18.     love (relationship)
9.        shame, inadequacy  
10.     disgust, repulsion    

   If there is a second contextualizing image in a dream, score on a separate line.



Dream ID#

1.  CI?

2. What is it?

3. Intensity
(rate 0-3)

4. What emotion?

5. Second Emotion
























Figure 2


Fig. 2. Change in dream image intensity.  Each bar represents the mean CI score for dreams after 9/11/01 minus the mean CI score before 9/11/01 in a given participant.

Table 1. Results Before vs. After 9/11/01. 















.003 (one-tailed)































Note: CI refers to the Contextualizing Image score, a measure of dream image intensity. (See Figs. 1 and 2.) “Length” refers to the number of lines of text in the printed dream — approximately fifteen words per line. “Dreamlike” refers to the score on “dreamlikeness” scale (13) running from 0 (equals totally thought-like) to 7 (equals totally dreamlike). “Attacks,” “buildings,” and “airplanes,” refer to the three ad-hoc content scales, each scored from 0 to 3.

Dreaming: Journal of the Association for the Study of Dreams. Vol 13(21)  June 2003.

  Copyright ©2003 Association for the Study of Dreams. All Rights Reserved