Home Curriculum Vitae Publications Conference Reports Forthcoming Extramural Colloquia Expert Testimony Teaching Healthcare The Human Ecology of Memory Research Archive Publications and Reports Rants


Flashbulb Memories: 

Historical and Personal Memories 

and Flashbulb Quality


Mary Mullane Swar & John F. Kihlstrom

University of California, Berkeley


View the illustrations accompanying this text.


The events of September 11 renewed interest in the phenomenon of flashbulb memories. In a classic paper, Brown and Kulik (1977) defined flashbulb memories as "memories of the circumstances in which one first learned of a very surprising and consequential (or emotionally arousing) event". People of a particular age often have flashbulb memories of the assassination of John F. Kennedy, or of Martin Luther King, or of Robert F. Kennedy. Other flashbulb memories that have been studied include the Challenger Disaster of 1986 and the Loma Prieta earthquake of 1989. And newspaper stories indicate that there are several studies in progress of flashbulb memories for the World Trade Center attacks.

Flashbulb memories were so named because it seems as if the mind has "taken a picture" of the circumstances in which the news was learned. Accordingly, most analyses have focused on the accuracy of flashbulb memories. For example, Ulric Neisser has written of his own flashbulb memory for learning of the Japanese attack on Pearl Harbor in 1941. His memory is that he was listening to a baseball game at the time, but of course no baseball games were being played in December. Instead, it has been suggested that Neisser was actually listening to a football game (Neisser himself has another interpretation of the mistake).

Many studies have collected subjects' memories immediately after the event, and compared these "on-line" records to flashbulb memories reported after some interval has transpired. The general conclusion of these studies is that, appearances to the contrary notwithstanding, people's flashbulb memories are not necessarily accurate representations of the event in question. Moreover, whatever accuracy they may have may derive not from some exotic "Now Print" mechanism related to emotional arousal, but rather from the rather mundane effect of repeated rehearsals as people share their experiences of learning the news.

Accurate or not, flashbulb memories are still memories, and are interesting nonetheless. Flashbulb memories, whether of the assassination of President Kennedy, the Challenger disaster, the World Trade Center attacks, or such personal, private, idiosyncratic moments as one's first kiss, may serve an important function in the individual's personality. Moreover, the sharing of such memories may be an important social activity. Neisser himself has suggested that flashbulb memories are less snapshots of the past than benchmarks in one's personal history. As he has put it, "The flashbulb recalls an occasion when two narratives that we ordinarily keep separate -- the course of history and the course of our own life -- were momentarily put into alignment.... [W]e remember the details of a flashbulb occasion because those details are the links between our own histories and ‘History’.... They are the places where we line up our own lives with the course of history itself and say ‘I was there’.

In this study, we have set aside the question of accuracy, and closely examined the internal characteristics of a specimen flashbulb memory, and its relation to emotion. In this way, we hope to identify features of what might be called "flashbulb quality" that could be used to identify any memory, of whatever event, as "flashbulb" in nature. In addition, we were stimulated by Neisser's analysis to undertake a preliminary investigation of the role that flashbulb memories might play in linking autobiography with history.



The subjects in this experiment were 368 students enrolled at the University of Arizona tested in the Fall semester of 1989, who served in the study in partial fulfillment of the research participation component of their introductory psychology course. Almost two-thirds of the subjects (235, or 63.9%) were women, and most (294, or 79.9%) were aged 18-21, with an average age of almost 19 years.

wpa02f1.gif (8364 bytes)The subjects were first asked to provide a brief narrative account of the circumstances in which they first learned of the Challenger disaster (the average subject was 19 years old, and thus about 15 or 16 years old at the time the incident occurred on January 28, 1986). Using a very simple scheme, we coded this free-recall material for various items of information described by Brown and Kulik and other researchers, such as the place where the learning occurred, what was going on at the time, who the informant was.

The free-recall narrative was followed by a series of specific questions along the same lines, again coded according to a very simple scheme. This inquiry was followed by a series of questions concerning the characteristics of the memory, such as modalities of imagery and number of rehearsals. Finally, we asked a series of questions about the event itself, with a particular focus on the person's emotional reaction to the event then and now.



On the basis of the material they collected, Brown and Kulik identified six features as "canonical" characteristics of a flashbulb memory:

PLACE: the place where the subject learned of the target event, in this case the Challenger disaster;
EVENT: what was going on at the time;
INFORMANT: who informed the subject of the event;
OWN AFFECT: the subject's own emotional reaction to the news;
OTHER AFFECT: the emotional reaction of the informant and others present;
AFTERMATH: what happened immediately after learning of the event.

In addition to these, we added three other features of interest:

TIME: the time the subject learned of the target event;
OTHERS PRESENT: other people who were with the subject at the time;
FIRST THOUGHTS: the subject's first thoughts when he or she learned of the target event.

Brown and Kulik identified as a "flashbulb" any memory in which the person claimed to remember the circumstances in which he or she learned of the event, plus at least one other characteristic. Only two subjects appeared to draw a complete blank on the event, failing to mention even one of the canonical features in their narratives, and even these subjects remembered at least something codable when provided with specific prompts. So, we can say that more than 99% of our sample had a flashbulb memory for the Challenger disaster.

Of course, this is a very liberal criterion, and raises the question of whether some memories have more flashbulb qualities than others. The average subject included 4 of Brown and Kulik's canonical characteristics in his or her free narrative (M = 4.21, SD = 1.34), and provided information on all six categories (M = 5.84, SD = 1.08) when probed with specific questions. Nor surprisingly, memory improved significantly in response to the specific probes, t(367) = 23.88, p < .001. When we examined the distribution of canonical characteristics, we found that our subjects free-recall narratives varied considerably in what might be called "flashbulb quality". There was little such variance in response to specific probes, of course, due to ceiling effects.

It occurred to us that the various features commonly ascribed to flashbulb memories could be used to create a sort of psychometric scale by which we could rate any episodic memory for its "flashbulb quality". A scale composed solely of Brown and Kulik’s six canonical characteristics yielded a reliability estimate (coefficient alpha) of .62. We then added a couple of other features that seemed relevant to us, or which had cropped up in the subsequent literature: the time the event occurred; other people who were present; and the person’s first thoughts after learning the news. This nine-item scale had a reliability of .67. These figures are not great, by the standards of modern personality questionnaires, but they do indicate that "flashbulb quality" can be measured with some degree of reliability.

In order to clarify the internal structure of our scale of flashbulb quality, we performed a factor analysis of the nine features. A principal-components analysis extracted three factors with eigenvalues greater than 1. Varimax rotation showed that six items loaded on the first factor: place, others, informant, event, other affect, and aftermath; all these items are concerned with the situational context in which the person learned of the Challenger disaster. The second factor was comprised of only two items: first thoughts and own affect; these have to do with the person’s internal reaction to the news. The third, weak, factor was a singleton consisting of the "time" item. A revised scale of flashbulb quality, consisting of the six items loading on the first factor, boosted the reliability coefficient, to .72, despite the drop from nine to six items.

In order to explore the correlates of flashbulb quality in more detail, we divided the distribution of flashbulb quality at the median (4 or fewer of the 6 canonical features versus 5 or 6), and then looked for differences on various cognitive and emotional attributes. Not surprisingly, subjects whose memories had relatively many flashbulb qualities rated their memories are clear than the others (M = 4.08 vs. 3.57, respectively), t(366) = 3.03, p < .005). Comparing "highs" to "lows":

There were no significant differences in the involvement of individual sensory modalities, such as vision or audition, or in an index of overall sensory involvement created by counting the number of modalities that received ratings of "3" or greater on a 0-6 scale.

Highs were no more likely than lows to have color imagery, or to have an "observer" memory, in which they saw themselves as well as their surroundings.

There was no difference in the number of times the subjects had thought about the Challenger incident, but there was a significant difference in the number of times they had shared their memories with others.

The groups were equally surprised by the incident, and rated it as equally consequential.

Highs rated themselves as re-experiencing the event more than lows, particularly with respect to their thoughts and emotions.

There were no group differences in particular emotional states, either then or now, or in an index of overall emotional involvement.

For our money the designation "flashbulb" really only applies to what is produced spontaneously in free-recall narratives. However, the specific probes can be useful in clarifying the nature of subjects’ memories. For example, it turned out that some of our subjects (71, or 19.3%) actually witnessed the Challenger disaster on TV. These subjects’ memories actually had significantly more flashbulb qualities (M = 4.63, SD = 1.23) than did those who learned about the disaster second-hand (M = 4.10, SD = 1.35), t(366) = 3.02, p < .005. The two groups did not differ in terms of overall emotional involvement or in reported rehearsal of the incident. However, they did differ in terms of the number of sensory modalities involved in their mental imagery, but the difference was not in the direction we might have expected: subjects who actually witnessed the event had somewhat less sensory involvement in the memories (M = 2.24, SD = 1.58) than those who learned about it later (M = 2.61, SD = 1.03), t(366) = 2.58, p < .01).

In the final part of our study, we shifted from the characteristics of flashbulb memories to their function. If one role of a flashbulb memory is to link personal autobiography with history, it occurred to us that the Challenger disaster might serve as a particularly effective cue for the retrieval of both personal and historical memories. Accordingly, before completing the flashbulb memory questionnaire a random half of our subjects were asked to list "every important historical event that appeared in the news between August 1, 1985 and July 31, 1986"; the other subjects were asked to list "every important event that happened to you, personally" during the same interval. The period was chosen to bracket the Challenger disaster, and also correspond, more or less, to the structure of the academic year. Half the subjects in each group were specifically reminded that the Challenger incident occurred on January 28, 1986. The event-memory questionnaire was administered prior to the flashbulb memory questionnaire because to do the reverse would have effectively reminded all of the subjects of the Challenger incident.

Of course, subjects who were reminded of the Challenger incident got that memory for free, as it were, so we had to analyze it separately. A 2x2 between-groups analysis of variance yielded significant main effects of both memory type (personal or historical), F(1, 364) = 62.09, p < .001, and reminder (present or absent), F(1, 364) = 42.08, p < .001. The Challenger incident was more likely to be recalled as a historical event than as a personal event, and more likely to be recalled after a reminder (no surprise there), but not a single subject recalled the Challenger incident spontaneously as a personal memory, and only 30% recalled it spontaneously as a historical memory. Although we find Neisser’s "benchmark" hypothesis extremely attractive, this is not what we would expect if the Challenger incident is a place where autobiography and history meet.

We also examined the impact of the Challenger reminder on overall memory -- again eliminating the Challenger incident from consideration. Again, the ANOVA yielded significant main effects of both memory type, F(1, 364) = 353.04, p .001), and reminder, F(1, 364) = 5.81, p < .05, but this time there was a most unusual interaction, F(1, 364) = 14.01, p < 001. Perhaps the most striking finding was how few historical events were remembered, regardless of reminder condition -- only about two on average (M = 1.99, SD = 2.13), compared to more than 10 personal events (M = 10.96, SD = 6.28). 

This is not because nothing happened that year: the period in question includes at least a dozen memorable news stories.

The main effect of the reminder was also surprising: collapsing across memory conditions, subjects who got the Challenger reminder actually recalled fewer events (M = 5.59, SD = 5.48) than those who did not (M = 6.96, SD = 7.16).

Finally, the significant interaction was also unexpected: while the Challenger reminder boosted memory for historical events, it actually reduced memory for personal events. Why this should be isn’t clear, but even if the Challenger benchmark doesn’t serve as a particularly effective retrieval cue, it may have helped people to calibrate the dating of their memories. That is, once the subjects were reminded that the Challenger incident occurred in the winter of 1986, they may have been able to eliminate some candidate memories as not falling within the specified time period.



Further research is required to address this question, as well as others that have arisen in the course of this study. For the present, however, we think our study suggests how research on flashbulb memories can go beyond the original questions of accuracy and special mechanisms. Even if they are not particularly accurate, flashbulb memories are still memories -- that is, mental representations of the past. It may be that our scale of "flashbulb quality" will prove useful in the study of autobiographical memory in general, by identifying memories that have particular personal, emotional, and interpersonal significance for the subject. Moreover, by serving as benchmarks and as a source of narrative exchange between people, flashbulb memories may play an important role in the cognitive economy of individual and collective memory.



Paper presented at the annual meeting of the Western Psychological Association, Irvine, Ca., April 2002.  Research supported by NIMH Grant #MH-35856.  We thank Martha L. Glisky for help in collecting the data at the University of Arizona.


This page last revised 04/08/10 02:58:53 PM.