MefTWry & Cognition /989, /7 (4), 435-443 Flashbulb memories: Special, but not so special SVEN-AKE CHRISTIANSON University of Umeb, Umea, Sweden This study deals with flashbulb memories associated with the assassination ofthe Swedish prime minister OlofPalme. A major goal ofthis research was to explore the consistency ofsuch memories through comparison of the subjects' recollections on two different occasions, one year apart. The results obtained indicated that flashbulb events are accurately recalled in terms of a narrative conception of the concomitant circumstances of the event, but that the event descriptions are not consistent with respect to the specific details of these circumstances. It was concluded that the loss of information during one year contradicts the notion that flashbulb memories persist in absolute accuracy over time, as has been claimed in previous studies. Rather, these memories appear to be reconstructions based on residuals ofthe circumstances concomitant with the specific event (i.e., that of'first hearing ofthe shocking news), and these memories followthe same pattern of recollectionas does recollectionofother autobiographical and laboratory-inducedemotional events. On the evening of February 28, 1986, as he and his wholeevent, including manyirrelevantdetails, relatively wife were walking home from a movie theater in down­ indiscriminately. As Colegrove reported (1899), such townStockholm, theSwedish primeminister, OlofPalme, vivid images seem to be easily recallable even after 33 was shot and killed in cold blood. The event's devastat­ years. Thus, althoughin manyexperimental and clinical ing impact on Swedishcitizens resembles the emotional studies on memory of emotionally traumatic events, recol­ traumaexperiencedby many Americanswhenthey were lection has been found impaired with respect to certain, first told of the assassination of John F. Kennedy. The especially peripheral or irrelevant details of the emotional mainpurposein thepresentstudywasto investigate Swed­ events(e.g., Christianson & Loftus, 1987, 1988; Clifford ish citizens' recollections of the circumstancesin which & Scott, 1978; Deffenbacher, 1983), it seemsthat no one they first heard the news of Palme's assassination, and, forgets the flashbulb memories.I more specifically, to study the consistency of these One may stillask, however, whetherthesevivid recol­ memories over time. lections really do remain unchanged in memory over time. Previous researchhasshown an impressive concordance Brownand Kulik's descriptionof a flashbulb memoryas in subjects' remembering of shocking national events such a "photographic picture" would lead us to suspect that as assassinations (see, e.g., Brown& Kulik, 1977;Cole­ such memories are completely accurate, and that nothing grove, 1899; Pillemer, 1984; Winograd & Killinger, is later lost or addedto the pictorialmemoryof the scene 1983). Not only does such emotionally shockingnews it­ in which a person finds him- or herself when the flash­ self remain very well preserved in people's memories, bulb is first fired. Once one has access to the "picture," so too do the subjects' memories of the circumstances one should thus have access to both importantand unim­ under whichthey first heard the news: the informant, the portantdetails of the flashbulb occasion. An uncritical ac­ location, the time, the nature of any ongoingactivity, the ceptance of suchvividdescriptions of eventsfromthe dis­ subject's own clothing, the subject's own affect, and so tant past has been questioned by Neisser (1982), on the forth. This detailed recall performance was referred to groundthat the accuracy of flashbulb memories has never as flashbulb memory by Brown and Kulik (1977), who been verified. After providing some examples of com­ attributed it to a neuropsychological "now print" mech­ plete fabrication in flashbulb memories, Neisser con­ anismthat is triggeredwhenan eventisemotionally arous­ cludes: "Apparently flashbulbs can be just as wrong as ing, surprising, and consequential, or when an event has other kindsof memories; they are not producedby a spe­ great''biologicalsignificance. " According to Brownand cial quasiphotographic mechanism" (p. 45). Further ar­ Kulik, this special memory mechanism preserves the gument against a special flashbulb-memory mechanism has recently been presented by McCloskey, Wible, and This research was supported by Grant F 158/87 from the Swedish Cohen (1988). In their evaluation of previous research Council for Research in the Humanities and SocialSciences. The research was completed while the author was a Visiting Fulbright Scholar at the on flashbulb memories, they conclude that there are no University of Washington. The author is indebted to Mats Carlsson, empirical or logical grounds for postulating a special Anita Fredriksson, Maria Nilsson, and Stefan Viktorsson for their in­ mechanism for thistypeof memories. Instead, McCloskey valuable help in conducting the research. Many thanks also to Eugene et al. suggestthat "flashbulb memories should be viewed Winograd and one anonymous reviewer for their constructive comments on an earlier draft of this paper. Requests for reprints should be sent as products of 'ordinary' memory mechanisms, and hence to Sven-Ake Christianson, Department of Psychology, University of as phenomena that may offer insights into the nature of Umea, RAdhusesplanaden 2, S-902 47 Urnea, Sweden. these mechanisms" (McCloskey et al., 1988, p. 171).2 435 Copyright 1989 Psychonomic Society, Inc. 436 CHRISTIANSON Nevertheless, despite the occasional criticism, it does sidered flashbulb memories as they are defined by Brown seem that we have a robust phenomenon, referred to as and Kulik (1977), one might expect the subjects to flashbulb memory, in which some information is retained recollect identical details ofthe circumstances on both test quite vividly from highly emotional public events. These occasions. That is, consistent and accurate flashbulb memories are reported to persist with little subjectively memories are here defined in terms of a nonsignificant experienced loss of perceptual clarity (see, e.g., Brown difference between recall responses in a 1986 test and a & Kulik, 1977). Still, from these subjective reports of repeated test in 1987. On the other hand, if these recollec­ flashbulb memories we cannot tell how accurate people tions are merely reconstructions, subjects might recall the really are in their descriptions of the flashbulb event and gist (central details) of the main event and reconstruct a its concomitant circumstances. The possibility remains narrative conception based on this core recollection. If that the descriptions of the circumstances may be have the latter hypothesis is true, then flashbulb memories are been altered or fabricated without conscious awareness. very much comparable to other, more private, autobio­ One way to verify the accuracy and consistency of flash­ graphical memories (see Rubin, 1982; Rubin & Kozin, bulb memories is to ask subjectsa series of questions about 1984), and also to well preserved central detail informa­ their recollection ofthe circumstances in which they first tion found in studies of laboratory-induced emotionally heard the shocking news, and then ask the same questions traumatic events (e.g., Christianson, 1984; Christianson again later on, in order to compare the coherence between & Loftus, 1987, 1988). the descriptions reported on the two occasions. This dou­ Since an intense emotional reaction to a flashbulb event ble assessment procedure was employed by Pillemer has been claimed as one of the most important theoreti­ (1984) in a study of peoples' memories of the assassina­ cal prerequisites for a flashbulb memory (see Brown & tion attempt on President Reagan on March 30, 1981, and Kulik, 1977; Pillemer, 1984; Schmidt & Bohannon, it was also used more recently by McCloskey et al. 1988), one major shortcoming of the data presented in (1988), in a study of peoples' recollections of the circum­ the McCloskey et al. (1988) and the Pillemer (1984) stances in which they first learned of the explosion of the studies is the lack of evidence of intense affect in their space shuttle Challenger on January 28, 1986. Pillemer respective subjects. In the Pillemer study, the average questioned the subjects 1 month, and again 6 months, after emotion and surprise ratings were only moderate (about the assassination attempt, and he concluded that memories 3 on a five-point scale)-hardly the strong emotional ofsuch an event are highly consistent over a 6-month in­ response claimed by Pillemer (1984), as well as Schmidt terval. On the other hand, in the McCloskey et al. study, and Bohannon (1988), who used the Pillemer results in the subjects were questioned a few days after the explo­ their defense ofthe "flashbulb-memory hypothesis. " In sion, and then again 9 months later; a considerable in­ the McCloskey et al. (1988) study, the subjects' emotional consistency was found between the responses given on reactions were unfortunately not measured, which makes the two occasions. These results suggest that flashbulb it difficult to determine the empirical and theoretical value memories are subject to inaccuracy and deterioration. oftheir evaluation of memory processes associated with Just a month after the Challenger tragedy, the citizens "flashbulb" events as opposed to other, "ordinary," of Sweden experienced their own national trauma-the emotionally surprising events. The issue of affect (and assassination of their prime