• "forgetting"
  
• "amnesia"
  
• "repression"
  
• "dissociation"
  

1. "Amnesia" is a descriptive construct. It directs our attention to the condition of being unable to remember experiences like childhood sexual abuse.
2. "Repression" and "dissociation" are explanatory constructs. They point to hypothesized psychological mechanisms that may be responsible for the condition of amnesia.
3. All constructs exist on a continuum from descriptive to explanatory. When it comes to empirical evidence of traumatic and recovered memories, dissociation is more descriptive of more empirical evidence than is repression.
4. The persistence and the accuracy of a memory are completely separate issues, and not keeping them separate as constructs leads to considerable confusion. People can have memories that are largely true or largely false whether or not those memories have been recovered or continuously available. Thus it is very misleading to discuss "recovered versus false memories of abuse" or present these as mutually exclusive categories.

It is not rare for people to say that at some point they came to remember a past abuse experience which they had not previously remembered.

• "He remembered."
  
• "She recovered a memory."
  
• "Before beginning therapy, the patient experienced delayed recall."
  
• "Some research subjects had a return of previously repressed memories."
• "Some subjects in the study made new associative linkages among dissociated memory fragments, and integrated these with their conscious, verbal and narrative autobiographical memories."

As noted above, no matter which constructs we use, we should never confuse constructs with empirical evidence. And no matter which constructs we prefer, this does not change the empirical evidence we already have (though our constructs can influence our interpretations of it). Finally, better constructs lead to better research questions and methods, better empirical evidence, better interpretations of evidence and still-better constructs. . . That's the self-perpetuating process of scientific progress, of course.

This Web page documents scientific progress by those who study child abuse memories:

1. A substantial body of empirical evidence of amnesia and delayed recall for abuse has existed for years.

2. Significant progress has occurred in how such empirical evidence is described and explained in ways that do not confuse and mislead people, particularly in shifting from "repression" to more descriptive constructs like dissociation and explanatory constructs from cognitive science and neuroscience.

Hypertext Table of Research Findings

Amnesia and delayed recall for sexual abuse experiences are NOT rare. This table presents published research studies on whether some people with histories of child sexual abuse experience periods of amnesia and delayed recall.

This is a HYPERTEXT table: Click on authors' names to go directly to abstracts of their studies; then click on your browser's "Back" button to come back to the table.

Methodological notes: 1) Data from "community" samples are more representative than data from "clinical" samples; community samples represent the general population, while clinical ones represent people in mental health treatment. 2) "Prospective" studies are better than "retrospective" ones; in the former, researchers follow and later question confirmed people with abuse histories, while the latter rely on subjects' reports of past abuse. 3) Each of these studies has some methodological flaws or limitations, but their relative consistency strongly suggests that these findings are not spurious.

A note on this table's limited scope: The table below presents only a small selection of studies. As early as 1997, Scheflin and Brown's review of the scientific research on recovered memories of sexual abuse (see next section) could present a table of 25 studies – every single one of which found periods of total and/or partial amnesia in a subpopulation of people with histories of sexual abuse.

*  There were 129 subjects with documented sexual abuse incidents, but 42 (38%) did not report those experiences, and it was concluded that some of those subjects actually did not remember (see Williams, Study 1, below). Clearly, some of those subjects might experience delayed recall later, which would suggest that the the 16% figure is an underestimate. (Also, 5 of the 80 who did remember were inadvertently not asked about whether they had ever experienced amnesia for the documented abuse incident.)

**  As discussed in the paper, it is highly unlikely that all subjects who did not report sexual abuse (despite it being substantiated in childhood) simply did not remember that the events had occurred. Instead, as in all such studies, some percentage did not report prior abuse for other reasons – including unwillingness to disclose the information to researchers, current interpretations of those experiences as not abusive, etc. – and it appears these influences are greater for males than females.

The Journal of Psychiatry and Law
Special Issue on Recovered Memories of Sexual Abuse

The Summer 1996 volume of The Journal of Psychiatry and Law, published in February of 1997, is an indispensable resource. Its six articles include Scheflin and Brown's comprehensive review of scientific studies of recovered memories of sexual abuse, and Dalenberg's study of the accuracy of sexual abuse memories recovered in psychotherapy (she actually conducted interviews with both victims and perpetrators, some of whom confessed). Directions for ordering this special issue, as well as the Fall 1995 special issue that focused on claims of false memories, are at the end of this section.

Abstract: Using data from a study with prospective-cohorts design in which children who were physically abused, sexually abused, or neglected about 20 years ago were followed up along with a matched control group, accuracy of adult recollections of childhood physical abuse was assessed. Two hour in-person interviews were conducted in young adulthood with 1,196 of the original 1,575 participants. Two measures (including the Conflict Tactics Scale) were used to assess histories of childhood physical abuse. Results indicate good discriminant validity and predictive efficiency of the self-report measures, despite substantial underreporting by physically abused respondents. Tests of construct validity reveal shared method variance, with self-report measures predicting self-reported violence and official reports of physical abuse predicting arrests for violence. Findings are discussed in the context of other research on the accuracy of adult recollections of childhood experiences.

Excerpts from the Discussion:
     ". . . In many ways, these findings indicate accuracy in retrospective self-reports and good discriminant validity. Individuals who were physically abused, based on official records, retrospectively reported the highest rates of childhood physical abuse in the sample. On the CTS (Severe Violence and Very Severe Violence subscales) and the SRCAP [Self-Report Measure of Childhood Physical Abuse], physically abused individuals reported significantly higher rates of physical abuse than did individuals who had experienced sexual abuse or neglect in childhood and individuals who were part of a matched control group. The extent of remembering (i.e., the percentage of individuals who had been physically abused who reported having been physically abused on one of the measures used here) is in line with previous research. These results also reveal that the extent of reporting a history of childhood physical abuse varied dramatically by the criterion (or measure) used.

     "At the same time, there is a problem in underreporting of physical abuse. A substantial group of individuals who were physically abused do not report having been physically abused in childhood. Of the 110 people in the sample who had documented cases of physical abuse in childhood, 60-62% reported abuse using the CTS-VSV and SRCAP. This means that approximately 40% of individuals with documented histories of physical abuse did not report. Whether these people did not report (as suggested by Della Femina et al., 1990) because of embarrassment, a wish to protect parents, a sense of having deserved the abuse, a conscious wish to forget the past, or lack of confidence in or rapport with the interviewer, we do not know. But these findings suggest that a substantial minority would not be included in retrospective self-report assessments of childhood physical abuse. A more lenient criterion (such as the CTS-Minor Violence subscale) would capture most of the physically abused people (see Table 2); however, this criterion also identifies 92% of the sexual abuse and neglect cases and 86% of the control participants as having been physically abused in childhood. Using the CTS-Minor Violence subscale, the rate of false positives (as presented in Table 5) approaches almost half the sample. These findings illustrate that the rate of false positives is directly related to the measure of childhood physical abuse used. . .

     "Henry et al. (1994) concluded that reliance on retrospective reports about psychosocial variables should be treated with caution. They suggested that "the use of retrospective reports should be limited to testing hypotheses about the relative standing of individuals in a distribution and should not be used to test hypotheses that demand precision in estimating event frequencies and event dates" (p. 92). We support their recommendation to use caution against overly simplistic interpretations that take retrospective reports at face value.

     "These methodological problems pose significant challenges to researchers in the field. Notwithstanding the real difficulties involved, there is a critical need to develop reliable and valid ways to assess histories of childhood victimization. . . ."

Widom, C. S. & Morris, S. (1997). Accuracy of adult recollections of childhood victimization: Part 2. Childhood sexual abuse. Psychological Assessment, 8, 412-421.

Abstract: Questions have been raised about the accuracy of retrospective self-reported information about childhood sexual abuse. Using data from a prospective-cohorts-design study, a large group of children who were sexually and physically abused or neglected approximately 20 years ago were followed up and compared with a matched control group. Accuracy of adult recollections of childhood sexual abuse was assessed using 4 different measures, completed in the context of a 2-hr in-person interview in young adulthood ( N = 1, 196). Results indicate gender differences in reporting and accuracy, substantial underreporting by sexually abused respondents in general, good discriminant validity and predictive efficiency of self-report measures for women, and some support for the construct validity of the measures. Implications for researchers and practitioners are discussed.

Excerpts from the Introduction:
     ". . . [A] significant risk of distortion and loss of information is associated with the recollection of events from a prior time period. If asked to recall childhood events, it is possible that respondents forget or redefine their behaviors in accordance with later life circumstances or their current situation. It is also possible that a person might redefine someone else's behavior in light of current knowledge. Unconscious denial (or repression of childhood traumatic events) may also be at work in preventing the recollection of severe cases of childhood abuse. Furthermore, given society's disapproval of various forms of family violence, a person may be embarrassed to report such experiences or unwilling to reveal such private information in the context of an interview setting. Thus, for a variety of reasons, there may be considerable slippage in accuracy in retrospective reporting. . . .

     "Empirical findings suggest that a person's cognitive appraisal of life events strongly influences his or her response (Lazarus & Launier, 1978). The same event may be perceived by different individuals as irrelevant, benign, positive, or threatening and harmful. It is likely that a child's cognitive appraisal of early childhood events will also determine at least in part whether they are experienced as neutral, negative, or harmful. The child's perception might reflect events occurring subsequent to the abuse experience as well as the child's perception of the experience. Theoretically, this is also important because long-term consequences may depend on the person's awareness or memory of the earlier abusive experience or experiences. Considering Lazarus and Folkman's (1984) discussion of the role of cognitive appraisal in mediating one's response to stress, it may be that children who do not define their early childhood experiences as abusive will show better outcomes.

     "There may also be gender differences in reporting or willingness to report childhood sexual abuse for a variety of reasons. Female psychiatric patients have been found more likely than male patients to report histories of sexual abuse. . ., female patients have been found more likely than male patients to reveal childhood sexual assault experiences to therapists (Jacobson & Richardson, 1987), and women have reported greater likelihood than men of being a victim of sexual assault (Burnam et al., 1988). Social pressures against reporting early childhood sexual experiences and embarrassment may lead to greater reluctance among men to report, whereas it may be socially more acceptable for women to report such histories. On the other hand, some of the apparent underreporting may be associated with the small number of male victims of sexual abuse in most studies (Finkelhor, 1990).

     "One approach to assessing the power or efficiency of retrospective self-report measures is to calculate the relative improvement over chance (RIOC). Loeber and Dishion (1983) devised this index to represent the improvement over chance as a function of the range of its possible predictive efficiency. . . .

     "A second approach to establishing the usefulness of retrospective reports of childhood sexual abuse is based on the construct validation process, one of the techniques used to establish the psychometric qualities of assessment instruments. In addition to establishing the validity of retrospective self-report measures using "known groups," construct validity attempts to assess how these self-report measures theoretically relate to other variables or indexes. That is, there are certain theoretical expectations about the way people who have a history of childhood sexual abuse should behave or should manifest certain outcomes. Based on logical relationships, then, tests of construct validity can offer evidence that these measures do or do not measure childhood sexual abuse, without providing definitive proof."

     "To validate our retrospective self-report measures of childhood sexual abuse, three outcomes frequently associated with childhood sexual abuse (depression, alcohol problems, and suicide attempts) will be assessed. Ideally, retrospective reports of childhood sexual abuse should relate to subsequent outcomes similar to the way official reports of childhood sexual abuse relate to these outcomes.

Excerpts from the Discussion:
     "We examined the accuracy of four retrospective self-report measures of childhood sexual abuse. In general, we found that women and men differ in the extent to which they recall or report having experienced childhood sexual abuse. Approximately 16% of men with documented cases of sexual abuse considered their early childhood experiences sexual abuse, compared with 64% of women with documented cases of sexual abuse. These gender differences may reflect inadequate measurement techniques or an unwillingness on the part of men to disclose this information. They may also reflect differences in the meaning of these behaviors for men and women, particularly viewed in a cultural context. Gender differences in reporting and in perceptions of early childhood experiences may reflect early socialization experiences in which men learn to view these behaviors as nonpredatory and nonabusive. Many of the sexual experiences considered to be sexual abuse (e.g., showing/touching sex organs, kissing in a sexual way) may be seen as developmental rites of passage, part of a learning process. Men reported more sexual experiences in which they touched the other person. Social pressures against reporting certain kinds of early childhood experiences may also lead to greater reluctance among men to report. Future research ought to examine whether the underreporting by men is due to embarrassment or to perceptions about sexual experiences.

     "In our examination of the validity of retrospective self-report measures of childhood sexual abuse using known groups, we also found gender differences in the discriminant validity of the four measures. Our results indicate good discriminant validity for the self-report measures used here for women but much less so for men. A higher percentage of women with official histories of childhood sexual abuse recall or report sexual abuse in young adulthood than do women with histories of physical abuse or neglect, who in turn report higher levels than nonabused and nonneglected controls (Table 3). On the other hand, men in our sample with documented cases of sexual abuse do not report higher levels of sexual experiences (any sex before age 12) than do men with documented cases of physical abuse or neglect or control men. Sexually abused men are significantly more likely to consider that they were sexually abused and to report more often having had sex against their will than are controls, but so are physically abused or neglected men. It is noteworthy that more physically abused or neglected men reported having had sex with an older person than did sexually abused men, none of whom reported having had this experience in childhood.

     "Overall, we found substantial underreporting of sexual abuse among known victims of childhood sexual abuse. This is particularly impressive because these are court-substantiated (documented) cases of childhood sexual abuse. Much attention has been paid to the lack of recall or failure to report histories of childhood sexual abuse among known victims of abuse. Although this lack of reporting is significant, it may not be surprising when viewed in a somewhat different context. Nonreporting by crime victims in the context of victimization surveys has been studied for a number of years (Garofalo & Hindelang, 1977), and problems with respondent embarrassment about the incident or "protective mechanisms," or simply memory decay or forgetting have been described. . . .

     "For women, we found strong relationships between retrospective self-report measures of childhood sexual abuse and the three outcomes examined here: DSM-III-R diagnoses of depression and alcohol abuse/dependence and suicide attempts. We also found that women with documented cases of childhood sexual abuse who were followed up prospectively into young adulthood were at increased risk for having alcohol abuse/dependence diagnosis and for making suicide attempts.

     "It was surprising that we did not find that women or men with documented cases of childhood sexual abuse were at increased risk of being diagnosed with depression according to DSM-III-R criteria, despite the widespread belief that childhood sexual abuse leads to depression. We did, however, find a significant relationship between retrospective self-report measures of childhood sexual abuse and depression diagnosis. Thus, this pattern of findings suggests that the relationship between childhood sexual abuse and depression is complicated and may depend on a person's cognitive appraisal of early life events (cf. Lazarus & Folkman, 1984). Individuals who meet the criteria for a DSM-III-R depression diagnosis (current or remitted) are more likely to recall having been sexually abused in childhood than individuals without depression diagnoses, although individuals with documented cases of sexual abuse in childhood who were followed up into young adulthood were not at increased risk of receiving a depression diagnosis. This was true for men as well as women in this sample. . . .

     "The underreporting we found means that there is a substantial group of people with documented histories of childhood sexual abuse who do not report these experiences when asked in young adulthood to do so. Whether this is due to loss of memory, denial, or embarrassment is not known. However, there are important implications from these these findings for other researchers and clinicians. For researchers, the underreporting of childhood sexual abuse poses a serious concern for epidemiological research, especially that which involves a large proportion of men. For clinicians, these findings reinforce the need to develop more sensitive techniques to elicit this information from men.

Diana Elliott

Diana Elliott is a psychologist at the University of California at Los Angeles (UCLA) School of Medicine. She has published two important studies of delayed recall of abuse and other traumatic experiences, based on data from a stratified random sample of the general US population. The most recent of these publications is an investigation of rates of partial and complete delayed recall for a variety of traumatic experiences, not just sexual abuse. It was published in the October 1997 Journal of Consulting and Clinical Psychology, the methodologically rigorous and most prestigious journal of the American Psychological Association. In that paper, extensively excerpted below, Elliott uses her findings to address the validity of the construct of psychogenic or dissociative amnesia and the claim that recovered memories are the product of questionable therapy practices.

Elliott, D. M. (1997). Traumatic events: Prevalence and delayed recall in the general population. Journal of Consulting and Clinical Psychology, 65, 811-820.

Abstract: A random sample of 724 individuals from across the United States were mailed a questionnaire containing demographic information, an abridged version of the Traumatic Events Survey (DM Elliott, 1992), and questions regarding memory for traumatic events. Of these, 505 (70%) completed the survey. Among respondents who reported some form of trauma (72%), delayed recall of the event was reported by 32%. This phenomenon was most common among individuals who observed the murder or suicide of a family member, sexual abuse survivors, and combat veterans. The severity of the trauma was predictive of memory status, but demographic variables were not. The most commonly reported trigger to recall of the trauma was some form of media presentation (i.e., television show, movie), whereas psychotherapy was the least commonly reported trigger.

Excerpts from the literature review:
     "In contrast to normal forgetting, theoretical writers in the area of trauma have suggested that some memory loss in trauma survivors may reflect dissociative avoidance strategies developed by the victim to reduce trauma-related distress. . . From this perspective, traumatic memory loss may be understood as a form of avoidance conditioning, whereby access to memory is punished by the negative affect that accompanies the recall, thereby motivating the development of memory-inhibiting mechanisms. Such avoidance strategies might interfere with memory at any point during rehearsal, storage, or retrieval of material. . . There are other reasons for memory loss, such as organic impairment, lack of significance of the event, and infantile amnesia. However, if access to events is lost because of avoidance conditioning, the more severe and chronic the trauma, the more painful the resultant affect should be, and thus, the more likely the victim's avoidance behavior would be reinforced. . .
     ". . . [S]ome memory disruption seems to occur at the retrieval level, rather than solely at rehearsal and storage levels. According to Tulving (1983), cues that assist in the recall of events are typically those that match the original encoded material. This suggests that dissociative avoidance strategies may be effective if they reduce the individual's responsiveness to relevant cues in the environment that otherwise may activate the original memory traces. However, when recognition cues are sufficient in number, intensity, or meaningfulness, they may overwhelm existing avoidance defenses, resulting in the emergence of previously unavailable memories. Extreme dissociative avoidance (i.e., in response to a highly aversive experience), however, might be relatively resistant to external cuing and, thus, less likely to remit in response to environmental triggers.
     "Understood from this perspective, dissociative amnesia for previous traumatic events would be best predicted by the severity of the trauma and most apt to be triggered by intrapersonal, interpersonal, or environmental cues that closely match the original trauma. Although certain forms of memory loss (e.g., infantile amnesia, normal forgetting, organic impairment) may best be predicted by demographic variables (e.g., age at time of trauma, length of time since the event, current age), avoidance-related traumatic memory loss should be less a function of demographic variables and more related to characteristics of the trauma" (p.812).

Excerpts from the results section:
     "Participants were most likely to report continuous memories of adult sexual assault that did not include penetration (94%), major motor vehicle accidents (92%), and natural disasters (89%). A history of partial memory loss was most common when an individual had witnessed murder or suicide of a loved one (38%), had been victim of child sexual abuse (22%), and had been a victim of child physical abuse (22%). A history of complete memory loss was most common among victims of child sexual abuse (20%), witnesses of combat injury (16%), victims of adult rape (13%), and witnesses of domestic violence as a child (13%)" (p.814).

     ". . . . [T]hose who reported delayed recall (partial or complete memory loss) of any trauma also reported significantly (a) more types of trauma, (b) more distress about the trauma (both at the time of the event and at the time of data collection), and (c) a younger age at the time of the earliest trauma.
     "The relationship between age and a history of memory loss could be due to the normal lack of recall for events occurring in the first 3 to 4 years of life (i.e., infantile amnesia) for those victimized in early childhood (Loftus, 1993). To examine this hypothesis, I completed a second analysis, deleting the 24 participants who reported trauma before the age of 5. This produced no change in the results, with memory loss more frequently reported when the trauma occurred at a younger age. . ." (p.814)

     ". . . Across traumas, participants reported that recall was most commonly triggered by some sort of media presentation (54%), an experience similar to the original trauma (37%), and a conversation with a family member. Recall of the trauma was least likely to have been triggered by a sexual experience (17%), or psychotherapy (14%)" (p.815).

Excerpts from the discussion section:
     "The findings of the present study suggest that a history of trauma is common in the United States. For example, 40% of respondents experienced a major motor vehicle accident or natural disaster, 43% had witnessed violence, and 50% had been victims of interpersonal violence. . . .
     "These data also suggest that delayed recall of traumatic experiences may not be uncommon, with some proportion of individuals reporting impaired recollection for virtually every type of trauma. This phenomenon appears to be more common among events considered particularly upsetting or distressing (e.g., among childhood sexual abuse survivors, those who witnessed the murder or suicide of a loved one, and veterans who witnessed combat injury) and less common for events that contained no interpersonal violence (e.g., major motor vehicle accidents, disasters, and having a child die under the age of 18)" (p.816).

     "Race was the only demographic variable that was even marginally associated with delayed recall of a trauma, and the race-memory relationship was mediated by the severity of the trauma experienced. However, several characteristics of trauma severity predicted memory status. Such data support an avoidance defense mechanism hypothesis as a partial explanation for the findings, because more traumatic events would appear more likely to be remembered, not forgotten, if no defensive response was involved" (p.817).

     "The extent to which the questions used in the present study may have been misunderstood by participants is unclear. However, a pretest of the questions used in this study indicated that 96% of the individuals understood the questions to refer to a period of time in which the individual was unable to access part or all of the memory of the traumatic event, as opposed to simply not thinking about the trauma. Additionally, participants in the present study went on to record their age at recall of each specific trauma and indicated what it was that cued their recall. Given these data, misinterpreting the questions to be about normal forgetting is not likely to be a sufficient explanation for the memory findings reported here" (p.818).

     "This study does not support the notion that delayed recall is limited to sexual abuse. It suggests that the phenomena occurs across a variety of traumas and is especially high for traumatic events involving interpersonal victimization" (p.818).

     "With regard to the claim [that therapy creates recovered memories], only 14% of the participants in this study who reported delayed recall of a trauma reported having their memory triggered during the course of therapy. Even if all the individuals who had ever been in treatment reported delayed recall of trauma (a conservative assumption), 86% of the sample, nevertheless, reported recovering memory through other means. This finding suggests that the process of psychotherapy, per se, does not intrinsically explain the recovered memory phenomenon. Rather, these data suggest that, like other posttraumatic stress responses, intrusion of previously avoided memory can be cued by environmental stimuli, perhaps in the same way as has been documented with posttraumatic flashbacks. . ." (p.818)

The final paragraph:
     "As previously noted, the best predictor of memory status was the severity of the trauma, rather than demographic variables. These findings suggest that the traumatic impact of the event – rather than childhood amnesia, normal forgetting, secondary gain, or iatrogenic treatment effects – provides a good conceptual fit to the data. Many authors (Herman, 1992; Terr, 1994; van der Kolk, et al., 1996) have noted that traumatic amnesia is a complex phenomenon that involves biological, cognitive, and psychological aspects that may vary from traditional notions of 'normal' memory. In this regard, future research might focus on such processes as they relate to normal versus traumatic encoding, forgetting, and recalling" (p.818).

Elliott, D. M., & Briere, J. (1995). Posttraumatic stress associated with delayed recall of sexual abuse: A general population study. Journal of Traumatic Stress, 8, 629-647.

Abstract: "This study examined delayed recall of childhood sexual abuse in a stratified random sample of the general population (N = 505). Of participants who reported a history of sexual abuse, 42% described some period of time when they had less memory of the abuse than they did at the time of data collection. No demographic differences were found between subjects with continuous recall and those who reported delayed recall. However, delayed recall was associated with the use of threats at the time of the abuse. Subjects who had recently recalled aspects of their abuse reported particularly high levels of posttraumatic symptomatology and self difficulties (as measured by the IES, SCL, and TSI) at the time of data collection compared to other subjects."

Bessel van der Kolk

Boston University psychiatrist Bessel van der Kolk is one of the foremost authorities on traumatic memory, particularly the possible roles of biological and dissociative phenomena in the processes of encoding and retrieval. (I have conducted research with Dr. van der Kolk and co-authored papers on traumatic memories with him.) The five papers below, four of which are available on the web, cover some of Dr. van der Kolk's key contributions to the study of traumatic memory.
     The first three papers are available on the web, and these document progress in the measurement of traumatic memory characteristics with the Traumatic Memory Inventory. The introduction to "Exploring the Nature of Traumatic Memories" addresses important historical and methodological issues in research and theory concerning traumatic memories, and how these necessarily differ from research and theory concerning normal memories. "Retrieving, Assessing, and Classifying Traumatic Memories," of which I am the primary author, introduces a new method for evoking and assessing the nature of traumatic memories, is rich with clinical data, and has extensive discussions of methodological and theoretical issues. "Dissociation and the Fragmentary Nature of Traumatic Memories" is a classic in the field, and essential reading for anyone who wants to understand how the psychological construct of dissociation sheds much more light on the nature of many traumatic memories than that of repression.

Abstract: "For over 100 years clinicians have observed and described the unusual nature of traumatic memories. It has been repeatedly and consistently observed that these memories are characterized by fragmentary and intense sensations and affects, often with little or no verbal narrative content. Yet, possibly because traumatic memories cannot be precipitated under laboratory conditions, the organization of traumatic memories has received little systematic scientific investingation. In our laboratory we have developed an instrument, the Traumatic Memory Inventory (TMI), which systematically assesses the ways that memories of traumatic experience are organized and retrieved over time. In this paper we report findings from our third study using the TMI, of 16 subjects who had the traumatic experience of awakening from general anesthesia during surgery. We assessed changes in traumatic memory characteristics over time and differences between memories of subjects with and without current Posttraumatic Stress Disorder. Our findings suggest the need for more rigorous methods for the assessment of the evolution of traumatic memories. In order to develop a comprehensive and integrated understanding of the nature of traumatic memory, we need to combine careful clinical observations with replicable laboratory methods, including those of cognitive science and neuroscience."

Excerpts: "The understanding of how people process traumatic events has, until recently, been entirely within the domain of clinical practice and observation. Traditionally, the fields of clinical psychology and psychiatry on the one hand, and cognitive science and neuroscience on the other, have had such widely divergent samples, methodologies and concepts on which they based their understandings of memory processes, that there has been a veritable confusion of tongues between these disciplines. During the past decade, when the observation that people may lose all memory for sexual abuse experiences and retrieve them at a later time was brought to the public's attention, many cognitive scientists took an incredulous stance. Yet for over a century this observation had been consistently reported in the psychiatric literature on other traumatized populations. Despite dozens of reports, starting with Pierre Janet (1889) in the 1880s, followed by Breuer and Freud (1893), repeated during the first World War (Meyers, 1915; Southard, 1919), the second World War (Sargant and Slater, 1941) and the Vietnam War (van der Kolk, 1987), most laboratory scientists disregarded the validity of these observations. In the past decade a small group of cognitive scientists began to take clinical reports seriously (Freyd, 1991, 1994; Morton, 1994; Schooler, 1994). However, because amnesia and delayed recall for traumatic experiences had never been observed in the laboratory, many cognitive scientists adamantly denied that these phenomena existed (e.g., Loftus, 1993; Loftus & Ketcham, 1994), or that retrieved traumatic memories could be accurate (Kihlstrom, 1995).
      "In both science and therapy we often are confronted with unexpected findings. Whether one is a laboratory scientist or a clinician, such phenomena ideally should provoke new insights and creative theoretical and methodological advances. Laboratory scientists' practice of "controlled" research may render them more prone to observe the phenomena that they set out to measure, while clinicians cannot help but be frequently confronted with unexpected phenomena that don't fit their constructs and models. This often forces them to suspend disbelief and to attend to the unfolding of clinical data for which they have no pre-existing explanations" (pp. 10-11).

      "Despite the power of these clinical observations, these phenomena have not been systematically studied in the laboratory. The problem is not that laboratory science cannot study traumatic memories, but that laboratory science cannot study traumatic memories under conditions in which the memories studied are for events that take place in the laboratory. The event encoded into memory simply cannot be a 'controlled' variable in the laboratory science sense, as in landmark work of Loftus and her colleagues with systematically altered films of car accidents (Loftus, 1975, 1979). This is so because, for ethical reasons, not scientific ones, the extreme terror and helplessness that precede the development of PTSD simply cannot be replicated in such a setting. Roger Pitman (personal communication, July, 1996) attempted to simulate a truly traumatic stressor by having college students watch "The Faces of Death," a film consisting of actual footage of deaths and mutilations of people and animals, in the laboratory. Even this stimulus, which is probably as extreme as any institutional review board would allow, failed to precipitate PTSD symptoms in these normal volunteers.       "Hence it appears inescapable that to study the nature of traumatic memories one must study the memories of people who have actually been traumatized..." (pp. 11-12)

      "Shobe and Kihlstrom (1997) recently published an article claiming that traumatic memories are qualitatively not any different from memories of ordinary events. Without actually having studied the memories of traumatized individuals themselves, they dismissed all existing observational studies of the memories of individuals with PTSD out of hand. Their rationale for doing so is found in the article's final section, "Clinical lore and scientific evidence."

Although their ideas about the underlying mechanisms are different, Terr, van der Kolk and Whitfield all agree on the outcome: Memories of trauma, or at least of certain forms of trauma, are encoded by processes, such as repression and dissociation, that make them difficult to retrieve as coherent verbal narratives. The result is that traumatic memories are primarily available as isolated, nonverbal, sensory, motor, and emotional fragments. If this conclusion were valid... (1997, p.74).

      "Shobe and Kilhstrom have reversed the order of things. First, clinicians working with traumatized individuals found themselves confronted with unexpected observations: incoherent memories of 'isolated, nonverbal, sensory, motor, and emotional fragments.' Second, once they were struck by the consistency of this observation, clinician-scientists looked for theoretical constructs to make sense of the data.
      "Initially, the constructs of repression and dissociation were the best they could find. It is not that pioneering students of traumatic memory ignored laboratory evidence, or that they did not search among laboratory scientists' constructs for ones that could help them explain the data they were encountering. It is just that when it came to delayed recall and the fragmentary nature of many traumatic memories, clinician-scientists encountered a conceptual void in the laboratory memory research literature. Laboratory scientists had studied memories for events they had created under controlled conditions, and thus had never encountered fragmentary traumatic memories. In short, laboratory scientists never had a reason to create constructs explicitly addressing fragmentary traumatic memories.
      "After first encountering inescapable empirical evidence of how traumatic memories can differ from non-traumatic ones, and second, searching for constructs to describe and explain their observations, more recent students of traumatic memory then set out to conduct systematic research on the characteristics of traumatic memory. Early studies focused on the controversial phenomena of amnesia and delayed recall... Laboratory memory scientists like Kilhstrom (1995) and Loftus (1993) have vigorously attacked this line of research. However, others including Freyd (1991, 1994, 1996), Morton (1994) and Schooler (1994) have taken seriously the observations of clinicians and clinician-scientists' research on traumatic memory. These researchers have led the way in applying cognitive science constructs to the full complexity of traumatic memories, including phenomena like delayed recall and fragmentation" (pp. 13-14).

Abstract: "The study of traumatic memories is still an emerging field, both methodologically and theoretically. Previous questionnaire and interview methods for studying traumatic memories have been limited in their ability to evoke and assess remembrances with the characteristics long observed by clinicians. In this paper, we introduce a new standardized method that incorporates a laboratory procedure for retrieving memories of traumatic events and a clinically informed measure for assessing these memories' characteristics. We present three case studies to demonstrate the data yielded by script-driven remembering and the Traumatic Memory Inventory - Post-Script Version (TMI-PS). We then discuss subjects' script-driven remembrances in terms of methodology, theoretical classification of traumatic memories, and the interplay between the two. Finally, we critique our method in detail and offer suggestions for future research. If validated as a method for evoking and assessing traumatic memories, and shown to yield reliable data, this integrative method shows great promise for advancing both clinical and cognitive research on traumatic memories."

Excerpt: "In recent years, much of the research on traumatic memories has focused on recovered memories, true or false, and much of the theory on speculations about encoding and storage processes responsible for amnesia and delayed recall. This state of affairs has largely been a function of social and cultural factors. Scientifically speaking, however, the cart may have been put before the horse. That is, even though research on episodic traumatic memories is dependent on subjects' reports of memories they have just retrieved, research has shed little light on the processes and contents of memory retrieval in traumatized individuals. In this paper, we present a new method for evoking traumatic memories and assessing some of their basic characteristics. Our method brings together a laboratory procedure for standardized retrieval of memories, and a semi-structured interview for assessing memory characteristics based on well-established observations by clinicians dealing with traumatized patients. We offer this easily adapted approach to promote controlled research on the characteristics of traumatic memories, particularly prospective studies of their transformations over time.
     "Endel Tulving's (1972) classic chapter on episodic and semantic memory begins, "One of the unmistakable signs of an immature science is the looseness of definition and use of its major concepts" (p.381). This certainly appears to be the case today for the scientific study of traumatic memories. Use of the unitary construct of "traumatic memory" is common, though clinical experience and recent empirical and theoretical work suggest that memories for traumatic experiences are complex and heterogeneous phenomena, which change over time in a variety of ways. At this early stage, it might be more helpful to use the super-ordinate and plural construct of "traumatic memories" and methodically build a definitional taxonomy - just as traditional memory researchers have done since Tulving's incisive statement nearly 30 years ago.
      "A primary goal of this paper is to demonstrate that progress toward an empirically derived taxonomy of traumatic memories will be advanced by more attention to the following: (1) memory retrieval or evocation methods, and (2) instruments for assessing memory characteristics. We also aim to show that the former must draw more from laboratory research, and the latter from clinical experience and understanding... We believe such changes in shared theoretical and methodological frameworks can foster the understanding, communication and collaboration needed to advance the field (pp. 34-35).

Abstract: "Since trauma arises from an inescapable stressful event that overwhelms people's coping mechanisms, it is uncertain to what degree the results of laboratory studies of ordinary events are relevant to the understanding of traumatic memories. This paper reviews the literature on differences between recollections of stressful and of traumatic events. It then reviews the evidence implicating dissociation as the central pathogenic mechanism that gives rise to posttraumatic stress disorder (PTSD). A systematic exploratory study of 46 subjects with PTSD indicated that traumatic memories were retrieved, at least initially, in the form of dissociated mental imprints of sensory and affective elements of the traumatic experience: as visual, olfactory, affective, auditory, and kinesthetic experiences. Over time, subjects reported the gradual emergence of a personal narrative that can be properly referred to as 'explicit memory.' The implications of these findings for understanding the nature of traumatic memories are discussed."

Excerpt: "Trauma and dissociation.  Dissociation refers to the compartmentalization of experience: elements of the experience are not integrated into a unitary whole, but are stored in memory as isolated fragments consisting of sensory perceptions or affect states. . . However, the word dissociation is currently used to describe four distinct, but interrelated phenomena: (1) the sensory and emotional fragmentation of experience. . . (2) depersonalization [feeling that you are not real] and derealization [feeling the world is unreal] at the moment of the trauma. . . (3) ongoing depersonalization or 'spacing out' in everyday life. . . (4) containing traumatic memories within distinct ego-states (Dissociative Disorder). . . . The precise interrelationships among these various phenomena remain to be spelled out: not all people who have vivid sensory intrusions of traumatic events also experience depersonalization, while only a small proportion of people who have both of these experiences will go on to chronically dissociate, or to develop a full-blown dissociative disorder" (pp.510-511).
     Christianson (1982) has described how, when people feel threatened, they experience a significant narrowing of consciousness, and remain merely focussed on the central perceptual details. As people are being traumatized, this narrowing of consciousness sometimes evolves into amnesia for parts of the event, or for the entire experience. Students of traumatized individuals have repeatedly noted that during conditions of high arousal 'explicit memory' may fail. The individual is left in a state of 'speechless terror' in which he or she lacks words to describe what has happened. . . However, while traumatized individuals may be unable to give a coherent narrative of the incident, there may be no interference with implicit memory: they may 'know' the emotional valence of a stimulus and be aware of associated perceptions, without being able to articulate the reasons for feeling or behaving in a particular way" (p.511).

Excerpts from the study's results:

• "Of the 36 subjects with childhood trauma, 15 (42%) had suffered significant or total amnesia for their trauma at some time in their lives" (p.516).

• "Twenty-seven of the 36 subjects with childhood trauma reported confirmation of their childhood trauma from a mother, sibling, or other source who knew about the abuse, from court or hospital records, or from confessions or convictions of the perpetrator(s)" (pp.516-517).

• "Subjects considered most questions about the [comparison] nontraumatic memory nonsensical: none had olfactory, visual, auditory, kinesthetic reliving experience related to such events as high school graduations, birthdays, weddings, or births of their children. They also denied having vivid dreams or flashbacks about these events. Subjects claimed not to have periods in their lives when they had amnesias for any of these events, and none of the subjects felt the need to make special efforts to suppress memories of these events" (p.517).

• "No subject reported having a narrative for the traumatic event as their initial mode of awareness (they claimed not having been able to tell a story about what had happened), regardless of whether they had continuous awareness of what had happened, or whether there had been a period of amnesia. . . [A]ll subjects, regardless of age at which the first trauma occurred, reported that they initially 'remembered' the trauma in the form of somatosensory or emotional flashback experiences. At the peak of their intrusive recollections all sensory modalities were enhanced, and a narrative memory started to emerge" (p.517).

• "[The score of subjects on the Dissociative Experiences Scale] was significantly correlated with the event-related variables of duration of the trauma. . . , presence of physical abuse. . ., and presence of neglect. . . Also, level of dissociation was correlated with affective reliving. . ., kinesthetic reliving. . . lack of current narrative memory. . . and with self-destructive self-soothing behaviors. . . Dissociation was not correlated with the self-soothing behaviors of talking things over, working, cleaning, sleeping or turning to religion" (p.517-518).

van der Kolk, B. A. (1994). The body keeps the score: Memory and the evolving psychobiology of posttraumatic stress. Harvard Review of Psychiatry, 1, 253-265.

Excerpt: "[Posttraumatic Stress Disorder], by definition, is accompanied by memory disturbances, consisting of both hypermnesias [inabilities to forget] and amnesias. . . Research into the nature of traumatic memories. . . indicates that trauma interferes with declarative memory, i.e. conscious recall of experience, but does not inhibit implicit, or non-declarative memory, the memory system that controls conditioned emotional responses, skills and habits, and sensorimotor sensations related to experience. There now is enough information available about the biology of memory storage and retrieval to start building coherent hypotheses regarding the underlying psychobiological processes involved in these memory disturbances. . ."

van der Kolk, B. A., & van der Hart, O. (1989). Pierre Janet and the breakdown of adaption in psychological trauma. American Journal of Psychiatry, 146, 1530-1540.

Abstract: "In the reappraisal of the work of Pierre Janet at the centenary of the publication of L'automatisme psychologique, the authors review his investigations into the mental processes that transform traumatic experience into psychopathology. Janet was the first to systematically study dissociation as the crucial psychological process with which the organism reacts to overwhelming experiences and show that traumatic memories may be expressed as sensory perceptions, affect states, and behavioral reenactments. Janet provided a broad framework that unifies into a larger perspective the various approaches to psychological functioning which have developed along independent lines in this century. Today his integrated approach may help clarify the interrelationships among such diverse topics as memory processes, state-dependent learning, dissociative reactions, and posttraumatic psychopathology."

Excerpt: "[Janet wrote that when] people become too upset to tell their story, these [traumatic] memories cannot be transformed into a neutral narrative: 'the person is unable to make the recital which we call narrative memory, and yet he remains in the difficult situation'. . . This results in a 'phobia of memory' . . . that prevents the integration ('synthesis') of traumatic events and splits off the traumatic memories from ordinary consciousness. . . The memory traces of the trauma linger as subconscious fixed ideas that cannot be 'liquidated' as long as they have not been transformed into a personal narrative and instead continue to intrude as terrifying perceptions, obsessional preoccupations, and somatic experiences, such as anxiety reactions" (p.1533).

Judith Lewis Herman

These four papers by Harvard psychiatrist Judith Herman and her colleagues Mary Harvey and Emily Schatzow address fundamental issues in the memory controversy. Judith Herman is the author of Trauma and Recovery, which is widely viewed as the best book yet written on psychological trauma and recovery; it includes an excellent chapter entitled "Remembrance and Mourning."
     The first paper below was published in 1987, before the False Memory Syndrome Foundation had come into existence. Herman and Schatzow present evidence that patients may recover verifiable  memories of childhood sexual abuse in the course of group treatment. In the second paper, Harvey and Herman show how the knowledge of clinicians should inform scientific research on traumatic memory. In the third paper, Herman articulates the "dialectic of psychological trauma" which characterizes both individuals' struggles with their traumatic memories and social controversies like that over recovered memories. She illuminates the ways that victims, perpetrators and bystanders respond to crimes ranging from organized political violence to the private crimes of sexual and domestic violence – and the moral obligations of mental health professionals who find themselves embroiled in such situations. I have provided an extended excerpt in which Herman addresses the utility of the construct of dissociation for understanding paradoxical qualities of traumatic memories. In the fourth paper and most recent paper, Herman and Harvey review 77 intake evaluations conducted in their own outpatient program, and report that trauma-specific reminders and recent life crises, not previous psychotherapy, are the typical precipitants of delayed recall.

Herman, J. L., & Schatzow, E. (1987). Recovery and verification of memories of childhood sexual trauma. Psychoanalytic Psychology, 4, 1-14.

Abstract: "Fifty-three women outpatients participated in short-term therapy groups for incest survivors. This treatment modality proved to be a powerful stimulus for recovery of previously repressed traumatic memories. A relationship was observed between the age of onset, duration, and degree of violence of the abuse and the extent to which the memory of the abuse had been repressed. Three out of four patients were able to validate their memories by obtaining corroborating evidence from other sources. The therapeutic function of recovering and validating traumatic memories is explored."

Excerpts: "The majority of the patients (64%) did not have full recall of the sexual abuse but reported at least some degree of amnesia. . . Just over one quarter of the women (28%) reported severe memory deficits [i.e., recalled very little from childhood, reported recent eruption of previously inaccessible memories, or had such recall during the course of group treatment]" (p.4).
     "The majority of patients (74%) were able to obtain confirmation of the abuse from another source. Twenty-one women (40%) obtained corroborating evidence from the perpetrator himself, from other family members, or from physical evidence such as diaries or photographs. Another 18 women (34%) discovered that another child, usually a sibling, had been abused by the same perpetrator. An additional 5 women (9%) reported statements from other family members indicating a strong likelihood that they had also been abused, but did not confirm their suspicions by direct questioning. The three following case examples illustrate corroboration of the incest histories by, respectively, admission of the perpetrator, testimony of other family members, and physical evidence.

"Andrea (Case Example 1) wrote a letter to her stepfather confronting him about the sexual abuse and demanding an apology. Her stepfather responded by phone. He acknowledged 'fooling around' with her but refused to apologize, stating that she knew she 'wanted it as much as he did.' He did not believe the abuse had been harmful because vaginal intercourse had not occurred, and added resentfully that he had respected her virginity, only to have her 'throw it away on a bum.' He concluded the conversation by exhorting her to stop blaming the family for the troubles she had brought upon herself.

"Bernadette (Case Example 2) disclosed the sexual abuse to her mother, who burst into tears and cried, 'Oh no! Not you too!' She then told Bernadette that after she left home, her younger sisters had complained that their father tried to molest them.

"After a heroic military career, Claudia's brother (Case Example 3) was killed in combat in Vietnam. Her parents continued to make pilgrimages to his grave, and had transformed their home into a shrine dedicated to his memory. His room with all of his belongings, had been left untouched. During a visit to her parents home, Claudia conducted a search of her brother's room. In a closet she found an extensive pornography collection, handcuffs, and a diary in which he planned and recorded his sexual 'experiments' with his sister in minute detail" (p.10).

Harvey, M. R., & Herman, J. L. (1994). Amnesia, partial amnesia, and delayed recall among adult survivors of childhood trauma. Consciousness and Cognition, 4, 295-306.

Abstract: "Clinical experience suggests that adult survivors of childhood trauma arrive at their memories in a number of ways, with varying degrees of associated distress and uncertainty and, in some cases, after memory lapses of varying duration and extent. Among those patients who enter psychotherapy as a result of early abuse, three general patterns of traumatic recall are identified: (1) relatively continuous recall of childhood abuse experiences coupled with changing interpretations (delayed understandings) of these experiences, (2) partial amnesia for abuse events, accompanied by a mixture of delayed recall and delayed understanding, and (3) delayed recall following a period of profound and pervasive amnesia. These patterns are represented by three composite clinical vignettes. Variations among them suggest that the phenomena underlying traumatic recall are continuous and not dichotomous. Future research into the nature of traumatic memory should be informed by clinical observation."

Excerpt: "One aim of this paper is simply to describe the variations in traumatic recall that are frequently witnessed in clinical settings by ethical, observant, and reliable psychotherapists. Another is to counter an increasingly adversarial relationship between memory researchers who are relatively less familiar with clinical realities than they might be and clinicians who feel placed on the defensive by sweeping accusations of professional malfeasance" (p.297).

Herman, J. L. (1995). Crime and memory. Bulletin of the American Academy of Psychiatry and the Law, 23, 5-17.

Abstract: "The conflict between knowing and not knowing, speech and silence, remembering and forgetting, is the central dialectic of psychological trauma. This conflict is manifest in the individual disturbances of memory, the amnesias and hypermnesias [inabilities to forget], of traumatized people. It is manifest also on a social level, in persisting debates over the historical reality of atrocities that have been documented beyond any reasonable doubt. Social controversy becomes particularly acute at moments in history when perpetrators face the prospect of being publicly exposed or held legally accountable for crimes long hidden or condoned. This situation obtains in many countries emerging from dictatorship, with respect to political crimes such as murder and torture. It obtains in this country with respect to the private crimes of sexual and domestic violence. This article examines a current public controversy, regarding the credibility of adult recall of childhood abuse, as a classic example of the dialectic of trauma."

Excerpt: "On the one hand, traumatized people remember too much; on the other hand, they remember too little . . . . The memories intrude when they are not wanted, in the form of nightmares, flashbacks, and behavioral reenactments. Yet the memories may not be accessible when they are wanted. Major parts of the story may be missing, and sometimes an entire event or series of events may be lost. We have by now a very large body of data indicating that trauma simultaneously enhances and impairs memory. How can we account for this? If traumatic events are (in the words of Robert J. Lifton) 'indelibly imprinted,' then how can they also be inaccessible to ordinary memory?"
     When scientific observations present a paradox, one way of resolving the contradiction is to ignore selectively some of the data. Hence we find some authorities even today asserting that traumatic amnesia cannot possibly exist because, after all, traumatic events are strongly remembered. Fortunately for the enterprise of science, empirical observations do not go away simply because simplistic theories fail to explain them. On the contrary, I believe that some of the most important discoveries arise from attempts to understand apparent paradoxes of this kind. I would like to offer two theoretical constructs that may help us clarify and organize our thinking in this area. The first is the concept of state-dependent learning; the second is the distinction between storage and retrieval of memory. . . .
     When people are in a state of terror, attention is narrowed and perceptions are altered. Peripheral detail, context, and time sense fall away, while attention is strongly focused on central detail in the immediate present. When the focus of attention is extremely narrow, people may experience profound perceptual distortions, including insensitivity to pain, depersonalization, time slowing and amnesia. This is that state we call dissociation. . . .
     Traumatic events have great power to elicit dissociative reactions. Some people dissociate spontaneously in response to terror. Others may learn to induce this state voluntarily, especially if they are exposed to traumatic events over and over. Political prisoners instruct one another in simple self-hypnosis techniques in order to withstand torture. In my clinical work with incest survivors, again and again I have heard how as children they taught themselves to enter a trance state.
     These profound alterations of consciousness at the time of the trauma may explain some of the abnormal features of the memories that are laid down. It may well be that because of the narrow focusing of attention, highly specific somatic and sensory information may be deeply engraved in memory, whereas contextual information, time-sequencing, and verbal narrative may be poorly registered. In other words, people may fail to establish the associative  linkages that are part of ordinary memory.
     If this were so, we would expect to find abnormalities not only in storage of traumatic memories, but also in retrieval. On the one hand, we would expect that the normal process of strategic search, that is, scanning autobiographical memory to create a coherent sequential narrative, might be relatively ineffective as a means of gaining access to traumatic memory. On the other hand, we would expect that certain trauma-specific sensory cues, or biologic alterations that produce a state of hyperarousal, might be effec