Trauma, by definition, is the result of exposure to an inescapably stressful event that overwhelms a person's coping mechanisms. Since it would be immoral to expose laboratory subjects to the sort of overwhelming stimuli that give rise to the dissociated sensory reexperiences characteristic of posttraumatic stress disorder (PTSD), we are uncertain to what degree the vast literature involving laboratory studies of less stressful events is relevant to understanding how people process traumatic memories.

Relying on clinical observations, it has been recognized for more than a century that "when people become too upset by their emotions, memories cannot be transformed into a neutral narrative" (Janet 1919). The ensuing terror "results in a phobia of the memory that prevents the integration (synthesis) of the traumatic events and splits the traumatic memories from ordinary consciousness," leaving them to be organized as visual perceptions, somatic preoccupations and behavioral reenactments (Janet 1894).

The dissociative nature of traumatic memories seems to be what distinguishes them from memories of everyday experience: dissociation at the moment of the trauma now has been established as the single most important predictor for developing PTSD (Marmar and others; Shalev and colleagues). The flashbacks and nightmares characteristic of PTSD can be seen as products of that dissociation. Dissociation means that emotional, sensory, cognitive and behavioral aspects of the traumatic experience are not integrated. For example, traumatized people may know what has happened to them, but they may have no feelings about it. Conversely, people may act disturbed without knowing what makes them behave that way. The British psychiatrist C.S. Meyers, who in 1915 coined the term shell shock, described soldiers' reactions to trauma during World War I as follows: "The normal personality [is] replaced by an 'emotional' personality. Gradually or suddenly an 'apparently normal' personality returns-normal, save for the lack of all memory of the [traumatic] events, normal, save for the manifestation of somatic, hysteric disorders indicative of mental dissociation."

He suggested that treatment of shell shock should "deprive the 'emotional personality' of its pathological, distracted, uncontrolled character, and effect...its union with the 'apparently normal' personality hitherto ignorant of the emotional experiences in question" (Meyers 1920). The extensive literature from World War I documents how many shell-shocked soldiers became amnestic for their traumas, which were later relived as nightmares, flashbacks and behavioral reenactments. E.E. Southard's Shellshock and Neuropsychiatry (1919) alone has 23 such detailed case reports.

The clinical accounts of trauma cases by Janet (1893), W.H.R. Rivers (1918), Kardiner (1941), Lenore Terr (1994) and myself (e.g., van der Kolk 1984, 1987) illustrate the ways in which traumatized individuals remember. During the past half-century psychiatrists and psychologists have largely abandoned detailed case descriptions in favor of enumerating the relationships between certain symptoms and any number of other variables. This method allows for findings that can be replicated by other scientists in studies of similar design, but it tends to lose some of the richness of the subjective human experience.

An Immediate Past

Ordinarily, memories of particular events are remembered as stories that change and deteriorate over time and that do not evoke intense emotions and sensations. In contrast, in PTSD the past is relived with an immediate sensory and emotional intensity that makes victims feel as if the event were occurring all over again. The longitudinal study of the psychological and physical health of 200 Harvard undergraduates who participated in World War II is a good illustration of how adults process traumatic events (Lee and cohorts). When these men were reinterviewed about their experiences 45 years later, those who did not have PTSD had considerably altered their original accounts: the most intense horror of the events had been bleached. In contrast, time had not modified the memories of those who had developed PTSD, who recalled their experiences with extreme vividness. Thus, paradoxically, the ability to transform memory is the norm, while the problem in PTSD is that the full brunt of an experience does not fade with time.

Accounts of the memories of traumatized patients consistently mention that emotional and perceptual elements tend to be as prominent as declarative components (van der Kolk and Fisler 1995). These patients see the event happening again-they hear it, smell it and have kinesthetic sensations. This dissociative nature of traumatic memory complicates the capacity to communicate about it. Research has shown that recall occurs in a state-dependent fashion; triggered by exposure to sensory or affective stimuli that match sensory or affective elements associated with the trauma. In some people the memories of trauma may have no verbal (explicit) component: the memory may be entirely organized on a perceptual level, without an accompanying narrative about what happened.