Can we salvage the concept of psychological trauma?

Richard J. McNally has some timely advice for those revising the diagnostic manuals

Traumatology is a vibrant field, yet the diagnosis of post-traumatic stress disorder (PTSD) remains controversial 30 years after its ratification in the third edition of the Diagnostic and Statistical Manual of Mental Disorders. As the American Psychiatric Association revises the diagnostic bible, this article considers the meaning of trauma and where the manuals have lost their way. With the concept of trauma having broadened, what are the implications for theory and practice? Where can we find potential solutions to the problem?

Debates in traumatology possess an emotional intensity seldom seen in other mental health specialties (e.g. Dobbs, 2009; McNally, 2007). This is perhaps because, unusually among the anxiety disorders, PTSD implies categories of victim and perpetrator that often entangle moral and scientific discourse. When someone develops PTSD, there is usually someone to blame. When someone develops panic disorder, there is no one to blame. This entangling of the scientific, the political and the moral ensures that the diagnosis of post-traumatic stress disorder remains controversial (e.g. Brewin, 2003; Rosen, 2004) decades on from its origins.  

This article concerns a primarily scientific issue: the debate about the meaning of trauma itself (and I address other PTSD diagnostic issues elsewhere: McNally, 2004; McNally, 2009a). The American Psychiatric Association (APA) plans to publish DSM-V in 2013, and the committee responsible for revisiting the diagnostic criteria for PTSD is currently grappling with this problem. The World Health Organization is likewise revisiting its criteria for PTSD in preparation for the next edition of the International Classification of Diseases (ICD-11; Friedman & Karam, 2009). What, then, are the shortcomings of current definitions?

Conceptual bracket creep
Most syndromes in DSM-III (APA, 1980) had diagnostic criteria that were agnostic about aetiology, but PTSD was one of the exceptions. It required exposure to a traumatic stressor as one of its defining criteria. The authors of the PTSD diagnosis assumed that a circumscribed set of extraordinary stressors uniquely possessed the capacity to cause the symptomatic profile of the syndrome. These stressors were not merely the routine difficulties of everyday life. Rather, they were members of a class of events that fell outside the perimeter of usual human experience, causing distress in nearly anyone. The text provided examples of canonical traumatic stressors, including combat, rape, natural disasters, and torture. It also disqualified stressors falling within the ambit of ordinary life, such as simple bereavement, marital discord, or developing a chronic illness.  

Yet surely establishing the stressors that do or do not possess the capability of causing PTSD is an empirical matter, not a conceptual one. In fact, many studies have reported that people can develop PTSD-like symptoms after exposure to stressors that fall short of the DSM definition of trauma (e.g. Mol et al., 2005). However, a person who meets all criteria for PTSD, but whose stressor does not qualify as traumatic, cannot receive the diagnosis. Concerns in the USA about denying these sufferers the diagnosis, and hence  reimbursable treatment for PTSD, motivated a dramatic expansion of the concept of trauma in later DSM editions.

The concept of trauma embodied in DSM-IV-TR (APA, 2000) has two parts, and only if people meet both parts do they count as trauma survivors. One, Criterion A2, requires that the person must have experienced extreme fear, horror, or helplessness as the event was occurring. The other, Criterion A1, certifies three groups of people as potentially eligible for the diagnosis of PTSD: direct recipients of serious threat or harm, such as rape victims or combat veterans; those experiencing vicarious exposure to the trauma experienced by others, such as bystanders at a drive-by shooting; and those experiencing ‘informational exposure’. This third group is new in DSM-IV. It consists of people who are ‘confronted with’ information about threats to others, such as horrified television viewers of the 9/11 terrorist attacks (Marshall et al., 2007). Although the authors of the original PTSD diagnosis surely would not have envisioned these viewers as trauma victims (McNally & Breslau, 2008), our current DSM concept of trauma certifies them as trauma survivors just as much as it does those who escaped the World Trade Center before its collapse. In fact, one survey indicated that about 4 per cent of American adults living far from the scenes of the terrorist attacks developed apparent PTSD (Schlenger et al., 2002), presumably while watching coverage of the attacks from the comfort of their living rooms. These people developed what Young (2007) wryly calls ‘posttraumatic stress disorder of the virtual kind’ (p.21).

Because of this bracket creep in the definition of trauma (McNally, 2003a), most people today qualify as trauma survivors. For example, in one epidemiological survey, 89.6 per cent of adults in southeastern Michigan were trauma survivors by DSM-IV criteria (Breslau & Kessler, 2001). Clinicians have reported PTSD symptoms in dental patients following extraction of a wisdom tooth (de Jongh et al., 2008), and have diagnosed PTSD in people whose stressors include giving birth to a healthy baby after a routine delivery (Olde et al., 2006), and exposure to rude sexual jokes in the workplace (McDonald, 2003). As Shephard (2004), the distinguished British historian of military trauma, concluded, ‘Any unit of classification that simultaneously encompasses the experience of surviving Auschwitz and that of being told rude jokes at work must, by any reasonable lay standard, be a nonsense, a patent absurdity’ (p.57).  

Some psychologists, for example Weathers and Keane (2007), have argued that the post-DSM-III expansion of the concept of traumatic stressor simply amounts to rendering explicit what was implicit in the original definition of trauma. However, the psychiatrists instrumental in ratifying PTSD in DSM-III disagree. For example, the self-described ‘midwife’ of the diagnosis argued that, ‘this broadening should be reconsidered. Giving the same diagnosis to death camp survivors and someone who has been in a motor vehicle accident diminishes the magnitude of the stressor and the significance of PTSD’ (Andreasen, 2004, p.1322). The chair of the DSM-III, Robert Spitzer, has likewise expressed his objection to bracket creep in the definition of trauma (Spitzer et al., 2007).

Conceptual bracket creep in the definition of trauma does not occur in the ICD-10 PTSD diagnosis. The ICD-10 conceptualises a traumatic event or situation as one that is ‘exceptionally threatening or catastrophic’ and one that ‘is likely to cause pervasive distress in almost anyone’. Importantly, ICD-10 requires a person to be present at the scene of the trauma to qualify as a trauma survivor.  

Why bracket creep is a problem
Conceptual bracket creep poses problems for our field. First, expanding the concept of trauma makes it difficult to elucidate the psychobiological mechanisms producing the symptoms of PTSD. To be sure, people exposed to minor traffic accidents, wisdom tooth extraction, bad jokes at work, and so forth may report PTSD-like symptoms. Yet it is questionable whether they are experiencing the same thing or have the same psychobiology as someone who has been raped, tortured, or experienced combat. Findings in the PTSD field often fail to replicate even across groups of people who have experienced diverse yet undeniably traumatic events. Expanding the range of trauma to include PTSD caused by television, for example, will only make matters worse.

Second, the more we broaden the concept of trauma, the less convincingly we can award causal significance to the stressor itself, and the more we must emphasise vulnerability factors in the aetiology of PTSD. To put this issue in perspective, we must distinguish between risk for PTSD in general, and risk for PTSD among those exposed to trauma (i.e. vulnerability factors). The severity of the trauma itself is often the most important predictor of PTSD in general, as the dose–response effect implies (March, 1993). Indeed, exposure to trauma is more than a mere risk factor; it is logical requirement for the syndrome to emerge. Without a stressor to re-experience, one cannot suffer from re-experiencing symptoms, for example. This is why we cannot dispense with Criterion A1 entirely. Although exceptions abound (McNally, 2003b), the more severe the trauma, the more likely someone will develop PTSD.

The study of vulnerability factors was practically taboo in the early days of our field because it allegedly involves ‘blaming the victim’ (Blank, 1985). Now that the PTSD diagnosis is firmly ensconced in the DSM, traumatologists are increasingly comfortable investigating why people vary in their response to the same trauma (Yehuda, 1999). Indeed, the study of risk factors is essential for understanding any disorder, including PTSD. Nevertheless, diagnosing PTSD in people exposed to relatively minor stressors results in risk factors moving into the causal foreground and the stressor receding into the causal background. If risk factors overwhelmingly account for the emergence of PTSD in response to minor stressors, this seems to undermine the rationale for having a diagnosis of PTSD in the first place.

However, to hypothesise that risk factors must bear the causal burden for explaining PTSD-like symptoms following minor stressors does not mean that risk factors are irrelevant in response to major stressors. In fact, Helzer (1981) found that Vietnam veterans with a ‘predisposition’ to psychiatric breakdown, indexed by family history of psychiatric illness, antisocial behaviour, pre-military substance abuse, and so forth, were more likely to develop post-military depressive symptoms regardless of their level of combat exposure than were those lacking risk factors. Rather than predisposition becoming less important at the highest levels of combat stress, it tended to be even more important for predicting depressive symptoms. In a prospective longitudinal study of a civilian cohort, Koenen et al. (2008) found that adults exposed to trauma almost never develop PTSD unless they had received another psychiatric diagnosis earlier in life.

Third, if vulnerability factors are chiefly responsible for suffering following minor stressors, then imaginal exposure therapy targeting the memory of the event may fail. Pre-existing neuroticism, lack of social support, and so forth, may be the reasons why a person reports persistent PTSD-like symptoms following exposure to a minor stressor. Accordingly, attempts to render the memory of the event less emotionally disturbing may be seriously misguided and result in treatment failure.

Fourth, by viewing more and more of life’s stressors through the lens of trauma, we may overmedicalise normal emotional responses to stressors and undermine resilience in the face of adversity (Shephard, 2004). If almost anything can qualify as a trauma, then the concept morphs into a metaphor for misfortune in contemporary life and loses whatever distinctive meaning it originally had. Moreover, inculcating expectations that emotional distress in response to horror invariably signifies PTSD, a chronic disease that can be managed, but never cured, may foster undue pessimism and needless functional impairment.

The future of the concept
At least two changes seem necessary to salvage the concept of trauma in DSM-V. Whether they will be sufficient is an entirely different matter.

At the very least, a person should be physically present at the scene of trauma to qualify as a trauma survivor. As in ICD-10, the person should be either a direct recipient of trauma or a personal witness to the trauma of another. Indirect exposure (e.g.via the media) should not certify someone as a trauma survivor. Anyone who does develop PTSD-like symptomsvia such indirect exposure should receive
a diagnosis of either anxiety disorder NOS (not otherwise specified) or a new V code diagnosis for acute nonpathological reactions to a stressor (McNally, 2009a; Spitzer et al., 2007).

We should abolish Criterion A2, which stipulates that a person must experience extreme fear, horror, or helplessness to count as exposed to a traumatic event. Criterion A2 poses three problems. First, it confounds the person’s response with the inciting event, the host with the pathogen. Second, the initial, peritraumatic response reflects either the expression of a personal vulnerability diathesis or the initial stages of the syndrome of PTSD itself (Breslau, in press). It is not, however, part of the traumatic event that incites the response. Indeed,
we cannot study the association between stressors and the responses they elicit if we incorporate peritraumatic reactions into the very definition of the stressor itself. Third, studying initial emotional and cognitive responses to trauma is a legitimate inquiry, yet confining the range of possible reactions to extreme fear, horror, and helplessness is pointlessly restrictive. Indeed, emotions of shame, betrayal, guilt, and rage may also result from trauma (e.g. Rubin et al., 2008). However, the definition of trauma should not incorporate peritraumatic emotional reactions to the event. 

These proposed changes may alleviate some of the empirical and conceptual problems plaguing the field of PTSD, but various unresolved complexities remain.

Physical versus psychic trauma
As Hacking (1995) observed, psychic trauma is a metaphor originating in surgical medicine. Just as a psychic trauma damages the body, psychic trauma damages the mind. However, the distinction between stressors that exert their impact via the physics of force striking the body and stressors that exert their impact via their meaning is fuzzy in practice. Many Criterion A1 stressors are physically traumatic as well as psychologically traumatic (e.g. rape, torture), producing pain as well as fear.

Other traumatic stressors are purely psychological, exerting their pathogenic influence via the meaning they have for the victim. Consider a person who encounters a threatening person with a handgun. The weapon incites fear in anyone who understands the meaning of a handgun and the lethal threat it presents. Does this imply that whatever a person interprets as a threat counts as a trauma
if it incites fear? Not necessarily. The handgun threat poses an objective lethal threat, and its psychologically traumatising qualities arise from this fact. The occurrence of purely psychological stressors does not require us to adopt an ‘anything goes’ attitude that certifies any event as a trauma as long as the person regards the event as stressor.

Memory and the dose–response effec As several authors have observed (e.g. Breslau et al., 2002; McNally, 2009b; Rubin et al., 2008; Young, 1995), memory for the traumatic stressor indispensably lies at the heart of PTSD. And such intrusive thoughts, avoidance of reminders, nightmares and flashbacks are about the trauma, not merely caused by the trauma (McNally, 2009a). Other PTSD symptoms do not refer to the trauma even though they result from it. Therefore, it is the memory of trauma that theoretically unifies an otherwise diverse list of symptoms including irritability, emotional numbing, loss of interest, foreshortened future, sleep disturbance, and exaggerated startle (Breslau et al., 2002; Young, 1995). The memory of the referent event, linked to Criterion A1, prevents the syndrome from dissolving.

This interpretation of PTSD, however, does not sit well with the concept of a dose–response effect in PTSD (March, 1993), when the frequency, duration and severity of trauma predicts either the probability of someone developing PTSD or the number and severity of PTSD symptoms. For example, in World War II doctors found that the average time to psychiatric breakdown was about three months of continual combat (Jones & Wessely, 2005). This implies that cumulative exposure to multiple traumatic events eventually produces a trauma-caused syndrome. Yet it is unclear how the clinician (or the patient) can tie each symptom to a memory of a specific referent trauma in these cases. In principle, someone might develop exaggerated startle following one trauma, emotional numbing after another, and nightmares after still another. In contrast to the single trauma case, where there is a one-to-many mapping between event and symptoms,we would have a many-to-many mapping between several events and symptoms.

Childhood sexual abuse
DSM-IV-TR emphasises events ‘that involved actual or threatened death or serious injury, or a threat to the physical integrity of self or others’ (APA, 2000, p.467) as qualifying traumatic stressors. Yet the text states that ‘For children, sexually traumatic events may include developmentally inappropriate sexual experiences without threatened or actual violence or injury’ (p.464). Reviewing the history of how experts have conceptualised sexual abuse, Davis (2005) remarked: ‘The PTSD framework as a general model for sexual abuse was by no means obvious’ (p.116). Indeed, the moral reprehensibility of sexual molestation notwithstanding, it is unclear how it fits the trauma paradigm if violence or threat of violence is absent. Many people, including myself (e.g. McNally, 2003b), have used the term ‘survivor’ of childhood sexual abuse. Calling someone a survivor implies that the person was in danger of losing his or her life. Yet this is seldom the case in most instances of sexual abuse.

Agent and victim
An interesting historical shift in our understanding of trauma concerns the near disappearance of the ‘self-traumatized perpetrator’ (Young, 2002). In the 1970s trauma theorists, such as Lifton (1973), wrote about Vietnam veterans haunted by memories of atrocities they inflicted upon noncombatants. Conventional moral categories would classify these veterans as perpetrators, not victims of trauma. The solution was to shift the moral blame to the government that sent these men to a war characterised by atrocity-producing situations. This enabled therapists to treat these veterans as victims without approving their actions during the war (Fassin & Rechtman, 2009).

The idea that one can be traumatised by one’s own actions underscores both the cognitive and moral complexity of trauma (McNally, 2003b). A certain level of cognitive development must be present before an organism can develop self-representations suitable for the self-evaluations that provide the foundation for the moral emotions of guilt and shame. Although animal fear conditioning research provided the original foundation for behavioural models of PTSD (Keane et al., 1985), rats and dogs lack the cognitive capability to experience guilt and shame. Human beings can be self-traumatised perpetrators, whereas rodents cannot. Nevertheless, it seems odd to call an agent of trauma a survivor of trauma. In any event, our current concept of trauma does not address the self-traumatised perpetrator.

Can Wittgenstein illuminate the concept of trauma?
Finally, attempts to provide necessary and sufficient criteria for defining the concept of trauma presuppose that we can formulate criteria for a unitary concept of trauma. Yet as Wittgenstein (1953) argued, most concepts resist definition in this way, instead having overlapping attributes with no single defining feature. At best, we may clarify correlated attributes often shared by traumatic stressors (e.g. life-threat, physical presence at the scene of trauma). Ultimately, facts may compel us to abandon the unitary concept of trauma.

Richard McNally
is Professor of Psychology at Harvard University
[email protected]


American Psychiatric Association (1980). Diagnostic and statistical manual of mental disorders (3rd edn). Washington, DC: Author.
American Psychiatric Association. (2000). Diagnostic and statistical manual of mental disorders – text revision (4th edn). Washington, DC: Author.
Andreasen, N.C. (2004). Acute and delayed posttraumatic stress disorders. American Journal of Psychiatry, 161, 1321–1323.
Blank, A.S., Jr. (1985). Irrational reactions to post-traumatic stress disorder and Viet Nam veterans. In S. Sonnenberg, A.S. Blank, Jr. & J.A. Talbott (Eds.) The trauma of war (pp.69–98). Washington, DC: American Psychiatric Press.
Breslau, N. (in press). Causes of posttraumatic stress disorder. In P.E. Shrout (Ed.) Causality and psychopathology. Washington, DC: American Psychiatric Publishing.
Breslau, N., Chase, G.A. & Anthony, J.C. (2002). The uniqueness of the DSM definition of post-traumatic stress disorder: Implications for research. Psychological Medicine, 32, 573–576.
Breslau, N. & Kessler, R.C. (2001). The stressor criterion in DSM-IV posttraumatic stress disorder. Biological Psychiatry, 50, 699–704.
Brewin, C.R. (2003). Post-traumatic stress disorder: Malady or myth? New Haven, CT: Yale University Press.
Davis, J.E. (2005). Accounts of innocence. Chicago, IL: University of Chicago Press.
de Jongh, A., Olff, M., van Hoolwerff, H. et al. (2008). Anxiety and post-traumatic stress symptoms following wisdom tooth removal. Behaviour Research and Therapy, 46, 1305–1310.
Dobbs, D. (2009). The post-traumatic stress trap. Scientific American, 300(4), 64–69.
Fassin, D. & Rechtman, R. (2009). The empire of trauma (R. Gomme, Trans.). Princeton, NJ: Princeton University Press. (Original work published 2007)
Friedman, M.J. & Karam, E.G. (2009). Posttraumatic stress disorder. In G. Andrews et al. (Eds.) Stress-induced and fear circuitry disorders (pp.3–29). Arlington, VA: American Psychiatric Association.
Hacking, I. (1995). Rewriting the soul. Princeton, NJ: Princeton University Press.
Helzer, J. (1981). Methodological issues in the interpretation of the consequences of extreme situations. In B.S. Dohrenwend & B.P. Dohrenwend (Eds.) Stressful life events and their contexts (pp.108–129). New York: Prodist.
Jones, E. & Wessely, S. (2005). Shell shock to PTSD: Military psychiatry from 1900 to the Gulf War. Hove: Psychology Press.
Keane, T.M., Zimering, R.T. & Caddell, J.T. (1985). A behavioral formulation of PTSD in Vietnam veterans. Behavior Therapist, 8, 9–12.
Koenen, K.C., Moffitt, T.E., Caspi, A. et al. (2008). The developmental mental-disorder histories of adults with posttraumatic stress disorder. Journal of Abnormal Psychology, 117, 460–466.
Lifton, R.J. (1973). Home from the war. New York: Touchstone.
March, J.S. (1993). What constitutes a stressor? In J.R.T. Davidson & E.B. Foa (Eds.) Posttraumatic stress disorder (pp.37–54). Washington, DC: American Psychiatric Press.
Marshall, R.D., Bryant, R.A., Amsel, L. et al. (2007). The psychology of ongoing threat. American Psychologist, 62, 304–316.
McDonald, J.J., Jr. (2003). Posttraumatic stress dishonesty. Employee Relations Law Journal, 28, 93–111.
McNally, R.J. (2003a). Progress and controversy in the study of posttraumatic stress disorder. Annual Review of Psychology, 54, 229–252.
McNally, R.J. (2003b). Remembering trauma. Cambridge, MA: Belknap Press/Harvard University Press.
McNally, R.J. (2004). Conceptual problems with the DSM-IV criteria for posttraumatic stress disorder. In G.M. Rosen (Ed.) Posttraumatic stress disorder (pp.1–14). Chichester: Wiley.
McNally, R.J. (2007). Revisiting Dohrenwend et al.’s revisit of the National Vietnam Veterans Readjustment Study. Journal of Traumatic Stress, 20, 481–486.
McNally, R.J. (2009a). Can we fix PTSD in DSM-V? Depression and Anxiety, 26, 597–600.
McNally, R.J. (2009b). Posttraumatic stress disorder. In P.H. Blaney & T. Millon (Eds.) Oxford textbook of psychopathology (2nd edn) (pp.176–197). Oxford: Oxford University Press.
McNally, R.J. & Breslau, N. (2008). Does virtual trauma cause posttraumatic stress disorder? American Psychologist, 63, 282–283.
Mol, S. S. L., Arntz, A., Metsemakers, J. et al. (2005). Symptoms of post-traumatic stress disorder after non-traumatic events. British Journal of Psychiatry, 186, 494–499.
Olde, E., van der Hart, O., Kleber, R. & van Son, M. (2006). Posttraumatic stress disorder following childbirth. Clinical Psychology Review, 26, 1–16.
Rosen, G.M. (Ed.) (2004). Posttraumatic stress disorder. Chichester: Wiley.
Rubin, D.C., Berntsen, D. & Bohni, M.K. (2008). A memory-based model of posttraumatic stress disorder. Psychological Review, 115, 985–1011.
Schlenger, W.E., Caddell, J.M., Ebert, L. et al. (2002). Psychological reactions to terrorist attacks. Journal of the American Medical Association, 288, 581–588.
Shephard, B. (2004). Risk factors and PTSD. In G.M. Rosen (Ed.) Posttraumatic stress disorder (pp.39–61). Chichester: Wiley.
Spitzer, R.L., First, M.B. & Wakefield, J.C. (2007). Saving PTSD from itself in DSM-V. Journal of Anxiety Disorders, 21, 233–241.
Weathers, F.W. & Keane, T.M. (2007). The Criterion A problem revisited. Journal of Traumatic Stress, 20, 107–121.
Wittgenstein, L. (1953). Philosophical investigations (G.E.M. Anscombe, Trans.). New York: Macmillan.
Yehuda, R. (1999). (Ed.). Risk factors for posttraumatic stress disorder. Washington, DC: American Psychiatric Press.
Young, A. (1995). The harmony of illusions. Princeton, NJ: Princeton University Press.
Young, A. (2002). The self-traumatized perpetrator as a ‘transient mental illness’. L’Évolution psychiatrique, 67, 630–650.
Young, A. (2007). Posttraumatic stress disorder of the virtual kind. In A. Sarat, N. Davidovitch & M. Alberstein (Eds.) Trauma and memory (pp.21–48). Stanford, CA: Stanford University Press.

BPS Members can discuss this article

Already a member? Or Create an account

Not a member? Find out about becoming a member or subscriber