In PTSD, devastating memories interact with vulnerable neural structures to produce an overwhelming emotional, episodic memory of an experience. The multiple decision-making system theory and its interacting support structures suggest a perspective on PTSD as interacting episodic and emotional memories.
After World War I, soldiers returning from the trenches of Europe were described as succumbing to “shell shock”: jumpy, cowering at the slightest noise, often unable to function in normal society. In World War II, the term was “battle fatigue.” The classic image is that of a father in 1950, sitting at the kitchen table, smoking his cigarette, staring out into his memory. This difficulty that some soldiers have is not a new phenomenon.1 In his description of soldiers coming back from the American Civil War, Jacob Mendes Da Costa described a phenomenon of sudden-onset heart palpitations, which he called “irritable heart” and came to be known colloquially as “soldier’s heart.”2 A 2006 historical study found that the best predictor of soldier’s heart in Civil War veterans was the percentage of the company killed, presumably a reliable proxy for experiencing devastating emotional trauma.3 The term used now, post-traumatic stress disorder, recognizes that this problem can arise from traumatic non-military experiences as well: victims of violent crime, sexual assault, terrorist attack, car crashes, and other disasters are also vulnerable to PTSD.4
At this point, it is not clear what the cause of PTSD is. The symptoms, however, are consistent and recognizable (as can be seen in the similarities of the descriptions of shell shock, battle fatigue, and soldier’s heart).5 These symptoms include uncontrollable re-experiences of the event. Based on our separation of episodic and semantic memories (Chapter 16), these re-experiences would be unquestionably episodic in nature. They are often vividly present and intrusive. Patients with PTSD also show a withdrawal from society and friends, as well as hyperarousal, increased vigilance, and increased startle responses, which we can take as evidence of Pavlovian action-selection (Chapter 8) and motivational (Chapter 13) learning. These startle responses can be extremely dangerous in well-trained soldiers, who have learned to associate violent action responses with those situations (Chapter 10).
Like addiction and problem gambling, PTSD seems to be an interaction between inherent vulnerability and external cause. For example, it is not true that everyone exposed to a difficult or traumatic situation gets PTSD.6 There is now evidence that there is a significant difference in hormonal levels and several brain structures between soldiers and other patients who end up with PTSD after an incident and those who don’t.7 Twins without PTSD (who share the same genetic code) of soldiers who got PTSD reacted more strongly to startling cues than twins of soldiers who didn’t get PTSD from similar experiences.8 Similarly, the hippocampus (a brain structure we’ve seen to be deeply involved in memory, see Chapter 16) is physically smaller in twins of soldiers with PTSD than twins of soldiers without.9 As with all of these vulnerabilities (like the genetic vulnerabilities we’ve discussed elsewhere), these are correlations. People with smaller hippocampi may be more likely to end up with PTSD than people with larger hippocampi; however, there are lots of other factors involved. Simply having a smaller hippocampus does not mean that you will get PTSD if you are ever exposed to a trauma.
Three theories have been proposed to explain PTSD: (1) that PTSD patients have stored a “flashbulb” memory, which cannot be forgotten,10 (2) that PTSD patients have difficulty extinguishing fear-related memories,11 particularly through altered fear processing,12 and (3) that PTSD patients have difficulty identifying the context of memories, allowing those memories to intrude on other contexts.13 A fourth possibility is that PTSD patients have difficulty consolidating memories from hippocampal-dependent episodic representations into hippocampal-independent semantic representations (Chapter 16). We will address each of these theories and then try to bring them back together into a coherent picture of decision-making and PTSD.
The first psychological theories of PTSD suggested that the traumatic event produced a flashbulb memory, a perfect photographic memory of an instance. The concept was that the stress and intensity of the memory storage overwhelmed the cortical storage systems, “burning it into the brain.”14 This theory, however, predicts that the memory will be recalled identically each time, which is not how these memories occur.
Although the original definition of flashbulb memories put forward by Roger Brown and James Kulik recognized that the flashbulb illuminated only part of the picture, the concept remained that the illuminated images remained steady. In fact, Jennifer Talarico and David Rubin found that, although they remained vivid, “flashbulb” memories (defined as memories that stayed for long periods of time) decreased in consistency over time, suggesting that they were being reconstructed each time.15 Talarico and Rubin suggest that the primary difference between flashbulb memories and more common, everyday memories, was the internal confidence in it, rather than in the properties of the recalled memory itself.
Interestingly, this suggests that although the traumatic memories are vivid and intrusive, they are constructed, much like ordinary memories, and also suggests a relationship between episodic memory and PTSD.
We have already seen (in Chapter 4) how animals extinguishing responses to learned associations change their behavior because of additional components entering into the equation, not because they have forgotten the learned association.16 The addition of a new process rather than the forgetting of the original association leads to spontaneous renewal of responses and cued recovery of those responses.
Central to this concept is the recognition that there are multiple memory (and action-selection) systems. In particular, there is a Pavlovian (emotional) action-selection system that works through direct connections from sensory thalamic systems to the amygdala.17 As we saw in Chapter 8, our Pavlovian responses produce visceral, emotional reactions to stimuli, sometimes even without our knowledge. But Pavlovian mechanisms also affect the motivational system (Chapter 13) as well, which influences the Deliberative and Procedural systems (see Chapter 6). This means that a Pavlovian, emotional response could label a memory particularly salient and particularly likely to be retrieved.
However, we should be wary of simply saying that PTSD patients have difficulty extinguishing their behaviors. There are several key differences between the sort of associative learning that occurs in typical extinction paradigms and PTSD. Extinction paradigms usually entail multiple exposures to the cueing stimulus to create the original association, and the reaction to the association is common among all animals exposed.18 Spontaneous reappearance of the association occurs after long time periods, not immediately after the event.19 Cued reappearance of the association occurs with a re-exposure to the specific cues that created the original association.20 In contrast, PTSD generally occurs in response to a single, traumatic event, the intruding memories occur quickly after the event and fade with time, and the intruding memories appear, even uncued, in situations with little or no relationship to the original association.21
Lynn Nadel and W. Jake Jacobs have suggested that PTSD patients have specific difficulties in separating contexts, which allows the memories to reappear (to intrude) in situations where they would normally not.22 Specifically, they propose that glucocorticoids in the hippocampus that arise during stress disable the hippocampus during memory storage, allowing the memory to become unlabeled spatiotemporally, thus allowing the memory to become, in a sense, unstuck in time. Stress affects the storage and retrieval of memories, primarily through the action of glucocorticoids (such as cortisol).23 In particular, Nadel, Jacobs, and their colleagues have found that stressful events prior to an experience facilitate the storage of emotional memories through a hippocampal–amygdala interaction, even while they impair hippocampal processing on hippocampally-dependent memory tasks.24 Although stress during retrieval impairs that retrieval (it’s harder to remember things under stress), stress immediately following training increases storage and later retrieval (it’s easier to remember things that happen to you when you are stressed).25
This interaction (between salient, value-marked memories and uncoupled hippocampal memory) has been suggested as a key factor in PTSD by several researchers.26 In these theories, PTSD occurs when stress levels enhance amygdala learning and suppress hippocampal learning.27 Nadel and Jacobs suggest that the effect of the massive stress event is to enhance learning in the amygdala, decrease learning in the hippocampus, and produce a highly value-laden but contextually-uncoupled memory.28 These theories, however, suggest that the PTSD patient will not be able to provide a spatiotemporal image of the event, particularly the patient’s location within the sequence of the event, which does not seem to fit with the descriptions of the vivid nature of the traumatic memory in PTSD.29
We saw in Chapter 16 how the normal progress of memory is a transition from episodic (you are there) to semantic (I remember that) memories. An intriguing possibility is that PTSD occurs not when the hippocampus is unable to create an episodic memory of the event, but rather when the hippocampal–cortical system is unable to consolidate the memory into semantic storage. As we discussed in Chapter 16, although early studies of consolidation suggested that memory transferred intact from hippocampally-dependent structures to hippocampally-independent structures over time, more recent studies have suggested that certain functions of memory remain hippocampally-dependent forever.30 Recent studies have suggested that the transferred memories have a fundamentally different character than recently stored memories—recently stored memories (dependent on the hippocampus) are stored in an episodic manner, while the consolidated ones are stored in a semantic manner extrahippocampally.
These theories are consistent with new views of the hippocampus as playing a key role in mental time travel, both of reconstructing the past and of imagining the future.31 If normal recovery from trauma entails a consolidation of a memory away from an episodic (you-are-there, return-to-the-past) memory into a semantic (narrative, detached) memory, then PTSD patients may be stuck in the episodic mode. They may be unable to accomplish this transformation and continue to reconstruct their past every time they access the memory, leaving the memory with an immediacy that intrudes on normal function.
This is, in fact, consistent with the original studies on flashbulb memories,32 which suggested that the normal “flashbulb” memory was stored, replayed and rehearsed, and decreased in its emotional impact with time. Importantly, Brown and Kulik were studying normal people remembering distant tragedies (e.g., Where were you when you heard the World Trade Center had been hit by terrorists?), not people with PTSD involved in the event itself (e.g., firefighters escaping from the falling buildings). Brown and Kulik hypothesized that eventual accounts of flashbulb memories were derived from continued covert and overt rehearsal of these stories, translating them into a narrative and perhaps even verbal form.
This hypothesis rests on the concept that the hippocampus in PTSD patients is weak or dysfunctional, not nonfunctional. Patients with complete hippocampal lesions do not show any episodic memory recall events at all, and when they recall past events or imagine future events, their descriptions are primarily semantic, not episodic.33 PTSD experiences are vivid and spatially and temporally located, with strong suggestions of mental time travel, all indications of episodic memories, and a dysfunctional (but not nonfunctional) hippocampus.A
Attempted treatment for PTSD has included pharmacological treatments, attempts to extinguish the memory by acquiring safety cues, and attempts to recontextualize the memory, effectively transferring it from the emotionally-tagged episodic to the more detached semantic systems.34 Although there is still a lot of work to do determining the vulnerabilities that engender PTSD, the mechanisms by which it creates dysfunction, and what might serve as potential treatments, it’s clear that the mental disorder known as PTSD has a physical manifestation in brain function.
What is clear is that PTSD entails a complex and dysfunctional interaction between memory systems. Somehow, the emotional and episodic memory systems have become entangled in a way that is difficult to disentangle. Because they are reconstructed each time, PTSD memories are not “flashbulb memories,” yet they are, in a very real sense, burned into the brain. PTSD includes problems with emotional memories, which affect Pavlovian motivational and action-selection systems, but PTSD also includes hippocampal interactions, particularly with episodic memory retrieval. PTSD patients seem to have a dysfunctional hippocampus, unable to transition memories from episodic to semantic representations with time.
• Jonathan Shay (1994) Achilles in Vietnam: Combat Trauma and the Undoing of Character. New York: Scribner.
• Frontline (2009). Soldier’s Heart. PBS.
• Joseph E. LeDoux (1996). The Emotional Brain. New York: Simon and Schuster.
• Chris Cantor (2005). Evolution and Posttraumatic Stress: Disorders of Vigilance and Defence. London: Routledge.