van der Kolk (1994)
The paper that named the idea. Bessel van der Kolk’s “The Body Keeps the Score: Memory and the Evolving Psychobiology of Posttraumatic Stress” (Harvard Review of Psychiatry, 1994) is the origin of the phrase that titles his 2014 trade book and this wiki’s Week 10 The Body Keeps the Score folder. It is the second trauma source ingested, after Payne, Levine & Crane-Godreau (2015) — and it is worth reading as the classic statement that the SE paper’s most striking claim (trauma is “in principle fully reversible”) is arguing against. See bessel-van-der-kolk.
Where Payne et al. bring a proprietary therapy and no data, van der Kolk brings no therapy of his own to sell here and a large (if 1994-vintage) evidence base — but the two share a single core commitment: the trauma is in the body and the nervous system, not in the narrative. They then diverge, sharply, on whether the resulting state can be undone. That divergence is the wiki’s new are-traumatic-memories-indelible debate.
The thesis: the body keeps the score
The whole paper is an argument for one sentence in its abstract: a trauma is “stored in somatic memory and expressed as changes in the biological stress response.” The lineage van der Kolk claims for it runs back a century — Janet (1889), Freud, Pavlov, Kardiner’s “physioneurosis,” Grinker & Spiegel’s catalogue of soldiers’ motor symptoms — all noticing that overwhelming experience leaves a physiological residue that behaves like a memory but is not available as one.
The mechanism he proposes: under sufficiently intense arousal, declarative (conscious, narrative) memory fails, and the experience is organized instead on a “somatosensory level (as visual images or physical sensations) that is relatively impervious to change.” Janet’s “visceral sensations (anxiety and panic) or visual images (nightmares and flashbacks)”; Piaget’s “failure of semantic memory leads to the organization of memory on a somatosensory or iconic level.” The traumatized person does not recall the event so much as re-live it in the body — “speechless terror,” in which “the emotional impact of the event may interfere with the capacity to capture the experience in words or symbols.” See traumatic-memory.
Two memory systems, and why trauma splits them (Figures 1-2, Table 2)
The paper’s spine is the declarative/nondeclarative distinction (Figure 1): DECLARATIVE (explicit) memory for facts and events, versus NONDECLARATIVE (implicit) memory for skills/habits, emotional associations, and conditioned sensorimotor responses. Van der Kolk’s claim: “trauma interferes with declarative memory… but does not inhibit implicit, or nondeclarative, memory.” The conditioned emotional response, the somatic sensation, the flashback — all implicit, all spared, all re-enacted rather than remembered.
The neural model (Figure 2, Table 2) is the 1990s dual-system account:
- The amygdala assigns “free-floating feelings of significance” to sensory input fast, and drives the body (hypothalamus, ANS, hpa-axis).
- The hippocampus does the slower job of categorizing experience in space and time — declarative memory, the “cognitive map.”
- Moderate amygdala activation enhances hippocampal consolidation (hypermnesia for stressful events); excessive activation inhibits hippocampal functioning, so “memories are then stored in sensorimotor modalities: somatic sensations and visual images.”
- Norepinephrine drives consolidation on an inverted-U — both too little and too much impair storage — which van der Kolk offers as the mechanism producing PTSD’s coexisting hypermnesias and amnesias.
This is the wiki’s first source to give the amygdala a memory-consolidation role, distinct from the fear/salience/somatic-marker readings already on that page.
The bimodal response, and the interoceptive hook
Van der Kolk insists the trauma response is bimodal: hyperarousal, hyperreactivity, and re-experiencing on one pole; numbing, avoidance, amnesia, and anhedonia on the other. The most interoceptively relevant claim in the paper sits at this hinge: traumatized people suffer a “loss of the capacity to use affect states as signals.” “In subjects with PTSD, feelings are not used as cues to attend to incoming information,” so they “go immediately from stimulus to response without psychologically assessing the meaning of an event.”
That is a deficit in the functional use of interoceptive/affective signals, and it connects the paper to the wiki’s core: it is the clinical face of the misinference the predictive-coding sources describe. “The misinterpretation of innocuous stimuli, such as unexpected noises, as potential threats” is interoceptive inference gone wrong — a prior of danger overriding benign input — described in 1994 without the vocabulary. Compare schema-guided-symptom-perception, cognitive-model-of-panic, and Barrett’s body-budget account of chronic misprediction.
Where the physiology is PTSD-specific: startle, cortisol, opioids
Three findings the paper treats as signatures, each now with its own page:
- Nonhabituation of acoustic startle (Table 1A). Shalev found 93% of PTSD subjects failed to habituate to loud tones vs 22% of controls; remitted patients still fail, raising the marker-vs-vulnerability question. See acoustic-startle-response.
- The paradoxical HPA profile. Unlike acute stress (and unlike major depression), chronic PTSD shows low urinary cortisol with enhanced negative feedback — glucocorticoid-receptor upregulation and hyperresponsiveness to low-dose dexamethasone (Yehuda). Prior trauma blunts the cortisol response to new trauma. See hpa-axis, which this ingest extends from a depression-typical high-cortisol picture to the PTSD-specific low-cortisol one.
- Stress-induced analgesia. Re-exposing PTSD subjects to a trauma-resembling stimulus produced naloxone-reversible analgesia equivalent to 8 mg of morphine, two decades after the trauma — evidence that numbing is opioid-mediated. See stress-induced-analgesia.
The claim that makes the debate: “emotional memories are forever”
The paper’s hardest theoretical commitment, and the one that puts it in direct tension with the SE paper already in the wiki: van der Kolk adopts LeDoux, Romanski & Xagoraris (1991) — cortical lesions prevent extinction of conditioned fear, so “once formed, the subcortical traces of the conditioned fear response are indelible, and… ‘emotional memory may be forever.‘” With Kolb (1987): PTSD is impaired cortical control over subcortical structures. Symptoms emerge when that inhibition weakens (drugs, alcohol, sleep, aging, reminders).
Two things to hold precisely:
- The claim is indelibility of the trace plus modifiability of expression: Figure 2’s caption says the emotional memories “are thought to be relatively indelible, but their expression can be modified by feedback from the prefrontal cortex.” So van der Kolk is not a pure fatalist — treatment restores cortical inhibition — but he denies the trace itself can be erased.
- This is exactly what Payne et al. (2015) deny: for SE, trauma is a reversible dynamical attractor state (core-response-network), and completion (biological-completion) resolves it rather than inhibiting it. The two positions carve the same clinical territory — persistent, arousal-triggered, body-held trauma — and disagree about whether the persistence is permanent-but-inhibitable or dynamical-and-reversible. New debate: are-traumatic-memories-indelible.
Note the irony the wiki should keep visible: SE leans on LeDoux’s animal work (escape-completion abolishing fear; threat+restraint as the pathogen) to argue for reversibility, while van der Kolk leans on LeDoux’s animal work (cortical lesions blocking extinction) to argue for indelibility. Same source tradition, opposite therapeutic morals. See joseph-ledoux, survival-circuits.
Treatment
The pharmacology section is dated (the case for SSRIs in PTSD, van der Kolk’s own fluoxetine trial, the observation that startle habituation worsened on fluoxetine even as symptoms improved). What survives is the framing: because trauma is held sub-cortically and re-lived through the body, the therapeutic goal is to help patients “live in the present” by relocating the trauma in time and restoring cortical modulation — medications are adjuncts that create enough safety to do that work. This is the same target interoceptive-exposure and SE both aim at from different directions; van der Kolk’s version is inhibitory (strengthen the brake), SE’s is transformative (shift the state).
How the wiki files it
The foundational clinical-neurobiological statement of trauma as a bodily phenomenon — the classic against which the wiki’s other trauma material is positioned. Stronger evidence base than the SE paper and no commercial stake, but a 1994 synthesis leaning heavily on unpublished data and animal analogy, and committed to an indelibility thesis that a later wiki source directly contests. Its lasting contribution to this wiki is threefold: the declarative/nondeclarative memory split (traumatic-memory), the amygdala’s memory-consolidation role (amygdala), and the “loss of affect states as signals” — an interoceptive deficit at the centre of PTSD that ties the trauma literature to the wiki’s predictive-coding spine.