DEEP READ · OFF-LIST

The Body Keeps the Score

Bessel van der Kolk · 2014

中文 →

In one sentence

The real trouble with trauma is not that you cannot forget what happened; it is that your body does not know it is over — heart rate, breathing, muscle tone and vigilance are still set to the moment of the event, still reading the present as the scene of the crime. Which moves the whole task of recovery off the thing everyone assumes it is, telling the story straight, and onto something clumsier and more basic: teaching this body that right now is safe.

Where it sits

Bessel van der Kolk is a Dutch-born American psychiatrist who began seeing Vietnam veterans at the Boston VA hospital in 1978 and has worked on essentially one subject ever since. He was part of the push that put PTSD into American psychiatric diagnosis in 1980, founded a trauma treatment centre, and ran clinical studies on EMDR, yoga and neurofeedback. This 2014 book is the ledger of that career, and a publishing freak: it climbed the bestseller lists years after release, peaking during the pandemic, and has spent close to three hundred weeks on the New York Times list — a fact that says something in itself. Its position: it redefined trauma from a disorder of memory into a condition of the body and nervous system, and moved the centre of gravity in treatment from talking to the body. And precisely because its reach has been so wide, it is also one of the most contested psychology bestsellers of the last decade — the criticisms section below is not optional reading.

The central claims

• Trauma is not a memory of the past; it is a physiological response still running in the present. "It's over" is a fact about the calendar. As far as the body is concerned, the alarm was never switched off.
• What trauma damages first is self-awareness and the sense of the body, and only secondly the narrative memory. Survivors frequently cannot say what they feel or where they are in their own bodies — which is why insight alone, coming top-down, cannot reach the lesion.
• Recovery therefore has to work bottom-up: breath, movement, rhythm, touch, voice, being in sync with other people. Once the body registers safety, language and meaning can catch up.

The core ideas, one at a time

What "keeping the score" actually means

Get the title straight or the whole book will be misread. "The body keeps the score" does not mean a recording is filed away in your tissues. It means an overwhelming experience resets the defaults of the nervous system — resting heart rate, breathing pattern, muscle tone, sensitivity to sounds and faces, the baseline of stress hormones — to a wartime setting, and leaves them there. So a person sitting on their own sofa can be running a defensive programme from thirty years ago: a slammed door and the whole body locks; a certain smell and the stomach clenches; a voice from behind and the pulse goes to 120.

Van der Kolk's summary is blunt: the essence of trauma is the present being occupied by the past. An ordinary painful memory carries a timestamp — you know it is a recollection. A traumatic fragment carries none; it arrives as image, smell, palpitation, rigidity, whole and uninvited, into now, and the body treats it not as playback but as live footage. This is also why the sentence that breaks families — "but you're safe, it's finished" — does nothing. You are addressing the reasoning part, and the reasoning part is not the one in charge.

The smoke detector and the watchtower: why "understanding it" isn't enough

This is the anatomical core of the book, and he gets it across with two metaphors. The amygdala is the brain's smoke detector: it processes danger signals extremely fast and extremely crudely, and its job is to sound the alarm, not to check whether the alarm is warranted. The medial prefrontal cortex is the watchtower: high up, good view, able to judge that it is only burnt toast and to stand the alarm down. In ordinary conditions the two cooperate. After trauma, the detector becomes hair-trigger while the watchtower goes offline precisely when the alarm sounds — so the moment a person is triggered, the capacity to appraise the situation vanishes and only fight, flight or freeze remain.

From which follows the book's most practically useful distinction. The top-down route regulates emotion through cognition — understanding, disputing thoughts, re-narrating. The bottom-up route changes the state of the body directly — breath, movement, rhythm, posture, voice. His claim is not that talking is useless, but that when the watchtower goes dark the instant the alarm rings, you cannot issue all your orders from the watchtower.

Speechless terror: not unwilling to say it, unable

The book's most-cited imaging finding comes from work he and colleagues did in the 1990s: survivors listened to a recorded account of their own trauma while being scanned, and during the induced flashback activity in Broca's area — the region that produces speech — dropped, while right-hemisphere and limbic regions handling emotion and imagery lit up. He named the state speechless terror.

This explains a scene that is common in clinics and routinely misread as resistance: a person talks fluently about work, about the weather, and then touches the thing and dissolves into broken syntax, lost focus, or simply goes blank. They are not declining to speak; in that moment the machinery that makes speech has genuinely dropped out. Which leads to a basic commitment of the book: the first task is not to get someone to tell the story, but to bring their arousal down to a range in which speech is possible. Reverse the order and "recounting the trauma" can become the second injury. (Honesty demands the caveat: that study had a very small sample and limited replication, and treating it as settled science is overuse — see the criticisms.)

Interoception and alexithymia: getting back on terms with your body

Interoception is the capacity to sense your internal state — heartbeat, breath, gut, the tightness of a shoulder, the unnameable sense that something is off. Trauma survivors tend to fail at one of two extremes: either drowned by body signals (a racing heart means catastrophe is coming) or shut off from them entirely, numb and blank about their own physical existence. Alongside this runs alexithymia — literally, no words for feelings: able to report "I'm bad," unable to say whether it is fear, anger or shame.

Why does this matter so much? Because you cannot regulate what you cannot feel. If you do not know your shoulders are clenched you cannot release them; if fear and rage are indistinguishable you have one blunt response available for both — drink, binge, explode, or shut down. So every "odd" therapy in the book (yoga, tai chi, choir, theatre, drumming) points at the same target: bring interoception back online, so a person can feel, then name, and only then choose. It is also the line most usable by an ordinary reader: healing is not thinking your pain through; it is repossessing your body.

The self-sensing system: the blank patch on the scan

One observation in the book is genuinely chilling. A set of midline brain regions — insula, medial prefrontal cortex, posterior cingulate among them — handles the sense of "who I am right now and what I am feeling," and is most active when a person is at rest doing nothing in particular. It is the physiological floor of selfhood. In scans of severely traumatised patients, this territory is markedly quiet — close to blank.

Translated: these people are not merely in a bad mood. The basic sensation that there is somebody here has faded. They describe the world as seen through frosted glass, fail to recognise themselves in a mirror, and, when asked how they feel right now, genuinely cannot answer — not will not. It is also why so many survivors chase intensity: extreme sport, dangerous relationships, pain, because only a hard enough input makes the blank patch flicker, and lets them feel briefly that they are alive. (Same caveat: these imaging studies are typically small, so take the direction and not the strength of the claim.)

Following this thread, van der Kolk arrives at what I think is the best definition in the book: the opposite of trauma is not happiness but agency — a person believing again that their own actions can change their situation. Helplessness is the core experience of trauma, so anything that restores "I did something and the world shifted slightly" counts as treatment: rowing, drumming, staging a play, catching the beat in a choir. All of it beats being told to cheer up.

Attachment, developmental trauma and the ACE study: childhood as dose, not background

The middle of the book turns to childhood, and this is where it goes further than the usual PTSD literature. Van der Kolk insists on a clinical fact: long-term harm inflicted by the people closest to you and a single external catastrophe are not psychologically the same thing. After a car crash, a parent's arms are the antidote; when the harm comes from the person who is supposed to provide safety, the child is caught in a loop with no exit — disorganized attachment, in which the only shelter is also the storm. The child takes the blame ("I'm bad"), because being at fault at least preserves a shred of control.

The theory underneath is attachment research: John Bowlby argued that an infant's bond to a caregiver is a survival need on a par with food, and Mary Ainsworth's "strange situation" made it observable — separate a one-year-old briefly from the mother, then reunite them. Securely attached children cry, seek, and settle quickly once held. Disorganized children do contradictory things: move towards the mother and freeze halfway, or circle in place, face blank. That image of freezing halfway is more or less the whole book: the impulse to seek help and the impulse to flee fire at once, and the system jams. Children living with chronic unpredictable danger also develop what clinicians call frozen watchfulness: unusually quiet, undemanding, and scanning adult faces without pause. Adults tend to praise such children for being good. It is in fact an expensive survival strategy.

The strongest evidence here is the ACE study (adverse childhood experiences): a survey of tens of thousands of adults in the 1990s found that the number of categories of childhood abuse, neglect and household dysfunction showed a clear dose–response relationship with adult depression, addiction and attempted suicide — and with heart and lung disease as well. Risk rises in steps as the count rises. The shock of it: the bill for childhood adversity is finally paid by an adult body. On this basis van der Kolk campaigned for a new diagnosis, developmental trauma disorder; it was not accepted into DSM-5 in 2011, though a related concept, complex PTSD, later entered the World Health Organization's ICD-11. (Two things to hold onto: ACE data rest on adult recall, and correlation is not causation.)

Reenactment, dissociation, and "why did she go back?"

What changes most in how you read other people are the book's accounts of two phenomena. The first is dissociation: when neither fight nor flight is possible, the body deploys its last option — shut down. The person goes numb, floating, watching themselves in a film. At the time this saves them; the cost is that it becomes the default, so that later stress takes them offline automatically and life arrives in fragments that will not line up.

The second is reenactment: survivors are drawn again and again into structurally similar situations — the same kind of partner, the same conflicts, the same humiliating position. This is not masochism and needs no mystical account: familiar pain is easier for the nervous system to register as predictable than unfamiliar safety, and to a system that has lived without control, predictability is a powerful attractant. The same logic reads self-harm and addiction as crude but effective regulation: the pain of a cut pulls a person out of numbness, alcohol muffles the alarm. Grasp this and your answer to "why doesn't he just leave?" changes permanently: it is usually not a failure of will but a survival programme that genuinely worked once and is still running.

The map of treatments — and what each one's evidence is worth

The final third is a catalogue: EMDR, trauma-sensitive yoga, mindfulness, neurofeedback, Internal Family Systems, Somatic Experiencing, theatre and choir, psychodrama-style structure work. Their shared logic is clear — all of them route around pure language and act directly on arousal, rhythm and bodily sensation.

The thread he prizes most is rhythm and synchrony. Infant–caregiver interaction is essentially a sustained exchange of beats: look, smile, sound, pause, answer. Trauma survivors often lose the ability to fall into step with other people, showing up either as withdrawal or as constant vigilance inside relationships. So choirs, drumming, dance and group martial arts are not decoration in this book: they let a person re-experience being in the same rhythm as others without having to discuss anything painful — the oldest form of safety there is. Combine that with agency — making a move in a group and being answered — and you often get someone out of isolation earlier than any amount of reasoning would.

But the evidence behind these methods varies enormously, and the book is not even-handed about that, so it is worth setting out separately:

The argument in outline

The book is really one line, walked seven times with different cases:

• The phenomenon: survivors' suffering looks less like remembering too much and more like a body that never stood down — hypervigilance, numbness, insomnia, irritability, unexplained physical symptoms.
• The mechanism: the smoke detector is hair-trigger, the watchtower goes offline at the worst moment, speech production is suppressed during flashbacks — so reason cannot reach the scene.
• The extension: chronic childhood trauma damages attachment, selfhood and body sense rather than just memory, and the ACE data show its effects running dose-by-dose into adult physical health.
• The consequences: dissociation, reenactment, addiction and self-harm are outputs of one dysregulated system, not moral failures.
• The core experience: what runs through all of it is helplessness — so the opposite of trauma is not happiness but agency, the restored belief that your actions move the world.
• The prescription: restore safety and regulation at the level of the body first (bottom-up), then do meaning and narrative (top-down). People are repaired in relationship and in rhythm.

What has it actually established? Most solidly this: trauma produces a whole-body dysregulation that crosses the mind–body line, so any approach that treats only thoughts is bound to fall short. On the strength of what? Three things — forty years of clinical observation, evidence from imaging and stress physiology, and large-sample epidemiology of the ACE kind. The further step, "therefore use these specific therapies," moves faster than the evidence, and that is the part a reader has to police personally.

Misreadings, criticisms, live disputes

• Misreading 1: "the body remembers, so memories are stored in muscle." There is no good evidence that tissue stores episodic memories. The defensible claim is that the set points of the nervous and endocrine systems have been changed and the body keeps running on the old settings. Taking the metaphor for anatomy is where this book gets abused most.
• Misreading 2: "talk therapy is useless and CBT is obsolete." The book does lean that way, and the evidence does not support it: trauma-focused CBT, prolonged exposure and cognitive processing therapy remain the best-supported first-line treatments. Read it as "more than talking," not "instead of talking."
• Misreading 3: "you must recover and retell every detail to heal." The book itself opposes forced retelling; and the wider assumption that repressed traumatic memories can be retrieved intact is fiercely contested in memory science. Elizabeth Loftus's experiments show memory can be suggested into existence, and Richard McNally's Remembering Trauma argues systematically that trauma is usually remembered all too well rather than sealed away. Be extremely wary of any therapy promising to dig up buried memories.
• Criticism 1: polyvagal theory is shaky. The book borrows Stephen Porges's framework to explain the social-engagement / fight-flight / shutdown sequence, and it tells beautifully; but its central claims in comparative neuroanatomy have been sharply attacked (Paul Grossman and others). Best treated as a serviceable metaphor, not a verified mechanism.
• Criticism 2: the imaging evidence is oversold. The Broca's-area finding rests on a single-digit sample and was exploratory work; the book presents it as the settled account of traumatic speechlessness. The move from one small study to one large conclusion happens more than once here.
• Criticism 3: the enthusiasm doesn't track the evidence. Neurofeedback, IFS and Somatic Experiencing get generous space and warmth on thin controlled-trial support, while the best-supported treatments are dispatched quickly. The author is himself a researcher in and advocate of several of these methods, which readers are entitled to know.
• Criticism 4: concept creep. This book, and the cultural wave it launched, have stretched "trauma" to cover nearly all distress. But one stable epidemiological finding is that most people exposed to severe traumatic events do not develop PTSD; recovery is the norm. Calling all pain trauma dilutes the situation of the genuinely severe cases, and can reinforce the story "I am permanently damaged" — the exact opposite of recovery.
• In fairness: almost every criticism above is about strength of evidence and calibration of language, not about the question the book is asking. The direction — that trauma has to be addressed at the level of the body — has broadly held up over the past decade. The book simply runs half a length ahead of the data.

Ten sentences