A brilliant provocation, an incomplete provocation
Karl Friston is probably one of the most important living neuroscientists. His work on predictive processing and the free energy principle has fundamentally changed how we understand perception, cognition, and the brain itself. That's why, when Friston co-authored an article provocatively titled The Body Does Not Keep the Score (Kotler, Mannino, Fox and Friston, 2026), people pay attention.
And, intellectually, the article is fascinating.
But it also reveals one of the recurring blind spots of highly abstract computational neuroscience: the temptation to explain lived human experience as if cognition alone were the main stage of reality.
The article argues that trauma is less a matter of "trauma stored in the body" and more a disorder of predictive processing: the brain becomes trapped in maladaptive predictions of threat, rigidity, and hypervigilance. In this framework, trauma is not literally filed away in the tissue. The body functions as a messenger of predictive error, rather than as an independent storage site.
On one level, this is correct.
On another level, it runs the profound risk of overlooking the phenomenological, relational, and embodied dimensions that made The Body Keeps the Score a clinically transformative work.
Because the real issue goes beyond whether trauma is "stored" in the fascia like a USB drive. That caricature was never the strongest version of the somatic theory of trauma.
The real question is this: can trauma reorganize the body's physiology so profoundly that the organism itself becomes configured around survival?
The answer is clearly yes.
And here Friston's framework becomes, at the same time, both enlightening and reductionist.
A controversy built on a false dichotomy
What makes the recent debate surrounding the article intellectually frustrating is not a lack of sophistication in the text. It's that the controversy itself rests on a false dichotomy that should never have existed.
The debate has been framed as if we had to choose between two mutually exclusive positions: either trauma is "stored in the body," or trauma is fundamentally predictive and computational.
That opposition crumbles the moment we move beyond slogans and examine how living systems actually work.
Because trauma operates as an organization of the organism, rather than as something stored somewhere.
And organizations don't exist in isolated locations. They exist through systems.
The problem is that contemporary discourse, especially digital discourse, continues to flatten complex theories into ideological camps. Computational neuroscience reacts against a simplified somatic mysticism. Somatic communities react against reductionist cognitivism. Then, each side caricatures the other until scientific and clinical reality disappears.
Friston's article reacts, in part, against a vulgarized version of the trauma discourse that became popular in the last decade: the idea that trauma is literally lodged within the tissues, waiting to be "released" through purely bodily interventions. Some of these interpretations have become biologically naive, therapeutically exaggerated, and scientifically weak.
But instead of correcting the excesses while preserving valid intuitions, the article risks swinging towards the opposite reduction: redescribing trauma primarily as maladaptive inference.
And this reveals a deeper problem of modern intellectual culture: our obsession with replacing one totalizing model with another.
Predictive processing explains the mechanism. It doesn't fully explain the experience.
The predictive model proposes that the brain constantly generates predictions about the world and updates itself by correcting errors. Trauma reduces this flexibility: the system overpredicts danger and interprets bodily signals through threat-oriented priors.
It's elegant neuroscience.
But elegance falls short of completeness.
A predictive model can explain how a nervous system maintains a traumatic organization. It doesn't fully capture what trauma feels like as lived bodily reality.
Because trauma is not experienced in the way that "incorrect probabilistic priors" would be experienced.
It is experienced as contraction, respiratory restriction, muscle shielding, collapse, hypervigilance, dissociation, altered perception of time, autonomic dysregulation, relational fear, loss of spontaneity.
In other words: the organism itself changes.
And this matters because Friston's framing risks reproducing an old Cartesian error in updated computational language: the subtle recentralization of the brain as the sovereign locus of human experience.
But the body operates as much more than output hardware for cortical prediction.
The organism is an integrated living system.
The body was never separate from the prediction
One of the strangest aspects of the controversy is that predictive processing itself already implies embodiment.
Prediction does not occur in an abstract, floating brain disconnected from physiology.
The brain predicts through the body.
Interoception, the perception of the body's internal states, occupies a central place in predictive processing frameworks. The nervous system constantly models heart rate, respiration, muscle tension, visceral states, inflammation, hormonal fluctuations, and autonomic activation.
In fact, according to researchers such as Anil Seth (2021) and Lisa Feldman Barrett (2017), the brain is fundamentally a regulatory engine for bodily survival.
Consciousness itself can emerge, to a large extent, from the predictive regulation of the organism.
This means that the body is situated within predictive processing, rather than outside of it. This is the entire reason why predictive processing exists.
That's why the phrase "the body doesn't keep score" becomes conceptually misleading: it implies a separation that modern embodied neuroscience has been dismantling for decades.
The organism predicts as an integrated system.
Trauma is reactivated, not stored
The most accurate formulation is probably this: trauma is dynamically re-actualized throughout the organism, rather than being statically stored somewhere.
This is profoundly different.
Trauma functions more as a persistent pattern than as a frozen object hidden in muscle tissue.
This pattern includes autonomic regulation, interoceptive expectation, emotional conditioning, procedural memory, defensive physiology, attentional bias, and relational anticipation.
The system reproduces the traumatic organization in real time, over and over again.
That's why people who have survived trauma often say, "I know I'm safe logically, but my body doesn't believe it.".
That phrase alone dismantles the false dichotomy.
Because what is happening is simultaneously predictive, physiological, embodied, emotional, and relational.
The nervous system learned survival at all levels of the organism.
The body is not a passive scorecard
One of the emerging counterarguments, associated with people influenced by Lisa Feldman Barrett, proposes this formula: "the brain keeps the score, the body is the scorecard.".
Again: partially true.
But biologically incomplete.
Because the body actively shapes brain states, beyond passively displaying them.
Modern neuroscience increasingly supports bidirectional embodiment. Interoception, vagal signaling, gut-brain interaction, hormonal states, immune activation, inflammatory cytokines, microbiome modulation, autonomic patterns—all of these profoundly shape cognition, mood, and perception.
Approximately 80% of the vagal fibers travel from the body to the brain, rather than in the opposite direction.
The organization operates as a recursive loop, rather than as a top-down command structure.
This means that trauma cannot be reduced to "bad predictions" generated solely by the brain.
Trauma becomes embodied because the body itself reorganizes around a physiology of survival.
Breathing changes. Posture changes. Inflammatory states change. Baseline autonomic tone changes. Muscle tension changes. Sleep changes. Digestive function changes. Hormonal rhythms change.
And eventually, the perception itself changes.
At a biological level, in the strictest sense. Beyond any metaphor.
The real question is not where trauma is "stored".
The whole question of "where trauma lives" is, in reality, conceptually immature.
Complex systems do not locate their organization in a simplistic way.
Where is a symphony stored? In the notes? In the orchestra? In the conductor? In the acoustic field?
The question itself becomes inappropriate.
Trauma functions more as a networked organizational state than as a stored entity.
It exists through neural pathways, autonomic tone, endocrine regulation, immune signaling, muscle patterns, attentional habits, relational expectations, and perception itself.
Trying to reduce this to "brain" or "body" reflects a fundamentally mechanistic misunderstanding of living systems.
The hidden Cartesianism of the debate
Ironically, both extremes remain trapped in Cartesian thinking.
The purely somatic caricature says: "the body stores trauma independently.".
The hypercomputational cartoon says: "the brain predicts trauma and the body obeys.".
Both secretly maintain the same division: brain versus body.
But modern systems neuroscience increasingly suggests that this division is artificial.
The organism is recursive.
The brain affects the body. The body affects the brain. The immune system affects mood. The microbiome affects cognition. Posture affects emotion. Attachment affects autonomic regulation. Inflammation alters perception. Breathing alters cognition.
There is no isolated sovereign center.
And that is precisely why trauma is so multidimensional.
Friston's model underestimates the embodiment of development
One of the deepest problems with purely predictive explanations is that they can flatten developmental trauma, turning it into informational abstraction.
A child raised in chronic unpredictability does much more than "learn inaccurate predictions".
Their entire organism calibrates itself around danger.
Attachment trauma reorganizes autonomic regulation, emotional tolerance, stress reactivity, interoceptive awareness, relational perception, sense of self, and the ability to feel safe.
Even before reflective cognition is fully operational.
This matters enormously.
Because trauma often operates as a physiological adaptation to environments that were genuinely unsafe, beyond any epistemic error.
The nervous system learned survival, without hallucinating threats from nowhere.
And many somatic clinicians sense something that Friston's framework risks minimizing: a body that continues to experience existence as dangerous does not become cognitively updated.
Why Friston's criticism still resonates
Despite these limitations, Friston's critique resonates for one important reason: parts of trauma culture have become conceptually neglected.
Terms such as "stored trauma", "trapped energy" or "somatic release" began to circulate without sufficient precision.
Some therapeutic spaces drifted towards biologically vague metaphors, disconnected from rigorous mechanisms.
And when scientific language becomes imprecise, the backlash is inevitable.
Friston attempts, in part, to restore conceptual rigor.
That impulse is legitimate.
But the danger lies in overcorrection.
Because when computational neuroscience becomes overly abstract, it risks losing touch with clinical reality and lived experience.
Clinical reality remains stubbornly embodied
Beyond theoretical language, trauma clinicians continue to observe the same phenomena: respiratory constriction, dissociation, muscle shielding, autonomic collapse, exaggerated startle responses, altered gaze behavior, impaired interoception, chronic inflammatory states, visceral dysregulation.
These signs are part of the traumatic organization itself, rather than being secondary decorations around cognition.
And healing systematically involves physiological processes: co-regulation, vagal security, movement, breathing, relational attunement, autonomic flexibility, embodied presence.
No single predictive framework completely resolves severe trauma if the organism continues to inhabit a chronic survival physiology.
Computational seduction
There is also a broader philosophical question.
Predictive processing has become so intellectually powerful that it risks becoming a totalizing framework, capable of describing everything while explaining less than it appears.
This is the danger of grand unified theories.
It all ends up being prediction, inference, error minimization, Bayesian updating.
But human suffering resists such elegant abstractions.
A dissociated person who has survived trauma does much more than fail to optimize predictive flexibility.
He is inhabiting a body that no longer feels habitable.
That distinction has real clinical consequences.
Because people do not heal solely through conceptual accuracy.
They heal through relational security, nervous system regulation, embodied integration, emotional processing, movement, attachment repair, and coherent experience.
Ironically, Friston's own article indirectly acknowledges this when discussing flow states and metastability.
But that creates a contradiction: if healing requires restoring embodied flexibility through physiological and experiential processes, then the body is clearly more than a passive messenger.
It is part of the traumatic system itself.
Van der Kolk never advocated mystical storage
One of the strange aspects of the debate is that the criticisms usually attack a simplified version of somatic discourse, an Instagram caricature, rather than the more solid clinical formulations.
Bessel van der Kolk (2014) never literally claimed that trauma lodges within muscle fibers like trapped ghosts.
His central intuition operated on a much more sophisticated level: trauma reorganizes embodiment.
That intuition remains overwhelmingly valid.
Individuals with PTSD exhibit altered autonomic regulation, exaggerated startle responses, changes in heart rate variability, disrupted interoception, altered cortisol rhythms, inflammatory dysregulation, chronic muscle contraction, and activation of procedural memory.
The body participates structurally in trauma, beyond the symbolic.
The deepest cultural issue: reductionism disguised as sophistication
What this controversy ultimately exposes is something broader about modern intellectual culture.
We continue to struggle deeply with complexity.
Academic culture rewards elegant explanatory comprehension: a single framework that explains everything.
But living systems resist simplification.
The predictive processing movement is incredibly powerful. But, like psychoanalysis in the early 20th century or behaviorism later on, it risks becoming imperialistic: expanding until every phenomenon ends up being redescribed in its vocabulary.
At that point, the explanation ceases to be genuine understanding and becomes linguistic absorption.
Yes, trauma involves prediction.
But saying "trauma is predictive rigidity" describes one layer of its organization, far from amounting to a complete explanation of trauma.
The deepest problem: modern neuroscience continues to fear subjectivity
At the deepest level, this debate reveals something broader about contemporary science.
There is a lingering discomfort with the experience.
Somatic language became popular because millions of people immediately recognized themselves in it, beyond any misunderstandings about neuroscience.
They understood: "I know this in my body.".
On an experiential level. Beyond any metaphor.
The danger of hypercomputational models is that they sometimes become too disconnected from phenomenology, too far removed from how human beings actually experience being alive.
Trauma operates beyond incorrect prediction.
It is altered existence.
Altered temporality.
Altered incarnation.
Altered relationality.
Altered ability to be present.
And no purely computational framework fully captures that.
The synthesis already exists
The irony is that the most advanced trauma science has been integrating these perspectives for years.
Researchers such as Stephen Porges (2011), Bessel van der Kolk (2014), Antonio Damasio (2010), Lisa Feldman Barrett (2017) and Jaak Panksepp (1998) have been working for years towards integrated body and mind models.
The future clearly does not lie in choosing sides: brain versus body, cognition versus physiology, prediction versus embodiment.
The future lies in organismic integration.
The body keeps track, because the whole organism does.
In the end, the title itself is probably the problem.
The Body Does Not Keep the Score It is rhetorically provocative but conceptually misleading.
Because the agency keeps absolute track.
It does so as a living, dynamic system continuously shaped by experience. Beyond static storage. Beyond mystical energy residue. Beyond isolated cortical inference.
Trauma survives when the organism remains organized around survival.
Healing occurs when the organism regains enough security, flexibility, and integration to stop predicting the world exclusively through the lens of threat.
And that process is, at the same time, neural, bodily, emotional, relational, predictive, physiological, existential.
Trying to isolate one dimension from the others says more about our intellectual habits than about the trauma itself.
The essential thing: integrating traumas
Beyond the theoretical debate, what matters clinically is quite concrete. The central issue is not discussing where the trauma is located. It's about integrating it.
Integrating trauma means allowing the experience, which was organized as a survival mechanism, to be processed, to settle, and to cease operating as an active background program. It means that what was split off, dissociated, encapsulated, or contained finds the conditions to become part of lived history, instead of continuing to be replayed in the present.
This integration never happens solely through cognition. It requires genuine security, regulation of the nervous system, embodied presence, relational attunement, time, and often, clinical support. It involves working with the body, with breath, with posture, with voice, with touch, with gaze. It involves working with history, with meaning, with the missing words. It involves working with connection, because much of what broke us was relational, and much of what heals is also relational.
Integration doesn't erase trauma. It puts it in its place. It restores the organism's flexibility to respond to the present from within the present, instead of from a past that continues to organize its physiology.
That's why the debate between "the body keeps track" and "the brain predicts" loses almost all its relevance when viewed from a clinical perspective. What truly matters is that trauma can be integrated. And that integration always involves the entire organism.
Friston is right about some things, and profoundly wrong about others.
Friston is right that trauma is deeply linked to predictive rigidity and a loss of adaptive flexibility. That insight is powerful and clinically valuable.
But where the framework risks becoming impoverished is in its implicit reduction of embodiment to predictive machinery.
The body functions as part of consciousness itself, beyond the mere theater where the brain would stage its simulations.
And until neuroscience fully reintegrates lived embodiment with computational models, it will continue to produce intellectually dazzling but existentially incomplete theories.
Sources and references
Kotler, S., Mannino, M., Fox, G. & Friston, K. (2026). The Body Does Not Keep the Score: Trauma, Predictive Coding, and the Restoration of Metastability. Frontiers in Systems Neuroscience, 20:1812957. Karl Friston: FRS, Professor of Neuroscience at University College London, creator of the free energy principle.
van der Kolk, B. (2014). The Body Keeps the Score: Brain, Mind, and Body in the Healing of Trauma. Viking, MD, psychiatrist, professor of psychiatry, Boston University School of Medicine.
Porges, SW (2011). The Polyvagal Theory: Neurophysiological Foundations of Emotions, Attachment, Communication, and Self-Regulation. WW Norton, PhD, neuroscientist, Indiana University.
Barrett, LF (2017). How Emotions Are Made: The Secret Life of the Brain. Houghton Mifflin Harcourt. PhD, Distinguished University Professor of Psychology, Northeastern University.
Seth, A. (2021). Being You: A New Science of Consciousness. Faber & Faber. PhD, Professor of Cognitive and Computational Neuroscience, University of Sussex.
Damasio, A. (2010). Self Comes to Mind: Constructing the Conscious Brain. Pantheon Books. MD, Professor of Neuroscience, University of Southern California.
Panksepp, J. (1998). Affective Neuroscience: The Foundations of Human and Animal Emotions. Oxford University Press. PhD, neuroscientist, Washington State University.
Varela, F.J., Thompson, E. & Rosch, E. (1991). The Embodied Mind: Cognitive Science and Human Experience. MIT Press. PhD in biology (Varela), PhD in philosophy (Thompson), PhD in psychology (Rosch).
Friston, K. (2010). The free-energy principle: a unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138.