Karuna Labs
Research Handbook · v1.0
For clinicians, researchers & trainees

The Predictive Brain in Chronic Pain.

A research handbook on predictive coding & active inference — mechanisms of the chronic pain cycle and their clinical applications.

For clinicians, researchers, and graduate trainees in pain medicine, psychology, physiotherapy, and digital therapeutics. Conditions in scope: primary nociplastic pain, chronic low back pain, fibromyalgia, migraine, IBS, chronic pelvic pain, persistent post-injury pain, and related psychophysiologic disorders.

PRIOR EVIDENCE PERCEPT ACTION
Karuna Labs  ·  Research synthesis
30 references  ·  3 figures
Predictive coding & active inference
in nociplastic and psychophysiologic pain
Abstract

Pain is inferred, not transmitted.

Chronic pain has historically been modeled as a faithful readout of peripheral tissue state. That model fails to explain the large fraction of patients whose pain persists, spreads, or intensifies in the absence of commensurate nociceptive input.

Predictive processing — the proposal that the brain is a hierarchical inference machine that generates perception by predicting the causes of its sensory signals and updating those predictions against incoming evidence — offers a mechanistic alternative.1,3

Under this framework, pain is not transmitted but inferred: it is the brain's best explanation for noisy, ambiguous bodily signals, weighted by prior expectation and by the estimated reliability (precision) of prediction and evidence.4,5 Chronic pain is reframed as a condition of aberrant inference: a high-precision prior predicting bodily danger that comes to dominate sensory evidence, sustained by a self-reinforcing loop of attention, fear, and avoidance that is itself a form of active inference.6,16

This reframing is not merely theoretical. It predicts, and recent randomized trials confirm, that interventions targeting the prior — its precision and the beliefs that sustain it — can produce substantial, durable reductions in pain. The landmark trial of Pain Reprocessing Therapy reported that two-thirds of patients with chronic back pain were pain-free or nearly so after treatment, with effects maintained at one year and mediated by changes in pain-related belief.12,13

The handbook develops the theory (Part I), applies it to the chronic pain cycle (Part II), and surveys the clinical applications it motivates (Part III).

Scope & clinical boundary

This handbook is a research and educational synthesis. It does not constitute medical advice. The framework applies to primary/nociplastic and psychophysiologic pain after appropriate medical evaluation has excluded structural and systemic disease. Clinical decisions remain the responsibility of qualified practitioners.

Contents

A map of the argument.

Part ITheoretical Foundations
  1. 01The Brain as a Prediction Machine
  2. 02Pain as Inference, Not Transmission
  3. 03Active Inference and Precision
Part IIThe Chronic Pain Cycle
  1. 04From Acute Protection to Chronic Prediction
  2. 05Anatomy of the Self-Reinforcing Loop
  3. 06Neurobiological Substrates
Part IIIClinical Applications
  1. 07Principles of Brain-Based Treatment
  2. 08The Intervention Toolkit
  3. 09Evidence Summary
  4. 10A Clinical Workflow
  5. 11Open Questions & Research Agenda
Back matterReferences
  1. R30 references with PMIDs & DOIs
IPart One

Theoretical foundations.

How the brain constructs perception, why that matters for pain, and the core machinery of predictive coding and active inference.

01 Chapter One

The Brain as a Prediction Machine.

For most of the twentieth century, perception was modeled as a bottom-up, feed-forward process: sensory receptors transduce energy, signals ascend a processing hierarchy, and at the top a percept is assembled. Predictive processing inverts this picture. On the predictive account, the dominant flow of information is top-down. The brain continuously generates a model of the most likely causes of its sensory input and sends those predictions down the cortical hierarchy; what travels back up is not raw data but prediction error — the residual mismatch between what was predicted and what was sensed.2,3

This architecture is hierarchical and reciprocal (Figure 1). Each level predicts the activity of the level below and receives error signals from it. Higher levels encode slower, more abstract, more context-laden expectations (the body schema, the meaning of a sensation, the expectation of harm); lower levels encode fast-changing sensory and interoceptive detail. Perception is the brain's equilibrium solution — the set of predictions that, for the moment, best explains away the incoming signal.1

Figure 1 · Hierarchical predictive coding
High-level priors BODY SCHEMA · MEANING · EXPECTATION Mid-level inference AFFECT · INTEROCEPTION · CONTEXT Low-level sensory NOCICEPTIVE · SOMATOSENSORY Sensory signals (the body) PREDICTION PREDICTION PREDICTION ERROR (π) ERROR (π) ERROR (π) TOP BOTTOM
Figure 1. Hierarchical predictive coding. Predictions descend the cortical hierarchy (left); only prediction error ascends (right). Precision (π) sets how strongly error at each level is allowed to revise beliefs above it.

The practical upshot is profound: we do not perceive the world directly. We perceive our brain's best model of the world, corrected only where the model is surprised. Andy Clark summarizes the view in a phrase that has become its motto — the brain is a guessing engine whose guesses are, most of the time, "good enough" that the sensory stream simply confirms them.3 Anil Seth's related framing, that ordinary perception is a "controlled hallucination" reined in by sensory evidence, captures the same inversion of the classical model.

Key definitions
Generative model
The brain's internal, hierarchical model of the causes of its sensations; the source of predictions.
Prior
A prediction or expectation carried into perception, learned from prior experience; encoded in higher levels of the model.
Prediction error
The mismatch between prediction and sensory input; the only signal that ascends to revise the model.
Precision
The estimated reliability (inverse variance) of a prediction or an error signal; precision sets how much influence each has — often implemented as attentional gain.
Free energy
A tractable upper bound on "surprise"; the quantity the brain is hypothesized to minimize, by updating beliefs or by acting.1
02 Chapter Two

Pain as Inference, Not Transmission.

If perception in general is inference, then pain in particular is inference. This is the conceptual hinge of the entire handbook. Nociception — the peripheral signaling of actual or potential tissue damage — is an input to the inferential process, not the experience itself. Pain is what the brain infers when it concludes that the body is in danger. As with any inference, the conclusion depends jointly on the evidence (nociceptive and interoceptive signals) and on the prior (the brain's expectation of bodily threat).

Formally, several groups have cast this as Bayesian inference over the hidden state "the body is in a damaging condition."5,6 The brain combines a likelihood (how probable the current sensations are, given pain or no pain) with a prior (how probable pain is, before looking) to produce a posterior — the perceived probability and intensity of pain. Two consequences follow immediately and are central to chronic pain:

  1. Priors can manufacture pain. A sufficiently strong prior expectation of pain can yield a painful percept from innocuous or absent peripheral input — a direct prediction of the model, and the computational signature of allodynia and of pain that outlives healing.5
  2. Learning reshapes the likelihood. Repeated pairing of sensations with pain broadens the range of bodily signals the brain reads as evidence of danger, so that progressively more benign inputs are inferred as painful.6

The same mathematics explains placebo and nocebo effects, which have been a stumbling block for transmission models but fall out naturally here: an expectation (prior) of relief or of harm shifts the posterior percept, and does so through identifiable descending pathways.4,27,28 Experimentally, cueing a high probability of intense pain biases perceptual decisions toward pain even when the delivered stimulus is held constant5 — a clean demonstration that the prior, not the stimulus alone, determines the experience.

Why this is not "the pain is imaginary"

Inference is not invention. A Bayesian percept is a real, neurally instantiated state with real suffering and real physiological consequences. The claim is about source, not legitimacy: the pain is generated centrally by a system doing exactly what it evolved to do — protect the body — on the basis of a miscalibrated prediction.

This distinction is clinically decisive, because reattributing pain to a reversible brain process (rather than to damage) is itself associated with recovery.13

"
The brain is a guessing engine whose guesses are, most of the time, good enough that the sensory stream simply confirms them.
— After Andy Clark3
03 Chapter Three

Active Inference and Precision.

A predictive system has two ways to reduce the mismatch between what it expects and what it senses. It can change the model to fit the world (perceptual inference), or it can change the world to fit the model (active inference): act so that the predicted sensations are the ones actually received.1

Reaching for a cup is active inference — the brain predicts the proprioceptive consequences of the arm already being at the cup, and movement unfolds to fulfill that prediction. Crucially, avoidance and guarding are also active inference. A brain that predicts danger from bending will generate the muscular guarding, postural bracing, and behavioral avoidance that prevent the feared sensation — and thereby prevent the disconfirming evidence that would update the prediction.16

The second pivotal construct is precision. Predictions and error signals are not weighted equally; each is scaled by its estimated reliability. Precision is the brain's confidence in a signal, and it functions as a gain control — neurobiologically linked to attention and to neuromodulators acting on the synaptic gain of error units.3,10

When the precision assigned to a prior is high relative to the precision of sensory evidence, the prior wins: the percept is dragged toward prediction. When sensory precision is high, evidence wins and the model updates. Pain is exquisitely sensitive to the balance of precision (Figure 3), not to nociceptive magnitude alone.

Figure 3 · The precision balance
Danger prior HIGH PRECISION Evidence LOW WEIGHT When the prior outweighs evidence: → the brain infers pain
Figure 3. Precision balance. When a high-precision danger prior outweighs sensory evidence, the brain infers pain even from harmless signals. Treatment aims to rebalance the beam: lower the prior's precision and raise the weight of corrective evidence.

This gives the framework its therapeutic leverage. Two levers move the percept: the content of the prior (what the brain expects) and its precision (how strongly it holds that expectation). Effective treatments, examined in Part III, can be read as operating on one or both — supplying disconfirming evidence to revise the content, and reducing threat and hypervigilance to lower the precision. The free-energy account also clarifies why insight or reappraisal can sometimes change pain rapidly: a high-level belief, once revised, re-weights everything beneath it.25

ConstructRole in pain & clinical handle
Perceptual inference
Update the generative model to match sensation — the route to durable belief change and recovery.
Active inference
Act on the body/world to make sensation match prediction — includes adaptive movement, but also maladaptive guarding and avoidance.
Precision (prior)
Confidence in the prediction; when pathologically high for a danger prior, drives persistent pain. A target for safety reappraisal and exposure.
Precision (sensory)
Confidence in incoming evidence; hypervigilance mis-assigns high precision to benign interoceptive noise.
IIPart Two

The chronic pain cycle.

How an adaptive protective system becomes a self-sustaining loop — and the neurobiology that implements it.

04 Chapter Four

From Acute Protection to Chronic Prediction.

Acute pain is adaptive. A nociceptive signal, inferred as bodily threat, motivates protective action and learning: the organism withdraws, guards, and updates its model so that the precipitating context is treated as dangerous in future. This is normal, useful Bayesian learning. The problem of chronic pain is the problem of a prior that fails to update after the tissue threat has resolved — what several authors call a "stubborn" or hyper-precise prior.6,29

Three processes convert an adaptive acute response into a maintained chronic one:

  1. Central sensitization — activity-dependent increases in the gain of central nociceptive circuits, so that the same input produces larger responses and the system reports pain to progressively weaker stimuli.17 In predictive terms, sensitization raises the precision the system assigns to nociceptive prediction error and lowers the threshold of the danger prior.
  2. Associative learning — through Pavlovian and operant conditioning, initially neutral cues (a posture, a place, a time of day, an emotion) become predictors of pain, broadening the set of triggers.14,22
  3. Avoidance as active inference — protective behaviors prevent the very experiences that could disconfirm the danger prior, so the prediction is shielded from correction and, through deconditioning, made more accurate over time.
The aberrant-inference thesis

Chronic primary pain is best modeled not as ongoing nociception but as aberrant inference: a generative model that has learned, with excessive confidence, to predict bodily danger, and that recruits attention, affect, and behavior in ways that continually confirm its own prediction. The pain is the posterior of a miscalibrated inference.5,6

05 Chapter Five

Anatomy of the Self-Reinforcing Loop.

The maintained condition is best understood as a positive-feedback loop in which each component increases the next and the last reinforces the first (Figure 2). The loop is the predictive-coding re-description of the classical fear-avoidance model of Vlaeyen and Linton,14,15 now given an explicit computational mechanism: every link is an instance of the brain working to minimize prediction error against a danger prior.16

Figure 2 · The self-reinforcing chronic-pain cycle
The chronic pain cycle 1 · PRIOR Danger prior "pain = damage" 2 · ATTENTION Hypervigilance body-scanning, bias 3 · PERCEPT Amplified pain the experienced symptom 4 · APPRAISAL Fear & catastrophizing ACC / insula 5 · BEHAVIOR Guarding · avoidance active inference 6 · CONSOLIDATION Deconditioning + sensitization PREDICTION-ERROR MINIMIZATION
Figure 2. The self-reinforcing chronic-pain cycle. A high-precision danger prior biases attention, amplifies the pain percept, drives fear and avoidance (active inference), and produces deconditioning and sensitization that reinforce the prior. Prediction-error minimization locks the loop.

Tracing the loop, link by link

  1. Danger prior. A high-precision belief that movement or sensation signals harm ("my back is fragile," "pain means damage"). Often seeded by an acute injury, reinforced by alarming explanations, imaging language, or prior trauma, which raises baseline precision.13,30
  2. Hypervigilant attention. The prior allocates precision to the body, so the system preferentially samples and up-weights bodily signals — attentional bias and body-scanning that make benign interoceptive noise salient.10
  3. Amplified pain percept. With prediction outweighing evidence, the posterior is dragged toward pain; sensitization ensures the percept is large. This is the experienced symptom.
  4. Fear and catastrophizing. The percept is appraised as threatening, recruiting affective circuitry (notably the anterior cingulate and insula) that further raises the precision of the danger prior.18,22
  5. Guarding and avoidance. Active inference generates protective behavior that prevents disconfirming experience — the patient does not test, and so does not learn, that the movement is safe.16
  6. Deconditioning & sensitization. Disuse, vigilance, and stress make the body genuinely more reactive, supplying evidence that appears to confirm the prior and closing the loop with higher precision than before.17

Because the loop is self-confirming, it is stable: it persists without any peripheral driver. But because it is a loop, it is also breakable at every node. Interrupting any link — down-weighting the prior, redirecting attention, reappraising the percept, reducing fear, or re-exposing the body to safe movement — reduces the prediction-error minimization that sustains the whole. This is the rationale shared by the therapies in Part III.

06 Chapter Six

Neurobiological Substrates.

The framework is not merely metaphorical; it maps onto identifiable circuitry. No single "pain center" exists. Instead, a distributed set of regions — historically the "pain matrix," more precisely the dynamic pain connectome — implements the hierarchical inference, with different hubs carrying prediction, error, precision, and affective valuation.18

RegionFunctional rolePredictive-coding interpretation
Primary & secondary somatosensory cortexS1, S2
Localization and sensory-discriminative processing; can become sensitized/hyperactive.
Lower-level sensory prediction and error; S2 is among the few sites whose stimulation directly evokes pain.
Anterior cingulate cortexACC
Affective-motivational dimension; suffering and unpleasantness.
Encodes the threat value / precision of the danger prior; lesion abolishes the unpleasantness while sparing detection.22
Insular cortex
Interoception and the felt bodily state; consistently altered in chronic pain.
Hub for interoceptive prediction; integrates bodily evidence with expectation.20,21
Prefrontal / default-mode network
Self-referential processing, rumination, meaning.
Highest-level priors — beliefs about the self and the body; source of "stubborn" expectations.
Descending modulatory systemPAG, RVM
Facilitation and inhibition of nociceptive transmission.
Pathway by which top-down predictions (incl. placebo/expectation) adjust ascending evidence.4

Two empirical anchors are worth highlighting. First, direct electrical stimulation of the human insula and S2 evokes pain, identifying these as nodes where bodily inference becomes painful experience.20 Second, the affective and sensory dimensions are dissociable: lesioning the ACC in animals leaves nociceptive withdrawal intact but abolishes the learned aversiveness of pain,22 exactly as a precision/valuation account predicts — the "hurt" and the "threat" are computed in part by different machinery. Multivariate fMRI work has further identified distributed, reproducible signatures of evoked pain,19 and treatment trials now use such imaging to show that successful psychological treatment shifts activity in precisely these prefrontal–limbic and interoceptive circuits.12

Interoception · the body's newsroom

Interoception — the sensing and regulation of the body's internal state — is the sensory stream most relevant to pain, and is itself organized predictively.9,10 Dysfunctional interoceptive inference, in which the brain mis-weights or misreads internal signals, is implicated across psychophysiologic disorders including chronic pain, anxiety, and functional gastrointestinal conditions.11

In loop terms, the interoceptive system is where benign internal "weather" is mis-forecast as a storm; therapies that retrain interoceptive interpretation (somatic tracking, interoceptive exposure) target this node directly.

IIIPart Three

Clinical applications.

Turning the framework into treatment: the principles that unify modern brain-based pain therapies, the specific interventions, the evidence, and a clinical workflow.

07 Chapter Seven

Principles of Brain-Based Treatment.

Every intervention examined below can be read as operating on one of two levers identified in Part I: the content of the danger prior, or its precision. Three principles follow directly from the model and unify otherwise disparate therapies.

  1. Reattribute the cause. Because the prior's content is the belief that pain signals damage, treatment begins by reattributing pain to a reversible, non-dangerous brain process. This is not reassurance for its own sake; reattribution to mind/brain processes is a statistical mediator of recovery in controlled data.13,30
  2. Supply disconfirming evidence safely. Because avoidance shields the prior from correction, recovery requires re-exposure to feared sensations and movements under conditions safe enough that the brain updates rather than defends. This is perceptual inference deliberately engineered.16
  3. Lower precision through safety. Because a high-precision danger prior dominates evidence, reducing threat — via reappraisal, attention retraining, autonomic down-regulation, and a trusting therapeutic relationship — lowers the gain on the prior so that corrective evidence can take hold.4
The common mechanism

Pain Reprocessing Therapy, graded exposure, somatic tracking, mindfulness, interoceptive exposure, and virtual embodiment differ in delivery but share one mechanism: they change what the brain predicts about the body, and how strongly it holds that prediction.

The framework thus offers not a single new technique but a mechanistic rationale that integrates the field and tells the clinician what each technique is for.

08 Chapter Eight

The Intervention Toolkit.

8.1  Pain neuroscience education & reattribution

Teaching patients that pain is constructed by the brain and does not require tissue damage directly targets the content of the danger prior. Modern "Explain Pain" approaches show that conceptual change in pain beliefs reduces threat and disability.30 In the Pain Reprocessing Therapy (PRT) trial, the degree to which patients came to attribute their pain to mind/brain (rather than bodily) causes statistically mediated their recovery,13 confirming that belief change is not incidental but mechanistic.

8.2  Pain Reprocessing Therapy & somatic tracking

PRT combines reattribution with somatic tracking: attending to feared bodily sensations with curiosity and an explicit appraisal of safety, rather than alarm. In predictive terms, somatic tracking presents the interoceptive signal to the system while withholding the threat valuation that would raise the prior's precision — supplying "safe" evidence and lowering gain simultaneously.

The randomized trial in chronic back pain found that 66% of PRT patients were pain-free or nearly pain-free post-treatment, versus 20% on placebo and 10% on usual care, with gains largely maintained at one year and corroborated by reduced activity in prefrontal and anterior insular regions.12

8.3  Graded exposure to feared movement

Derived from the fear-avoidance model, graded exposure systematically re-exposes the patient to avoided movements, beginning with those rated least threatening, to generate disconfirming evidence that updates the danger prior.14 Exposure-based treatment of pain-related fear improves function and reduces catastrophizing, and the active-inference reading clarifies why graded, predictable exposure works where abrupt confrontation does not: belief updating requires that prediction error be both present and survivable, so that the system revises rather than defends.16

8.4  Mindfulness & attention retraining

Mindfulness training reduces the precision the system grants to the danger prior by altering the global balance between prior-driven and evidence-driven processing — a mechanism articulated explicitly in predictive accounts of meditation.25 By cultivating non-reactive, present-moment attention to sensation, the practitioner decouples the sensory signal from automatic threat valuation, attenuating the hypervigilance and catastrophizing links of the loop.

8.5  Interoceptive retraining

Because chronic pain involves mis-weighted interoceptive inference, therapies that retrain the interpretation of internal signals — interoceptive exposure, breath and body-awareness practices — target the loop at its sensory root,11 helping the brain "fact-check" the body's alarms and reassign benign signals to the benign category.

8.6  Virtual embodiment & immersive VR

Immersive virtual reality can present the brain with vivid, first-person evidence that movement is safe and that the body is whole and capable, exploiting embodiment and presence to drive perceptual updating. VR for chronic pain is thought to act through distraction (transiently lowering sensory precision) and, more durably, through embodiment — manipulating the experienced body to revise the body schema and the danger prior.

A double-blind, placebo-controlled trial of a behavioral-skills VR program (EaseVRx) in chronic low back pain demonstrated superiority over sham VR with durable benefit,23 and an open-label study of virtual embodiment training reported significant within-session reductions in pain intensity and improvements in fear-avoidance and disability measures, motivating controlled trials.24

Matching technique to mechanism

Targets prior CONTENT: pain neuroscience education, reattribution, graded exposure, virtual embodiment (disconfirming evidence).

Targets prior PRECISION: somatic tracking, mindfulness, autonomic regulation, therapeutic safety (lower threat gain).

Targets the SENSORY node: interoceptive retraining, VR distraction (re-weighting evidence).

09 Chapter Nine

Evidence summary.

The table below collates representative controlled evidence for the framework's clinical predictions. It is illustrative rather than exhaustive; effect sizes vary with population, comparator, and outcome, and several modalities still await large, well-controlled, active-comparator trials.

Intervention / claimKey findingSource
Pain Reprocessing Therapy (chronic back pain, RCT)
~66% pain-free / near pain-free vs 20% placebo, 10% usual care; maintained at 1 yr; fMRI changes in prefrontal/insular regions.
Ashar et al.
JAMA Psychiatry 202212
Reattribution as mechanism
Shift in attributing pain to mind/brain causes mediated recovery.
Ashar et al.
JAMA Netw Open 202313
Expectation / prior changes pain
Probabilistic cueing biases pain perception with stimulus held constant; placebo via descending pathways.
Hechler 2016;
Büchel 20145,4
Therapeutic VR (chronic low back pain, double-blind RCT)
Behavioral-skills VR superior to sham VR; durable benefit.
Garcia et al.
JMIR 202123
Virtual embodiment training (open-label)
Significant within-session pain reduction; improved fear-avoidance and disability.
Trost et al.
JMIR Form Res 202424
Graded exposure to feared movement
Reduces pain-related fear, catastrophizing, and disability.
Vlaeyen & Linton
2000; 201614,15
Central sensitization as substrate
Activity-dependent gain increases lower thresholds and amplify responses.
Woolf
Pain 201117

Read together, these findings support the framework's central clinical claim: that interventions which change the brain's predictions about the body — and the precision of those predictions — can reduce or resolve chronic primary pain, and do so through the mechanistic pathway the theory specifies.

10 Chapter Ten

A Clinical Workflow.

The framework yields a pragmatic sequence for assessment and treatment. It presupposes appropriate medical evaluation; the workflow applies to primary (nociplastic) pain after structural and systemic disease have been suitably addressed.

Step 1 · Triage & diagnosis

Confirm a primary/nociplastic presentation. Features pointing to a predominantly predictive (centrally generated) mechanism: pain inconsistent with anatomy, that spreads or moves, varies with stress/attention/emotion, began at a stressful period, or coexists with other psychophysiologic conditions. Exclude red flags and structural disease first.30

Step 2 · Reattribution & education

Establish, with the patient, a brain-based formulation: the pain is real and protective, generated by a sensitized prediction rather than ongoing damage, and therefore reversible. Reattribution is both the foundation and an active ingredient.13

Step 3 · Lower precision

Reduce threat and hypervigilance through somatic tracking, mindfulness, autonomic regulation, and a safe therapeutic alliance, so that the danger prior loses its grip on the percept.12,25

Step 4 · Supply disconfirming evidence

Through graded exposure and, where available, virtual embodiment, give the brain repeated, survivable experiences of safe movement and benign sensation, driving the prior to update.16,23

Step 5 · Consolidate & generalize

Translate gains into daily function and identity — the patient's functional goals — reframing flares as transient mispredictions to be met with the same skills rather than as evidence of damage. Durability is the test of genuine model change, and is observed in the strongest trials.12

Patient-facing formulation · one paragraph

"Your pain is completely real, and it is your brain doing its job: predicting danger and protecting you. After an injury or a stressful period, the brain can learn to predict pain so strongly that it keeps generating it even after the tissues are fine — the alarm stays on. The encouraging part is that what the brain learns, it can unlearn. Treatment retrains the prediction: we lower the sense of threat, and we give your brain safe, repeated proof that movement and sensation are okay, until it turns the alarm back down."

11 Chapter Eleven

Open Questions & Research Agenda.

The framework is generative precisely because it is falsifiable and quantitative. Priorities for the next phase of research include:

  1. Computational phenotyping. Fitting Bayesian/active-inference models to individual patients to estimate prior precision and learning rates, enabling mechanism-based subtyping and treatment matching.6
  2. Objective markers. Validating neural (fMRI signatures, EEG measures such as alpha asymmetry), behavioral (movement guarding), and psychophysical readouts of prior precision and its change with treatment.19
  3. Active-comparator trials. Rigorous, same-content controlled trials for VR/embodiment and somatic therapies to isolate specific from expectation effects.23
  4. First-person measurement. Developing structured phenomenological methods to quantify subjective experience and link it to model parameters — a standing gap in the field.
  5. Trauma & precision. Testing the hypothesis that adverse experience raises the baseline precision of bodily-threat priors, and whether trauma-informed care modifies it.11
  6. Generalization across PPD. Determining how far the same inferential mechanism, and the same treatment levers, extend across migraine, IBS, fibromyalgia, and pelvic pain.11

The throughline is a shift in the explanatory target of pain medicine — from the tissues to the inference — and a corresponding shift in treatment, from managing a symptom to retraining a prediction. The accumulating controlled evidence suggests the shift is not only conceptually coherent but clinically consequential.

Bibliography

References.

Numbered to match the footnote citations throughout. Identifiers (PMID / DOI) are provided for retrieval via PubMed.

1.
Friston K. The free-energy principle: a unified brain theory? Nat Rev Neurosci. 2010;11(2):127–138. doi:10.1038/nrn2787 · PMID:20068583
2.
Rao RP, Ballard DH. Predictive coding in the visual cortex: a functional interpretation of some extra-classical receptive-field effects. Nat Neurosci. 1999;2(1):79–87. doi:10.1038/4580 · PMID:10195184
3.
Clark A. Whatever next? Predictive brains, situated agents, and the future of cognitive science. Behav Brain Sci. 2013;36(3):181–204. doi:10.1017/S0140525X12000477 · PMID:23663408
4.
Büchel C, Geuter S, Sprenger C, Eippert F. Placebo analgesia: a predictive coding perspective. Neuron. 2014;81(6):1223–1239. doi:10.1016/j.neuron.2014.02.042 · PMID:24656247
5.
Hechler T, Endres D, Thorwart A. Why harmless sensations might hurt in individuals with chronic pain: about heightened prediction and perception of pain in the mind. Front Psychol. 2016;7:1638. doi:10.3389/fpsyg.2016.01638 · PMID:27826270
6.
Eckert AL, Pabst K, Endres DM. A Bayesian model for chronic pain. Front Pain Res (Lausanne). 2022;3:966034. doi:10.3389/fpain.2022.966034 · PMID:36304671
7.
Lersch FE, Frickmann FCS, Urman RD, et al. Analgesia for the Bayesian brain: how predictive coding offers insights into the subjectivity of pain. Curr Pain Headache Rep. 2023;27(11):631–638. doi:10.1007/s11916-023-01122-5 · PMID:37421541
8.
Tabor A, Burr C. Bayesian learning models of pain: a call to action. Curr Opin Behav Sci. 2019;26:54–61. doi:10.1016/j.cobeha.2018.10.003
9.
Seth AK, Friston KJ. Active interoceptive inference and the emotional brain. Philos Trans R Soc Lond B Biol Sci. 2016;371(1708):20160007. doi:10.1098/rstb.2016.0007 · PMID:28080966
10.
Barrett LF, Simmons WK. Interoceptive predictions in the brain. Nat Rev Neurosci. 2015;16(7):419–429. doi:10.1038/nrn3950 · PMID:26016744
11.
Khalsa SS, Adolphs R, Cameron OG, et al. Interoception and mental health: a roadmap. Biol Psychiatry Cogn Neurosci Neuroimaging. 2018;3(6):501–513. doi:10.1016/j.bpsc.2017.12.004 · PMID:29884281
12.
Ashar YK, Gordon A, Schubiner H, et al. Effect of pain reprocessing therapy vs placebo and usual care for patients with chronic back pain: a randomized clinical trial. JAMA Psychiatry. 2022;79(1):13–23. doi:10.1001/jamapsychiatry.2021.2669 · PMID:34586357
13.
Ashar YK, Lumley MA, Perlis RH, et al. Reattribution to mind-brain processes and recovery from chronic back pain: a secondary analysis of a randomized clinical trial. JAMA Netw Open. 2023;6(9):e2333846. doi:10.1001/jamanetworkopen.2023.33846 · PMID:37773495
14.
Vlaeyen JWS, Linton SJ. Fear-avoidance and its consequences in chronic musculoskeletal pain: a state of the art. Pain. 2000;85(3):317–332. doi:10.1016/s0304-3959(99)00242-0 · PMID:10781906
15.
Vlaeyen JWS, Crombez G, Linton SJ. The fear-avoidance model of pain. Pain. 2016;157(8):1588–1589. doi:10.1097/j.pain.0000000000000574 · PMID:27428892
16.
Hood A, Stahl D, et al. The fear-avoidance model as an embodied prediction of threat. Cogn Affect Behav Neurosci. 2024;24(5):781–802. doi:10.3758/s13415-024-01199-4
17.
Woolf CJ. Central sensitization: implications for the diagnosis and treatment of pain. Pain. 2011;152(3 Suppl):S2–S15. doi:10.1016/j.pain.2010.09.030 · PMID:20961685
18.
Kucyi A, Davis KD. The dynamic pain connectome. Trends Neurosci. 2015;38(2):86–95. doi:10.1016/j.tins.2014.11.006 · PMID:25541287
19.
Wager TD, Atlas LY, Lindquist MA, et al. An fMRI-based neurologic signature of physical pain. N Engl J Med. 2013;368(15):1388–1397. doi:10.1056/NEJMoa1204471 · PMID:23574118
20.
Mazzola L, Isnard J, Peyron R, Guénot M, Mauguière F. Somatotopic organization of pain responses to direct electrical stimulation of the human insular cortex. Pain. 2009;146(1–2):99–104. doi:10.1016/j.pain.2009.07.014 · PMID:19665303
21.
Labrakakis C. The role of the insular cortex in pain. Int J Mol Sci. 2023;24(6):5736. doi:10.3390/ijms24065736 · PMID:36982807
22.
Johansen JP, Fields HL, Manning BH. The affective component of pain in rodents: direct evidence for a contribution of the anterior cingulate cortex. Proc Natl Acad Sci USA. 2001;98(14):8077–8082. doi:10.1073/pnas.141218998 · PMID:11416168
23.
Garcia LM, Birckhead BJ, Krishnamurthy P, et al. An 8-week self-administered at-home behavioral skills-based virtual reality program for chronic low back pain: double-blind, randomized, placebo-controlled trial conducted during COVID-19. J Med Internet Res. 2021;23(2):e26292. doi:10.2196/26292 · PMID:33484240
24.
Trost Z, Anderson Z, et al. Effects of embodiment in virtual reality for treatment of chronic pain: pilot open-label study. JMIR Form Res. 2024;8:e34162. doi:10.2196/34162 · PMID:38421701
25.
Laukkonen RE, Slagter HA. From many to (n)one: meditation and the plasticity of the predictive mind. Neurosci Biobehav Rev. 2021;128:199–217. doi:10.1016/j.neubiorev.2021.06.021 · PMID:34147532
26.
Hebb DO. The Organization of Behavior: A Neuropsychological Theory. New York, NY: Wiley; 1949.
27.
Ongaro G, Kaptchuk TJ. Symptom perception, placebo effects, and the Bayesian brain. Pain. 2019;160(1):1–4. doi:10.1097/j.pain.0000000000001367 · PMID:30086112
28.
Geuter S, Koban L, Wager TD. The cognitive neuroscience of placebo effects: concepts, predictions, and physiology. Annu Rev Neurosci. 2017;40:167–188. doi:10.1146/annurev-neuro-072116-031132 · PMID:28399689
29.
Edwards MJ, Adams RA, Brown H, Párées I, Friston KJ. A Bayesian account of "hysteria". Brain. 2012;135(11):3495–3512. doi:10.1093/brain/aws129 · PMID:22641838
30.
Moseley GL, Butler DS. Fifteen years of explaining pain: the past, present, and future. J Pain. 2015;16(9):807–813. doi:10.1016/j.jpain.2015.05.005 · PMID:26051220

"From the tissues, to the inference."

This handbook is a research and educational synthesis. It does not constitute medical advice, and the framework it describes applies to primary/nociplastic and psychophysiologic pain after appropriate medical evaluation has excluded structural and systemic disease.

Clinical decisions remain the responsibility of qualified practitioners.

© Karuna Labs Inc.  ·  The Predictive Brain in Chronic Pain  ·  v1.0