Why Your Brain Overpredicts Pain: How the Brain’s Guesses Shape Chronic Pain & Why Touch Might Matter
The Brain as a Prediction Machine
We notice this system most clearly when it gets something wrong. Reach for a step in the dark only to find one fewer than anticipated, and your body lurches before you understand why. Nothing touched you; the disruption came from the mismatch between what your brain predicted and what actually occurred. This gap is a prediction error, a central concept in Karl Friston’s account of how the nervous system continuously generates and updates predictions about the world (Friston, 2010).
In healthy pain processing, this system works like a well-tuned thermostat. When your hand touches a hot surface, your brain generates a prediction: “This will hurt.” The incoming sensory signal confirms or contradicts that prediction. If the two match, the prediction error is small and there is little to update, because the model was already right. If they don’t match, for example, if the surface is much hotter or cooler than expected, the larger error signals that the model needs adjusting, so the next prediction is more accurate. Either way, you pull your hand back, and pain serves its purpose: protection (Barrett & Simmons, 2015).
When Predictions Go Wrong: One Account of the Chronic Pain Spiral
Researchers working within the predictive coding framework, such as Hechler and colleagues (2016) and Tabor and colleagues (2017), have described chronic pain as a failure of this predictive system to update. The problem, in this framework, is not that the brain has stopped predicting but that it has stopped reacting. After an injury, whether from surgery, an accident, or illness, the tissue heals, and yet the brain’s model of that body part keeps forecasting danger. Each time sensory input arrives, it meets the same mismatch: the brain predicts severe pain, while the signal coming from the tissue may be minimal or absent. In healthy learning, the mismatch would drive the prediction down. Here it doesn’t. It’s worth being clear that this is one framework among several rather than a settled explanation, and that much of the supporting human evidence is correlational.
Why does this happen? Several factors appear to contribute. For instance, there is a process called central sensitization, where repeated nociceptive input can increase the excitability of neurons in the spinal cord and brain. One component of it, known as wind-up, describes how a stimulus delivered repeatedly at the same intensity produces a progressively larger pain response, as though the system were turning up its own gain (Woolf, 2011). Fear and avoidance behaviors learned during an acute injury may also help sustain the problem: avoiding a movement removes the opportunity to discover that the movement is safe, so the prediction never gets tested (Vlaeyen & Linton, 2000). Once the brain gets locked into a pain-prediction pattern, it becomes remarkably resistant to change, even when objective evidence (like imaging) shows the original injury has healed.
This is not imaginary pain. The brain’s mispredictions are very real, producing genuine pain signals and maintaining a cycle of suffering. For someone with chronic back pain, a simple movement can trigger intense pain even when imaging shows little structural damage. Prediction isn’t an alternative to physical pathology, but rather one of several factors that shape the experience, and it can amplify pain well beyond what is happening in the tissue.
What the Scans Actually Show
Modern brain imaging has revealed something striking about how activity is organized in chronic pain. However, it’s worth being precise about what these machines actually measure. Functional MRI does not record electrical activity directly. It tracks the blood-oxygen-level-dependent (BOLD) signal, a slow, indirect proxy for changes in neural activity. What researchers can do is ask how irregular that signal is over time, like how unpredictable it looks from one moment to the next, and quantify that irregularity as entropy. In a study drawing on more than 30,000 middle-aged and older adults in the UK Biobank, people with chronic pain showed higher brain entropy than controls in the prefrontal cortex and several other regions, with more widespread pain associated with higher entropy in the occipital cortex (Del Mauro et al., 2025).
Entropy measures how surprising a signal is from one moment to the next. That well-tuned thermostat from earlier produces a low-entropy record: dull, regular, easy to predict. A storm produces a high-entropy one, where the last gust tells you nothing about the next. On this measure, chronic pain brains look more like the storm. Brain regions responsible for sensory processing, attention, and emotion show heightened entropy, meaning their activity becomes less predictable from moment to moment. These are associations, not causal findings, and the direction isn’t consistent. In an earlier study using the same entropy measure in two Human Connectome Project samples, the same research group (Del Mauro et al., 2024) found no overall relationship between pain intensity and brain entropy, and two opposite patterns once the sample was split by age: in young adults, greater pain intensity went with lower entropy in regions handling the sensory side of pain, while in middle-aged and older participants it went with higher entropy. What the entropy literature currently supports is that chronic pain is accompanied by altered brain dynamics, not that disorganized activity produces pain.
Imaging studies also point to changes in how brain regions communicate. The pain system isn’t a one-way street: signals ascend from the body while descending pathways from regions including the prefrontal cortex and brainstem modulate them, and predictive coding accounts treat these two directions as a single recurrent loop rather than separate channels (Büchel et al., 2014). In chronic pain, several studies suggest that descending modulation becomes less effective. A meta-analysis of 30 studies found that people with chronic pain showed weaker conditioned pain modulation, a laboratory measure of the body’s built-in pain inhibition, than pain-free controls (Lewis et al., 2012). An fMRI study of people with fibromyalgia found that the rostral anterior cingulate cortex, a key region in the descending pain-control system, failed to respond to painful pressure as it did in controls (Jensen et al., 2009). Weaker descending control would leave incoming signals less well checked against the brain’s expectations. The result, on this account, is a system less able to tell a genuine threat from a false alarm.
Can Touch Reprogram Pain Predictions?
If chronic pain stems from a prediction gone wrong, then part of the answer might involve providing unexpected, non-threatening sensory input that helps revise the brain’s model. This is where manual therapies, including osteopathic manipulative treatment (OMT), massage, and other hands-on approaches, may play a role.
Here’s the logic. When a therapist applies gentle, sustained touch to a painful area, the patient receives sensory input that is organized, predictable, and, importantly, not harmful. If the brain predicted pain and the actual sensation is tolerable, that mismatch carries information, and over repeated sessions the model may begin to shift. Bialosky and colleagues (2009, 2018) have argued for something close to this: that manual therapy works less by mechanically correcting tissue than through a cascade of neurophysiological and cognitive effects, including changes in central pain processing and in what the patient expects.
Expectation isn’t a soft addition to this account; it can be measured. In an experimental study, healthy participants given a negative expectation before spinal manipulation reported greater pain sensitivity afterward than those given neutral or positive framing (Bialosky et al., 2008). The clinician relationship matters too: among 182 people with chronic low back pain, a stronger therapeutic alliance predicted better outcomes across treatment types (Ferreira et al., 2013). Pain neuroscience education, like teaching patients how pain is actually produced, has been associated with reductions in pain and disability in systematic reviews (Louw et al., 2016), though more recent overviews have criticized the quality of that evidence and cautioned against overstating the effect (Martinez-Calderon et al., 2023). Two claims are worth keeping separate here, since they often get run together: that expectations shape pain perception, which is well supported, and that expectations specifically amplify the effects of manual therapy, which is not yet settled.
What This Means for Living With Chronic Pain
Understanding chronic pain as a problem that involves misprediction, not tissue damage alone, reframes how we approach treatment and self-management. None of this is a treatment plan. But it changes what the problem looks like, and that changes what’s worth trying:
1. Your pain is real. Expectations don’t cause it, but they can shape it. The brain’s predictions shape pain intensity. This doesn’t mean the pain is “all in your head,” but rather it means that managing thoughts, reducing fear, and cultivating realistic expectations are legitimate and powerful therapeutic tools.
2. Movement and gentle touch may help retrain predictions. Gradual, supervised movement and tactile therapies aren’t only comfort measures; they give the nervous system repeated evidence that the affected area can be used without harm, which may shift pain-related predictions over time. A Cochrane overview of exercise for chronic pain found generally favorable effects on pain and physical function, though the underlying evidence was of low quality and results varied considerably across conditions (Geneen et al., 2017).
3. Education and trust may improve outcomes. Understanding the mechanisms behind chronic pain has been linked to reduced catastrophic thinking and fear avoidance, and working with clinicians who explain pain science and build trust in treatment is associated with better results (Ferreira et al., 2013; Louw et al., 2016).
4. The brain’s capacity to reorganize doesn’t disappear with chronicity. Cortical changes associated with chronic pain have been shown to normalize alongside reductions in pain in several treatment studies, which is part of why interventions that target the brain’s representation of the painful body part are being pursued at all (May, 2008; Moseley & Flor, 2012). Meaningful improvement is possible for many people, though outcomes vary considerably and this is not a guarantee.
The Future of Precision Pain Medicine
Although our understanding of chronic pain through a predictive-coding lens is very much growing, a lot of work remains. Future research must validate whether entropy signatures measured by brain imaging can reliably predict treatment outcomes. Clinical trials combining neuroimaging, standardized entropy analysis, and rigorous outcome measurement are essential. We also need to test whether interventions explicitly informed by predictive-coding theory, combining manual therapy, movement, and cognitive training, outperform conventional treatments.
If entropy normalization were validated as a biomarker, it could change how chronic pain is managed. Rather than proceeding by trial and error, clinicians might measure baseline neural entropy, choose interventions on that basis, and track change objectively. There is precedent for the general approach: Baliki and colleagues found that corticostriatal functional connectivity measured shortly after a back injury predicted which patients would go on to develop chronic pain (Baliki et al., 2012). Whether entropy can do comparable work is still an open question.
In the meantime, the practical implication is smaller and stranger than a biomarker. If chronic pain is partly a prediction the brain won’t revise, then part of the work is giving it something new to learn from: a movement that turns out not to hurt, a touch that turns out to be safe, an explanation that makes the alarm less frightening. Not a cure. Evidence, delivered one instance at a time, to a system that stopped updating.
References
Baliki, M. N., Petre, B., Torbey, S., Herrmann, K. M., Huang, L., Schnitzer, T. J., Fields, H. L., & Apkarian, A. V. (2012). Corticostriatal functional connectivity predicts transition to chronic back pain. Nature Neuroscience, 15(8), 1117–1119. https://doi.org/10.1038/nn.3153\
Barrett, L. F., & Simmons, W. K. (2015). Interoceptive predictions in the brain. Nature Reviews Neuroscience, 16(7), 419–429. https://doi.org/10.1038/nrn3950\
Bialosky, J. E., Beneciuk, J. M., Bishop, M. D., Coronado, R. A., Penza, C. W., Simon, C. B., & George, S. Z. (2018). Unraveling the mechanisms of manual therapy: Modeling an approach. Journal of Orthopaedic & Sports Physical Therapy, 48(1), 8–18. https://doi.org/10.2519/jospt.2018.7476\
Bialosky, J. E., Bishop, M. D., Price, D. D., Robinson, M. E., & George, S. Z. (2009). The mechanisms of manual therapy in the treatment of musculoskeletal pain: A comprehensive model. Manual Therapy, 14(5), 531–538. https://doi.org/10.1016/j.math.2008.09.001\
Bialosky, J. E., Bishop, M. D., Robinson, M. E., Barabas, J. A., & George, S. Z. (2008). The influence of expectation on spinal manipulation induced hypoalgesia: An experimental study in normal subjects. BMC Musculoskeletal Disorders, 9(19). https://doi.org/10.1186/1471-2474-9-19\
Büchel, C., Geuter, S., Sprenger, C., & Eippert, F. (2014). Placebo analgesia: A predictive coding perspective. Neuron, 81(6), 1223–1239. https://doi.org/10.1016/j.neuron.2014.02.042\
Del Mauro, G., Li, Y., Yu, J., Kochunov, P., Sevel, L. S., Boissoneault, J., Chen, S., & Wang, Z. (2025). Chronic pain is associated with greater brain entropy in the prefrontal cortex. The Journal of Pain, 32, Article 105421. https://doi.org/10.1016/j.jpain.2025.105421\
Del Mauro, G., Sevel, L. S., Boissoneault, J., & Wang, Z. (2024). Divergent association between pain intensity and resting-state fMRI-based brain entropy in different age groups. Journal of Neuroscience Research, 102(5), Article e25341. https://doi.org/10.1002/jnr.25341\
Ferreira, P. H., Ferreira, M. L., Maher, C. G., Refshauge, K. M., Latimer, J., & Adams, R. D. (2013). The therapeutic alliance between clinicians and patients predicts outcome in chronic low back pain. Physical Therapy, 93(4), 470–478. https://doi.org/10.2522/ptj.20120137\
Friston, K. (2010). The free-energy principle: A unified brain theory? Nature Reviews Neuroscience, 11(2), 127–138. https://doi.org/10.1038/nrn2787\
Geneen, L. J., Moore, R. A., Clarke, C., Martin, D., Colvin, L. A., & Smith, B. H. (2017). Physical activity and exercise for chronic pain in adults: An overview of Cochrane Reviews. Cochrane Database of Systematic Reviews, 2017(4), Article CD011279. https://doi.org/10.1002/14651858.CD011279.pub3\
Hechler, T., Endres, D., & Thorwart, A. (2016). Why harmless sensations might hurt in individuals with chronic pain: About heightened prediction and perception of pain in the mind. Frontiers in Psychology, 7, Article 1638. https://doi.org/10.3389/fpsyg.2016.01638\
Jensen, K. B., Kosek, E., Petzke, F., Carville, S., Fransson, P., Marcus, H., Williams, S. C. R., Choy, E., Giesecke, T., Mainguy, Y., Gracely, R., & Ingvar, M. (2009). Evidence of dysfunctional pain inhibition in fibromyalgia reflected in rACC during provoked pain. Pain, 144(1–2), 95–100. https://doi.org/10.1016/j.pain.2009.03.018\
Lewis, G. N., Rice, D. A., & McNair, P. J. (2012). Conditioned pain modulation in populations with chronic pain: A systematic review and meta-analysis. The Journal of Pain, 13(10), 936–944. https://doi.org/10.1016/j.jpain.2012.07.005\
Louw, A., Zimney, K., Puentedura, E. J., & Diener, I. (2016). The efficacy of pain neuroscience education on musculoskeletal pain: A systematic review of the literature. Physiotherapy Theory and Practice, 32(5), 332–355. https://doi.org/10.1080/09593985.2016.1194646\
Martinez-Calderon, J., Ho, E. K.-Y., Ferreira, P. H., Garcia-Muñoz, C., Villar-Alises, O., & Matias-Soto, J. (2023). A call for improving research on pain neuroscience education and chronic pain: An overview of systematic reviews. Journal of Orthopaedic & Sports Physical Therapy, 53(6), 353–368. https://doi.org/10.2519/jospt.2023.11833\
May, A. (2008). Chronic pain may change the structure of the brain. Pain, 137(1), 7–15. https://doi.org/10.1016/j.pain.2008.02.034\
Moseley, G. L., & Flor, H. (2012). Targeting cortical representations in the treatment of chronic pain: A review. Neurorehabilitation and Neural Repair, 26(6), 646–652. https://doi.org/10.1177/1545968311433209\
Tabor, A., Thacker, M. A., Moseley, G. L., & Körding, K. P. (2017). Pain: A statistical account. PLOS Computational Biology, 13(1), Article e1005142. https://doi.org/10.1371/journal.pcbi.1005142\
Vlaeyen, J. W. S., & Linton, S. J. (2000). Fear-avoidance and its consequences in chronic musculoskeletal pain: A state of the art. Pain, 85(3), 317–332. https://doi.org/10.1016/S0304-3959(99)00242-0\
Woolf, C. J. (2011). Central sensitization: Implications for the diagnosis and treatment of pain. Pain, 152(3, Suppl.), S2–S15. https://doi.org/10.1016/j.pain.2010.09.030
About the Author
Akshaya Sahasra Ganji is an undergraduate at Florida International University on a BS/DO track, with research experience in neurology and health equity. Her work focuses on how the brain processes pain and on the policy side of equitable access to care. She writes to make neuroscience legible to people outside it. Outside the lab, she loves to read and is interested in films.