Therapeutic neuroscience
The pillar: how a predicting brain builds pain and the self, and how both can be retrained.
A neurosurgeon can work inside an awake patient's skull while the patient chats about the weather. The brain being touched registers nothing. The same organ is also the only place a stubbed toe, a migraine or ten years of back pain has ever been felt. Both facts are true, and together they explain more about chronic pain than either does alone.
Does the brain have pain receptors?
No. Brain tissue itself has no pain receptors, which is why surgeons can operate on an awake brain. Headaches come from the meninges, blood vessels, muscles and nerves around it. Yet the brain is where every pain is assembled: it weighs danger signals from the body against expectation, memory and context, and produces the experience. Therapeutic neuroscience starts from that distinction.
The organ that cannot be hurt is the organ that produces every hurt, and in chronic pain the way it does so changes.
No. Nociceptors, the free nerve endings that respond to damaging heat, pressure or chemicals, are found in skin, muscle, joints, bone, the organs, the coverings of the brain and the walls of its blood vessels. The gray and white matter of the brain itself has none. You can press on the cortex, cut it or pass an electric current through it, and the person will feel no pain from the tissue being touched.
This is the reason awake brain surgery exists. For tumors near speech or movement areas, surgeons numb the scalp and the dura (the tough outer membrane) with local anesthetic, then wake the patient. The patient counts, names pictures or moves a hand while the surgeon works, so that the team can tell in real time which tissue is safe to remove. Wilder Penfield built the first maps of the human cortex this way in the 1930s and 1940s, and the technique is routine today.
Everything around the brain. The meninges are richly supplied with nociceptors, as are the arteries inside and outside the skull, the muscles of the scalp and neck, the sinuses and the eyes. Most of these structures send their danger signals through the trigeminal nerve, which is why migraine research centers on what is called the trigeminovascular system. When a headache throbs, you are feeling signals from blood vessels and membranes, interpreted by a brain that is itself silent. Our chronic headache and migraine page covers what that means for long-running headaches.
A danger signal from the meninges is not a headache until the brain has turned it into one. That conversion, from signal to experience, is the whole subject of this page. It is also where the brain adds expectation, memory and mood, which is why two people with identical scans can have very different headaches.
Several, and none of them does the job alone. Imaging studies over the past three decades have found a consistent set of regions that become active when something hurts. Each contributes a different piece of the experience.
| Region | What research suggests it contributes |
|---|---|
| Thalamus | The relay. Danger signals climbing the spinal cord pass through here on the way to the cortex, and the thalamus helps decide what gets through |
| Primary and secondary somatosensory cortex (S1, S2) | Where and how strong. S1 holds the sensory map of the body; S2 adds detail and links the sensation to memory of past sensations |
| Insula | The felt state of the body. Integrates signals from inside the body (interoception) with emotion. With S2, the region where direct electrical stimulation has evoked pain |
| Anterior cingulate cortex (ACC) | The unpleasantness and the urge to do something about it. The suffering part of pain rather than the sensing part |
| Prefrontal cortex | Meaning, expectation and attention. Decides what the sensation signifies and what to do, and drives descending pathways that can turn the signal down |
The insula deserves a closer look. In patients being evaluated for epilepsy surgery, who have electrodes placed in the brain for clinical reasons, Mazzola and colleagues found that direct electrical stimulation of the posterior insula could produce painful sensations, and that the pain was mapped to specific body parts depending on where the current was applied. Across the stimulation studies that have been reported, the insula and the neighboring S2 region of the operculum are the main cortical areas from which pain has been evoked. Even there, most stimulations produced sensations that were not painful.
In the 1990s Ronald Melzack proposed that pain is the output of a widely distributed network he called the body-self neuromatrix, rather than a direct readout of injury. The phrase 'pain matrix' became shorthand for the regions in the table above. Then a problem emerged: the same regions activate for loud sounds, bright flashes and anything else that demands attention. They mark salience, not pain specifically.
Researchers now tend to describe a dynamic pain connectome: pain emerges from shifting interactions among the salience network, the default mode network and the brain's own descending pain-control system, and the pattern changes from moment to moment with attention and mood. The practical point is that there is no pain center to remove or switch off. There is a conversation among regions, and the conversation can be changed. Our commentary on brain stimulation and neural silencing looks at what happens when researchers try to intervene in it directly.
From the brain, as a protective conclusion drawn from many inputs at once. Signals from damaged or threatened tissue are one of those inputs, usually the most important one. So are expectation, memory, mood, what you are paying attention to and what the pain would mean for your life. The conclusion is pain: an experience designed to make you protect something.
Two words need separating here. Nociception is the signal: a nerve ending in the body detecting something potentially damaging and passing the message up. Pain is the experience the brain produces. Most of the time they travel together, which is why they feel like the same thing. They are not. A soldier can have a serious wound and feel little until the fighting stops. A person can have a scan full of alarming findings and no pain at all. And a person can have severe pain in a limb that no longer exists.
A useful way to hold this: nociceptors are weather stations, and the brain is the forecaster. The stations report temperature and pressure. The forecaster issues a forecast, and the forecast is what you act on. It takes the station readings seriously, but it also weighs the season, yesterday's weather and what happened the last time the pressure dropped. A good forecaster is sometimes wrong in both directions, and a forecaster who has lived through one bad storm may start calling for rain on clear days.
Phantom limb pain is the clearest evidence that the forecast can run without the stations. After an amputation there are no nociceptors in the missing hand, yet the brain's map of the body still contains a hand, and that map can still generate burning, cramping or crushing pain in it. Melzack built the neuromatrix theory largely on this observation: if pain can be produced with no input from the limb, the pain must be produced centrally.
One influential framework, predictive processing, describes the forecaster in more detail. On this account the brain is constantly predicting the state of the body and correcting those predictions against incoming signals, and pain is what it feels like when the system concludes that protection is needed. The framework has growing support but is far from settled, and it does not replace the biology of nerves and tissue. Those are real inputs, and they need real medical assessment.
None of this makes pain imaginary. Every pain anyone has ever felt was produced this way, including the pain of a broken arm. The output is real whether or not the input is. What changes is the set of questions worth asking. 'What is wrong with the tissue?' remains one of them. 'What has the brain concluded, and why?' becomes another.
Research suggests that as pain persists, the brain activity linked to it shifts away from the sensory regions and toward the regions that handle emotion, learning and the sense of self. The back still hurts in the back. What changes is how the brain is representing that hurt.
The clearest demonstration comes from a study by Hashmi and colleagues, who followed people with a new episode of back pain for about a year and scanned them repeatedly. At the start, pain-related activity sat in the expected sensory regions. In the people whose pain recovered, it stayed there and faded. In the people whose pain became chronic, the activity migrated toward the medial prefrontal cortex and amygdala, circuits involved in emotion, threat learning and self-referential thought. The pain had not moved. Its representation had.
An earlier study from the same group looked at the default mode network, the set of regions most active when the mind is at rest and turned inward. In healthy people this network quiets when attention goes to a task. In a small group of people with chronic back pain, several of its regions failed to quiet normally during a simple attention task. The authors' interpretation was that chronic pain had changed the baseline operation of the network that constructs the resting self. This was a small study, and findings in this area have been mixed since, so it is best read as a clue rather than a settled fact.
These shifts are learning. The same neuroplasticity that lets a pianist's fingers get faster lets a pain system get more sensitive, more anticipatory and more tied to emotion. Amplification in the spinal cord and brainstem, described on our central sensitization page, is one part of that learning. The cortical shift described here is another. And because it is learning, research suggests that at least some of it can be revised, which is the premise behind graded sensorimotor retraining and related approaches.
The regions in the table above were not built for pain. The insula builds the felt sense of the body from the inside: heartbeat, breath, warmth, tension, the background hum that tells you a body is here and it is yours. The medial prefrontal cortex and posterior cingulate, the core of the default mode network, are where self-referential thought happens: what I am like, what I can do, what has happened to me. Pain is produced in the same places where 'me' is produced.
This is the thread that runs through the whole therapeutic neuroscience series. Our sense of self is constructed, moment to moment, from predictions about the body and the story we hold about it. In chronic pain the construction changes. The body map that the insula and somatosensory cortex maintain starts to read as damaged, so that an ordinary stretch arrives pre-labeled as a threat. The narrative that the default mode network keeps running starts to center on the pain: someone who used to be a runner becomes someone whose knee goes. Research suggests this is not just a figure of speech. The shift toward emotional and self-referential circuits described above is where the activity now sits.
A self organized around a body that feels unsafe behaves accordingly. It braces before moving, scans for symptoms, declines invitations, plans the day around the pain. Each of these is reasonable given the forecast. Each also feeds the forecaster more evidence that the body is in trouble, and keeps the loop closed.
That is why treatment at the level of tissue alone can miss. If pain has moved into the networks that build the self, then the treatments most likely to help are the ones that give the self-model new evidence: education that changes what a sensation means, graded movement that proves a feared action safe, attention training that changes how the body is read, and embodiment work in virtual reality that lets the brain watch its own body move further than the forecast allowed. None of these approaches ignores the body. They aim at the place where the body is interpreted.
First, your pain is real. It is produced by your brain, which is true of every pain anyone has ever had, and that makes it neither less real nor less physical. If someone has told you that a brain-based explanation means the pain is 'in your head', they have the neuroscience backwards. The brain is where all pain is, including the pain they would feel if they slammed a car door on their hand.
Second, it changes the target. If pain is a protective conclusion rather than a damage readout, then changing what the brain concludes is a legitimate treatment goal, not a consolation prize. Pain neuroscience education, pain reprocessing therapy, graded exposure to feared movement and VR embodiment training all work at this level, and they are designed to be used alongside medical care, not instead of it. The goal is less pain and more function. Many people improve substantially with this kind of work, and some do not, and anyone who promises otherwise is selling something.
Karuna's program is therapeutic neuroscience put into a 12-week structure. Patients work from home with a provided headset, doing short daily VR embodiment exercises designed to create safe prediction errors around movement and retrain the brain's body map, with a weekly one-on-one session with a pain coach and clinician review before the program starts and oversight throughout. Karuna's outcome data, with its limits spelled out, is on the how it works page. The free consultation exists to work out whether the program fits your pain, including telling you if it does not. Karuna takes workers' compensation referrals, and the care team confirms coverage case by case.
A brain that produces pain is still a brain that responds to real danger, and some signals should not be reinterpreted at home. See a doctor promptly, or call emergency services, for any of the following:
For clinicians, one practical note. The sentence 'pain is produced by the brain' is accurate and, said cold, is heard by most patients as 'you think I am making it up'. The nociceptor fact in the first section of this page is the bridge: start with the organ that feels nothing and assembles everything, and the rest of the explanation tends to land.
A headache is pain produced by the brain in response to signals from pain-sensitive structures around it: the meninges, the blood vessels inside and outside the skull, the scalp and neck muscles, the sinuses and the nerves that supply them. The brain tissue itself is silent. In migraine, the trigeminal nerve and the blood vessels it supplies are thought to be the main source of the signal.
Tension-type headache and migraine that recur for months are discussed on our chronic headache and migraine page. A headache that is sudden, severe and unlike any you have had before needs same-day medical review.
'Controls' is a better word than 'feels', because the brain does turn pain up and down. The prefrontal cortex, the anterior cingulate and a brainstem region called the periaqueductal gray drive descending pathways that reach back down to the spinal cord and dampen or amplify incoming signals before they ever reach consciousness. This is the mechanism behind the soldier who feels nothing until the fighting stops, and behind much of the placebo effect.
In chronic pain this descending control system is thought to work less well, which is part of why ordinary signals get through at full volume. Treatments that change expectation, attention and perceived safety are, in part, attempts to re-engage it.
The pain matrix is an older label for the group of brain regions that reliably activate when something hurts: the thalamus, S1 and S2, the insula, the anterior cingulate and parts of the prefrontal cortex. It grew out of Melzack's neuromatrix theory, which proposed that pain is the output of a distributed body-self network rather than a direct response to injury.
The regions are real. The idea that they form a pain-specific system has not held up, because the same regions respond to any attention-demanding event. Most researchers now talk about a dynamic pain connectome, a shifting pattern of interaction among several networks, rather than a fixed matrix.
Yes. Phantom limb pain is the plainest case: real, often severe pain in a body part that no longer exists. Pain also persists in many people long after an injury has healed, and it can appear in people whose tests show nothing that would explain it. In each case the brain is producing a protective experience that the current state of the tissue does not call for.
Pain without a clear injury is still real pain, produced by the same machinery as pain with one. It also still needs medical assessment first, because 'nothing found' and 'nothing there' are not the same, and some serious conditions take time to show.
Research suggests chronic pain changes the brain, but 'damage' is the wrong word. Imaging studies have found differences in activity, connectivity and, in some studies, gray matter density in people with long-standing pain, and some of these differences lessen when pain improves. That pattern fits learning and adaptation rather than injury.
None of these findings can diagnose or measure pain in an individual. They are group averages, and they describe a nervous system that has adapted to a long threat, which is also a nervous system that can adapt again.
No. Every pain is brain-based, including the pain of a fracture, so the explanation does not sort pain into physical and psychological. What it does say is that the brain weighs many inputs, including tissue signals, expectation, stress, memory and attention, and that in chronic pain the weighting has shifted toward protection.
Thoughts and emotions are inputs to that process, in the same way that a cut is an input. Treating them is treating a real mechanism of real pain, and it works best alongside, not instead of, proper medical care.
Talk with our care team about your pain, your history, and whether KVET™ is right for you. Free, and from the comfort of home.