Proprioception and Interoception: What the Body Knows About Itself

Proprioception lets you find your hand without looking

Proprioception works with your eyes closed, and you can prove it in ten seconds. Put this page down, close your eyes, and put your left hand somewhere behind your back without looking. Then bring your right index finger up and touch the tip of your left thumb.

Most people manage it, or come close, on the first try. Nothing about that is obvious. Your eyes were shut, so vision told you nothing. Your left hand was out of sight the entire time, and you had not planned where to put it. Yet somewhere in your head there was an accurate answer to the question of where that thumb had ended up. The answer was accurate to within a centimetre or two, updated continuously, and available instantly.

You were reading an internal report. Your body was measuring itself and sending the results upward, and your brain was assembling those results into a working model of a hand it could not see. Proprioception builds that model fresh, moment by moment, from live measurements taken inside the tissue.

Interoception is the second report

Right now you have some sense of whether you are hungry, whether your chest feels tight, whether your gut is comfortable, whether you are short of breath. That report is interoception. It comes from measurements taken inside the organs rather than inside the muscles, traveling up a different road.

Two reports, then. Proprioception describes the body as a shape moving through space. Interoception describes the body as a set of conditions being maintained. Neither made the list of five senses you learned as a child. Between them they carry almost everything your nervous system knows about the thing it is responsible for.

Now consider what happens when a report is wrong rather than missing. Think of the last time you walked down a staircase in the dark, reached your foot for a step that was not there, and felt your whole body lurch. For a fraction of a second the model said one thing and the world said another. That lurch is what a bad report feels like from inside, and this page is about the many quieter versions of it.

Where the names proprioception and interoception come from

Both senses have proper names. Proprioception was named by Charles Sherrington in 1907, and interoception took its modern form from A. D. Craig in 2002. Both names were coined by people who were looking for something else.

Sherrington named proprioception in 1907

Sherrington was an English physiologist who spent his working life on reflexes and who later shared the 1932 Nobel Prize in Physiology or Medicine for that work. He was trying to answer a plain question. A reflex is an automatic response, and a response needs a trigger, so what was triggering the reflexes that keep an animal standing up? Nothing outside the animal was touching it. Nothing was making a sound. The trigger had to be coming from inside.

Sherrington sorted the body’s sense organs by where their information comes from. Some face outward, toward light and sound and touch. Some sit in the gut. And a third group, he argued, lies deep in muscle, tendon, and joint, and reports on nothing except the body itself. In 1907 he published a paper naming that third group the proprioceptive system, from the Latin for one’s own. These were the organs by which the body senses itself, and their reflex traffic was continuous rather than occasional.

That is proprioception: the sense of limb position, limb movement, muscular force, and effort, generated from inside the moving parts.

Craig defined interoception in 2002

The second name is much more recent in its modern form. A. D. Craig, a neuroanatomist who spent decades tracing where sensory fibers actually go, was following a set of small nerve fibers that carry temperature and pain. Conventional teaching sent them to the touch areas of the cortex. Craig found that they did not go there. They ran to a different destination.

There they joined traffic arriving from the gut, the heart, the vessels, the muscles, and the skin. The whole convergent stream described one thing, which was the physiological condition of the body. In 2002 he set the argument out in a review asking how it is that we have feelings from our bodies at all. His proposal was that this system builds a representation of the material self.

That is interoception: the sense of the body’s internal state, covering temperature, effort, hunger, thirst, air hunger, gut sensation, and the felt tone of the tissues.

Proprioception tells the brain where the body is. Interoception tells the brain how the body is. Everything else on this page follows from the fact that both are reports, and reports can be wrong.

Input quality names the fidelity of the body’s self-report

Input quality is the fidelity of the information the nervous system holds about its own body. Proprioception and interoception are the two main channels that carry that information, and tone is regulated from what they report.

This library treats tone as the integrated, coupled organization the nervous system maintains across the whole body, together with its capacity to move where the moment demands and return to balance afterward. Health is the width of that range. That is the pillar this library sits under, and input quality is the part of tone that supplies its information.

Input Quality is the fidelity of the information the system holds about itself. Not how loudly the body responds; that is gain. Not which value it defends; that is the set point. Input quality asks a narrower question: how good is the report, and do the channels agree?

A regulator can only be as good as its information. Everything else tone does runs downstream of the report.

Input quality has to be named separately because it is invisible in the output. A system running on a degraded report does not look like a system with a sensor problem. It looks like a system with a control problem. The muscle guards, the balance wobbles, the pain persists, the gut complains, and every one of those is a plausible local fault. The fault may instead be that the information arriving was never good enough to regulate on.

One canonical point belongs here, because it is the most misunderstood idea in this whole model. In the Unified Model of Tone, specificity means correspondence, not force. An input is specific when it matches what the system is actually able to receive at that moment, in that tissue, in that state.

How hard the input is delivered is a completely separate axis, running from the lightest contact through to surgery. A large force with no correspondence is not specific. A very small one with exact correspondence is. Input quality is where that distinction does its work, because correspondence is a statement about information. Section 13 returns to it with the measurements behind it.

The four receptor families that build proprioception

Proprioception runs on four families of receptors, and each family measures a different physical quantity. Muscle spindles measure length, Golgi tendon organs measure force, joint receptors cover the extremes of range, and skin measures surface strain.

A sensory receptor is a nerve ending shaped so that some physical event, a stretch, a squeeze, a temperature, a chemical, changes the electrical state of the ending and makes it fire. Firing is the only language a nerve has. A receptor is a translator that turns a physical quantity into a firing rate.

The muscle spindle

Buried inside your muscles are small bundles of specialised fibers wrapped in a capsule, lying parallel to the working fibers. Stretch the muscle and you stretch the spindle, and the sensory endings coiled around it fire faster. The spindle is a length meter, and because its firing changes as the length changes, it is also a speed meter.

Uwe Proske and Simon Gandevia, two physiologists who have spent careers on this question, reviewed the whole field in a survey of the proprioceptive senses. Their conclusion matters for everything below. Position and movement are not read off any single receptor. They are computed from populations of afferents, and the result is then referred to a central map of the body to work out where the limb sits in space. There is a model in the middle, and the receptors feed it.

The Golgi tendon organ

Where muscle joins tendon sit encapsulated endings woven among the collagen strands. Pull on the tendon and the strands squeeze the ending, and it fires. This one is a force meter rather than a length meter. A physiologist named Jami gathered the evidence in a review of tendon organ properties and their central actions, which established how faithfully these endings track the force a muscle is actually producing.

Joint receptors

The capsules and ligaments around joints carry their own endings. The intuition is that these are the position sensors, since they sit at the joint. The evidence says otherwise. Proske and Gandevia, revisiting the question in a shorter review of the kinaesthetic senses, concluded that joint receptors are not major contributors at most joints. They contribute at the extremes of range. The heavy lifting is done elsewhere.

Skin

Skin is a proprioceptive organ. Bend a finger and the skin over the joint stretches in a specific pattern, and that pattern is itself information about the angle. Edin and Johansson set out to test whether the brain uses it. They anaesthetised part of a finger, so the joint itself reported nothing, and then deformed the sentient skin above it to imitate the strain of a bend.

In their report on skin strain patterns, every subject felt a movement that had not happened, and mirrored it with the opposite hand. When the skin strain was damped with a ring, subjects failed to perceive real movements that did take place.

Four instruments, then, measuring length, speed, force, angle, and surface deformation, all at once, everywhere in the body, continuously. The proprioceptive report your brain reads is what those measurements add up to.

The two pathways that carry interoception to the brain

Interoception runs on receptors in the walls of the organs, and its traffic travels two main roads: the vagus nerve and a dedicated spinal pathway ending in the insula.

The instruments sit in the walls of the gut, the airways, the heart, the great vessels, the liver, and the tissues themselves. Some are stretch receptors that fire when a hollow organ fills. Some are chemical sensors that fire when acidity or oxygen or a signaling molecule changes. Some are temperature sensors. Some report on inflammation. Every one of them is measuring a condition rather than a position.

The vagus nerve carries the visceral report

A large share of interoceptive traffic travels on the vagus nerve, which most people have heard of as a calming nerve and which is better understood as a sensory cable. Berthoud and Neuhuber gathered what is known about its sensory arm in a review of the functional and chemical anatomy of the afferent vagal system. The tracing work they surveyed follows fibers to the oesophagus, the airways, the heart, and the aorta.

Through the abdominal branches those fibers reach the entire gastrointestinal tract, the liver, the biliary system, and the pancreas. The recording studies they surveyed, run mostly in rats and cats, identified mechanical, chemical, temperature, and osmotic sensors along that route. This library has a separate page on that nerve, so the point here is narrow. The vagus is one of the main roads on which the body’s internal report travels.

The spinal pathway ends in the insula

The rest of the interoceptive traffic travels up the spinal cord in a thin layer at the very back of the cord’s grey matter, the layer Craig had been tracing. Its destination is the insula, a patch of cortex folded away inside the side of the brain.

Craig and his colleagues tested this directly by cooling people’s skin inside a brain scanner while measuring where activity followed. In the resulting study of thermosensory activation, graded cooling produced activity in the insula rather than in the classical touch cortex. How intense the cold felt tracked activity further forward in the same region.

So there is a road for where the body is, and a road for how the body is, and they arrive in different places. Both are afferent. Both are the raw material for regulation. And both can carry a report that no longer matches the body.

The neck is the body’s densest proprioceptive source

The neck carries more proprioceptive instrumentation per gram than almost any other region of the body, and the quality of its report shapes balance, gaze, and the sense of where the world is.

The general principle came from an unexpected direction. Peck and colleagues were studying pairs of muscles that cross the same joint, one large and one small, working in parallel. Mechanically, the small one contributes almost nothing. So why is it there? They counted spindles in both members of such pairs, in humans and in dogs.

In their comparison of spindle concentrations, every pair examined showed the same thing. The small muscle carried a significantly higher spindle density than its large partner. Their proposal was that these small muscles exist to sense rather than to pull.

Spindle density peaks between skull and spine

The neck is full of them. Kulkarni and colleagues counted spindles in the tiny suboccipital muscles that sit between the skull and the top two vertebrae. The tissue came from human foetuses, so the figures are not adult figures and the ranking is what carries.

In their quantitative study of human fetal suboccipital muscle, spindle density reached 242 per gram in the inferior oblique, 190 in the superior oblique, and 98 in rectus capitis posterior. No tendon organs were found at all. These muscles are almost purely sensory tissue with a small mechanical side job.

The pattern holds further down. Boyd-Clark and colleagues examined the deep neck muscles from C5 to C7 at autopsy, across sixteen people aged four to seventy-seven. Their study of spindle distribution and density found that longus colli carries a high spindle density, clustered rather than scattered. The density did not change with age. Jing-Xia Liu and colleagues went further into the tissue itself, and in a morphological study of spindles in the deep neck muscles found structural features unusual among human muscles.

Why does the density sit there? Because the neck carries the head, and the head carries the eyes and the balance organs. To know what the eyes and the inner ear are reporting about the world, the brain must first know exactly where the head is sitting on the body. Neck proprioception supplies that term.

Three experiments that change the neck signal

Three experiments show how much rides on that term. A group led by de Jong, testing what neck receptors contribute to balance in people, injected local anaesthetic into one side of the neck and recorded what followed. In their report on ataxia and nystagmus induced by neck injection, silencing that input produced a broad, staggering gait and a drop in muscle tone on the injected side.

People felt as though they were falling toward that side. Nystagmus appeared in the animals tested but not in the human subjects, so the human finding is ataxia rather than disordered eye movement. Nothing had touched the inner ear or the brain.

Biguer and colleagues went the other way and added a false neck signal instead of removing a true one. They vibrated the muscles on one side of the neck. In their study of neck vibration and the representation of visual space, a small target viewed in the dark appeared to shift sideways, and people pointed accordingly.

Marco Bove and colleagues ran the same manipulation on blindfolded people marching in place. In their study of neck vibration during stepping, the whole body rotated steadily away from the vibrated side, at about one degree per second, without anyone noticing they had turned.

Change the neck signal and you change where the world seems to be. That is how much of the brain’s spatial arithmetic passes through this one region.

Neck injury disturbs balance, gaze, and head positioning

The clinical consequence is direct. Julia Treleaven, a physiotherapist and researcher on neck disorders, reviewed how these connections behave when the neck is injured. Her survey of sensorimotor disturbances in neck disorders argued that altered cervical input changes the integration, timing, and tuning of sensorimotor control. Measurable disturbances in head repositioning, eye movement control, and standing balance follow, along with reported dizziness and unsteadiness.

The brain believes the proprioceptive report, not the world

When proprioception and the world disagree, the brain acts on proprioception. There is a clean way to test this, and it has been run many times.

Tendon vibration injects a false report

The trick is a vibrator. Press a small vibrating device against a tendon and the spindles inside that muscle fire as though the muscle were being stretched, whether or not it is. You have injected a false measurement into a real channel.

Goodwin, McCloskey, and Matthews, physiologists working on where the sense of movement comes from, were the first to run this systematically. Their aim was to settle a long argument about whether muscle receptors contribute to conscious position sense at all. In their study of vibration-induced illusions of movement, vibrating a tendon produced a vivid felt sensation of the joint rotating, in the direction that would have stretched the vibrated muscle. The limb had not moved. The person was certain it had.

Roll and Vedel then did the harder version, recording directly from single sensory fibers in awake human volunteers while running the same illusion. In their combined vibration and microneurography study, the illusory speed grew as vibration frequency rose from 10 Hz to about 70 or 80 Hz, then fell away above that. Recording from the spindles showed why: the primary endings were locking onto the vibration cycle. The false report and the false perception had the same shape.

A psychologist named Lackner, who studied spatial orientation, then pushed the method to its limit. He had people hold a part of their own body while a nearby tendon was vibrated. In his report on proprioceptive influences on body shape, the perceived shape and orientation of the body changed within seconds. If the fingers were pinching the nose, the nose seemed to grow. The body map is a live construction, rebuilt continuously from the proprioceptive input it is given.

Vision can overwrite the hand

The same lesson arrives from the other direction. Botvinick and Cohen wanted to know what happens when vision and touch disagree about where a hand is. They hid a person’s hand, put a rubber hand in view, and stroked both together. In their brief report on rubber hands, people came to feel the touch in the rubber hand and to locate their own hand toward it.

The nervous system regulates its model of the body, not the body directly. When the report is false, the regulation is faithful to the falsehood.

What happens when proprioception is lost entirely

Remove proprioception and movement collapses, even when the motor system is intact. A small group of people who lost the sense entirely shows what it was doing all along.

An illness can destroy the large sensory nerve fibers while sparing the small ones and sparing the motor nerves entirely. The person keeps pain, temperature, and muscular fatigue. The muscles work normally. What disappears is light touch and proprioception below the neck.

Jonathan Cole and E. M. Sedgwick studied one such man over many years, asking what remains of the sense of effort and of movement when the peripheral report is gone. Their findings, published as a study of force and movement perception without large sensory afferents, are precise.

Allowed to move his forearm and to look, he could discriminate weights nearly as well as controls. With his eyes shut, he could only tell 200 grams from 400. A crude sense of effort survived. A limited motor memory survived, enough to hold a posture or repeat a simple movement. No novel movement was possible without watching.

John Rothwell and colleagues at Queen Square had earlier examined manual motor performance in a deafferented man, documenting the same paradox in the hand. The motor system was intact. The instructions could be issued. Without the returning report, the movements decayed.

Partial loss is the common case

Most people who lose proprioception lose it partially rather than totally, and the commonest cause is a peripheral neuropathy. Simoneau and colleagues asked what that partial loss costs in standing balance. They compared seventeen people with diabetes and significant sensory neuropathy against seventeen with diabetes and no neuropathy and seventeen with neither, and measured sway on a force platform.

In their study of postural instability in diabetic sensory neuropathy, the neuropathy group showed between 66 and 117 per cent more instability than controls depending on the condition tested. Instability tracked the sensory measures, not diabetes itself. The deficit was largest when vision or vestibular cues were also removed.

The loss became severe only when the other channels were taken away too, and that arrangement is the subject of the next two sections.

How input quality fails

Input quality fails in five distinguishable ways: the report can be degraded, noisy, mismatched, absent, or never accurately established. Each produces a different clinical picture, and each calls for a different response.

1. Degraded

The channel still works and carries less. Receptors thin out, conduction slows, discrimination coarsens. Simoneau’s neuropathy group is the sharp version. The slow version is ordinary ageing. Daniel Goble and colleagues reviewed what happens to proprioception across the lifespan in a survey of proprioceptive sensibility in the elderly. They describe measurable degeneration, the functional consequences that follow, and the adaptations the nervous system makes to compensate. Degradation is quiet. Nobody feels their spindles reporting less.

2. Noisy

The signal arrives with the message buried in scatter. Noise is different from loss, because a noisy channel still demands attention while providing less usable information. Martin Paulus and Murray Stein reviewed the interoceptive case in an account of interoception in anxiety and depression, and their model turns on exactly this.

They describe increased but noisy interoceptive input, amplified by belief, with the system then leaning harder on top-down expectation because the incoming signal predicts so poorly. Noise in equals confidence out, and the confidence is misplaced.

3. Mismatched

Every channel works and they contradict each other. The classic example is seasickness. Reason, a psychologist studying why motion makes people ill, set out a neural mismatch model of motion sickness adaptation. In it, the sickness comes from a discrepancy between the motion signals arriving and the ones the nervous system expected from experience.

Below deck on a boat, the inner ear reports movement and the eyes report a still cabin. Nothing is broken. Both reports are accurate. The system cannot hold both, and the result is nausea, sweating, and misery.

Mismatch is the failure mode clinicians miss most often, because every individual test comes back normal.

4. Absent

The channel is gone. Cole’s patient is the pure case, and a dense nerve injury, a spinal cord lesion, or a limb amputation are the common ones. What matters here is what the brain does with the space.

Merzenich and colleagues, studying how fixed the cortical body map really is, amputated a digit in adult monkeys and remapped the hand area afterward. In their report on cortical map changes following digit amputation, the territory that had represented the missing finger was taken over by its neighbors. The map does not leave a gap. It closes over.

That closure has consequences. Flor and colleagues measured the amount of cortical reorganisation in people who had lost an arm and compared it against how much phantom pain they reported. In their study of phantom limb pain as a correlate of cortical reorganisation, the relationship was very strong, and it held for painful phantom sensations rather than for non-painful ones.

5. Never established

The map was never accurately built. Development lays down the body’s self-representation through movement and feedback, and when either is disturbed early, the representation that results is imprecise from the start. Jason Wingert and colleagues tested joint position sense and movement detection in people with cerebral palsy against age-matched controls.

In their study of joint position sense and kinaesthesia in cerebral palsy, there were no group differences when the limb was visible. With vision removed, errors rose significantly in almost every limb, and the errors were systematically biased in one direction.

With vision available, performance looked normal. The deficit only appeared when the system had to rely on its own internal report. A great many people are running on exactly that arrangement without knowing it.

None of the five failures is felt

Notice what none of these five produce. None produces a hole in experience. You do not feel a degraded spindle population, a noisy visceral channel, or a body map that closed over an absence. The report keeps arriving, complete and confident. The system keeps regulating on it. Whatever the report says is, for the nervous system, simply what is true.

Sensory reweighting, and what it costs

A nervous system with a degraded channel shifts its trust to the other channels. The shift is called sensory reweighting, and it explains why serious input quality problems can hide for years.

Peterka, a scientist studying how humans stay upright, built the definitive demonstration. He tilted the floor and the visual surround under people at many different amplitudes while measuring how the body swayed, and compared healthy subjects against people who had lost vestibular function entirely.

In his analysis of sensorimotor integration in postural control, healthy subjects changed their behavior as the disturbance grew, relying progressively more on vestibular information as the other cues became unreliable. Those without a vestibular system could not perform that reweighting, and their responses stayed rigidly linear.

Kelvin Oie and colleagues then showed that the reweighting happens across several channels at once. In their study of simultaneous re-weighting of vision and touch, changing the reliability of one channel altered how heavily the system used the other. The channels behave as a set of weights that sum to a decision, and the system retunes those weights continuously.

Fingertip touch steadies balance through information, not force

How little it takes to change those weights is startling. John Jeka and Lackner asked people to stand in a difficult heel-to-toe position and lightly touch a rigid bar with one fingertip, using less force than the weight of a small apple.

In their study of fingertip contact and postural control, that touch reduced sway as effectively as being allowed to see the room. Under a tenth of a kilogram of force cannot hold anyone up. The touch was supplying information, and the timing showed it: the sway reduction preceded the contact forces rather than following them.

The same principle can be run deliberately in the other direction. Attila Priplata and colleagues asked whether a tiny amount of added randomness could push weak signals over threshold. They put vibrating insoles under the feet of people with diabetic neuropathy and people who had had a stroke. The vibration was set below the level anyone could feel. In their study of noise-enhanced balance control, every sway measure improved, and the improvement was largest in those who had swayed most to begin with.

So reweighting works, and it is not free. Leaning on vision means falling apart in the dark. Leaning on a hand on the wall means a corridor without walls becomes a hazard. Leaning on conscious attention means the moment attention is elsewhere, the compensation drops. A person who has quietly reweighted for a decade looks fine until the day the substitute channel is removed.

Chronic pain degrades the brain’s map of the body

Input quality degrades from the top down as well as from the receptors up, because chronic pain coarsens the brain’s map of the painful region.

Lorimer Moseley, a pain scientist, asked people with chronic back pain to do something that sounds trivial. Lying face down, they were asked to say where on their own back they were being touched, and to draw the outline of their back as they felt it. In his report on distorted body image and tactile dysfunction in chronic back pain, they were worse at locating touch than controls. Their drawings of their own backs were distorted, with the painful region often missing or blurred.

Whether that pattern holds across conditions has been tested. Mark Catley and colleagues pooled sixteen studies covering five chronic pain conditions in a systematic review and meta-analysis of tactile acuity in chronic pain. Acuity was reduced in arthritis, complex regional pain syndrome, and chronic low back pain. It was not reduced in burning mouth syndrome. The authors were explicit that the results were somewhat inconsistent and that every included study carried a high risk of bias.

That is a mixed result, and it is exactly what the tone model expects. An input meets a state, and the state is different in every person. A single condition label sitting over dozens of different tonal organizations should not produce a uniform sensory finding, and it does not. Read the record expecting one condition to do one thing to everybody and it looks noisy. Read it as input meeting tone and it looks the way it should.

The direction of the arrow matters too. Degraded input coarsens the map; a coarsened map makes the next input harder to place. The loop runs both ways, which is why how long a problem has been running changes what you are dealing with.

How input quality is measured, and what measurement cannot do

Input quality is measurable, and the instruments are cruder than the sense they measure. Four tests carry most of the clinical weight.

Joint position sense error

Blindfold someone, move a limb or the head to a target, return it to neutral, then ask them to find the target again. The distance they miss by is the error. Revel and colleagues built the standard head version and tested it on thirty people with neck pain against thirty healthy controls. Their study of cervicocephalic kinaesthetic sensibility found the patients significantly worse at relocating the head.

The follow-up literature is not tidy. de Vries and colleagues assembled fourteen studies in a systematic review of joint position sense error in neck pain. The findings differed by group, with traumatic neck pain more consistently associated with error than non-traumatic. Jia Han and colleagues compared the available methods in a critical review of proprioception testing, and concluded that the three main approaches measure related but distinguishable things. There is no single number called proprioception.

Posturography

Stand on a plate that measures where your weight falls, and record how that point wanders. Peterka’s work is the sophisticated version. The clinical version compares standing with eyes open against eyes closed, on firm ground against foam, which is a direct test of how much a person is leaning on vision.

Guskiewicz reviewed the concussion application in an assessment of postural stability testing after concussion, finding measurable deficits lasting several days. He also noted that symptom severity, cognitive scores, and postural stability were often not related to each other, which is a caution against reading any one of them as the whole picture.

Two-point discrimination

Touch the skin with two points and find the smallest separation at which two are still felt as two. Flavia Mancini and colleagues mapped this across the whole body in a systematic survey of spatial acuity for pain and touch. They confirmed the fingertip as the sharpest region, and showed that acuity for pain follows its own gradients. The test is cheap and it is coarse, and it is influenced by attention and instruction.

Heartbeat detection

For interoception the standard task is Schandry’s, published in a 1981 paper on heartbeat perception and emotional experience: sit still and count your own heartbeats without taking a pulse. Sarah Garfinkel and colleagues then showed that this measures only one of three separable things. In their study distinguishing interoceptive accuracy from awareness, objective performance, self-reported sensitivity, and metacognitive insight into one’s own accuracy were all distinct in a sample of eighty.

The limits run deeper than that. Giorgia Zamariola and colleagues examined 572 participants in an analysis of why heartbeat counting scores are problematic. Over 95 per cent of the score reflected under-reporting. The correlation between actual and reported beats was only 0.16 overall, and the score was structurally bound to a person’s heart rate. Christopher Ring and Jasper Brener reached a related conclusion in a comparison of methods showing that heartbeat counting is unrelated to heartbeat detection.

The most widely used measure of interoceptive accuracy is contested, and any claim resting on it is only as strong as the instrument. That is a limit on the instruments rather than on the sense.

Why a precise input works, and why force is a separate question

Because regulation runs on the body’s self-report, delivering a more accurate afferent signal is a real intervention. The mechanism is unglamorous, and one experiment establishes the first link in the chain.

Pickar and Wheeler set out to answer a mechanical question. Do the proprioceptors in the deep spinal muscles actually respond to a brief mechanical load applied to a vertebra? They worked in anaesthetised cats, isolated a dorsal root so they could record from single sensory fibers, and applied loads shaped in force and time like a manual thrust. In their recordings from muscle proprioceptors under manipulative-like loads, muscle spindles and tendon organs in the paraspinal tissue responded to the load.

The result was recorded in the lumbar spine of anaesthetised cats, and it establishes one link in a chain: paraspinal proprioceptors do fire in response to a brief load shaped like a manual thrust. Whether the same holds in a waking human spine has not been recorded.

What follows from that link is this model’s reading rather than the study’s finding. On that reading, a brief, well-placed mechanical input to spinal tissue is an information event. It delivers a burst of fresh afferent traffic into a channel thick with receptors. What the system then does with that traffic is the part nobody has measured.

Specificity is correspondence

This is where the model’s definition of specificity earns its keep. Specificity is correspondence between the input and what the system can actually receive. The relevant question is whether the input matches the state of that tissue at that moment, well enough for the system to register it as information rather than as noise or as threat. How hard is a separate question.

Magnitude is a separate axis entirely, and it runs from a contact you could barely feel through to surgery. A large force with no correspondence is not specific, and it can be perfectly forceful while telling the system nothing it can use. A very small one with exact correspondence is specific. Jeka’s fingertip is the proof that lives outside any clinical setting: under a tenth of a kilogram, carrying no mechanical support whatsoever, changed standing balance as much as vision did.

The information lives in the correspondence between input and state. Force is a separate axis.

The same logic explains why the most useful place to work is often not the painful place. What follows is clinical observation rather than established mechanism. Tension patterns layer, and a presenting pattern frequently sits on top of a deeper one, so releasing the deeper one is often what moves the presenting one.

There is rarely a single lynchpin. There are several points of critical tension carrying different amounts of potential, and the art is finding the one that yields the largest change for the smallest, best-matched input at that moment.

The conditions dominated by input quality

The conditions where input quality carries most of the weight share one signature: the proprioceptive or interoceptive report degrades or contradicts itself while every local tissue tests normal.

Every condition in this library combines several parts of tone. These are the conditions where the quality of the body’s self-report does most of the work.

Balance and coordination is the definitive case. Standing upright is a computation performed on three streams of information, and the output degrades whenever any stream does. Peterka’s reweighting result and Simoneau’s neuropathy result are both, at bottom, statements about balance.

Vertigo and dizziness is largely a mismatch problem. When the inner ear, the eyes, and the neck disagree about which way the head is moving, the experience is spinning, nausea, and fear, and every individual test can be normal.

Neck pain belongs here for the reason section 06 gave. One of the densest sources of position information in the body is also the region where guarding, injury, and long-held posture most reliably change the quality of the report. That is why disturbances in head repositioning and balance travel with it.

Brain injury damages the integration rather than the receptors. The report arrives and cannot be assembled, which is why postural stability testing detects deficits that a symptom questionnaire misses.

Cerebral palsy is the never-established case, and Wingert’s vision-dependent result is the reason. Any account that treats it purely as a motor condition is describing half of it.

The senses as a topic sits directly on input quality, since proprioception and interoception are the two members of the set that were left off the list.

Pain and tinnitus both involve a map that has reorganised around a change in input, which is the Merzenich and Flor territory, and both are combinations rather than pure cases.

What actually improves proprioception and interoception

Input quality can be trained, and the trial record is uneven in exactly the way the tone model predicts, because every programme is an input meeting a different tone.

Balance and vestibular training have randomised evidence

Balance training improves balance, and it prevents injuries. Timothy McGuine and James Keene randomised 765 high school soccer and basketball players to a balance training programme or to standard conditioning. In their trial of balance training and ankle sprain risk, sprain rates fell from 9.9 per cent to 6.1 per cent, cutting the risk by more than a third. In previously uninjured athletes the difference did not reach significance, which is itself informative about who has a report worth repairing.

In older adults the picture is more modest. Tracey Howe and colleagues assessed the Cochrane evidence in a review of exercise for improving balance in older people. Gait, balance, coordination, and functional training were moderately effective immediately after the intervention, on weak evidence. They found insufficient evidence to conclude anything about vibration plates or computerised balance programmes. The effective programmes tended to run three times a week for three months with dynamic standing work.

Where a channel has actually failed, retraining the integration works better. Michelle McDonnell and Susan Hillier assessed the trials in a Cochrane review of vestibular rehabilitation. They found moderate to strong evidence of benefit for unilateral vestibular loss. The odds ratio for improvement in dizziness was 2.67, across four studies and 565 participants. One exception held: for benign paroxysmal positional vertigo, repositioning manoeuvres beat exercise in the short term.

Training aimed at the neck specifically has been tested. Konstantin Beinert and Wolfgang Taube randomised thirty-four people with subclinical neck pain to five weeks of balance training or to staying active. In their study of balance training and cervical sensorimotor function, head repositioning accuracy improved and pain fell in the training group only. The correlation between the two changes was weak.

Map retraining and interoceptive training

Retraining the map directly is promising and unproven. Samuel Kälin and colleagues reviewed six randomised trials of sensory discrimination training for chronic low back pain in a systematic review of the approach. Most studies found statistically significant improvement in pain and function. Only two reached clinical relevance, dosage varied widely, and the authors called for a properly powered trial before anyone concludes anything.

On the interoceptive side, one trial stands out. Lisa Quadt and colleagues randomised 121 autistic adults to interoceptive training, which pairs heartbeat detection with feedback on how accurate they were, or to an active control. In the ADIE trial, trait anxiety was lower in the training group at three months, with 31 per cent meeting recovery criteria against 16 per cent in the control group. Attrition was heavy, with only half the sample providing data at that point.

The null that matters

Sahib Khalsa and colleagues tested the widespread assumption that meditation sharpens interoception. They compared fifteen experienced meditators against fifteen matched non-meditators, using drug infusions to raise cardiac arousal so the test was not limited to resting conditions. In their finding that meditation is not associated with improved interoceptive awareness of the heartbeat, there was no evidence of better detection or accuracy at any dose. Meditators did localize cardiac sensations very differently, across the chest, abdomen, neck, back, and head.

The null is informative. A practice widely believed to sharpen interoception did not improve the one thing measured, on a measure that is itself contested. What changed was the organization of the experience rather than the acuity of the sensor, which is a statement about tone rather than about the receptor.

How Input Quality relates to the other dimensions

Input quality borders the other dimensions of tone, and each border is a distinct question. The ontology stays useful only if the lines are kept sharp.

Set point, gain, oscillation, coupling

Set point is the value the system is defending. Input quality is what the system knows about where it currently sits. This is the closest neighbor of all, because a degraded sensor and a drifted target produce identical behavior from outside. If the report is wrong, a correctly defended value lands in the wrong place, and no test of the target will reveal it.

Gain is how loudly the system responds relative to the input. Input quality is about the input; gain is about the response to it. A faint, accurate signal met with enormous gain and a loud, false signal met with normal gain look similar in the output and are opposite problems.

Oscillation is the rhythm and range a single system moves through. Movement generates afferent information, so the two dimensions feed each other, but the border holds. Oscillation asks whether the system is moving through its range. Input quality asks what the system learns while it does.

Coupling is whether separate systems stay in step with each other. The distinction matters here more than anywhere. Mismatch between the eyes, the inner ear, and the neck is a coupling failure at the level of channels; the fidelity of any one of those channels is an input quality question. Oscillation is within a system, coupling is between systems, and input quality is upstream of both.

Prediction owns the model, input quality owns the evidence

Prediction is the system acting on a model rather than on the world. This border is the subtlest one on the page. Lisa Feldman Barrett and Kyle Simmons put the case for the other side in an argument that interoceptive experience largely reflects predictions about the expected state of the body. On their account the ascending signals constrain the prediction rather than build the experience.

Take that seriously and read the line this way. Prediction owns the model and how heavily it is trusted. Input quality owns the evidence the model is corrected against. A poor report and an over-weighted prediction produce the same result, which is a system acting on something other than its actual state.

Load is what holding a state costs and what accumulates when the system cannot stop paying. Running on a degraded report is expensive, because compensation demands attention and effort that an accurate report would not. Input quality names the deficit. Load names the bill.

Constraint and slack is where the system has room to move, mechanically and neurally. The two are physically entangled, since a segment that does not move generates little fresh afferent traffic. They separate cleanly in principle. Constraint asks whether the movement is available. Input quality asks whether the movement is being reported.

Time course describes how the other dimensions change with the age of the problem. It matters enormously here, because a report degraded for a month sits on an intact map and a report degraded for a decade sits on a map that has already reorganised around the absence.

An accurate input is information, not a repair

In the Unified Model of Tone, afferent fidelity is a general term in every regulation the body performs. Everything above is established science. The claim in this section is the model’s own, stated as such.

Tone is the integrated organization of the body’s interacting state, and an organization is assembled from information. Degrade the information anywhere and the organization assembled from it is degraded in kind, at every scale that depends on that region. The 53 sources on this page measure pieces of that claim with different instruments.

This is why the model treats a precise, well-matched input as an information event rather than a repair. Nothing is being put back. Something is being reported accurately, possibly for the first time in years, into a system that will reorganise around better data because that is what it does continuously anyway. The body has a distributed capacity to regulate itself, integrated at highest density by the nervous system and running in every tissue. It does not need instruction. It needs an accurate picture.

This is also where restoring and masking separate. An input that improves the report restores, and the change shows up across systems because regulation is shared. An input that quiets the output while leaving the report unchanged masks, and the organization underneath keeps paying.

Frequently asked