The Senses and the Nervous System

Every sense converts a physical event into the same electrical code

In 2014 Sanjeev Ranade and colleagues deleted one protein from the sensory neurons of mice, and light touch stopped registering. Piezo2 is where pressure becomes electricity, and every other sense has its own version of that step.

Start with the wire. A nerve cell is a living wire with a cell body, a long fiber, and endings that meet other cells. It carries one kind of message, a brief electrical pulse called a spike. A spike has no color, no pitch, and no texture. It is the same small event whether it left an eye or a fingertip. What differs is which wire carried it, how fast the pulses came, and where the wire lands.

A receptor is the translator standing at the end of that wire, and its whole job is turning a physical event into spikes. In the eye, a particle of light bends a pigment molecule inside a rod or a cone, and the bend changes the cell's electrical state. In the ear, sound moves fluid, the fluid bows tiny hair bundles, and the bending tugs pores open. In the skin, pressure deforms the membrane of a nerve ending until a channel pops open.

That last translator now has a name. Ranade's group deleted a candidate channel from the sensory neurons of mice and asked whether light touch still registered. It did not. They reported that Piezo2 is the major transducer of mechanical force for touch. Pressure opens a hole in a membrane, charged particles rush through, and the wire fires.

Every one of these translators produces the same output. An eye and an ear are built nothing alike, and they send the brain the same kind of thing. The brain itself sits inside the skull in complete darkness and silence. It has never seen light and never heard a sound. It receives reports, and it builds every experience a person has ever had out of those reports.

The senses run well past the familiar five

You know where your hand is with your eyes closed, and you know how hard a muscle is pulling before you look. Uwe Proske and Simon Gandevia gathered decades of work on the proprioceptive senses in 2012, covering body shape, position, movement, and muscle force. Add balance from the inner ear and the continuous reporting of the body's interior, and the familiar five are a small fraction of the traffic.

Every channel shares one architecture: a translator, a wire, a relay, and a patch of cortex. Every channel also arrives with a volume control that something else sets.

The senses report change rather than steady conditions

The retina never tells the brain how bright the light is. Stephen Kuffler showed in 1953 that a retinal output cell reports where the light changes, and ignores an even field almost entirely.

Kuffler was recording from single output cells in the retina of a cat and expected something like a light meter, a cell that fired harder for a brighter spot. He got something stranger. Each cell listened to a small patch of retina, and that patch was built in two rings.

Light in the middle drove the cell up, and light in the surrounding ring drove it down. Flood the whole patch evenly and the two effects canceled. Kuffler had found a retina organized to report contrast. An edge lights it up. An even field barely moves it.

Nine years later David Hubel and Torsten Wiesel pushed their electrodes past the eye into the visual cortex of the cat, hunting for cortical cells that fired for spots of light. Spots did almost nothing. What woke the cells was a bar of light at a particular angle, moving in a particular direction. Their report on receptive fields and functional architecture in the cat visual cortex showed the cortex building edges and orientations out of the retina's contrast signals, then stacking the answers into columns.

Put the two results together and one rule covers both. The nervous system is a difference detector. It spends its bandwidth on change and lets steadiness fade.

You can feel this in ten seconds. Notice the shirt on your shoulders. It has been there all day and you have not felt it for hours, because a constant pressure stops being news. Walk into a bakery and the smell is overwhelming, then within minutes it is gone, though the air has not changed. Sit under a refrigerator hum and it disappears until the moment it switches off, and the silence is what you notice.

A system tuned to change has to keep deciding how much change counts. That decision is a setting. Settings can be well placed, and settings can jam.

The sensory map is redrawn by the traffic it receives

Michael Merzenich mapped the hand region of adult monkeys in 1984, removed a single finger, and found the neighboring fingers occupying the empty territory months later. Adult sensory cortex had been assumed to be fixed.

The map came out of the operating room. In the mid twentieth century the neurosurgeon Wilder Penfield was operating on people with severe epilepsy and needed to know which patch of cortex did what before removing any tissue. He touched the exposed brain with a small electrode while the patient stayed awake and talked.

Stimulate here and the patient feels the thumb. Stimulate there and the lip tingles. Drawn out, the results became the sensory homunculus, a distorted little body with a huge face and hands and a tiny trunk. The neuroanatomist Marco Catani traced the figure and its afterlife in a 2017 history of the sensory homunculus.

Merzenich tested the assumption directly. He and his colleagues mapped the hand region, removed the digit, and mapped it again after months had passed. The territory that had belonged to the missing finger did not fall silent. Neighboring fingers moved in and took it, as their report on cortical map changes after digit amputation in adult monkeys describes.

Losing input costs a channel its cortical ground

Vision told the same story from the other direction. Wiesel and Hubel closed one eyelid in young kittens to find out what a developing cortex does without input. When they later recorded from the visual cortex, most cells no longer answered the deprived eye at all, though that eye was structurally sound. Their study of single cells in the striate cortex of kittens deprived of vision in one eye showed input winning and losing cortical ground.

The most striking case came from people. In 1996 Norihiro Sadato and colleagues scanned blind adults who read Braille fluently, expecting touch to occupy touch cortex. It did. It also lit the visual cortex. Their finding that Braille reading activates the primary visual cortex in blind subjects means an unused visual region had been recruited by the fingers.

The territory follows the traffic. A map written by input is a map that can be rewritten by input.

Where a sensory channel sits, how much cortex it commands, and how loudly it speaks are all held in place by ongoing use. None of that is static enough to be called hardware in the way a lens or an eardrum is hardware.

Every sensory channel carries an adjustable amplifier

Gain decides how large an output a given input produces. The eye works from starlight to noon sun by moving that setting, and no fixed setting could cover both.

Turn gain up and a whisper arrives as a shout. Turn gain down and a shout arrives as a whisper. The input has not changed at all.

Your senses run enormous gain ranges as a matter of routine. Step out of a dark cinema into afternoon sun and you are briefly blinded, then within minutes the same street is comfortable. Walk into a quiet library from a loud street and at first you hear nothing, then you can hear a page turn.

How fast does the setting move? In 2001 a group including the physicist Adrienne Fairhall recorded from a motion-sensitive neuron in the fly and kept changing the statistics of the motion the animal saw. The cell rescaled itself, adapting its code to the input within about a tenth of a second. The neuron was renegotiating its own units while the stimulus ran.

That is not a quirk of insects. Matteo Carandini and David Heeger gathered evidence across species, sensory systems, and brain regions, and argued that one operation appears everywhere. A neuron's raw drive gets divided by the pooled activity of its neighbors, so the output reflects a ratio rather than an absolute. They called normalization a canonical neural computation, documented in vision, hearing, and smell. Every sense divides. Every sense therefore carries a level.

Almost every sensory complaint in the clinic is one of those two failures, or a third: a system that has stopped updating the setting at all.

The brain wires back down to the sense organ and turns it down

Bundles of fibers leave the brainstem, travel back out to the cochlea, and quiet the ear's own amplifier cells. John Guinan spent a career measuring how far that traffic lowers cochlear gain in humans.

The ear is the cleanest example, and most people have never heard of it. These olivocochlear fibers end on the outer hair cells of the cochlea, the snail-shaped organ that turns sound into spikes. Guinan's 2006 review of their anatomy, physiology, and measurable effects in humans covers the reduction of cochlear gain and the improved detection of a signal buried in noise. Your brain turns your ear down.

Touch and pain have a version of the same arrangement, proposed before anyone could see it. In 1965 the psychologist Ronald Melzack and the neurophysiologist Patrick Wall were trying to explain everyday facts the pain science of the day could not.

Why does rubbing a banged shin help? Why does the same injury hurt differently in different circumstances? They proposed that the spinal cord contains a gate, and that large touch fibers and signals descending from the brain open and close it before the message ever reaches awareness. Their gate control paper reorganized the field.

State the principle plainly. Sensing is a two-way conversation. Signals climb, instructions descend, and what a person finally experiences is the settlement between them. No point in the chain delivers raw world upward untouched.

So the useful question about a sensitive person is rarely whether the organ is broken. It is where the conversation has settled. A brake that never engages leaves the amplifier wide open. A brake clamped on leaves the world faint. Both are regulation, and both are adjustable in principle.

A percept is assembled from several channels, weighted by their reliability

Mark Meredith and Barry Stein found single midbrain cells in 1983 that fired far harder for a weak light and a weak sound together than for the two responses added. You do not have five senses feeding five experiences.

Meredith and Stein were recording in the superior colliculus, the structure that orients the eyes and head toward events. They presented a weak light, then a weak sound, then both together. Individually each stimulus produced a modest response. Together the cell fired far beyond the two responses added. Their report on interactions among converging sensory inputs in the superior colliculus described a cell whose job is combination rather than reception.

Combination is not optional, and knowing about it does not switch it off. In 1976 the psychologists Harry McGurk and John MacDonald were studying how infants match voices to faces. A dubbing error showed them that a face mouthing one syllable, paired with a recording of a different syllable, produces a third syllable in the listener's ear.

Their paper on hearing lips and seeing voices is famous because the illusion survives full explanation. Close your eyes and the sound corrects itself. Open them and vision overrules your ears again.

The combination follows a reliability rule

The vision scientists Marc Ernst and Martin Banks had people judge the height of a ridge by looking at it, by feeling it, and by both, then degraded the visual information deliberately. As vision got noisier, the brain leaned on the hand. They showed in 2002 that humans combine sight and touch in a statistically optimal fashion, weighting each channel by how reliable it currently is. The nervous system tracks how trustworthy each channel currently is and spends its confidence accordingly.

Push the weighting far enough and the body itself moves. Matthew Botvinick and Jonathan Cohen hid a person's hand, placed a rubber hand in view, and stroked both with brushes in perfect time. Within a minute people report feeling the touch in the rubber hand. Their demonstration that rubber hands feel touch that eyes see shows that the body a person inhabits is assembled from agreeing channels, and can be reassigned when those channels agree on a lie.

Standing up is the same computation, running all day

The bioengineer Robert Peterka tested human balance by moving the floor and the visual surround while measuring sway. He showed in 2002 that upright stance depends on continuously reweighting vision, the inner ear, and body sense, with the weights shifting whenever a channel becomes unreliable. Balance is a negotiated agreement among channels rather than a single sense.

The clinical consequence is immediate. A person whose neck stops reporting reliably will have a visual problem. A person whose inner ear was injured will develop a visual dependence, and busy visual environments will be the hardest places to stand. That is exactly the pattern behind persistent dizziness, and it follows from the weighting rather than from any damaged organ.

Much of what you feel is what prediction failed to cancel

Roger Sperry surgically rotated the eyes of fish and newts in 1950, and the animals spun in endless circles. Every motor command sends a copy to the sensory side, and the correction had nothing left to subtract from.

Sperry was studying how an animal tells its own movement apart from movement in the world. He argued in his account of the spontaneous optokinetic response produced by visual inversion that every motor command sends a copy to the sensory side, so the expected consequences can be subtracted out. With the eyes reversed, the subtraction ran backwards.

That copy is called corollary discharge, and you can demonstrate it on yourself in a second. You cannot tickle yourself. Sarah Blakemore, Daniel Wolpert, and Chris Frith built a small robot to find out why. A person moved one hand, and the robot delivered the tickle to the other hand, either instantly or after a short delay.

Instantly delivered touch felt dull. A delay of a fifth of a second brought the tickle back. Their study of central cancellation of self-produced tickle sensation showed the brain subtracting what it predicted and feeling the remainder.

Expectation moves what the body reports

Karl Friston generalized this into a single principle. Brains build a model of their causes, generate predictions, and act to reduce the difference between prediction and evidence. His statement of the free energy principle as a unified brain theory is a framework under active argument in the field. The observation underneath it is measurable: expectation shapes sensation.

Because expectation shapes sensation, expectation can be measured moving it. Tor Wager and Lauren Atlas reviewed the mechanisms behind placebo effects in 2015 and found changes in the brain's own pain-modulating circuitry rather than changes in what people were willing to say. Their account of how context and learning shape the experience of pain shows a nervous system adjusting its own sensory reports according to what it anticipates. The same forecasting machinery runs outside the senses, in prediction.

The third failure follows from all of this, and it produces the most stubborn symptoms. If perception is a model checked against evidence, the system can fail by locking the model. The prediction stops updating. A limb that no longer exists still reports, a silent room still rings, and dizziness continues long after the injury that started it has healed. Nothing is broken at the receptor. The model stopped listening.

The state of the body sets the threshold for the outside world

Beneath sight, sound, and touch runs a stream of reports from inside the body: heartbeat, breath, gut, temperature, blood chemistry, tissue state. Louisa Edwards showed in 2001 that a single heartbeat is enough to move a spinal reflex threshold.

The neuroanatomist Bud Craig traced where these signals go. He followed a pathway from small fibers in the body's tissues through the spinal cord and thalamus into the insula, a region folded deep in the side of the cortex. He argued in his 2002 account of interoception that this is a full sensory system in its own right, with its own receptors, its own route, and its own cortex. It reports the state of you.

Reading that stream is not a single skill. Sarah Garfinkel and colleagues tested three things people mean by body awareness. How accurately can a person count their own heartbeats? How much do they believe they are good at it? And how well does their confidence track their actual performance? Their 2015 finding that accuracy, sensibility, and awareness come apart matters in the clinic. A person can be flooded with body signals and read them badly, or read them well and feel little.

The heartbeat moves a reflex and a fear judgment

With each beat a pressure pulse stretches receptors in the great arteries, those receptors fire, and their firing reaches the brainstem. Edwards and colleagues timed a small electrical stimulus to different moments of that cycle and measured the leg withdrawal reflex. They found the nociceptive flexion reflex modulated across the cardiac cycle. The same stimulus met a different threshold depending on the instant it arrived.

Emotion rides the same pulse. Garfinkel, Hugo Critchley, and colleagues showed fearful and neutral faces timed to the heartbeat while measuring detection and brain response. Fearful faces were judged more intense, and amygdala responses were altered, when the images arrived on the beat. Their 2014 report that sensitivity to fear stimuli depends on individual heartbeats puts the body's own rhythm inside the perception of threat.

The same stimulus meets a different threshold depending on what the body is doing that second.

If a heartbeat shifts a spinal reflex and a fear judgment within a single second, the state of the body is never outside perception. It is one of the channels being weighted. That is why the same sound is tolerable on a rested morning and intolerable on a sleepless afternoon. The fidelity of these inner reports, and of the proprioceptive traffic beside them, is input quality.

The network that reads the body's state also commands it

Eddy Benarroch assembled the anatomy in 1993 and named it the central autonomic network. The insula, the amygdala, the hypothalamus, and the brainstem read the body's state and command it, using one set of regions for both jobs.

Two great outflows run from the brain to the organs. The sympathetic division readies the body for effort and threat. The parasympathetic division, carried largely by the vagus nerve, slows the heart and turns the body toward rest and digestion. Benarroch's account of the central autonomic network binds those regions into one integrated system, so a memory, a posture, and a signal from the gut all enter the machinery that sets a sensory threshold. The autonomic nervous system is the anatomy of the two outflows.

Julian Thayer and Richard Lane took the next step in 2000. They proposed that the flexibility of this network governs how well a person regulates emotion and attention, and that a rigid system loses that flexibility. Their model of neurovisceral integration ties the vagal brake to the capacity for adjusting to changing demands.

All of this leaves a track that can be measured cheaply. Healthy hearts do not beat like metronomes. The interval between beats varies from beat to beat, and that variability reflects the pull between the two outflows.

An international task force set the standards for measuring heart rate variability in 1996, and the measure has indexed cardiac autonomic state ever since. Reading it as a window onto tone is the Unified Model of Tone's interpretation of that established measure, and it predicts why one number tracks so many unrelated sensory complaints.

A system that has judged the environment unsafe runs its gain high

Stephen Porges proposed that the nervous system continuously and unconsciously evaluates safety and danger, a process he named neuroception and described in his 2007 account of the polyvagal perspective. Comparative physiologists dispute parts of the evolutionary argument. The regulatory reading holds either way. Under threat a missed signal costs more than a false alarm, so sensory gain rises.

The reach of this integration keeps widening. In 2021 a team led by Tamar Koren tagged insular neurons that were active during gut inflammation in mice, then reactivated those neurons later. The inflammation returned to the same tissue. Their finding that insular cortex neurons encode and retrieve specific immune responses places immune state inside the same cortical territory that reads the body and shapes perception.

Tone is the organization every sensory setting is held within

Medicine names each sensory part and each sensory measure. Cochlear gain, the spinal gate, thalamic relay, cortical normalization, reliability weighting. The organization running across all of them has no name in that vocabulary.

The Unified Model of Tone names it. Tone is the integrated organization of the body's interacting state at a given moment, the way its mechanical, electrical, chemical, fluid, and neural processes are related as one bound state. In the senses, that organization is held across receptor, spinal cord, brainstem, thalamus, and cortex at once. Cochlear gain and the spinal gate are two voices inside it.

Health has a definition in those terms. Health is the width of the range the system can move through and return from: gain raised for a faint signal, lowered for a loud one, and brought back. A nervous system in good tone hears the page turn in the library and tolerates the street outside. When that range narrows and the system starts defending one fixed setting, the narrowing is what shows up as illness. Sensory dysregulation is that collapse, and the receptor stays intact throughout.

Health is the width of the sensory range. Sensory dysregulation is that range collapsed into a setting the nervous system now defends.

Perception is a chord, not a note

Ernst and Banks weighting sight against touch in 2002, Peterka reweighting balance in the same year, and McGurk's dubbed syllable all say the same thing from three directions. There is no single sensory value. There is a pattern across coupled channels, and the pattern is what a person perceives. A sensory symptom is therefore never the property of one organ. It is what the whole pattern currently produces.

That changes what a cause can be. If a sensory symptom belongs to the organization, its cause need not live in any single channel. It can live in the tuning: in how far one channel's gain has drifted, how tightly the channels are bound to each other, and how long the setting has been held. A search for the broken part is not built to find that kind of cause, which is why so many of these patients are told their tests are normal.

Tinnitus with a normal hearing test is amplified silence

Tinnitus is the perception of sound with no external source, and a large share of the people who have it show a normal audiogram. Roland Schaette and David McAlpine tested that combination in 2011 with electrodes on the scalp.

Tinnitus is common and it can be severe, as Berthold Langguth and colleagues describe in their 2013 review of its causes and clinical management. The hearing test says the ear is fine. The person still hears a ring, a hiss, or a hum.

Schaette and McAlpine recruited people with tinnitus and clinically normal hearing, plus matched controls without tinnitus, and used the auditory brainstem response. Clicks are played into the ear while scalp electrodes record the volley of activity traveling up the hearing pathway. That volley appears as a series of small waves. Wave one comes from the auditory nerve, the ear's own output. Wave five comes from the midbrain, further up the line.

Their result was precise. In the tinnitus group, wave one was reduced while wave five was normal. The ear was sending less. The center was still delivering its usual amount. The amplification between them had risen to make the numbers come out even.

The periphery had gone quieter, most likely from a loss of nerve fibers that a standard hearing test cannot see. The central system did what an amplifier with automatic level control does and raised the gain to restore its expected output. Once the gain is high enough, the system's own background activity crosses the threshold of perception. Silence gets amplified until it becomes a sound.

Benjamin Auerbach, Richard Salvi, and colleagues generalized that reading across the animal and human literature in 2014. Their review argues that central gain control underlies both tinnitus and hyperacusis, the condition in which ordinary sounds are painfully loud. One setting, two symptoms, depending on how far it moves and what it amplifies. Tinnitus is central gain heard as a sound, and the auditory case is worked through there.

Nothing was broken. The regulation moved, and a compensation that overshot became audible.

A person told their hearing test is normal has been told the truth about the receptor and nothing at all about the regulation. The regulation is where the symptom is.

The same stuck setting appears under four different diagnoses

Misophonia, photophobia in migraine, persistent dizziness, and the sensory differences of autism sit in four specialties. Each carries the same shape: an intact receptor and a setting that stopped moving.

Misophonia sits in the region that reads body state

Misophonia is the condition in which specific sounds, usually chewing, sniffing, or tapping, produce immediate rage or panic. Sukhbinder Kumar and colleagues scanned people with misophonia in 2017 while playing trigger sounds, unpleasant sounds, and neutral sounds.

Trigger sounds produced an exaggerated response in the anterior insula, altered connections between that region and the frontal cortex, and a rise in heart rate and skin conductance. Their study of the brain basis for misophonia found the reaction sitting in the region that reads body state, dragging the autonomic system with it. The sound is not louder. The gate around it has changed.

Light worsens migraine at a shared thalamic cell

Rodrigo Noseda, Rami Burstein, and colleagues asked why light worsens a migraine, including in blind patients who retain light-sensitive retinal cells without conscious sight. Recording in the thalamus, they found retinal signals converging on the same neurons that carry pain traffic from the coverings of the brain.

Their 2010 demonstration of a neural mechanism for the worsening of headache by light shows two channels feeding a shared cell. Photophobia is a convergence rather than an exaggeration of the eye. Migraine is where that convergence becomes a whole attack, worked through in headaches.

Dizziness persists when the weighting never reverses

A committee led by the psychiatrist Jeffrey Staab defined the criteria for persistent postural-perceptual dizziness in 2017, a condition in which an acute vestibular event resolves and the dizziness does not. Symptoms worsen with upright posture, motion, and complex visual environments such as supermarket aisles.

Weighted by reliability, an injured inner ear earns a reasonable emergency shift toward vision. In these patients the shift never reverses. The person is still balancing as though the inner ear cannot be trusted, long after it can. Vertigo and balance and coordination carry the vestibular side.

Sensory differences in autism are measurable in the wiring

Elysa Marco and colleagues reviewed the neurophysiology of sensory processing in autism in 2011 and found consistent differences in how sensory signals are handled, alongside considerable variability between individuals. Julia Owen and colleagues then imaged the wiring. Using diffusion imaging in children with sensory processing difficulties, they found measurable differences in white matter microstructure in tracts carrying sensory traffic. These children are describing something that shows up on a scan.

John Rubenstein and Michael Merzenich proposed in 2003 that some autism features follow from an increased ratio of excitation to inhibition in key circuits. That means too much accelerator relative to brake in the cortex itself. The hypothesis has been influential, and the evidence for it points in more than one direction. Sensory difference is defining rather than incidental in spectrum disorders, where the developmental picture belongs.

The model reads all four presentations as one process under four diagnoses. In each the receptor does its job and the setting has stopped moving. Where it froze high, the world assaults. Where it froze locked, the model outlives the event that set it.

Taking input away raises sensitivity rather than calming it

Two weeks of earplugs made healthy adults judge ordinary sounds louder, and five days of blindfolding produced hallucinations in ten of thirteen volunteers. Common sense says an overloaded system needs less input.

The blindfold study came first. Lotfi Merabet and colleagues blindfolded sighted volunteers continuously for five days to study how fast the adult brain reorganizes. They were not studying hallucination and found it anyway. Ten of the thirteen volunteers reported visual hallucinations during prolonged blindfolding, ranging from simple flashes and shapes to detailed scenes, with onset generally after the first day. Deprived of input, the visual system raised its own excitability until internal activity became visible experience.

Formby moved a loudness threshold in both directions

Formby's experiment is the more useful one, because it moved the setting both ways on purpose. The audiologist Craig Formby and colleagues recruited a small number of listeners with normal hearing and split them into two groups. One group wore earplugs for two weeks, lowering the background sound reaching the ear. The other wore devices delivering low-level noise for two weeks, raising it. Then both groups had their loudness judgments and loudness discomfort levels retested.

The results moved in opposite directions. After two weeks of earplugs, sounds were judged louder and discomfort levels shifted downward. After two weeks of added background noise, sounds were judged quieter and tolerance improved. Formby and colleagues described this in 2003 as adaptive plasticity of loudness induced by chronic attenuation and enhancement of the background. The ears never changed. The setting did, in whichever direction the environment pushed it.

Two weeks of quiet did not calm a loud system. It taught that system to amplify.

The result is about sound, in a brief report, in a small group of adults with normal hearing. It is the only direct evidence of a threshold moving with protection. The earplugs at the dinner table have that finding behind them. The sunglasses indoors have an extension of it, and the model predicts the visual channel behaves the same way for the same reason. Relief from a filter in the hour is entirely real, whatever the filter does to the setting over months.

Accommodations that let a person work, learn, or stay in a classroom are functional supports. Autistic people, children with sensory processing differences, people with diagnosed hyperacusis, and people in an active photophobic migraine often depend on them. Any change to one belongs in a plan made with the clinician who examines you, built from graded and tolerable steps.

Formby's design also does something the model needs. It shows a sensory threshold moving up in one group and down in another, from the same starting range, under opposite inputs. A threshold that can be pushed both ways is a threshold that can be brought back toward the middle from either side.

Neurons already restore their own activity from either side

Gina Turrigiano and Sacha Nelson found the machinery in 2004. Raise a neuron's activity for hours and it scales every incoming synapse down. Silence it and it scales them all up, without erasing what the network learned.

Turrigiano and Nelson were working on a problem that had bothered the field for years. Learning strengthens synapses. If strengthening were the only rule, activity would climb until the network seized or saturated. Something had to hold the rate inside a range without erasing what was learned.

They found it. Raise a neuron's activity artificially over hours and it scales all of its incoming synapses down. Silence the neuron and it scales them all up. The adjustment is multiplicative, so the relative pattern of learned strengths survives while the overall level returns toward target. Their 2004 review of homeostatic plasticity in the developing nervous system describes machinery whose only job is restoring a value toward a range from whichever side it drifted.

Sleep is the nightly version of the same operation

Giulio Tononi and Chiara Cirelli proposed that waking experience drives a net strengthening of synapses, and that sleep renormalizes them to keep the total sustainable. Their 2014 account of sleep and the price of plasticity has substantial support and is still being argued.

A separate line of work led by Lulu Xie found that sleep drives metabolite clearance from the adult brain, with the fluid spaces around cells expanding during sleep in mice. Sleep is a nightly recalibration, which is why a sensory threshold on a sleepless afternoon is a different threshold. Sleep is where that recalibration is set out.

The model's position follows directly. What a single neuron does with its firing rate, the model claims the whole coupled system does with its sensory range. Too loud comes down. Too quiet comes up. The target is a range rather than a direction.

Findable causes come first, and a filter manages input rather than range

Findable causes must be found first, and a regulation story is never a reason to skip a workup. Sudden hearing loss, a curtain across the visual field, and one-sided tinnitus each send a person somewhere else the same day.

Sudden hearing loss is a medical emergency and needs urgent assessment. Sudden vision loss, a curtain across the visual field, or new flashes and floaters need urgent eye care. Tinnitus in one ear only, with hearing loss on that side, needs investigation for a tumor of the hearing nerve.

New numbness, weakness, or loss of coordination needs a neurological evaluation. Loss of smell, unexplained sensory loss in the feet, and severe tolerance changes can follow vitamin deficiency, diabetes, thyroid disease, medication effects, infection, or demyelinating disease. Each of these is a real cause with a real test.

Deficiency, tumor, infection, and demyelination are sensory problems with causes of their own, and each needs treating on its own terms. After those are excluded, a large group of people remain with severe symptoms and normal results, and that group has been explained badly.

Masking the input and restoring the range are different aims

The distinction is mechanical rather than moral. A filter lowers the input arriving at a system whose gain is already too high. Formby's earplug result is the reason to take that seriously, since two weeks of attenuation in healthy ears made sound less tolerable afterward. The model predicts that reducing input without restoring range drives the setting further in the direction it already went.

Medication and protection have a real place. Hearing protection at a concert or a job site prevents damage and should be worn. Dark and quiet during an active migraine is sensible care. Acute injury deserves rest.

Medications that reduce migraine frequency, treat anxiety, or manage pain help many people function, and stopping a prescribed drug on the strength of something you read is a bad idea. The argument here is about aim. A drug or a filter manages the output. The other aim is the range itself: what the system can tolerate and return from.

The model predicts convergence toward the middle from both sides

Loudness discomfort levels, light thresholds, two-point discrimination, and heartbeat detection can each be measured in both directions. A filter or a sedative moves all of them one way for everyone, which makes the model's prediction separable in an ordinary clinic.

The design, and how to read each result

Take a sensory threshold that can be measured reliably in both directions. Loudness discomfort levels work well, as do light sensitivity thresholds, two-point discrimination on the skin, and heartbeat detection accuracy for the inner senses. Recruit two groups from the same clinic. One group starts too sensitive, with thresholds far below the normal range. The other starts too insensitive, with thresholds far above it. Give both the same intervention, one intended to restore regulation rather than block input. Measure again.

The model predicts convergence. The over-sensitive group becomes more tolerant. The under-sensitive group becomes more perceptive. Both move toward the middle of the normal range, and the size of each shift scales with how far the person started from that middle.

Three other outcomes are read the same way, fixed in advance. A uniform one-way shift in every participant marks the intervention as a mask rather than a restoration. No movement in either group means the intervention is inert on this axis. Divergence, with both groups moving further from the middle, marks the input as a disturbance of the mechanism rather than a restoration of it.

Formby's two-week study is why this is more than a thought experiment. It already moved loudness tolerance down in one condition and up in another within two weeks in healthy adults. The measure is sensitive, the timescale is short, and the equipment is standard.

Two mixed trials, read through the model

Tinnitus retraining therapy combines counseling with sound therapy and has been widely used. A large randomized trial reported by Roberta Scherer and Craig Formby in 2019 compared it against standard of care and against partial versions of the protocol in service members. It found that the full protocol did not outperform standard of care on tinnitus-related quality of life.

Case-Smith's review splits down the middle. The occupational therapist Jane Case-Smith and colleagues reviewed sensory interventions in autism in 2015 and drew a line between two practices filed under one name. Sensory integration therapy is delivered by a therapist who tailors the activity to the individual child.

Two randomized trials of it produced positive effects on individualized goals, with effect sizes running from 0.72 to 1.62. Sensory-based intervention, the classroom-level equipment protocols such as weighted vests and brushing, produced few positive effects across the studies reviewed.

An input meets a tone. A protocol delivered uniformly to a heterogeneous population meets many different tones and produces many different outcomes, which average toward nothing. Some participants in the tinnitus trial had a peripheral loss driving central gain. Some had a locked model with an intact periphery.

Some arrived with an autonomic system running hot and a sleep debt that would defeat any sound protocol. Group them together, apply one recipe, and the scatter cancels itself. The half of the autism review that worked is the half where a clinician adjusted the input to the child in front of her.

The model reads that scatter as information. It predicts that responders and non-responders in the pooled trials differ measurably in their starting tone, and that stratifying by starting state would reveal the convergence a pooled average erases. That is a reanalysis of data that already exists rather than a new trial.

The model states what restored range would look like on those instruments. Loudness discomfort levels widen and light tolerance climbs. Two-point discrimination sharpens where it had blurred, and heartbeat detection improves where the inner senses had gone quiet. Heart rate variability rises and sleep consolidates. Each of those is a measure of width rather than a measure of relief, which is the difference the model is built on.

How the senses relate to the rest of the library

The senses are where tone is easiest to measure, because a threshold is a number and a threshold moves. Each neighboring page carries one part of why it moves.

Three foundations of tone carry the weight here.

• Input quality is the fidelity of what arrives: a spike train is all the brain ever gets, and a cochlea sending fewer fibers or a neck reporting badly changes what perception has to work with.
• Gain is the amplifier between that report and the experience, the setting Fairhall watched rescale within a tenth of a second and Carandini and Heeger found dividing in every sense.
• Coupling is why the channels cannot be treated one at a time, since sight is weighted against touch, vision against the inner ear, and every one of them against the state of the body.

The remaining foundations each carry a piece.

• Prediction is what the tickle experiment measures and what a locked sensory model fails to update.
• Set point is the expected output the central auditory system defended when the ear went quiet.
• Oscillation is the cardiac cycle that shifted a spinal reflex threshold within one beat.
• Load is what holding a high gain all day costs, and why the same restaurant is survivable in the morning.
• Constraint covers the white matter differences imaged in children with sensory processing difficulties: the wiring the regulation has to work within.
• Time course separates a week of light sensitivity after a concussion from ten years of it.

The condition pages divide the symptoms.

• Tinnitus is central gain heard as a sound.
• Vertigo is the vestibular channel failing outright, and balance and coordination is the reweighting that follows.
• Headaches is where light and pain converge on one thalamic cell.
• Spectrum disorders is where sensory difference is defining rather than incidental.
• Dysautonomia is the same regulation failing where instruments read it most directly, and sleep is the nightly recalibration that sets the next day's thresholds.

Two more pages hold the machinery. Neurophysiology is the spike, the synapse, and the relay that every sensory channel is built from. The autonomic nervous system is the anatomy through which a sensory threshold and a heart rate turn out to be the same capacity.

Frequently asked