Brain Activity and the Nervous System

What a recording of the brain actually shows

An electroencephalogram measures the summed electrical field of millions of cortical cells shifting their charge in step. That one physical fact governs everything a brain recording can and cannot say.

Start with a single word you will need for everything that follows. A nerve cell is a living wire. It holds a small electrical charge across its outer skin, the membrane, the way a battery holds a charge across its terminals. When the cell is pushed hard enough, that charge flips and a pulse runs down the wire. Then the cell resets and waits.

Any wire carrying a charge creates a small electrical field around it. One cell makes a field far too faint to notice. Millions of cells lying side by side in the same orientation, shifting their charge in step, make a field large enough to be read through bone and skin. That is all an electroencephalogram, or EEG, is. Sensors on the scalp read the summed electrical field of large populations of cortical cells rising and falling together.

The first person to see it was not looking for thought. In 1875 an English physician and physiologist named Richard Caton laid the wires of a galvanometer, an instrument for detecting tiny currents, directly onto the exposed brains of rabbits and monkeys. He wanted to know whether the brain produced electricity at all. It did. He recorded feeble currents that varied continuously, shifted when the animal was shown a light, and disappeared at death. The current was there before he stimulated anything.

Half a century later, a German psychiatrist named Hans Berger spent years trying to record the same thing from an intact human head. He was chasing a question about mental energy. In 1929 he published the first human EEG and described a steady rhythm of roughly ten cycles per second over the back of the head, which he named the alpha wave.

It grew strong when his subject closed the eyes and broke up when the eyes opened. His finding sat largely ignored until other laboratories confirmed it, and the story of how those two men opened the field is now the standard opening of every textbook on the subject. Berger's original 1929 report is the founding document of clinical electroencephalography.

What the colored maps measure

The other common picture of brain activity is not electrical at all. Functional MRI produces colored maps of the brain and is described in the press as showing thought. It shows blood.

The reason it works at all was proposed in 1890. Two physiologists at Cambridge, Charles Roy and Charles Sherrington, were asking whether the brain's blood supply is fixed or whether it follows the brain's own work. They concluded that the brain possesses an intrinsic mechanism that varies local blood supply with local activity. Modern work has filled in how that happens, and it is more complicated than a simple demand signal.

Glial cells and several signaling pathways are involved, and the control point may sit further down the vascular tree than anyone assumed. Reviewing the evidence in 2010, David Attwell and colleagues concluded that blood flow may be controlled by capillaries as well as by arterioles. The mechanism is still being argued about, which is worth remembering the next time a colored map is presented as settled.

Hold on to the gap between the measurement and the thing measured. An EEG reads the summed field of cells firing in step, which makes it blind to cells that are busy out of step. Functional MRI reads a blood signal that lags the electrical event by seconds. Both are real measurements of brain activity. Neither is the organization itself, only its shadow.

The brain is never off

The brain generates almost all of its own activity, and the discovery arrived by accident, in data everyone else was throwing away.

In 1995 a biophysicist named Bharat Biswal, working at the Medical College of Wisconsin, was trying to clean up functional MRI images. Resting scans were full of slow drifting fluctuations that everyone treated as noise to be filtered out. Biswal set out to characterize that noise so it could be removed properly.

He found that it was not noise. In a brain doing nothing at all, the slow fluctuations in the left motor cortex rose and fell in tight correlation with the right motor cortex. They tracked other regions of the same motor system too. The resting brain was organizing itself into functional networks with nobody asking it to.

Six years later a group at Washington University led by the neurologist Marcus Raichle noticed something related and stranger. Across many experiments, a particular set of regions consistently became less active whenever a person was given a task. Those regions were doing their most intense work when the subject was lying still and unoccupied. Raichle called this a default mode of brain function, an organized baseline the brain returns to whenever the world stops making demands.

The energy arithmetic that settles it

Then came the arithmetic, and the arithmetic is what makes this decisive. The brain is about two percent of body weight and consumes roughly twenty percent of the body's energy at rest.

Where does it go? In 2001 two neuroscientists, David Attwell and Simon Laughlin, built an itemized energy budget for excitatory signaling in the grey matter of rodent brain. Action potentials took an estimated 47 percent of it, the postsynaptic effects of glutamate 34 percent, holding the resting charge 13 percent, and recycling the transmitter 3 percent.

Their conclusion was that signaling accounts for a large fraction of what the brain spends. The scope is precise: a model of excitatory transmission in rodent gray matter, not a measured budget for a whole human head, so the percentages do not transfer straight across. The shape of the answer does. The brain is not idling between events. It is spending on something.

Raichle and his colleague Mark Mintun then set the task-driven changes against that background and found them tiny. The additional energy a region uses when it is put to work is a small fraction of the energy it was already using. Raichle later laid the two pictures side by side in a paper called Two views of brain function. A century of experiments, he argued, had tacitly assumed the brain is a reflex machine waiting for input. The evidence says brain function is mostly intrinsic.

State it flatly. The baseline is the phenomenon. What the brain does on its own, continuously, without instruction, is the main event of brain activity. What the world adds is a small edit on top of it.

How a rhythm gets made

Two forces generate every brain rhythm: excitation, which pushes cells toward firing, and inhibition, which pushes them away. Wire the two into a loop and the loop cycles.

The first force is excitation. One cell releases a chemical onto the next and pushes its charge closer to the flipping point. Excitation is the accelerator. The second is inhibition. A different class of cell releases a chemical that pushes the charge away from the flipping point, making the target harder to fire. Inhibition is the brake.

Neither is good or bad. A brain with only accelerator would seize. A brain with only brake would fall silent. Rhythm is what happens when the two are wired into a loop, so that excitation builds, recruits inhibition, gets shut down, and the cycle repeats.

Every brain rhythm in the catalog is a version of this argument between accelerator and brake, running at a particular speed. The full physics of rhythm, why living systems oscillate at all, belongs to the oscillation page; what matters here is where a brain rhythm comes from and what it is doing.

Watching the loop run in living cortex

You can watch this piece of brain activity directly. In 1993 a neurophysiologist named Mircea Steriade, working at Laval University, put fine electrodes inside individual cortical cells of anesthetized cats to see what the cortex does when nothing is happening. He found the cells alternating slowly between a depolarized state and a hyperpolarized one, slower than one cycle per second, entire populations moving together. Neuroscience now calls these up states and down states. The cortex, left alone, breathes.

Speeds vary enormously. In 2004 two neuroscientists, Gyorgy Buzsaki and Andreas Draguhn, surveyed the whole catalog and made a structural point. Cortical networks oscillate in a family of bands that span five orders of magnitude in frequency, from slower than one cycle a minute to hundreds of cycles a second. The same family is preserved across mammals. A pattern that conserved is part of the design.

The most misunderstood member of that family is the alpha rhythm Berger found. For decades alpha was read as the sound of a cortex doing nothing, an idling rhythm. The Austrian cognitive neuroscientist Wolfgang Klimesch spent years testing that reading against memory and attention data. He argued instead that alpha is the signature of active, timed inhibition, a gate that opens and closes. He later extended the argument to show that alpha controls access to stored information rather than marking its absence.

That correction matters far beyond alpha. A strong rhythm is the brain holding a policy about what gets through, and the policy is enforced by timed inhibition.

Every brain measure is one voice from a chord of coupled rhythms

Clinical language treats brain activity as if it had one dial: too fast, too slow, too much theta, not enough beta. The electrophysiology says the rhythms are nested inside one another, and that changes what any single number can mean.

In 2006 a group at Berkeley led by Ryan Canolty and Robert Knight recorded directly from the cortical surface of people undergoing surgical monitoring, asking whether fast and slow rhythms were related. They were, tightly.

• Bursts of very fast gamma activity occurred at specific phases of the slow theta rhythm, as though the slow wave were carrying the fast one on its back.
• Canolty and Knight later reviewed the whole phenomenon and argued that cross-frequency coupling is a general mechanism for coordinating activity across scales.
• How that coupling is measured, and the traps in measuring it, belong to the coupling page.

Coordination across space works the same way. In 1989 four neurophysiologists in Frankfurt recorded from cat visual cortex. Charles Gray, Peter Konig, Andreas Engel and Wolf Singer wanted to know how separate columns of cells respond to different parts of one object.

They found the columns oscillating at roughly forty to sixty cycles per second and locking into synchrony with each other when they were responding to the same object. Distant patches of cortex were binding themselves together by phase, without any wire carrying the word same.

Coupled oscillators, and what one number can mean

Physics has a name for a population of things that adjust their timing to one another until they fall into step. They are coupled oscillators. The computational neuroscientist Michael Breakspear and colleagues took the standard mathematics of coupled oscillators, the Kuramoto model. They showed that it reproduces cortical dynamics with real neurobiological meaning, including the way cortex sits between full synchrony and disorder.

Now the consequence for anyone who has ever been handed a brain map. A single number pulled off a recording, alpha power, a theta to beta ratio, a peak frequency, is one voice read out of a choir. It is real. It is also a projection of a coupled system onto one axis, which is why the same number can mean opposite things in two people.

The coupled reading also settles what these measures are. Alpha power, gamma coupling, the slow oscillation and the background slope move together because each is a partial read-out of one organizing state. That is why no one of them, silenced or boosted alone, behaves like a lever. A cause visible only in projection still needs an experiment to pin it down, and the model stakes one below.

Membrane, circuit and cortex read out one variable

The charge across a cell membrane. The accelerator to brake balance of a circuit. The up and down states of a population. The nested rhythms binding distant regions. The Unified Model of Tone reads all of them as one variable measured at different depths.

Mainstream neuroscience studies these as separate literatures, with separate journals and separate methods. The model reads them as tone: the organization the nervous system holds across excitation and inhibition, rhythm and background, moment to moment. Every recording of brain activity is a partial reading of it. It is the same reading at every scale. The tone of a membrane, the tone of a circuit, the tone of a thalamocortical loop, the tone of a whole cortex.

State the model's position plainly. Tone is the organizing state the rhythms are expressions of, and it decides what any incoming signal becomes once it arrives. The same light, the same sound, the same touch, the same shock of bad news meets a different tone in two people and becomes two different events inside them. That is the model's core operating principle, and it is the reason two people with identical scans can live in different worlds.

The components of this account are not new. Every piece named on this page was measured by somebody else, published in a mainstream journal, and cited here. The contribution is the unification, and the unification is a substantive claim rather than a vocabulary change.

Four things, it says, are one organization read at different resolutions. The membrane potential of a cell. The excitation to inhibition ratio of a circuit. The phase relationships between regions. The descending regulation of the organs. It follows that a change in one predicts change in the others.

That is a claim specific enough to be tested, and its edge should be stated. A body of work in which cortical excitability, autonomic regulation and cortical map organization are recorded in the same people and load on one common factor confirms it. The unified reading is the assertion, and that shared factor is the finding that establishes it.

A healthy brain holds itself at the edge between silence and seizure

If health is a wide, flexible range of organization, that phrase has to mean something measurable in brain activity. It does. The measurement is called criticality, and cortical tissue tunes itself to it.

In 2003 two researchers at the National Institute of Mental Health grew sheets of rat cortex on a sixty-channel electrode array. John Beggs and Dietmar Plenz then watched what the tissue did with no stimulation at all. Activity came in cascades. One patch would fire, recruit neighbors, and the cascade would either die out or spread.

They measured how big the cascades were and found the sizes followed a power law with an exponent of minus three halves, the mathematical signature of a system poised at a critical point. Small cascades were common, enormous ones rare, and every size in between occurred.

The number that matters is the branching parameter. It describes, on average, how many further cells one firing cell recruits. Below one, activity fades and information dies before it travels. Above one, activity multiplies and the tissue runs away into a seizure. Beggs and Plenz measured a value close to one, and their simulations showed that this exact setting maximizes information transmission while preventing runaway excitation. The tissue had tuned itself to the edge.

Later work asked what that means for a whole living brain. The computational neuroscientists Gustavo Deco and Viktor Jirsa modeled resting brain activity and showed that a brain near criticality is multistable, able to visit many organized states rather than sitting in one. Criticality buys flexibility. It lets a system go anywhere quickly without falling apart.

What criticality can and cannot tell a clinic today

Whether any of this can be read off a patient is a separate question, and the answer today is no. The child neurologist Vincent Zimmern surveyed the human literature and found criticality measures applied across epilepsy, anaesthesia, sleep medicine, developmental pediatrics and psychiatry. His verdict was blunt.

Studies inside a single domain contradict each other, with some reading seizures as departures from criticality and others as expressions of it. Samples run to two or three patients. Older ways of fitting a power law produce signatures that are not there. Criticality is a research program with a genuine measurement attached, not a clinical marker.

Read it through the model. Criticality is the measurable face of a flexible, poised, self-correcting range. A brain at criticality can be moved and comes back. A brain that has slid away from it is either damping everything toward silence or amplifying everything toward runaway, and both feel like illness from the inside.

Health is the width of the range brain activity can move through and still return. Tone held in that wide range is health. Tone that drifts out of it, toward the silent side or the runaway side, is what manifests as illness even while every scan stays clean.

The background the field discarded carries the excitation to inhibition balance

Every EEG contains named rhythms sitting on a smooth downward slope of power, the aperiodic component, and for most of a century that slope was subtracted away as noise. It was reporting the accelerator to brake ratio the whole time.

The correction came from a laboratory at the University of California, San Diego. In 2017 Richard Gao, Erik Peterson and Bradley Voytek asked whether the discarded slope carried information. They built a computational model and then tested its prediction against published recordings from rats and macaques. The steepness of the slope tracked the balance of excitation and inhibition in the tissue, and it moved in the predicted direction when general anesthetic shifted that balance.

The same group then built the tool that made this usable. It is a method for separating any power spectrum into its periodic peaks and its aperiodic background. Failing to separate them, they showed, had been producing errors in the literature for years. What looked like a change in alpha power was often a change in the background underneath it.

The slope also changes with age. Voytek and colleagues examined intracranial recordings across a thirty-eight year age range, alongside scalp recordings from younger and older adults. The slope was flatter in older brains, and flatter slopes tracked poorer cognitive performance.

Excitation to inhibition balance as an axis of disorder

The balance the slope reads had already been proposed as a general axis of disorder. In 2003 the developmental neurobiologist John Rubenstein and the neuroscientist Michael Merzenich made a proposal, and they scoped it carefully. Some forms of autism, they argued, are caused by an increased ratio of excitation to inhibition in sensory, mnemonic, social and emotional systems. Not all forms, and not one system.

That was a hypothesis about correlation until it was tested causally. In 2011 Ofer Yizhar and colleagues in Karl Deisseroth's laboratory at Stanford used optogenetics. The technique inserts light-sensitive proteins into chosen cells so that a pulse of light switches them on. They used it to raise the excitation to inhibition ratio in mouse prefrontal cortex directly. Raising it degraded information processing and produced social deficits, and raising inhibition alongside it partly restored function.

Here is the model's reading, stated as the model's. What the dominant paradigm discarded as background was the tone. The field looked for peaks because peaks look like events, and events are what a reflex machine has. The organizing state was in the part of brain activity that had no peaks at all, and it took the field the better part of a century to go back and look at it.

The cortex is built out of body traffic

Brain activity is assembled, continuously, out of the signal traffic arriving from the body. The fastest way to prove that is to lie to it.

Inside every muscle there are small sensors called spindles that report how fast the muscle is being stretched. In 1982 two French neurophysiologists, Jean-Pierre Roll and Jean-Pierre Vedel, applied a small vibrator to the tendon of a subject's biceps or triceps, out of sight, with the arm held completely still. The subject felt the elbow moving, smoothly and continuously, in the direction that would stretch the vibrated muscle. The illusion was strongest around seventy to eighty cycles per second.

Roll and Vedel then pushed a fine electrode into a living nerve to record spindle endings directly, in the leg rather than the arm, to find out what the vibration was doing to the traffic. Vibration of the same small amplitude drove the primary spindle endings in time with its own cycle. They fired once per cycle up to thirty or fifty cycles a second, then followed it in a sparser pattern above that. The secondary endings barely responded.

The arm is not moving. The person feels it move, vividly, because the brain's picture of the body is constructed from the incoming stream rather than read off the limbs. Feed the stream a lie and the construction changes.

The same is true of the inner body. The neuroanatomist Bud Craig traced a pathway up to the insular cortex. It carries signals about the physiological condition of the tissues: temperature, chemical state, muscle burn, visceral state. Craig argued that this stream constitutes a distinct sense of the material self. The brain's representation of your inner state is built out of afferent traffic in exactly the way its representation of your limbs is.

Change the stream and the cortex rewrites itself

In 1991 Tim Pons and colleagues at the National Institute of Mental Health recorded from macaques whose arm sensory nerves had been cut twelve or more years earlier. The cortical territory that once represented the arm had not gone quiet. It had been taken over by the face, across a stretch of cortex an order of magnitude larger than anyone expected.

Four years later the German psychologist Herta Flor asked whether that reorganization has a felt consequence in people. Using magnetoencephalography, which reads the magnetic fields produced by brain currents, she mapped the cortex of arm amputees and compared the maps against their pain. The amount of phantom limb pain scaled with the amount of cortical reorganization. The reorganization was not incidental to the suffering.

And the rewriting does not require an injury. In 1996 the physical therapist Nancy Byl, working with Michael Merzenich and William Jenkins, trained owl monkeys to perform a rapid, repetitive, highly stereotyped hand task, hundreds of trials a day. The animals developed a movement control disorder. When the cortex was mapped, the hand representation had smeared together, with receptive fields ten to twenty times larger than normal and the boundaries between fingers dissolved. Ordinary repetition, applied without variation, had degraded the map into focal dystonia.

This is the model's engine, demonstrated causally rather than by correlation. Cutting the input changed the cortex. Changing the training changed the map. Changing the map tracked the pain. The relationship between afferent traffic and cortical organization runs in both directions, in more than one species.

Migraine aura is a region losing its organization and getting it back

A wave of cortical silence that crawls at three millimeters per minute was first seen in rabbits in the 1940s. It is the clearest single picture of a collapse of organization, and of how one repairs itself.

In the early 1940s a Brazilian physiologist named Aristides Leao was working on the electrical activity of the living cortex, trying to understand epilepsy. He stimulated rabbit cortex expecting to provoke the spreading excitation of an experimental seizure. He got the opposite.

A wave of electrical silence rolled out from the stimulated point in every direction, flattening the ongoing brain activity as it passed, and it traveled slowly, at roughly three millimeters per minute. Behind the front, over several minutes, the activity came back on its own.

Three millimeters a minute is far too slow to be a nerve signal. Nerve conduction is measured in meters per second. Nothing was traveling. An organized state was collapsing locally, dragging its neighbors down with it, then reassembling.

Leao's wave stayed a laboratory curiosity for decades, suspected of explaining migraine aura but impossible to confirm in a living human head. In 2001 a team at Massachusetts General Hospital led by Nouchine Hadjikhani solved that. They put subjects who reliably get visual aura into a high-field scanner and recorded continuously until an aura began.

The blood signal changed in a slow front that crept across the visual cortex at three and a half millimeters per minute. It followed the map of the visual field exactly as the patient described the shimmer moving across their sight.

Read it through tone. Nothing broke during that aura. No cell died, no vessel blocked, no lesion formed. A region's organization collapsed, held excitation and inhibition in a single saturated state where it could no longer compute, and then rebuilt itself. The symptom was the collapse of a regulated range in one patch of cortex, moving as a front. And then the range came back, which tells you the capacity to reorganize was never lost.

One dysregulation wearing four diagnostic names

In 1999, patients with neurogenic pain, tinnitus, Parkinson's disease and depression were found carrying the same abnormality of resting brain activity. Four specialties, four checklists, one slipped setting.

The finding came from Rodolfo Llinas, a neuroscientist at New York University, working with Urs Ribary, Daniel Jeanmonod, Eugene Kronberg and Partha Mitra. Their instrument was magnetoencephalography, which reads brain magnetic fields from outside the head without any stimulation at all. They recorded spontaneous resting activity in healthy controls and in four groups of patients whose diagnoses have nothing to do with one another.

All four patient groups showed the same abnormality. Compared with controls, they had increased low-frequency theta rhythm at rest, together with a widespread increase in coherence between low and high frequency oscillations. Llinas and colleagues proposed a mechanism. The loop between the thalamus, the relay at the center of the brain, and the cortex has two modes. When thalamic cells sit at their normal charge, the loop passes information faithfully.

When those cells are held too negatively charged, by too much inhibition or too little excitation, they switch into a bursting mode that locks the loop into a slow theta rhythm. Around the edges of the slowed territory, fast gamma activity breaks out where inhibition is asymmetric, and that ectopic fast activity is what produces the positive symptoms: the pain, the ringing, the tremor.

They named it thalamocortical dysrhythmia. Four diagnoses, assigned by four different specialties, using four different symptom checklists, resting on one shared disturbance of the brain's baseline organization. Not one shared lesion. A shared setting.

This is the paradigm's own near miss at the model's central claim, and it deserves credit. Llinas and colleagues got to the doorstep. What kept dysrhythmia a syndrome instead of a principle is the habit of the framework itself, which sorts patients by presenting complaint and expects each complaint to have its own cause.

Sorted that way, a shared regulatory state can only ever look like a coincidence between four diseases. Read as tone, it is the ordinary case. The organization slipped, and the symptom depends on which part of the system was carrying the most load when it did.

Why the brain activity studies disagree, and why the model expects it

Three well-known literatures on brain activity contradict themselves: migraine excitability, the theta to beta ratio in ADHD, and neurofeedback trials. The Unified Model of Tone predicts all three messes from one principle.

Start with migraine. In December 2007 the journal Cephalalgia carried two reviews in the same issue on the same question, and the field has been setting them against each other ever since. The neurologist Sheena Aurora and the vision scientist Frances Wilkinson surveyed the evidence from magnetic stimulation, visual testing and imaging.

They concluded that the migraine brain is hyperexcitable, with its threshold for triggering activity set too low. In the pages immediately before, Gianluca Coppola, Francesco Pierelli and Jean Schoenen reviewed largely the same evidence and put the weight elsewhere. The migraine cortex fails to habituate to repeated stimulation, so responses build instead of settling, and they preferred the word hyperresponsive.

Then they declined the argument. The dispute between hyper and hypo excitability, they wrote, is a semantic misunderstanding, and calling the migraine brain hyperresponsive fits most of the available data. Their proposed culprit was thalamocortical dysrhythmia, the same slipped loop found under four unrelated diagnoses. Two teams, two vocabularies, one organization underneath, and the disagreement is mostly in the naming.

The theta to beta ratio and the sham-controlled trial

Now ADHD. For years the theta to beta ratio was promoted as an objective marker for the diagnosis. It measures slow activity relative to fast at the top of the head, and a device based on it reached the market. In 2013 Martijn Arns, C. Keith Conners and Helena Kraemer pooled nine studies covering more than seventeen hundred children.

The average effect looked large, but the studies disagreed with each other so strongly that the average was judged misleading and an overestimate. The gap between groups had been shrinking across the decade, driven by the ratio rising in the children who did not have the diagnosis. Their conclusion was precise: the ratio cannot be considered a reliable diagnostic measure, and yet a substantial subgroup of patients does deviate on it, in whom the measure carries prognostic value.

Then the hardest result. In 2017 the clinical psychologist Michael Schonenberg and colleagues at Tubingen ran a triple-blind randomized trial of EEG neurofeedback in adults with ADHD, comparing real training against a sham condition and against group therapy. Symptoms fell substantially in all three groups and stayed down at six months. Real neurofeedback was no better than sham.

The temptation is to explain these away as underpowered or poorly designed. The model reads them differently: all three findings are what it predicts.

A label is not a tone

A population sorted by diagnosis is not a population sorted by tone. Take a group of people who all carry one label and one shared symptom. You have gathered a person whose cortex runs over-aroused, a person whose cortex runs under-aroused, and a person whose peaks are normal while the background underneath has flattened.

All three have the same complaint. All three arrived by a different route through the same organization. Average their electrical measures and the deviations cancel, which is exactly the heterogeneity Arns and colleagues could not explain away.

Give all three the same one-directional training protocol and it corrects the tone of some, worsens the tone of others, and the group difference dissolves toward zero, which is exactly what Schonenberg found.

Ask whether the migraine cortex is set too excitable or responds too much to repetition, and the model answers that both descriptions are projections of one slipped organization onto two different measurement axes. Coppola and colleagues reached the edge of that themselves when they called the quarrel semantic and pointed at the thalamocortical loop.

Input meets tone. The same protocol, delivered to two different organizations, is two different interventions. A literature that pools by label will keep producing mixed results, and the mess is the fingerprint of the thing it is failing to measure.

How a temporary setting becomes the new normal

A slipped setting that corrected itself in an hour would never reach a clinic. Settings persist because the loops that would correct them are themselves part of what slipped.

The neuroendocrinologist Bruce McEwen at Rockefeller University gave the general answer. Systems that adapt by shifting their set points pay a price when the shift never reverses, and he called that accumulated cost allostatic load. The mechanism that protects you in the short term becomes the mechanism that harms you when it is left switched on.

There is a clean cellular example in the spinal cord. In 1983 the physiologist Clifford Woolf was studying withdrawal reflexes in rats after a peripheral injury. If the cord were a simple relay, the reflex would grow only because the damaged tissue was sending more signal.

Woolf showed that the excitability of the cord itself had increased, a central component of post-injury hypersensitivity, and that it outlasted the input that caused it. The gain had been turned up and left up. Neuroscience now calls this central sensitization, and it is a working model of how a nervous system can hurt without anything remaining broken at the site.

The same persistence, read at the whole-body scale

The neurologist Eduardo Benarroch at the Mayo Clinic described a central autonomic network. It runs from the insula and amygdala down to the brainstem, and together those regions govern heart, vessels, gut and breath. The psychophysiologists Julian Thayer and Richard Lane built on it in 2000 with a neurovisceral integration model. Autonomic, attentional and affective regulation run on one set of negative feedback and inhibitory processes, and anxiety disorders are what that inhibition looks like when it fails.

Nine years later they filled in the anatomy and the read-out, tracing inhibitory pathways from the prefrontal cortex to the amygdala and on to the medullary neurons that set heart rate. Losing that control shows up at once as poorer emotional regulation and as reduced heart rate variability. Cortical organization and bodily regulation are one system with two read-outs.

Then there is the nightly re-tuning that is supposed to reset all of it. In 2011 Vladyslav Vyazovskiy, working with Chiara Cirelli and Giulio Tononi at Wisconsin, recorded from cortical neurons in freely behaving rats kept awake for long stretches.

Local patches of cortex began going offline while the animal was awake and behaving, one region at a time, with slow waves appearing locally while the overall recording still looked awake. Performance on a reaching task degraded in step. Sleep and waking are local states of the same organization rather than global switches, and they can come apart.

Sleep also does physical maintenance. In 2013 Lulu Xie and colleagues in Maiken Nedergaard's laboratory at Rochester found that during sleep the space between brain cells expands substantially and the clearance of metabolic waste speeds up. A system that never fully drops into that state does not get its housekeeping done.

The widest reach of the loop was shown in 2021. Tamar Koren and colleagues in Asya Rolls's laboratory at the Technion tagged the insular cortex neurons that were active in mice during an episode of gut inflammation. Later, with the animals healthy, they switched those same neurons back on artificially. The inflammation returned, in the same tissue, in the same pattern. The cortex had stored an immune state and could reinstate it on its own.

Read all of it as one thing. A setting of brain activity persists because the loop that would correct it is itself part of what slipped. The gain that came up did not come back down, the descending regulation that would lower it is weakened, and the nightly process that re-tunes the whole arrangement is running locally and incompletely. Nothing is broken. The regulation has settled into a narrower, more expensive range and is now defending it.

Two different aims that look similar from outside

Anticonvulsants, sedatives and stimulants all change brain activity, reliably and often life-savingly. The distinction the Unified Model of Tone draws is about aim, not virtue: what is the intervention trying to do to the range?

Anticonvulsants raise the threshold for runaway excitation. Sedatives increase inhibition. Stimulants raise arousal in circuits that are running low. These drugs work, and for a person with epilepsy or severe attentional impairment they can be the difference between a functioning life and a wrecked one. Nothing here is medical advice or a reason to change any prescription.

Both are legitimate. They answer different questions. A person in status epilepticus does not need a wider range in the next four minutes. They need the excitation stopped. A person whose cortex has held one narrowed setting for eleven years has a different problem, and a permanent push in one direction is a lifelong management strategy rather than a resolution.

There is a useful example of a restoration aim stated openly in the clinical literature. The neurobiologist M. Barry Sterman spent his career at UCLA on sensorimotor rhythm training in epilepsy, the oldest and best established application of EEG operant conditioning. Reviewing the field with Tobias Egner in 2006, he set out what good practice requires.

Map a patient's quantitative EEG against a normative database before and after treatment. The purpose is to guide the protocol and to document progress toward EEG normalization. Read the goal carefully. The stated target is a return of the measured organization toward the normal range, individually determined, verified against a comparison.

That is the shape of a restoration claim, and ordinary clinic equipment can check it.

What a restoring input does to brain activity that a pushing one cannot

The Unified Model of Tone makes one prediction about brain activity precise enough to separate an input that restores regulation from one that masks a measure. A genuine restoration of tone moves a dysregulated measure toward the healthy middle from either side, and the change holds once the input stops.

Take any measure of brain activity that has a healthy middle rather than a healthy direction. There are several. The individual alpha peak frequency, meaning where a person's dominant rhythm actually sits. The aperiodic exponent, the steepness of the background slope that tracks excitation to inhibition balance. The theta to beta ratio. Heart rate variability, as the autonomic read-out of the same regulation.

Now phenotype people before an intervention and split them by where they start. Some sit above the healthy middle on the chosen measure. Some sit below it. Give both groups the same input, in the same session, with the same operator.

Why the naive version of the test fails

The obvious first version of the prediction is that a drug pushes everyone one way while a restoration moves people toward the middle from both sides. That version is wrong, and the pharmacology says so plainly. Stimulants have baseline-dependent effects. In 1979 Trevor Robbins and Barbara Sahakian argued that the calming of an overactive child by a stimulant is not paradoxical once behavioural pharmacology is taken into account.

A drug acts on a rate of behavior rather than on a person. Weber tested it directly in 1985 and found that boys with lower baseline response rates increased while boys with higher baselines decreased or did not change. Teicher and colleagues reported in 2003 that methylphenidate alters activity and attention in a manner inversely related in magnitude and direction to the baseline. A drug can move two people opposite ways.

So the claim has to be sharper, and it can be. Rate dependency is a graded tilt on one pharmacological action. Its direction only reverses near the middle of the range. It has to be dug out from under regression to the mean before it can be seen, and it ends when the drug clears. The model predicts a different shape.

Give the same non-pharmacological input, in one session, to a group split at the healthy middle on the chosen measure. Where a person started should predict which way they move better than the intensity of the input does. The high measure trends down, the low measure trends up, and the change holds after the input stops, because what changed was the organization rather than the chemistry in the blood.

The three patterns that mark a mask

State the reading in advance, because a prediction without a stated reading is not a prediction. Phenotype the subjects first, split them at the middle, correct for regression to the mean, and follow them past the session. Three patterns mark an input that pushed the measure. A uniform shift in one direction with no convergence.

A baseline position that does not predict the direction of change once the statistical artifact is removed. A convergence that collapses the moment the input stops. Any of the three and the input held a number in place. It did not return the range that generates the number.

The framework is broad on purpose, because it claims that one organizing variable is being measured under many names. That breadth rests on a single narrow discrimination that any competent laboratory with an EEG amplifier, a heart rate monitor and a pre-post design can run.

What can be read from brain activity now

Four read-outs of brain activity exist today, each sampling the same organization from a different side: the rhythms, the aperiodic background, direct cortical excitability, and heart rate variability.

The rhythms come first. Where a person's dominant alpha frequency sits, how the fast rhythms are coupled to the slow ones, whether the cortex responds to a repeated stimulus by settling or by building. These are the coupled voices of the chord, and each is a partial view.

The background comes second, and it is the newer and in some ways better read-out. The steepness of the aperiodic slope tracks excitation to inhibition balance and can be extracted from ordinary recordings that have already been collected, using the separation method built for exactly that purpose.

Cortical excitability can be measured directly rather than inferred. In 1985 a medical physicist named Anthony Barker, working at Sheffield with Reza Jalinous and Ian Freeston, discharged a magnetic pulse through a coil held against the scalp over the motor cortex.

The pulse induced a current in the cortex underneath and the subject's hand twitched, painlessly and without any electrode touching the brain. Transcranial magnetic stimulation grew out of that afternoon, and the strength of pulse required to produce a twitch is a direct measure of how excitable a cortex is at that moment.

Fourth is the body's side of the same regulation. Heart rate variability, the beat to beat variation in the interval between heartbeats, is a validated index of autonomic regulation with established metrics and normative values, as reviewed in detail by Fred Shaffer and J. P. Ginsberg. That is settled science. Reading heart rate variability as a window onto tone is the model's interpretation, offered as an interpretation, and the two claims stay separate.

Two load-bearing cautions

The first: findable causes must be found. A tumor, a stroke, an infection, a metabolic derangement, a structural malformation, a genetic epilepsy syndrome are real, detectable, and treated on their own terms. Nothing in a tonal reading substitutes for that workup. The model addresses the enormous remainder, the cases where the imaging is clean and the pathology is regulatory. Deciding which case is in front of you is a medical judgment made by a physician with the images in hand.

The second: the tone reading is a way of reading evidence rather than a treatment protocol. What it changes is where you look. The baseline is the phenomenon, and the background under the peaks carries the organization. The same measure can mean opposite things in two people.

A group averaged by diagnostic label will hide the very thing that would explain it. A brain that can move to meet the moment and settle again is more than a normal report. It is a nervous system with its range back.

How brain activity relates to the rest of the library

Brain activity is the recording; the foundations of tone are what the recording is a recording of. The library keeps them on separate pages because collapsing them is what makes explanations vague.

Two instrument pages sample the same organization from the body's side. Heart rate variability is the autonomic read-out that any test of the bidirectional prediction would carry alongside the EEG. The autonomic nervous system is the anatomy through which cortical organization reaches the heart, the vessels and the gut.

Four condition pages put this page's physiology to work.

• Migraine is where Leao's slow wave and thalamocortical dysrhythmia stop being laboratory findings and become a person's Tuesday.
• Tinnitus is one of the four diagnoses Llinas found in 1999 sharing a single slipped baseline.
• Depression is another of the four, read by its own literature as a disorder of network organization rather than of any single region.
• Sleep is the nightly re-tuning of the whole arrangement, and the page where local sleep and waste clearance get their full treatment.

Frequently asked