Homeostasis, Allostasis, and the Body’s Set Point

A thermostat shows how a set point works

Your body holds its temperature near 98.6 degrees, its blood pH near 7.4, and its blood sugar inside a narrow band. Each defended value is a set point, and homeostasis is the defending.

A thermostat runs the same loop in a simpler machine. Set the dial to 68 and the system knows one thing: the difference between 68 and the room. Let the room cool to 65 and that three-degree gap is the whole instruction. The furnace runs until the room is back at 68, then stops.

Let the room warm to 72 and the gap runs the other way, so the air conditioning kicks on and pulls the room back down to 68. Nothing in the system understands warmth. It holds one number, reads another, and works from either direction to make them match.

Move the dial to 78 and the furnace does not object. Same machinery, same effort, new target. If an open window cools the room toward 68, the system now fights that as an error. Nothing is broken. The target moved, and everything downstream moved with it.

The body runs this loop for temperature, blood pressure, pH, blood sugar, and dozens of other defended values at once. The interesting failures are almost never a broken furnace. They are a moved dial. High blood pressure with no findable cause behaves exactly like a thermostat set to 78: intact machinery, working correctly, defending the wrong number.

A body is also better than a thermostat. A thermostat waits for the room to change. A body moves its own dial ahead of demand, raising blood pressure before you stand and body temperature before dawn. That difference is the second half of this lesson, and it is where homeostasis becomes allostasis.

What a set point actually is

A set point is the reference value inside a negative feedback loop, and homeostasis is that loop defending it. Every regulated value in the body sits inside a loop with four parts, and this page uses the loop by name.

First, a sensor. Something has to feel the current value: a stretch receptor in an artery wall, a temperature-sensitive neuron, a cell that reads how much sodium is dissolved in the blood. Second, a reference. Somewhere the system holds the value it is aiming for.

Third, a comparator, which is the piece that subtracts one from the other and produces an error signal. Fourth, effectors: the muscles, glands, vessels, and behaviors that change the actual value until the error shrinks. That is a negative feedback loop, and the reference inside it is the set point.

Bernard, Cannon, and Selye each built part of homeostasis

The idea of homeostasis arrived in stages, and the people who built it were each looking for something specific.

Claude Bernard was a French physiologist working in the middle of the nineteenth century. His question was why a mammal can walk from a warm room into freezing air and stay recognisably the same animal inside. His answer appeared most fully in his 1878 lectures on the phenomena of life.

A body maintains its own internal sea, the fluid that bathes every cell, at conditions far steadier than the world outside. The constancy of that internal environment, he argued, is the condition of a free and independent life. An animal that must match the weather is a prisoner of the weather.

Walter Cannon was an American physiologist at Harvard, and he wanted a word for the machinery Bernard had described. He coined one in 1926 and set the idea out most fully three years later, in a review titled Organization for Physiological Homeostasis.

Homeostasis, from the Greek for similar and standing, was his term for the coordinated defense of steady internal conditions. His emphasis on organization was deliberate. He was not describing one loop; he was describing many loops that agree with each other, mobilized together, sometimes ahead of the demand.

Hans Selye was an endocrinologist in Montreal, and he was not looking for any of this. He was injecting rats with a series of crude tissue extracts, hoping to find a new hormone. What he found instead was that every injection, and cold, and injury, and toxins, produced the same three changes in the same order.

In a short 1936 letter to Nature he reported a syndrome produced by diverse nocuous agents. The adrenal glands enlarged, the lymphatic tissues shrank, and the stomach lining ulcerated, whatever had been done to the animal. The defense was general, and the defense cost something. That second observation is the seed of everything in this page about load.

From homeostasis to allostasis

Allostasis is the upgrade to homeostasis: the brain moves the set point ahead of predicted demand instead of waiting to correct an error. Homeostasis explains a thermostat very well. It explains a body less well, and the gap shows up in ordinary life.

Your blood pressure rises before you have finished standing up, not after you feel faint. Your body warms in the hours before dawn, in advance of the day, rather than in response to it. Insulin release begins at the sight and smell of a meal, before a single molecule of sugar has crossed into the blood. None of that is error correction. All of it is anticipation.

Peter Sterling is a neuroscientist at the University of Pennsylvania. He spent his career on how the retina achieves so much with so little energy, and he came to regulation from the direction of efficiency. Working with the epidemiologist Joseph Eyer, he proposed that the standard picture had it backwards, and he set out the full argument in a paper called Allostasis: a model of predictive regulation.

His point is blunt. Correcting an error after it happens is expensive and slow. A system that can predict a demand and prepare for it makes smaller errors, less often, and needs less spare capacity everywhere.

So the brain, in this account, does something a thermostat cannot. It tracks many variables at once, combines them with what it has learned, and sets the targets. Allostasis means stability through change: the values move, and the moving is the point. A blood pressure that rises for a sprint and settles afterward is a system holding the right value for the moment.

Allostatic load is the cost of a held adjustment

Credit belongs where it belongs. Allostasis is Sterling and Eyer’s idea, not this model’s. So is its cost side. Bruce McEwen, a neuroendocrinologist at Rockefeller University, took the concept and asked what happens when the adjusted state is held too long, called on too often, or never switched off.

He named the accumulated wear allostatic load, and he was explicit that the load comes in several distinct forms, including a system that cannot shut down after the demand passes. The concept is theirs. What this model contributes is the claim about how the targets across the whole body are related to each other.

Where set point ends and tone begins

In the Unified Model of Tone, set point is a foundational dimension of tone: the value the organization is currently defending, a coordinate inside the healthy range and never the range itself.

This library is built on a single variable. Tone is the integrated, coupled organization the nervous system maintains across the whole body, and its ability to move where the moment demands and return to balance afterward. Health is the width of that range. The pillar page on tone makes that case in full.

A set point is not that. A set point is one value the organization has chosen to defend. It is a coordinate inside the range, and the range is not made of one coordinate.

Health is a wide state-space, not a set point. Defending a value flexibly is health. Defending it rigidly is a different condition wearing the same number.

That distinction has to be held firmly, because the two ideas collapse into each other easily and the collapse ruins both. If tone were a set point, then health would mean sitting at 120 over 80 and never leaving, and a body that never left would be dying. If a set point were tone, then every regulated value would be its own separate affair and there would be nothing to unify.

What the model claims is the middle position. There are many defended values across the body, running on different clocks and different tissues, and they are set by one organization rather than by a committee of independent thermostats. That is why they drift together, why they recover together, and why a person can present with a blood pressure problem, a sleep problem, and a digestive problem that no single organ explains.

Set point is one of the foundations of tone, the small set of properties along which any nervous system can be described. Any condition combines two or three of them. Set point comes first because the others all reference it, and section fifteen draws the line against each in turn.

A single neuron defends its own firing rate

Homeostasis is a property of cells before it is a property of brains. The cleanest proof that set points are real happens inside one cell.

A neuron is a cell that talks in electrical pulses. It receives from thousands of other neurons through junctions called synapses, and each synapse has a strength: how much of a push that particular input delivers. Learning works by changing those strengths. Useful inputs are strengthened, unhelpful ones weakened. That process has an obvious danger built into it. If strengthening keeps happening, the cell fires more, which strengthens it further, and the circuit runs away into silence or into seizure.

Gina Turrigiano is a neuroscientist at Brandeis University, and this was exactly her question. How does a brain that is constantly rewriting synaptic strengths stay stable at all? In a 1998 experiment she grew cortical neurons in a dish, blocked their activity for two days, and then measured every synapse. She found activity-dependent scaling of quantal amplitude: all of the excitatory synapses had grown stronger by roughly the same multiplicative factor. When she drove the activity up instead, they all shrank by a common factor.

The elegance of the finding is in the word multiplicative. The cell turned every input up or down proportionally, which preserved the relative pattern among them. Whatever the cell had learned survived intact. Only the overall volume changed. Turrigiano later gathered the field’s work under the phrase the self-tuning neuron, and the mechanism is now standard textbook material as synaptic scaling.

The critical detail for this page is that the correction runs both ways. Too quiet and everything comes up. Too loud and everything comes down. That is a defended value, in one cell, with no brain required.

The firing-rate set point holds in a living animal

Keith Hengen, then working in Turrigiano’s laboratory, asked whether the same thing happens in a living, behaving animal. He implanted electrodes in the visual cortex of rats and recorded individual neurons continuously while the animals ate, slept, and moved, then deprived one eye of vision.

Firing rates dropped, then climbed back over the following days despite the deprivation continuing. In a 2016 report he showed that each neuron returned to a precise, cell-autonomous set point, and that the returning happened during active waking and was suppressed during sleep.

Each cell has its own number. It knows the number. It gets back to it from either direction. Everything larger in this page is that same architecture at a bigger scale.

The same set point architecture repeats at three scales

A reflex corrects blood pressure inside one heartbeat, the kidney corrects it over days, and body weight is defended across years. The architecture is identical at every scale, and only the clock changes.

Take one reflex first. In the walls of the large arteries in your neck and chest sit stretch sensors called baroreceptors. They feel how hard the blood presses with each beat and report it continuously to the brainstem, the stalk where the brain meets the spinal cord.

When pressure rises, the brainstem eases the accelerator and applies the brake, so the heart slows and vessels relax. When pressure falls, it does the reverse. The whole correction happens within a single heartbeat. That is the baroreflex, and it is the fastest defended value in the body.

Blood pressure is defended on two clocks

Take one organ system next. Arthur Guyton was an American physiologist who spent decades on the question of what actually determines blood pressure over months and years rather than seconds. His answer, summarised in a 1991 review on the special role of the kidneys and body fluids, was that the kidney sets the long-run level.

When pressure rises, the kidney excretes more salt and water, the volume of fluid in the circulation falls, and pressure follows it down. The loop is slow, working over hours and days, and Guyton argued it has the final authority over the long-term number.

Notice the layering. A fast neural loop and a slow renal loop defend the same blood pressure set point on different timescales. The layering is how one value gets held across a sprint and across a decade at once.

Body weight is a defended value

Then take the whole body. Douglas Coleman was a geneticist at the Jackson Laboratory, and he wanted to know whether the obesity of certain mutant mice came from a missing signal carried in the blood. He tested it surgically, joining the circulations of two animals so that they shared blood, a technique called parabiosis.

When he joined an obese mouse to a diabetic mouse, the obese mouse stopped eating and lost weight while its partner did not change. He concluded that one strain was producing a circulating satiety signal and the other could not read it. Two decades later that signal was identified as the hormone leptin.

Rudolph Leibel and his colleagues at Rockefeller University then asked whether human body weight is defended in the same way. They kept eighteen obese and twenty-three never-obese volunteers in a research setting and deliberately moved their weight up ten percent and down ten percent, measuring energy expenditure at each level.

They found changes in energy expenditure resulting from altered body weight that pushed in the opposing direction each time. Below the usual weight, the body burned about six kilocalories per kilogram of fat-free mass per day less than its new size predicted. Above it, the body burned about eight more. The correction ran in both directions, exactly as it did in Turrigiano’s dish.

The baroreflex curve separates set point from gain

The baroreflex curve puts a set point on paper. The midpoint of the curve is the pressure the reflex defends, the slope is the gain, and the midpoint moves.

Plot blood pressure along the bottom of a graph and the reflex’s answer up the side, measured as heart rate or as the traffic in the sympathetic nerves. The resulting line is an S. At very low pressures the reflex is already doing everything it can, so the line is flat. At very high pressures it is equally maxed out at the other end, so the line is flat again. In between, the line is steep, and small changes in pressure produce large corrections.

In 1972, B. B. Kent and colleagues published a mathematical model to assess changes in the baroreceptor reflex. It fitted that S-shape with four numbers, so that two experiments could be compared honestly. Two of those numbers matter here. One is the midpoint, the pressure the curve is centerd on, which is the set point. The other is the slope, how sharply the output changes for each unit of pressure, which is the gain.

The midpoint is which value the system is defending. The slope is how loudly it argues. They are two different numbers on one line, and they move independently.

This one figure does the boundary work for the whole ontology. A reflex can be centerd on the right pressure and respond too weakly, which is a gain problem. It can respond briskly and sharply around the wrong pressure, which is a set point problem. Reading a flat response as a high set point, or a drifted center as a weak reflex, is the commonest confusion in this territory. The page on gain takes the slope; this page takes the midpoint.

Baroreceptors reset their own set point

Now the central fact. That midpoint is not fixed. Baroreceptors re-center on whatever pressure they have been feeling. Mark Chapleau and his colleagues at the University of Iowa reviewed the mechanisms of resetting of arterial baroreceptors and separated it into three timescales. There is instantaneous resetting within a single cardiac cycle, acute resetting after a brief sustained rise, and chronic resetting after a long one or after the vessel wall has changed.

How local is that? Philip Munch, with Michael Andresen and Arthur Brown, took an aortic arch out of a rat. They kept it alive in a bath with its nerve attached, cut off from any brain and any hormone.

Raising the holding pressure produced rapid resetting of aortic baroreceptors in vitro with a time constant of three to five minutes. The threshold shifted along the pressure axis and the slope did not change. The sensor moved its own set point, and left its gain alone, with nothing else in the room.

Resetting is also used on purpose. Roger Dampney, a physiologist in Sydney, assembled the evidence that the reflex’s midpoint is deliberately shifted upward during exercise and defense reactions, describing resetting of baroreflex control during natural behaviors. During a sprint the reflex keeps correcting just as sharply, around a higher target. That is allostasis inside a single reflex, and it is completely normal.

The forty-year argument over whether the reflex sets the pressure

The field argued about this for forty years, and the argument is worth telling because it is where the concept earns its keep.

In 1973, Allen Cowley, working with Jean-François Liard and Arthur Guyton, cut the baroreceptor nerves in dogs and then recorded pressure continuously for days to see what the reflex had been contributing. Their answer, published as the role of the baroreceptor reflex in daily control of arterial blood pressure, was that the twenty-four hour average barely moved while the variability exploded. The conclusion drawn for a generation was that the reflex steadies pressure and does not set it.

Terry Thrasher, a physiologist at the University of Maryland, thought the experiment had answered a slightly different question. Cutting a nerve removes the sensor entirely. He instead left one carotid sinus innervated and reduced the pressure it was exposed to by tying off the artery upstream, so the sensor stayed connected and simply reported a lower value.

Mean pressure rose from about 104 to about 127 millimetres of mercury and stayed there for the seven days of unloading, and he titled the result unloading arterial baroreceptors causes neurogenic hypertension. Untying the artery normalized everything.

Read that result carefully, because it is the mechanism of this page in one animal. Nothing was damaged. A sensor was given a lower number to report, and the body raised the pressure to make the report come out right. Give it the true number back and the pressure returns.

The clinical version was recorded even earlier. In 1956 James McCubbin, with John Green and Irvine Page at the Cleveland Clinic, measured baroreceptor nerve traffic in dogs with long-standing renal hypertension and reported on baroceptor function in chronic renal hypertension. The receptors had reset to the elevated pressure. The brain was receiving a normal report from an abnormal pressure, and so the brain had no reason to object.

Thomas Lohmeier and Radu Iliescu later reviewed the whole question and argued that the baroreflex acts as a long-term controller of arterial pressure. Chronic activation of the reflex lowers pressure durably rather than fading away. The reflex never entirely surrenders to the drifted value. That residual grip is what any attempt at restoration has to work with.

Fever and pregnancy prove the brain moves the target

A set point is a physical, locatable reference the brain moves on purpose. Fever raises the temperature reference through one receptor in one hypothalamic nucleus, and pregnancy lowers the defended plasma osmolality by about ten milliosmoles per kilogram.

Fever is not a failure of cooling

When you run a fever you feel cold. You shiver, your hands go pale, you pull up a blanket. Those are the body’s heat-conserving responses, and they are running while your temperature is already above normal. If fever were a cooling failure that would make no sense at all.

It makes complete sense as a moved dial. Kazuhiro Nakamura, a physiologist in Kyoto, mapped the circuitry in detail in a review of central circuitries for body temperature regulation and fever. Signals from an infection cause prostaglandin E2 to be produced in the brain.

It acts on the preoptic area of the hypothalamus, a small structure that holds the temperature reference. The reference goes up. Everything downstream then works to reach the new target, which is why you shiver and constrict your vessels at 39 degrees.

Michael Lazarus, working with Clifford Saper at Harvard, pinned it to a single receptor in a single nucleus. Using genetic tools in mice, they deleted one prostaglandin receptor, EP3, from one small structure called the median preoptic nucleus. Their paper reported that EP3 prostaglandin receptors in the median preoptic nucleus are critical for fever responses. Fever was abolished.

A set point, then, has an anatomical address.

Pregnancy is a reset that ends

John Davison, an obstetric physician in Newcastle, worked with the nephrologist Marshall Lindheimer on what happens to the body’s water regulation during pregnancy. They studied eight women repeatedly: before conception, at three points during gestation, and again after delivery. Each time they infused salt solution and measured both the thirst response and the release of vasopressin, the hormone that tells the kidney to retain water.

Their report on serial evaluation of vasopressin release and thirst in human pregnancy found both thresholds falling by early pregnancy. Plasma osmolality, the concentration of dissolved particles in the blood, settled about ten milliosmoles per kilogram below the non-pregnant value and stayed there throughout.

By any laboratory standard those women were running low. They were not sick. The target had been lowered deliberately to support the expansion of blood volume that pregnancy requires. Davison and Lindheimer traced the trigger to human chorionic gonadotrophin, the hormone of early pregnancy, and by the postpartum study visit the thresholds had returned to their non-pregnant values.

Hold those two examples together. A reset can be exactly right, and a right reset ends when its occasion ends. That is the standard against which the failures in the next section should be read. A sympathetic bias set during a hard year is the same machinery. The difference is that it outlives the year.

How a set point goes wrong

A set point fails in five distinct ways: it drifts, it rigidifies, it goes undefended, it forms wrong, or several drift together. Each failure has its own signature, and collapsing them into the single word dysregulation loses everything useful.

One. The target drifts, and is now defended at the wrong value

This is the failure that gives set point its clinical weight. Nothing is damaged. The loop works. The loop is aiming at the wrong number and defending it competently.

Essential hypertension is the clearest case. The word essential means no cause was found, which is true of roughly nine in ten people with high blood pressure. The baroreceptors have reset, as McCubbin recorded. The kidney has resettled its pressure-volume relationship. The brainstem’s baseline drive has come up. Every component is doing its job around a target that has moved, and the twenty-four hour record shows it.

Paolo Verdecchia and colleagues followed 1187 people with essential hypertension and showed that ambulatory blood pressure independently predicts prognosis. Among those whose pressure stopped dropping at night the event rate ran nearly three times higher, a difference that reached significance in the women of that cohort and not in the men. The population evidence in the next subsection carries the claim without that qualification.

Body weight behaves the same way, and the evidence is unusually strong. Priya Sumithran and Joseph Proietto in Melbourne put fifty people through a ten-week very-low-energy diet, then measured the appetite hormones again a full year later. They found long-term persistence of hormonal adaptations to weight loss: leptin still suppressed, ghrelin still elevated, subjective hunger still raised, twelve months on.

Erin Fothergill and Kevin Hall at the National Institutes of Health followed fourteen contestants from a televised weight-loss competition. They found persistent metabolic adaptation six years later, with resting metabolic rate still several hundred kilocalories per day below what body composition predicted.

Thyroid regulation shows a subtler version. Stig Andersen in Denmark sampled sixteen healthy men monthly for a year and found narrow individual variations in serum thyroid hormones. Each man occupied a personal band roughly half the width of the population range. A person whose thyroid target has drifted can move a long way from their own value while every result stays inside the laboratory’s normal limits.

Glucose has a defended target too, and the brain is part of setting it. Michael Schwartz and colleagues at the University of Washington described a brain-centerd glucoregulatory system in a review of cooperation between brain and islet in glucose homeostasis and diabetes. They argue that the defended blood sugar level is set jointly by the brain and the pancreas rather than by the pancreas alone.

Two. The target is defended too rigidly

A value held at one number no matter what the day asks is a failure, even when the number is respectable.

The cleanest measurement of this is the night. Healthy blood pressure falls by ten to twenty percent during sleep, because the demand falls and the target follows it down. Takayoshi Ohkubo and colleagues followed 1542 residents of Ohasama in Japan for an average of 9.2 years. They found a linear relationship between the nocturnal decline in blood pressure and cardiovascular mortality, present in people whose twenty-four hour averages were high and in people whose averages were not.

Read that carefully. The daytime number can be acceptable and the failure to move at night still carries risk. What has been lost is the permission to leave the level, the allostatic half of the machinery.

Three. The target is no longer being defended at all

The opposite failure is a value the body has stopped holding. Stand up and gravity pulls roughly half a litre of blood into your legs within seconds. A working system detects the fall, tightens the vessels, and speeds the heart before you notice anything. When that defense fails, the person feels it immediately as lightheadedness, greying vision, racing heart, or collapse.

This is the territory of the autonomic disorders, and it belongs on this page because what fails there is the defense of the set point itself. Rigidity and abandonment look like opposites and they are the same axis read at two ends.

Four. The target never formed properly

Set points are calibrated, and much of the calibration happens early.

Dong Liu, working with Michael Meaney at McGill, measured how much mother rats licked and groomed their pups in the first ten days of life, then tested those pups as adults. The offspring of high-licking mothers showed lower stress-hormone responses, higher expression of glucocorticoid receptor messenger RNA in the hippocampus, and better shut-off of the stress response. The paper on maternal care, hippocampal glucocorticoid receptors, and stress responses reported that each measure tracked the frequency of maternal licking. Early handling had set the adult target.

David Barker, an epidemiologist at Southampton, found the human parallel in birth records. His work on the fetal origins of coronary heart disease reported that people whose birth weights sat at the lower end of the normal range had higher rates of later coronary disease. He argued that blood pressure, insulin responses, and hormonal settings are programmed before birth.

A target that never formed correctly behaves differently from one that drifted later, and this is where Set Point hands off to time course.

Five. Several targets drift together

Set points rarely fail alone, and the cumulative version predicts outcomes better than any single value.

Teresa Seeman and colleagues worked with McEwen on the MacArthur studies of successful aging. They built an index from ten measures spread across the cardiovascular, metabolic, and hormonal systems, scoring how many sat in the highest-risk quartile. Their report on the price of adaptation found that higher scores predicted greater decline in physical and cognitive function and higher incidence of cardiovascular disease, independent of standard risk factors.

That result is important for a reason beyond risk scoring. If each defended value were an independent thermostat, counting how many are off would add little. It adds a great deal, which is what a single organization defending many values would predict.

Why a drifted set point produces no symptom

A drifted set point produces no symptom, because homeostasis defends the drifted value as competently as it defended the right one. That single fact explains a great deal of ordinary clinical experience.

A symptom is an error signal reaching awareness. When a value drifts away from where it should be, and the reference drifts with it, there is no error left to signal. The system rests, comfortable, holding exactly what it has been told to hold.

That is why high blood pressure has no feeling, and why it is described as silent in every public-health leaflet ever written. It is why a regained weight feels correct rather than alarming to the body maintaining it. It is why a person can be measurably far from their own best state and report feeling fine, and mean it.

There is no symptom of a moved target, because the body is no longer registering an error. It is defending the drift.

What does show early is dynamics. Before the average changes, the movement around the average changes. Peter Rothwell in Oxford analyzed several large trial datasets and found that visit-to-visit variability in blood pressure predicts stroke independently of the mean. The maximum pressure reached carries information the average does not. Ohkubo’s nocturnal dip is the same lesson on a shorter clock.

Stated in the model’s terms, the body’s registration of itself has stopped tracking its own best state. The organization is still coherent. It is coherent around the wrong value, and it will resist a return with the same competence it uses to resist a departure. That competence is the reason restoration takes time and repetition rather than a single push.

How a set point is measured, and what measurement cannot do

A set point is measured through its defense, and several ordinary clinical tools already read it. Each instrument also has a stated limit, and the limits teach as much as the readings.

For blood pressure the tools are already in ordinary use. Twenty-four hour ambulatory monitoring outperforms the office reading, per Verdecchia. Within that recording, the nocturnal dip and the morning rise are read as separate quantities. Across visits, the drift and scatter are their own measure, per Rothwell. In a laboratory the baroreflex curve itself can be fitted, giving Kent’s midpoint and slope as two numbers.

Continuous glucose monitoring turns a single fasting number into a shape. Heather Hall and Michael Snyder at Stanford put continuous monitors on people who had no diabetes diagnosis and sorted the resulting traces into patterns they called glucotypes. Even among individuals classed as normoglycemic by standard tests, glucose reached prediabetic ranges fifteen percent of the time and diabetic ranges two percent of the time. The single fasting value had been hiding the behavior.

Thyroid function has a personal band, per Andersen. Body temperature has a personal baseline. And the allostatic load index, per Seeman, counts how many defended values across systems have drifted at once.

No instrument reads a set point directly

A set point is never observed directly. There is no probe that reads the reference. Every number above is a measure of how the system behaves when it is pushed off the target, and the target is inferred backwards from the defense. You measure the argument and reason back to the position.

That inference is genuinely contested, and four serious challenges deserve to be carried rather than buried.

Andrej Romanovsky, a thermal physiologist, argues that no unified thermal set point exists at all. In a review arguing that some concepts in thermoregulation have changed, he describes the system as a collection of independent effector loops, each with its own threshold. No single comparator anywhere holds one number.

John Speakman and a large group of obesity researchers argue that body weight fits a different model. In a survey of set points, settling points and some alternative models, they note that an actively defended set point struggles to explain the population-wide weight increase since the 1980s. A passive settling point between intake and expenditure accounts for the environmental data better.

Douglas Ramsay and Stephen Woods argue that allostasis is used inconsistently. Their paper clarifying the roles of homeostasis and allostasis holds that the useful distinction lies in the relationships among multiple regulatory loops rather than in the value of any one variable.

Theresa Beckie reviewed fifty-eight studies using the allostatic load index and found considerable heterogeneity in how allostatic load is operationalised, with biomarkers and cutoffs varying enough between studies to make comparisons difficult.

Each of those findings is accurate, and none of them weakens the model. Read them together and they describe a distributed organization defending several loosely coupled values through many effector loops with no central comparator. That is what this model claims. Romanovsky’s many-loops finding is a problem for a body run by one thermostat. It is a prediction of a body run by one organization.

The claim carries a testable edge worth naming here. Defended values across systems that drift and recover with shared structure rather than independently are being held together by one organization, and that shared structure confirms this framing. Seeman’s index and Ramsay and Woods’s proposed research programme point at the same test from opposite directions.

A reference range describes a crowd, not your set point

Homeostasis defends your personal value, not the population’s average. A result can therefore sit comfortably inside the laboratory range and still be far from the value your body is built to hold.

Every laboratory result you have ever received was compared against a range built from a population. That range has to be wide enough to contain almost everyone. Your own defended value is a single point inside it.

Eugene Harris, a clinical chemist, worked out what follows in 1974. In a paper on the effects of intra- and interindividual variation on the appropriate use of normal ranges, he set out the consequence.

When a person’s own fluctuation is small relative to the spread between people, a result can sit comfortably inside the population range while sitting far outside that person’s own. The ratio between the two spreads is now called the index of individuality, and when it is low, the population range tells you very little about the individual.

Andersen proved it for thyroid hormone with the year-long monthly sampling described earlier. Each man’s personal interval was about half the width of the group’s. A result can move a long way from that man’s own value without ever leaving the laboratory’s normal limits.

Personal baselines are measurable at scale

Ziad Obermeyer and colleagues proved it for temperature at scale. Analysing 243,506 readings from 35,488 outpatients who had no infection, they estimated individual baseline body temperatures. Mean temperature was 36.6 degrees Celsius, and a person’s baseline varied systematically with age, sex, and other characteristics. The individual baseline also carried information about subsequent mortality.

Even the textbook number has moved. Myroslava Protsiv and Julie Parsonnet at Stanford compared three American cohorts spanning Civil War veterans to the present day. They reported that human body temperature has been decreasing since the industrial revolution, falling monotonically by about 0.03 degrees Celsius per birth decade. The famous 37 degrees is a nineteenth-century measurement of a nineteenth-century population.

Normal is a statement about a crowd. Your set point is a statement about you, and the two answer different questions.

Nothing in this argues against reference ranges, which catch a great deal and should keep being used. It argues for one addition, which is a baseline of your own, taken when you are well, so that a later reading has something personal to be compared against.

Conditions that are mostly a set point problem

Set point carries most of the weight in blood pressure, endocrine, cardiovascular, autonomic, sleep, and idiopathic conditions. Every condition in this library combines a few foundations of tone, and these are the conditions where the defended value itself does most of the explaining.

Blood pressure is the definitive case. Essential hypertension is a pressure defended at a higher value by a system with nothing broken in it, which is why no cause is found in the great majority of people.

Endocrine conditions are set point problems by construction. A hormone axis is a defended value with a slow clock, and Andersen’s finding about personal thyroid bands is precisely why subclinical thyroid disease is such a contested category.

Cardiovascular regulation more broadly turns on defended values: resting heart rate, vascular resistance, circulating volume. The dynamics degrade before the averages do.

Dysautonomia is the failure of defense rather than the drift of a target. The value that should be held on standing is not held, and the person feels it within seconds, which is the opposite presentation from silent drift and the same axis.

Sleep depends on targets that are supposed to move on a daily schedule, including core temperature and cortisol. A circadian system that cannot move its targets on time produces a person who cannot fall asleep, cannot wake, or both.

Pregnancy is the physiological proof that a large, coordinated reset can be correct, and that a correct reset ends when its occasion does.

Idiopathic conditions are where the set point idea earns the most. A condition defined by the absence of a lesion is exactly what a drifted set point looks like from the outside, because a competently defended wrong value leaves nothing to find.

What actually changes a defended value

A set point moves when the conditions of regulation change, and the interventions with evidence share that pattern. The record is real, uneven, and reported here with its nulls attached.

Aerobic training lowers resting pressure, and it lowers it most in the people who started highest. Véronique Cornelissen and Neil Smart pooled 93 randomised trials covering 5223 participants and reported the effects of exercise training on resting blood pressure. Endurance training reduced systolic pressure by about 8 millimetres of mercury in the hypertensive groups, about 2 in the prehypertensive groups, and not measurably in those with normal pressure. The effect scaled with the distance from the middle.

Breathing near six cycles a minute retunes the same loop. Paul Lehrer and Richard Gevirtz reviewed the mechanism behind heart rate variability biofeedback. They concluded that the best-supported explanation is a strengthening of the baroreflex loop, with breathing at the resonance frequency driving large oscillations through it.

Treating sleep apnoea moves the twenty-four hour reading in people whose pressure resists drugs. Imran Iftikhar and colleagues pooled six studies of continuous positive airway pressure in resistant hypertension with obstructive sleep apnoea. They found a reduction of about 7 millimetres of mercury in ambulatory systolic pressure, with a similar effect across the four randomised trials.

Weight and device trials show the defense outlasting the push

Moving a set point is not a slogan, and the weight literature is where that becomes undeniable. Leibel showed the defense engaging within weeks, Sumithran showed the hormonal defense still running at twelve months, and Fothergill showed the metabolic defense still running at six years. Anyone who tells you a defended value can be reset easily is not reading that record.

Device work carries a similar caution. The Rheos Pivotal Trial implanted an electrical stimulator on the carotid sinus of 265 people with resistant hypertension, aiming to activate the baroreflex directly. John Bisognano and colleagues reported that baroreflex activation therapy lowered blood pressure and met three of its five co-primary endpoints, while missing the endpoints for acute responder rate and for procedural safety. The reflex can be pushed. Pushing it is neither simple nor free.

The pattern across all of this is consistent and is worth stating plainly. Inputs that work do not override the value. They restore the conditions under which the system re-evaluates it, and then the body moves the number itself. That is slower than a drug and it is a different kind of change.

How Set Point relates to the other dimensions

Set point borders each of the other dimensions of the Unified Model of Tone, and the ontology stays useful only while the borders stay sharp. Here is each neighbor and where the line runs.

Gain is how loudly the system responds relative to the input. On the baroreflex curve, set point is the midpoint and gain is the slope. A system can be centerd correctly and respond weakly, or respond fiercely around a drifted center.

Oscillation is the rhythm and range a single system moves through. Set point names the value; oscillation names the movement around it. Munch’s finding is the cleanest illustration, since the threshold shifted while the shape of the response did not.

Coupling is whether separate systems stay in step with each other. Oscillation is within a system and coupling is between systems. That line matters here, because several targets drifting together is a coupling observation, while the drift of any one of them is a set point observation.

Prediction is the system acting on a model rather than on the world. Allostasis is where the two dimensions meet: prediction supplies the forecast, and set point is the value the forecast tells the body to hold. Prediction owns the model. Set point owns the target.

Load is what holding a state costs and what accumulates when the system cannot stop paying. Selye and McEwen sit on that page. Set point asks which value is being defended; load asks what the defending costs.

Constraint and slack is where the system has room to move, mechanically and neurally. A target the body cannot reach because the tissue will not allow the movement is a constraint problem wearing a set point costume, and the two need separating clinically.

Input quality is the fidelity of the afferent signal, meaning what the system knows about itself. This is the closest neighbor of all, and Thrasher’s experiment is the reason. He changed nothing but the report, and the defended value followed. A degraded sensor and a drifted reference produce the same behavior from outside.

Time course describes how the other dimensions change with the age of the problem. A target drifted for twenty days behaves differently from one drifted for twenty years, and Barker’s work marks the extreme case of a target set before birth.

One organization sets all of the body’s set points

That organization is tone, and tone is what drifts when a blood pressure, a sleep pattern, and a digestion change together. Everything above is established physiology. This claim is the model’s own.

The claim is that the defended values across the body are not a committee of independent thermostats. They are coordinates set by one organization, which is tone, and that organization is what drifts. A person whose blood pressure has resettled higher, whose sleep architecture has flattened, whose digestion has changed, and whose resting muscle tone has risen does not have four disorders that happened to arrive together. Those are four readings of one organization holding a different overall setting.

This is where the difference between restoring and masking becomes concrete.

A drug that lowers blood pressure reliably prevents strokes and saves lives, and working in one direction is a feature of a tool designed to be dependable. The point here is about mechanism. Managing an output is a different operation from moving the value the body is aiming at. The second has barely been attempted, because the variable it acts on has never been named.

The bidirectional test that tells restoring from managing

The two operations make different predictions, and one trial design separates them.

The model predicts bidirectional restoration. A correction that genuinely restores tone should move a dysregulated value toward the middle of its healthy range from whichever side it was displaced, bringing a high value down and a low value up. A drug pushes one way regardless of where the person started. Two people receiving the same restorative input should converge; two people receiving the same drug should both move in the drug’s direction.

The idea has a clinical ancestor. Joseph Wilder was a physician working in neurology and psychiatry, and in a 1957 summary of decades of observation he documented what he called the law of initial value. The response to a stimulus depends heavily on the level the system started from, and at extreme starting levels responses can reverse direction entirely. Wilder described the phenomenon. The model claims a mechanism for it and stakes itself on a specific version.

The test has to be built properly, because sorting people by their starting side produces some convergence from regression to the mean alone. It needs a sham arm matched for contact and attention. It needs the site and the input specified in advance, from a measure taken before the outcome is known.

The direction of the predicted result must be stated before the data arrive. If the treated arms converge no more than the sham arms, the input is not restoring regulation. If the same input moves everyone the same direction regardless of starting side, it is pushing the output, and half the sample ends further from the middle than it started.

What the model adds beyond the established terms

Homeostasis, allostasis, allostatic load, and baroreflex resetting are established science, and every one is credited above to the person who built it. What the model adds is the unification and the composition. The body’s defended values are set by one organization rather than independently. That organization is measurable before the outcome is known. Conditions themselves can be described as combinations of a small number of these foundations. A relabeling makes no new prediction. This framework makes the one stated above.

The framework also stays testable. The convergence prediction above sorts a restoring input from one that manages a number. The independence test in section eleven asks whether one organization holds the defended values at all. Both are within reach of ordinary methods.

There is also the standard caution about causation, and it applies. Almost everything in the measurement section is observational, and observation cannot establish which way the arrow runs. That is precisely why the model does not rest its case there. It rests it on a prospective, sham-controlled test with the direction of the result named in advance, which is the only kind of evidence that can settle the question.

One boundary stands throughout. Not every drifted value is a regulation problem. Kidney disease, adrenal tumours, thyroid disease, medication effects, and sleep apnoea all move defended values through mechanisms of their own, and they need finding and treating on their own terms. What set point addresses is the large remainder where the workup comes back clean, the loop is intact, and the body is competently defending a value it should not be defending.

Frequently asked