Allostatic Load: What It Costs the Body to Hold a State

Load is the invisible cost of holding a state

Load is the part of a state that never appears in the picture of the state. Pick up a heavy suitcase and hold it at your side. For the first minute, almost nothing happens. Hold it for an hour and the hour will change you.

Now take two photographs. One at minute one, one at minute fifty. Set them side by side and they look nearly identical. Same person, same posture, same bag, same floor. Nothing in either image has moved.

What the second photograph cannot show is the forearm burning. It cannot show the shoulder locked up to the ear, the neck recruited to help, or the breath gone shallow and high in the chest. It cannot show the heart beating harder, or the sugar being pulled out of the blood to keep the grip closed. All of that is happening. None of it is visible in the picture.

That gap has a scientific name. Allostatic load is the cost the photograph cannot show. This library calls it load for short, because the cost belongs to the nervous system doing the holding.

Notice the second thing the photograph hides. The suitcase does not have to be heavy. A light bag held for a week costs far more than a heavy one held for a second. Duration is the multiplier, and weight is only the rate. This is why people are so often wrong about their own bills. They look for a large event and find only ordinary weeks.

And then the third thing, which is where the biology begins. You can put a suitcase down. A body cannot always put its state down. Suppose the nervous system has decided that this posture, this vigilance, this blood pressure and this level of inflammation are the correct answer. It will hold that answer whether or not anything is currently asking. The grip stays closed. The bill keeps arriving.

From homeostasis to allostasis to allostatic load

Allostatic load is the third word in a sequence that has to be built in order. Homeostasis names the body's steadiness, allostasis names its predictive regulation, and load names what the predicting costs.

Start with Walter Cannon, a physiologist at Harvard who spent his career on a single puzzle. The world outside a body swings enormously. Temperature, food, water, oxygen, danger. Inside the body, almost nothing swings. Blood sugar, blood salt, temperature and acidity stay inside narrow bands, hour after hour, for a lifetime.

Cannon wanted to know what does the holding. He named that steadiness homeostasis in 1926 and set out the organization behind it three years later, describing it as an organized defense rather than a passive equilibrium.

The mental picture for homeostasis is a thermostat. You set it to twenty degrees. It measures the room, and whenever the room drifts, it corrects back toward twenty. It has one job and one number, and it does nothing until the error appears.

Sterling names allostasis, McEwen and Stellar name the cost

Now the second word. Peter Sterling, a neuroscientist who studies how nervous systems are designed, argued that the thermostat picture describes the body badly. A body does not wait for the error. It anticipates. It raises blood pressure before you stand up, releases fuel before you run, and shifts hormones before the season changes. Sterling named this capacity allostasis and defined it as predictive regulation: the system moves its own settings ahead of demand rather than defending a fixed number.

Change the picture accordingly. Homeostasis is a thermostat holding twenty degrees. Allostasis is a house that starts heating at four in the afternoon because it expects you home at five. The second design is far smarter, and it is far more expensive. Predicting costs something. Pre-adjusting costs something. Being wrong costs something too, because the house heated an empty room.

That expense is the third word. Bruce McEwen was a neuroendocrinologist who spent decades on how stress hormones act on the brain. In 1993 he and Eliot Stellar, a physiological psychologist, asked a clinical question. Why do people exposed to broadly similar demands fall ill at such wildly different rates? Their answer relocated the damage. They proposed that the harm comes from the body's own adaptive machinery rather than from the event, and they named the accumulated cost of that machinery allostatic load.

Load is the invoice for range. A body that only defended fixed values would be cheap to run and unable to adapt. Because it predicts and pre-adjusts, it buys range, and the bill is what range costs.

This is the point where the established term and the model's term meet. In the Unified Model of Tone, tone is the integrated organization the nervous system holds across the whole body, and health is the width of the range that organization can move through.

Several foundational dimensions describe that organization, and load is one of them. The others tell you what the system is holding: which value it defends, how loudly it responds, what rhythm it runs, what it predicts, how much room it has. Load tells you what holding it costs.

Load is the price tag on a state rather than a state itself. That distinction is why the model tracks cost as its own quantity. Load is the term that converts a sensible adaptation into a disease over time.

Every adaptation is purchased

Nothing the nervous system does to protect you is free, and allostatic load is the ledger of what protection costs.

Raising blood pressure to carry you up a staircase is purchased. Tightening a set of muscles around a sore rib so the rib is not disturbed is purchased. Staying alert in a house where the mood can turn is purchased. Turning immunity up to meet an infection is purchased, and turning it down afterward is purchased as well. Every one of these is the right answer to a moment, and every one of them draws on the same finite supply of fuel, tissue and attention.

The chemistry makes this concrete. The neurobiologist Robert Sapolsky and his colleagues reviewed decades of work asking what stress hormones actually do, and found that the answer depends entirely on timing. In their reading, the same glucocorticoid hormone can prepare the body for a demand, permit a response, stimulate it, or suppress it, depending on when it arrives relative to the challenge. A molecule with four opposite jobs is a manager. And a manager who never leaves the building becomes the problem.

So the sentence this page turns on is simple. The response that saves you is the same response that bills you. Medicine has always studied the saving. Allostatic load is the study of the billing.

Hans Selye discovers the general cost of adaptation

Allostatic load has a prehistory, and it begins with Hans Selye's discovery that adaptation itself carries a general cost. Selye was an endocrinologist working in Montreal in the 1930s, and he was not looking for stress. He was looking for a new ovarian hormone.

His method was ordinary for the time: inject rats with crude extracts from ovaries, then open the animals up and see what changed.

Something changed reliably, and it was the wrong something. He kept finding the same three things. The adrenal glands, two small hormone factories sitting on top of the kidneys, were enlarged. The thymus and other lymph tissue, where immune cells are made and trained, had shrunk. And the lining of the stomach carried bleeding ulcers.

Then the finding that made his career. He injected other extracts and got the same three changes. He exposed rats to cold, to surgical injury, to exhaustion, to toxic substances, and got the same three changes again. In 1936 he published a short note in Nature describing a syndrome produced by diverse nocuous agents. The insult varied. The response did not.

The general adaptation syndrome hides cost in its middle stage

He later named the response the general adaptation syndrome and gave it three stages, which he laid out for clinicians in the British Medical Journal in 1950. First alarm, the immediate mobilization. Then resistance, in which the body settles into a costly holding pattern and looks, from outside, fine. Then exhaustion, when the capacity to keep holding runs out and the animal deteriorates.

That middle stage is the suitcase at minute fifty, and it is Selye's permanent contribution to the study of load. He established that there is such a thing as a general cost of adaptation, that it is silent while it accumulates, and that it eventually presents as damage.

Where non-specificity failed

His central theoretical claim, though, was non-specificity: the response is the same regardless of what caused it. That claim did not survive, and it is worth being exact about why.

In 1971 John Mason, a psychiatrist and physiologist who spent years measuring hormone output in humans and monkeys under controlled conditions, published a reexamination of the idea of non-specificity in stress theory. His argument was that Selye's stressors had something in common besides being physically noxious. They were novel, unpredictable, threatening, and outside the animal's control.

When those psychological features were removed or controlled for, the supposedly universal response shrank or changed. What looked like a fixed reaction to any insult behaved much more like a reaction to a particular kind of meaning. The neuroscientists David Goldstein and Irwin Kopin later traced how the concept had to be rebuilt after that, with specificity restored to the response pattern.

Here is the model's reading. Selye was right that a general cost exists, which is why allostatic load became a necessary concept. He was wrong that the cost is indifferent to what meets the body. An input does not have a fixed effect. It has an effect that depends on the organization it lands in.

The same cold, the same injury, the same week of bad news arrives at a differently organized nervous system in every animal, and the bill differs accordingly. Non-specificity had to fail, because there is no such thing as an input without a recipient. This library calls that principle input meets tone, and Selye's own data were the first large demonstration of it.

McEwen and Stellar relocate the damage to the response

Allostatic load turns from protection into damage under four specific conditions, and McEwen's 1998 paper in the New England Journal of Medicine remains the canonical statement of them.

Once you accept that the response does the wearing, a new question opens. Under exactly what conditions does a protective response turn into a destructive one? McEwen answered that in 1998. He described four ways the same mediators shift from protective to damaging, and those four conditions are the backbone of everything later in this page.

Read those four together and a pattern appears that is easy to miss. Only the first is about the size of the demand. The other three are about the behavior of the responder. That is the move McEwen and Stellar made, and it is why allostatic load is a claim about regulation rather than a claim about hardship.

It also explains the clinical fact that started their inquiry. Two people go through a similar decade. One is fine. One is not. If damage came from events, that would be a puzzle. If damage comes from how the system responds to events, it is exactly what you would expect.

The allostatic load index, marker by marker

The allostatic load index puts a number on cost: ten biological markers across several systems, one point for each marker sitting in the sample's highest-risk quartile.

An idea this broad earns its keep only if someone can put a number on it. In 1997 Teresa Seeman, an epidemiologist of aging, working with Burton Singer, John Rowe, Ralph Horwitz and McEwen, did exactly that. Their paper built an operational measure of allostatic load from ten biological parameters in a community study of older adults, and the design of that measure teaches more than the results do.

Two tiers of markers

The ten markers sit in two tiers, and the tiers mean different things.

The first tier is the chemicals of the response itself. Cortisol, the main glucocorticoid, measured in overnight urine. Adrenaline and noradrenaline, the catecholamines, which are the fast arm of the response. And DHEA-S, an adrenal steroid that partly counterbalances cortisol. These are what the body is currently spending.

The second tier is what the spending has already done. Systolic and diastolic blood pressure. The ratio of waist to hip, which tracks where fat has been deposited. Two cholesterol measures. And glycated hemoglobin, which reports the average blood sugar over the preceding months. These markers are the residue the spending leaves behind.

The scoring rule, and the limits

Now the scoring rule, which matters more than the list. You take the whole sample, and for each marker you find the quartile carrying the highest risk. A person scores one point for every marker on which they fall into that quartile. The score is the count of points. Nothing is averaged, and nothing is weighted by how far into the quartile you sit.

That design encodes a specific hypothesis, and it is the hypothesis this page exists to teach. Modest dysregulation across many systems at once matters more than dramatic dysregulation in one. A person with six markers slightly off scores six. A person with one marker badly off scores one. The index bets that the first person is in more trouble, and it is the first index in medicine built to see a whole body drifting rather than an organ failing.

Later versions extended the battery. The stress researcher Robert-Paul Juster, working with McEwen and Sonia Lupien, reviewed how the biomarker panel expanded to include inflammatory markers such as C-reactive protein, interleukin-6 and fibrinogen, along with immune and neural measures.

The limits are specific. There is no single agreed formula. A study of the American national health survey data found twenty-one different published ways of calculating an allostatic load score from the same underlying dataset, using different markers and different cut points. That variability makes scores hard to compare across studies and easy to tune.

Three further limits belong in plain view. The index is a research instrument rather than a test a person can walk in and order. It tells you that a body is running expensively, and it does not tell you which system is driving the expense or why. And it is a snapshot of residue, so it lags the regulation that produced it, sometimes by years. What it does exceptionally well is what nothing else was doing: it makes cost itself a measurable quantity.

Load forecast death and decline in the MacArthur cohort

Higher baseline scores in 1,189 adults aged seventy to seventy-nine predicted mortality over seven years, and the score moved in both directions within a few years. The results are worth stating precisely, because precision is what separates this from a slogan.

The measure was tested in the MacArthur Studies of Successful Aging, a long-running study of older adults living in the community.

Higher baseline load predicted death within seven years

The original 1997 paper found that higher allostatic load scores went with poorer cognitive and physical function at the time of measurement, predicted larger declines in both later, and predicted increased cardiovascular disease risk. In 2001, Seeman, McEwen, Rowe and Singer followed 1,189 men and women aged seventy to seventy-nine for seven years. Higher baseline load predicted death within those seven years, and predicted decline in both cognitive and physical functioning, independently of standard demographic factors and baseline health.

The same paper carries a null. The association with new cardiovascular events was only marginal, which sits beside the raised cardiovascular risk reported in 1997. The model expects that spread. The composite score tracks a whole body running out of range, and a whole-system cost maps loosely onto any single disease endpoint. The score forecast death and physical decline better than the metabolic markers alone and better than the primary stress chemicals alone.

The finding was extended in 2002, when the physician and epidemiologist Arun Karlamangla and colleagues confirmed allostatic load as a predictor of functional decline in the same cohort.

The score moved, and mortality tracked the direction

Then comes the study that matters most for the model, and it is easy to overlook. In 2006 Karlamangla, Singer and Seeman measured the same ten markers twice, two and a half years apart, in 171 high-functioning older adults. One design note belongs before the numbers. The score used here was a weighted summary of the ten markers rather than the plain quartile count taught in section 06, so the unit being counted differs.

They looked at who was alive four and a half years after the second measurement. Among people whose score had gone up, fifteen percent had died. Among those whose score had gone down, five percent had died. Adjusted for age and starting score, each one-unit rise in that weighted score carried a mortality odds ratio of 3.3, with a confidence interval running from 1.1 to 9.8.

The foundations of that result are narrow, and they belong in the record next to it. It rests on 171 people from a single cohort of unusually high-functioning adults aged seventy to seventy-nine. The p value was .047. The confidence interval nearly touches 1.0 at the bottom, so the data bound the effect loosely. And the design is observational, so it does not settle the direction of cause on its own.

What it establishes is the shape of the variable. The allostatic load score moves in both directions inside a few years, in people already in their seventies, and the direction it moves tracks with how long they live. Load behaves as a current state of the system rather than a record of the past, which is precisely what the model claims about tone in general.

Why prediction costs the brain measurable energy

Allostatic load has a literal fuel bill, because the brain takes about twenty percent of the body's resting energy and spends most of it on ongoing internal signaling. Three findings build the case, and together they turn cost from a figure of speech into a measured quantity.

Step one. Your brain is roughly two percent of your body weight. It takes about twenty percent of the energy your body uses at rest. Marcus Raichle, a neurologist, and Debra Gusnard, a neuroimaging researcher, both working on brain imaging and metabolism at Washington University, appraised that budget and reported something counterintuitive.

Almost all of it goes to ongoing internal activity, and the extra energy consumed when you perform a task is a small increment on top of a very large baseline. The brain stays busy running itself even between jobs.

Where the energy goes

Step two. Where inside the brain does the energy go? David Attwell and Simon Laughlin, a neuroscientist and a neurobiologist working on the economics of signaling, built an energy budget for the grey matter of the rodent brain. Their estimate: about forty-seven percent goes to action potentials, the electrical pulses nerve cells fire to talk to each other.

About thirty-four percent goes to the effects of those messages on the receiving cell. About thirteen percent goes to simply maintaining the resting voltage across the membranes, which is the cost of being ready to fire at all. And the total depends steeply on how fast the cells are firing.

Put those two steps together and you have a machine whose running cost is set by how much signaling it is doing, not by how much it is achieving.

Uncertainty is the expense

Step three closes the loop. In 2017 Achim Peters, an endocrinologist, working with McEwen and the neuroscientist Karl Friston, defined stress formally as uncertainty and drew out the consequence. Reducing uncertainty means updating an internal model until it fits the world, and that updating requires cerebral energy.

They note that the vertebrate brain takes priority over the rest of the body when energy is scarce, a principle they call the selfish brain. From there they argue that a brain which cannot resolve its uncertainty develops a persistent energy crisis. The name they give the systemic consequence of that crisis is allostatic load.

A nervous system running a poor model of the body is, in the most literal biological sense, an expensive nervous system. It pays every second for a mismatch that never closes.

This is where load and prediction meet without merging. Prediction, which has its own page in this library, describes a system acting on a model rather than on the world. Load describes what running that model costs when it will not resolve. A wrong prediction is a standing order on the body's fuel.

It also explains something patients report constantly and clinicians struggle to place: exhaustion out of all proportion to activity. If the ordinary background operation of the system has become expensive, then rest does not restore, because the cost is not coming from the day. It is coming from the organization.

The four failure modes that build allostatic load

A response that never shuts off, slow recovery, an inadequate response, and repeated hits without recovery do not look alike from outside, yet each ends in the same accumulating cost.

The response never comes down

The demand ends. The chemistry keeps running. A large part of the reason is that the nervous system does not need the event to be present to keep responding to it.

The psychologist Jos Brosschot and colleagues gave this a name and a body of evidence, showing that worry and rumination prolong stress-related physiological activation long after the situation is over. A meeting lasts an hour. The response to it can last a day, and can begin again at three in the morning with no meeting anywhere.

Recovery is slow

The intuition here is usually wrong. People assume the damaging thing is how hard the system reacts. The better predictor is how long it takes to come back. Andrew Steptoe and Michael Marmot, epidemiologists studying stress and health in a working population, followed 209 healthy men and women aged forty-five to fifty-nine for three years. They measured cardiovascular responses during two mildly stressful tasks and for up to forty-five minutes afterward.

Poor recovery of blood pressure after the task predicted a rise in resting blood pressure three years later, independently of baseline pressure, age, sex, social position, body mass and smoking. The adjusted odds of a meaningful systolic rise were 3.50 for the poor recoverers, with a confidence interval from 1.19 to 10.8. Reactivity mattered less consistently than recovery did.

Recovery time is the load-relevant variable. A system that surges hard and returns cleanly is running a healthy range. A system that surges modestly and takes an hour to return is paying rent on every disturbance in its day.

The response is inadequate, or was never properly established

This is the failure mode most people never hear about, and it is where the model earns its living. A chronically loaded system does not always show high stress hormones. It often shows low ones.

The evidence is consistent and initially baffling. Christine Heim, Ulrike Ehlert and Dirk Hellhammer are psychologists and biological stress researchers.

• Reviewing patients with chronic pain, chronic fatigue and related conditions, they described low cortisol output as a feature of stress-related bodily disorders rather than an exception.
• Eva Fries and colleagues later set out the case for reading this hypocortisolism as an adaptation, a downregulation that protects tissue from prolonged glucocorticoid exposure and carries its own costs.
• The neuroscientist Rachel Yehuda found the same paradox at the sharp end, reporting lower urinary cortisol in patients with post-traumatic stress disorder than in comparison groups.

The psychologists Gregory Miller, Edith Chen and Eric Zhou put numbers on the whole pattern by pooling more than a hundred studies. They found that cortisol output tends to be raised soon after a stressor begins and lower when the stressor is old, with the daily rhythm flattening as time passes. The picture is a system that starts by paying and ends by refusing to pay, with the flattened rhythm as the signature of both.

McEwen's fourth condition explains why the low state is not relief. When one mediator underperforms, another compensates. Cortisol restrains inflammation, so a system that cannot mount cortisol properly is a system with inflammation left off the leash. The bill did not shrink. It moved departments.

Repeated hits without recovery

There is one experiment that teaches this better than any review, and it is small, old and beautifully clean.

In 1995 Clemens Kirschbaum and colleagues at Trier took twenty healthy men and put each of them through the identical stress test on five consecutive days. The test was public speaking and mental arithmetic in front of an audience. They measured cortisol in saliva each time. Then they let the data sort the men rather than sorting the men in advance.

The group split cleanly in two. Thirteen men showed a cortisol rise on day one and essentially nothing on days two through five. Their systems learned the room. The remaining seven produced a large response on every one of the five days. Same task, same audience, same arithmetic, opposite bills.

Hold that result next to Selye. The demand was standardized to the minute, and the cost was not. Nothing about the input can explain the difference, because the input was identical by design. The difference lived in the men. This is input meets tone stated as a single experiment, and it is the reason this library treats load as a property of the responder rather than a property of the world.

Allostatic load damages whichever tissue is thinnest

Allostatic load is a whole-system quantity, and no one dies of a whole-system quantity. The account gets settled in a particular tissue, and which tissue it is varies from person to person.

Immune, cardiovascular, and mitochondrial routes

One route runs through the immune system, and it inverts a common assumption. Sheldon Cohen, a psychologist who studies stress and infection, and colleagues tested whether prolonged stress makes tissues stop listening to the hormone that would normally end an inflammatory response. In work published in 2012 they reported two results and proposed a route between them. In one sample of 276 adults, recent long-running threatening stress went with reduced sensitivity to glucocorticoid signaling.

That reduced sensitivity in turn went with a higher risk of developing a cold after controlled exposure to the virus. In a second sample of 79 infected adults, the same reduced sensitivity went with greater local production of inflammatory signaling molecules. Stress to inflammation is the path the authors argue for across the two studies rather than a single chain measured end to end. The brake was still being applied. The wheel had stopped answering it.

A second route runs through the cardiovascular system. Mika Kivimäki and Steptoe, epidemiologists working with large occupational cohorts, reviewed the evidence linking chronic stress exposures to the development and progression of cardiovascular disease, describing modest but consistent associations across many populations.

A third route reaches all the way down to the machinery that produces energy in the first place. Martin Picard and McEwen systematically reviewed the effects of psychological stress on mitochondria, the structures inside every cell that convert food and oxygen into usable fuel. Across a large body of mostly animal work, chronic stress was associated with changes in mitochondrial structure and function. The organelle that pays the bill is itself altered by the paying.

Notice the shape of this evidence, because the shape is the argument. Immune, cardiovascular and metabolic literatures each found the same fact inside their own boundary and gave it a local name. The model's claim is that these are one fact. A system running expensively will fail wherever it is thinnest, which is exactly why load presents as a different diagnosis in every person who carries it.

Why a childhood shows up in a fifty-year-old body

Allostatic load accumulates across a lifetime, which is why childhood adversity shows up as measurable inflammation in an adult body. If load accumulates, then a life should leave a mark that can be counted, and the largest attempt to count it began by accident in a preventive medicine clinic.

Vincent Felitti was an internist directing a department of preventive medicine at a large health plan in San Diego. With Robert Anda, a medical epidemiologist at the Centers for Disease Control, he sent a questionnaire about childhood experience to adults who had just completed a standard medical examination.

It went to 13,494 people and 9,508 returned it. The questions covered seven categories of adversity, including psychological, physical and sexual abuse, violence against the mother, and living with household members who were addicted, mentally ill, suicidal or imprisoned.

The ACE gradient

The 1998 result showed a graded relationship between the number of categories and adult disease. Graded means each additional category raised the risk further, which is the pattern you expect from an accumulating quantity rather than from a single traumatic switch. People reporting four or more categories carried four to twelve times the risk of alcoholism, drug use, depression and attempted suicide compared with those reporting none.

Smoking, poor self-rated health and sexually transmitted infection ran two to four times higher. Physical inactivity and severe obesity ran modestly higher. And the count showed a graded relationship to ischemic heart disease, cancer, chronic lung disease, skeletal fractures and liver disease.

The finding replicated. Karen Hughes and colleagues pooled thirty-seven studies covering 253,719 participants and found elevated risk across twenty-three outcomes in people reporting four or more adversities against people reporting none. That is a comparison of the two ends rather than a test of the gradient in between.

Their numbers are specific. The associations were weak for physical inactivity, overweight and diabetes, moderate for smoking, heart disease and respiratory disease, and strongest for problematic drug use and for violence toward self and others. They also found substantial disagreement between studies for almost half the outcomes.

The Dunedin answer

Surveys of adults remembering childhoods leave room for reporting bias, for shared disadvantage, and for adult behavior doing the real damage. The Dunedin study closes that gap from a different direction.

Andrea Danese, a psychiatrist, working with Carmine Pariante, Avshalom Caspi, Alan Taylor and Richie Poulton, used the Dunedin study, a New Zealand birth cohort followed since 1972 with contemporaneous records rather than adult recall.

They found that children who were maltreated had elevated C-reactive protein at age thirty-two, a standard marker of low-grade inflammation, with a risk ratio of 1.80. The effect held after adjusting for other early-life risks, for stress in adulthood, and for adult health and health behavior. The maltreatment was recorded when it happened. The inflammation was measured twenty years later.

Here is the model's reading, and it is a genuine departure from the usual framing. The body keeps no archive it consults. The history is constituted in the present organization: the way the system is currently tuned, what it currently expects, what it currently guards, what it currently spends. That organization is expensive, and the expense is what shows up in the blood at thirty-two.

Which is why the shadow is long, and why nothing has to be consciously remembered for the load to keep accruing.

The conditions where load is the dominant dimension

Trauma, autoimmune conditions, long COVID and their neighbors each carry load for a different reason. Every condition in this library is a combination of two or three parts of tone.

Trauma is the clearest case, because it is the condition in which the response outlives the event by decades. The organization that was correct during the danger is still being funded long after the danger. The low-cortisol findings above sit here, and they are the reason trauma so often presents as depletion rather than as alarm.

Mental health conditions carry load through the failure-to-shut-off route. Worry and rumination keep the response running in the absence of the stimulus, and the metabolic cost of an unresolved internal model is paid continuously rather than in episodes.

Autoimmune conditions are load conditions in a strong and specific sense. The rheumatologist Rainer Straub has argued that chronic inflammation and ageing both amount to an energy shortage negotiated between the brain and the immune system. An activated immune system consumes fuel that the rest of the body then does without. An immune system held at readiness is a suitcase that is never set down.

Endocrine conditions sit here because the hormones in the primary tier of the allostatic load index are the endocrine system. Blood sugar handling, thyroid set, reproductive hormones and adrenal output are all downstream of how long the system has been holding a costly state.

Long COVID is a load condition presenting as a fuel condition. A study in which patients performed exercise and were biopsied afterward found metabolic disturbances and exercise-induced muscle damage that worsened after post-exertional symptom flares. The defining feature of the illness is that ordinary activity costs more than it returns.

Addiction is the case where load and set point interlock. George Koob and Michel Le Moal are neuroscientists studying the neurobiology of dependence. They described addiction as a progressive shift in the brain's reward and antireward systems. The value the system defends moves, and a growing effort is required to reach ordinary comfort. That growing effort is load by another name.

Cost leaves no lesion

Fibromyalgia, unexplained symptoms and idiopathic conditions belong here for a structural reason. Load leaves no lesion. A body that is running expensively has nothing to biopsy, so the testing comes back normal while the person is measurably unwell. That result is what a cost looks like to an instrument built to find damage.

Sleep, gut health, anxiety and overall quality of life each carry a strong load term, because each is both a payer and a payee. Poor sleep raises the bill, and a high bill damages sleep. Those loops are the reason load tends to entrench rather than resolve.

How load differs from the other dimensions of tone

The boundary runs the same direction every time. The others describe what the system holds. Load describes the price of holding it.

Set point is the value the system is defending. Load is what defending it costs. A raised set point can be cheap if the system is comfortable there and expensive if it is fighting to hold it, so the two are independent readings of the same moment.

Gain is how loudly the system responds relative to the size of the input. High gain usually raises load, because a large response costs more than a small one. The two come apart in the third failure mode above, where a low-gain system carries an enormous bill through the compensations it triggers elsewhere.

Oscillation is the rhythm and range a single system moves through. Load is the metabolic price of that movement, and also the price of refusing to move. A flattened daily cortisol rhythm is an oscillation reading. The exhaustion that accompanies it is a load reading.

Coupling is whether separate systems stay in step with one another. Keep that line sharp: oscillation is within a system, coupling is between systems. Load is neither. It is what either arrangement costs to maintain.

Prediction, constraint, input quality, and time course

Prediction is the system acting on a model rather than on the world. Section 08 above is the meeting point, and the division of labor is clean. Prediction says what the system believes. Load says what the belief costs when it will not resolve.

Constraint and slack is where the body has room to move, mechanically and neurally. The physiotherapy researchers Paul Hodges and Kylie Tucker proposed that the body responds to pain by redistributing activity within and between muscles, producing stiffer, more guarded movement with less variability. That redistribution is a constraint change. The fuel it burns, and the tissue it eventually wears, is load.

Input quality is the fidelity of the signal coming back from the body. It feeds load directly. Poor information raises uncertainty, and uncertainty is what the brain spends energy trying to reduce. A system that cannot trust its own sensors will run its models harder.

Time course is asymmetric to all the others by design, because it describes how they age. Load is what makes time course matter, since a cost that is trivial in an acute state becomes the whole clinical picture in an entrenched one.

The inputs that move allostatic load, nulls included

Sleep, movement, slow breathing, meditation, connection and medication each move load-linked markers, and the evidence for each is stated here at the strength it supports and no higher. The MacArthur data already pointed at the important question, with the limits from section 07 in view. The score moves in both directions within a few years, and in that one small cohort the direction it moved tracked with survival.

Sleep lowers load and repairs the system that pays

Karine Spiegel, Rachel Leproult and Eve Van Cauter, physiologists studying sleep and metabolism, restricted eleven young healthy men to four hours in bed for six nights and measured what happened. Glucose handling deteriorated to a degree the authors compared with the effects seen in normal ageing, and evening cortisol rose. McEwen later framed sleep restriction directly as an allostatic load. Sleep is the only intervention on this list that reduces the bill and repairs the payer at the same time.

Movement improves the shut-off

Elisabeth Zschucke and colleagues randomized young men to thirty minutes of treadmill exercise or a placebo version of it, then gave them a laboratory stress task ninety minutes later. The exercise group showed a reduced cortisol response to the stressor, which the authors read as improved negative feedback in the stress axis. The interesting part is the mechanism. Exercise adds a demand and improves the shut-off, which is the second failure mode above.

Slow breathing exercises the return loop

Paul Lehrer and Richard Gevirtz explained the mechanism of heart rate variability biofeedback. Breathing at roughly six breaths per minute drives the blood pressure reflex at its resonant frequency, and exercises the loop that returns the system to baseline. A meta-analysis by Veronika Goessl and colleagues pooled twenty-four studies and found reductions in self-reported stress and anxiety. The trials were mostly small, and blinding a breathing exercise is close to impossible.

Meditation moves the markers

Here the null has to be stated plainly, because the field is oversold. Madhav Goyal and colleagues reviewed the randomized trials for the Agency for Healthcare Research and Quality. They found moderate evidence of small improvements in anxiety, depression and pain. They found no evidence that meditation programs were better than active comparisons such as exercise, medication or behavioral therapy. On the physiology, Michaela Pascoe and colleagues took a harder test.

They admitted only randomized trials comparing meditation against an active control, forty-five of them, and found lower cortisol, blood pressure, heart rate, triglycerides, C-reactive protein and tumor necrosis factor alpha in the meditation groups. The physiology and the symptoms are not yet telling the same story. Load-linked markers move, and they move even against an active comparison. The psychological benefit has not been shown to beat exercise or medication.

Connection reduces uncertainty

Julianne Holt-Lunstad and colleagues pooled 148 studies of social relationships and survival and found that people with stronger relationships had a fifty percent greater likelihood of surviving over follow-up. That is an association in observational data, and it is a large one. In the vocabulary of this page, other people reduce uncertainty, and uncertainty is the thing the brain is paying to resolve.

Medication lowers a marker

A drug that lowers blood pressure, blood sugar or inflammation lowers a marker in the secondary tier of the allostatic load index, and that is worth having. Real damage is prevented that way every day. What it does not do by itself is change what the system is trying to hold, which is why the marker generally returns when the drug stops. The distinction is mechanism and aim rather than merit.

How the model reads load, and how to tell restoring from masking

The Unified Model of Tone reads allostatic load as part of tone itself, named load, and the reading commits to a prediction one trial can read. Everything above is established science, and it is remarkable how far it gets on its own. What the model adds is a reading of what all of it is measuring.

The model's claim is that load is one dimension of tone rather than a separate phenomenon that happens to bodies under stress. This library defines tone as the integrated, coupled organization the nervous system maintains across the whole body, together with its capacity to move where the moment demands and return afterward.

Health is the width of that range. Load is what the current organization costs to hold, and a system with a narrow range is expensive by definition, because it is paying to stay somewhere the moment no longer requires.

That reframes the whole clinical picture. A raised allostatic load score is a statement that the body is currently funding an organization that no longer fits, and the interesting question is which point of tension is holding that organization in place. There is rarely one such point. There are usually several, in varying degrees of potential, and they layer: the place where symptoms appear is frequently downstream of the place where the pattern is actually held.

This is also where the model draws its sharpest practical line, and it draws it generously. Lowering a marker and restoring a range are different aims. A treatment that quiets a signal while the underlying organization keeps running has changed the reading and not the cost. A change that lets the system release the state it was holding lowers the reading because the cost fell.

Practitioners in every field are already working with this variable. Most have had no framework that names it, which is the only reason the distinction has stayed invisible for so long. Symptomatic relief has real and sometimes life-saving value, and nothing here is an argument against it.

Restoring regulation versus masking a symptom, and how to tell

One prediction separates an input that restores regulation from one that pushes a load-linked number.

The model predicts bidirectional restoration. If a load-linked measure is dysregulated, an input that genuinely restores tone should move it toward the healthy middle from either side. The high cortisol profile should come down. The flattened, depleted profile should come up. The same is predicted for heart rate variability, for blood pressure, for recovery time. A drug pushes in one direction, which is what a drug is for. A restoration of range should not care which side the system was stuck on.

John Thayer and Esther Sternberg made a closely related argument in 2006, describing the vagus nerve as a common regulator of several allostatic systems at once. That is the kind of shared control this prediction requires. An input that moves load-linked markers only in one direction has pushed an output, helping whichever group it happens to point at and carrying the other group further from the middle. Markers that move together as one organization, rather than independently, are reading a shared setting.

What the composition adds

Allostatic load already exists as a standalone concept about stress. Treated as part of tone, it composes with set point, gain, oscillation and the rest, so a condition can be stated as a specific combination and each combination predicts a different clinical shape. That is a structural claim, and it is testable in a way a synonym is not.

Observational evidence leaves the direction of cause open on its own. That is why this page carries the graded dose-response in the adversity data, the prospective design in Dunedin, the identical-input design in the Trier experiment, and the split-group test above. None of these is proof. Together they make a purely confounded reading of the whole literature hard to hold.

What remains genuinely open is the question the model most wants asked. If load is one property of a single organization, then restoring that organization should lower load across several systems at once, in the same person, measured together. Nobody has run that study properly. The claim is stated here so it can be.

The suitcase is the whole lesson. Nothing in the photograph is wrong. The cost is simply not in the photograph, and a system that has been holding for long enough will look, to every instrument built to find damage, entirely fine.

Frequently asked