← Brain Structures
Brain Structure · Maps to the Law of Rhythm

The HPA Axis

The body keeps a hormone clock and a hormone alarm running on the same wire — and most of human suffering downstream of stress is what happens when the alarm forgets how to turn itself off.
Read · 19 minPillar · Neuroscience

The hypothalamic-pituitary-adrenal axis is the central stress-response system of the vertebrate body. Everyone has heard "cortisol is the stress hormone" — and that phrase is doing the same damage that "everything vibrates" does for the Law of Vibration. It's half-right in a way that hides the actual machine. Cortisol is not the enemy. It is a metabolic signal you cannot live without; people whose adrenals fail (Addison's disease) die without it. The story isn't "cortisol bad." The story is a feedback loop — an exquisitely engineered thermostat with a self-shutoff — and what happens to a body and a mind when that shutoff erodes. That erosion is the through-line connecting chronic stress to depression, PTSD, metabolic disease, and the slow structural remodeling of the brain itself. Let's pull it all the way down.

01

What it is — anatomy, location, connectivity

The HPA axis is three glands wired in series, each releasing a hormone that tells the next one what to do, with the final hormone reaching back to silence the first two. Three nodes:

1. The hypothalamus. A small structure at the base of the brain, sitting just above the brainstem and below the thalamus, that serves as the master interface between the nervous system and the endocrine system. The relevant cells live in the paraventricular nucleus (PVN) — a cluster of neurons that synthesizes corticotropin-releasing hormone (CRH), also called corticotropin-releasing factor (CRF), the peptide that opens the whole cascade. The PVN also makes arginine vasopressin (AVP), which potentiates CRH's effect on the pituitary. The hypothalamus is not the start of the signal — it's the integrator. It receives inputs from the amygdala (threat), the hippocampus (context and memory), the prefrontal cortex (appraisal), the brainstem (visceral state), and the body's own bloodstream (it samples circulating cortisol directly).

2. The pituitary gland. A pea-sized gland hanging below the hypothalamus on a stalk, often called the "master gland." CRH and AVP travel a very short distance — through a private blood supply called the hypophyseal portal system — to the anterior pituitary, where they trigger cells called corticotrophs to release adrenocorticotropic hormone (ACTH) into general circulation. This portal vein is the elegant part: the hypothalamus doesn't shout into the whole bloodstream, it whispers down a dedicated pipe to the gland next door.

3. The adrenal cortex. ACTH rides the bloodstream down to the adrenal glands, which sit atop the kidneys. The outer layer — the cortex — responds to ACTH by synthesizing and releasing glucocorticoids, principally cortisol in humans (corticosterone in rodents, which matters because most of the foundational science was done in rats). Specifically ACTH acts on the zona fasciculata of the cortex. Note this is a different system from the adrenal medulla (the core), which releases adrenaline/noradrenaline for the fast fight-or-flight surge. The HPA axis is the slow arm of the stress response; the sympathetic-adrenal-medullary (SAM) system is the fast arm. Adrenaline acts in seconds; cortisol acts over minutes to hours.

The defining feature of the whole structure is negative feedback: cortisol, the end-product, circulates back to the pituitary, the hypothalamus, and critically the hippocampus and prefrontal cortex, and tells the upstream nodes to stop. It is a thermostat. The furnace's own heat tells the thermostat to shut the furnace off. Hold that image — almost every dysfunction of this axis is a failure of that shutoff.

02

What it does

Three jobs, in plain terms:

It keeps a daily rhythm. Cortisol is not just an emergency signal; it has a strong circadian rhythm baked in even when nothing stressful is happening. It peaks in the early morning around waking, drops across the day, and bottoms out around midnight. This is the body's "get up and mobilize energy" signal — cortisol raises blood glucose, primes metabolism, and readies the system for the demands of the day. This rhythm is the "clock" half of the system, governed ultimately by the brain's master clock, the suprachiasmatic nucleus.

It mounts the slow stress response. When the brain appraises a threat — physical or psychological — the axis fires, cortisol rises, and the body shifts into a mobilized state: glucose liberated into the blood, cardiovascular tone up, immune and inflammatory responses modulated, digestion and reproduction down-prioritized, and, importantly, memory and attention biased toward the threat. This is adaptive. An organism that couldn't surge cortisol under threat would be dead.

It regulates itself back down. The single most important function — and the one that breaks — is termination. A healthy stress response is defined as much by how fast it shuts off as by how high it climbs. The negative-feedback loop exists to make the response self-limiting, so that a transient threat produces a transient hormonal pulse and then a return to baseline.

The framing that organizes all of this comes from Bruce McEwen, the Rockefeller neuroendocrinologist who reframed stress physiology around the concept of allostasis — "stability through change." The body doesn't maintain rigid set-points; it actively adjusts them to meet anticipated demand. Cortisol is a primary allostatic mediator. The system is supposed to surge and recover. The pathology is in chronic activation, which McEwen named allostatic load — the cumulative wear-and-tear of a stress response that runs too often, too long, or fails to shut off. We'll come back to this; it's the central concept.

03

The mechanism — molecular to behavioral

Walk it from the receptor up.

The molecular bottom: two receptors, not one. This is the detail almost everyone misses, and it's the key that unlocks the whole feedback story. Cortisol acts on two different intracellular receptors with very different affinities — work largely associated with the Dutch neuroendocrinologist Ron de Kloet and colleagues:

  • The mineralocorticoid receptor (MR) — high affinity for cortisol. Because it binds cortisol tightly, MR is largely occupied even at low, basal cortisol levels (the trough of the daily rhythm). MR is densely expressed in the hippocampus.
  • The glucocorticoid receptor (GR) — substantially lower affinity (commonly cited as on the order of ten-fold lower). GR only becomes substantially occupied when cortisol is high — at the morning peak and during stress.

The functional consequence, which de Kloet's group articulated as the MR/GR balance hypothesis: the two receptors mediate different phases. MR, occupied at baseline, maintains the tonic, ongoing housekeeping and sets the sensitivity of the system. GR, recruited only when cortisol spikes, mediates the stress response and — crucially — drives the negative feedback shutoff and the longer-term recovery and adaptation. So the same hormone, at different concentrations, flips the system between "maintain" and "respond-then-recover" by which receptor it occupies. Both MR and GR, when bound, are transcription factors: they translocate to the nucleus and change gene expression. This is why cortisol's effects are slow and lasting — it's rewriting which genes a cell transcribes, not just flipping a switch.

The loop, step by step. Threat appraised → amygdala drives, hippocampus/PFC restrain → PVN releases CRH (+AVP) into the portal system → anterior pituitary releases ACTH into blood → adrenal cortex releases cortisol → cortisol acts body-wide (glucose, immune, cardiovascular) AND feeds back on pituitary, hypothalamus, and hippocampus via GR → upstream CRH and ACTH are suppressed → cortisol falls → system returns to baseline. The feedback operates on multiple timescales: a very fast, non-genomic suppression within minutes, and a slower, genomic suppression over hours.

The circadian layer. Superimposed on this is the daily rhythm, and it's not a smooth sine wave — cortisol is released in pulses (ultradian rhythm), roughly hourly bursts whose amplitude is largest in the morning. The suprachiasmatic nucleus (the master circadian clock) drives the PVN to shape this. One feature of the morning peak has its own name and its own literature: the cortisol awakening response (CAR) — a sharp rise in cortisol (commonly reported on the order of 50% above the waking level) over roughly the first 30–45 minutes after waking, on top of the already-rising morning curve. The CAR is thought to reflect the system gearing up to meet the anticipated demands of the day, and it's sensitive to anticipated stress (it tends to be higher before a demanding day). It has become a widely used, if methodologically finicky, biomarker — finicky because it depends heavily on exact sample timing relative to the moment of waking.

Acute vs. chronic — the hinge. Acutely, cortisol is protective and the feedback brings it home. The damage is in chronicity. When the system is activated relentlessly — chronic psychological stress, trauma, poverty, illness — you get sustained high cortisol, and over time the very feedback machinery degrades. The mechanism that ties this to the brain runs through the hippocampus, which is both a primary brake on the axis (it inhibits the PVN) and densely packed with GR, making it highly sensitive to cortisol. Sustained high cortisol is associated with hippocampal dendritic remodeling and, in chronic conditions, reduced hippocampal volume. The vicious circle McEwen and Robert Sapolsky described: chronic cortisol damages the hippocampus → a damaged hippocampus brakes the axis less effectively → cortisol climbs higher and shuts off slower → more hippocampal damage. The brake gets eaten by the thing it's supposed to stop. This is the "glucocorticoid cascade hypothesis," proposed by Sapolsky, Krey, and McEwen in 1986 — originally framed around aging, and still one of the most influential models in the field, though (see the debates) the strict version has been tempered over the decades.

Behavioral output. Cortisol crosses the blood-brain barrier readily (it's lipid-soluble) and acts directly on brain regions that govern mood, memory, fear, and motivation. Acute cortisol sharpens encoding of emotionally salient memory (why you vividly remember the moment of a crisis). Chronic cortisol impairs hippocampal-dependent memory, biases the brain toward threat and habit over flexible goal-directed action, and is tightly intertwined with the neurobiology of anxiety and depression. The hormone loop terminates, ultimately, in experience — in how safe or besieged the world feels.

04

The landmark science

Hans Selye (1930s–1950s) is the origin point. Selye, working in Montreal, exposed rats to a wide range of noxious stimuli and found they produced a non-specific, stereotyped response regardless of the specific stressor — including enlarged adrenal glands, shrunken thymus/lymphatic tissue, and stomach ulcers. He called this the General Adaptation Syndrome (GAS), with three stages: alarm, resistance, exhaustion. Selye effectively imported the engineering word "stress" into biology and is why we use it the way we do. His non-specificity claim has been substantially revised (responses turn out to be more stimulus-specific than he thought), but the GAS framework launched the field.

Geoffrey Harris (1940s–1950s) established the neurovascular link — that the brain controls the pituitary not by nerves alone but through the hypophyseal portal blood vessels, carrying hypothalamic releasing factors. This is the conceptual foundation of neuroendocrinology: the idea that the brain secretes hormones to command a gland.

Roger Guillemin and Andrew Schally spent decades isolating and characterizing the hypothalamic releasing hormones, winning the 1977 Nobel Prize in Physiology or Medicine for that work (shared with Rosalyn Yalow for the radioimmunoassay that made measuring these tiny hormone quantities possible). CRH itself — the top of the cascade — was isolated and characterized by Wylie Vale and colleagues in 1981, a long-sought 41-amino-acid peptide.

Bruce McEwen discovered, in 1968 (with Weiss and Schwartz), that the hippocampus concentrates corticosterone — the first demonstration that a glucocorticoid acts directly on a brain region governing cognition and emotion, not just on peripheral metabolism. This finding seeded decades of work on cortisol and the brain, the MR/GR distinction, and the allostasis/allostatic-load framework (McEwen & Stellar, 1993).

Ron de Kloet and colleagues developed the MR/GR balance hypothesis through the 1980s–1990s, establishing that the two receptors are differentially occupied across the cortisol range and mediate the maintenance-vs-response/recovery split described above.

Robert Sapolsky brought it together at the organismal level — both in lab work on glucocorticoids and hippocampal vulnerability (the glucocorticoid cascade hypothesis, with Krey and McEwen, 1986) and in long-term field studies of wild baboons in Kenya, where he showed that an animal's social rank and social experience shape its cortisol profile, demonstrating that "stress" in social mammals is profoundly psychological, not merely physical. His book Why Zebras Don't Get Ulcers is the canonical popular synthesis: a zebra's stress response is brief and total, then over; ours runs on imagined and remembered threats, chronically, and that chronicity is what makes us sick.

The dexamethasone suppression test (DST). Dexamethasone is a synthetic glucocorticoid that, in a healthy person, binds GR at the pituitary and suppresses the next morning's cortisol — because the brain reads the dexamethasone as "plenty of cortisol on board, stand down." Originally a tool for diagnosing Cushing's syndrome (cortisol excess). In the 1980s, work associated with Bernard Carroll and colleagues found that a substantial subset of patients with melancholic/severe depression fail to suppress — they are "DST non-suppressors," meaning their negative-feedback brake is impaired. This was an early, much-celebrated attempt at a biological test for a psychiatric diagnosis. It did not pan out as a diagnostic (sensitivity and specificity were too low — many depressed patients suppress normally, and non-suppression appears in other conditions), but as evidence that the HPA feedback loop is dysregulated in depression, it was a genuine landmark and remains influential. The refined version — the dexamethasone/CRH combined test — proved more sensitive to the feedback impairment.

05

How it shapes behavior & experience

The HPA axis is not "in the background." It is, much of the time, the weather of your inner life.

The morning is a hormonal event. That surge of cortisol around waking — and the CAR on top of it — is part of why mornings feel the way they do, why the first 30 minutes after waking carry a specific physiological charge. A blunted, flattened cortisol rhythm (low morning, no decline across the day) is associated with burnout, chronic fatigue states, and chronic stress — the "clock" has lost its amplitude. The felt sense of "I can't get going in the morning, and I'm wired at night" can be a flattened or inverted cortisol curve.

Acute stress sharpens, chronic stress dulls. Under acute threat, moderate cortisol improves memory encoding and focus — the inverted-U relationship between arousal and performance (the Yerkes–Dodson idea) has a glucocorticoid substrate. This is why you remember exactly where you were during a crisis. But push cortisol too high or too long and the curve falls off the other side: retrieval fails (the classic "blanking out" under extreme exam stress is partly cortisol impairing hippocampal retrieval in real time), and chronic elevation grinds down the very memory machinery.

It tilts the brain from flexible to defensive. Sustained cortisol biases the brain toward amygdala-driven, habit-based, threat-focused processing and away from prefrontal, flexible, goal-directed action. Behaviorally this looks like rigidity, reactivity, narrowed attention, and the inability to "think straight" under chronic stress. It is not a character flaw; it is an axis stuck on.

Early life sets the gain. Some of the most striking work — Michael Meaney and colleagues at McGill — showed that rat pups raised by high-licking-and-grooming mothers develop, for life, a better-regulated HPA axis (more hippocampal GR, more efficient feedback, calmer stress responses) than pups of low-nurturing mothers — and that this is transmitted through maternal behavior, not just genes (cross-fostering reproduces it), via lasting epigenetic changes to GR gene expression. This is one of the foundational demonstrations of behavioral epigenetics: early caregiving literally tunes the set-point of the stress system. The human parallels (early adversity, attachment) are studied intensely and are plausible, though the direct molecular translation to humans is still debated and harder to prove.

06

When it misfires

Cushing's syndrome — too much cortisol. Whether from a tumor (often a pituitary adenoma pumping out ACTH — "Cushing's disease") or from prolonged high-dose steroid medication, chronic cortisol excess produces a recognizable picture: central weight gain, a rounded face, muscle wasting, thinned skin, hypertension, glucose intolerance, bone loss — and, tellingly, psychiatric symptoms: depression, anxiety, cognitive impairment, sometimes psychosis. Cushing's is a natural experiment showing what sustained high cortisol does to the brain and body. Hippocampal volume loss in Cushing's is at least partly reversible once cortisol is corrected — an important data point in the debate over whether stress-related hippocampal changes are permanent.

Addison's disease — too little cortisol. Primary adrenal insufficiency (the disease that killed before glucocorticoid replacement existed, and that JFK lived with) causes fatigue, weight loss, low blood pressure, salt craving, and characteristic skin darkening (because low cortisol releases the feedback brake and ACTH precursors — which share a pigment-stimulating fragment — rise). Untreated, an Addisonian crisis is fatal. This is the clearest possible refutation of "cortisol is bad" — without it you die.

Depression — feedback that won't shut off. A consistent finding across decades is that a meaningful subset of people with major depression, especially the melancholic subtype, show signs of an overactive HPA axis with impaired negative feedback: elevated cortisol, hypersecretion of CRH (elevated CRH has been found in cerebrospinal fluid and post-mortem hypothalamic tissue of depressed individuals), and DST non-suppression. The interpretation: the brake (GR feedback) is desensitized, so the system runs hot. This doesn't mean cortisol causes all depression — the relationship is bidirectional and heterogeneous, and many depressed people have entirely normal cortisol — but HPA dysregulation is one of the most robust biological correlates of severe depression we have.

PTSD — and the counterintuitive twist. Here the field had to revise its assumptions. You'd expect a trauma disorder to show high cortisol. But the work of Rachel Yehuda and colleagues found that PTSD is frequently associated with low circulating cortisol alongside enhanced negative-feedback sensitivity (PTSD patients tend to over-suppress on a low-dose DST — the opposite of depression). The proposed mechanism is increased GR sensitivity/number, an over-tuned brake. The picture is more complicated than "stress = high cortisol," and it cleanly distinguishes the HPA signature of PTSD from that of depression even though they often co-occur. Yehuda's group also reported intergenerational effects — altered glucocorticoid regulation in children of trauma survivors — which is provocative and influential, though the epigenetic-inheritance interpretation remains genuinely contested and is not settled science.

Allostatic load — the slow burn. McEwen's central clinical insight: it isn't any single stress hormone elevation that harms you, it's the cumulative dysregulation — too-frequent activation, failure to shut off, failure to mount a response when needed, or compensatory over-activation of one system because another is blunted. Allostatic load links chronic stress to a spread of outcomes: visceral fat and metabolic syndrome (cortisol drives central adiposity and insulin resistance), hypertension and cardiovascular disease, immune dysregulation, impaired wound healing, and accelerated cognitive aging. The HPA axis is a major conduit by which psychological chronic stress becomes physical disease. That is the whole reason it sits at the center of psychosomatic medicine.

07

The debates & frontier (and pop-culture distortions)

The glucocorticoid cascade hypothesis got tempered. The original 1986 Sapolsky–Krey–McEwen model — chronic cortisol causes irreversible hippocampal neuron death, driving a feed-forward spiral of aging and worsening feedback — was hugely influential but turned out too strong. Later work showed that much of what looked like "neuron death" is reversible remodeling (dendrites retract and regrow), that adult neurogenesis in the hippocampus is suppressed by stress but can recover, and that hippocampal volume changes in humans are often reversible with treatment (as in treated Cushing's, or depression remission). The contemporary view is more dynamic and more hopeful than the original cascade: the hippocampus is plastic, not simply degenerating. McEwen himself led much of this revision.

Cause or consequence in depression? The chicken-and-egg problem is unresolved. Does HPA dysregulation drive depression, or does depression (and its lifestyle correlates — poor sleep, inactivity) drive HPA dysregulation? Probably both, in a loop, and probably only in some subtypes. CRH-receptor antagonists were a major pharmaceutical bet to treat depression and anxiety by targeting the top of the axis; the early trials were largely disappointing, which itself is informative — it suggests the axis is a correlate and modulator more than a simple master switch you can flip with one drug.

The epigenetic-inheritance frontier. Meaney's rat work is solid; the leap to human transgenerational epigenetic inheritance of stress (Yehuda's Holocaust-survivor studies, the "Dutch Hunger Winter" cohort) is where things get contested. The popular telling — "your grandmother's trauma is written into your DNA" — vastly outruns the evidence. There are real, replicated animal findings and suggestive human correlations, but clean demonstration of inherited (versus prenatally transmitted or socially transmitted) epigenetic stress marks in humans is genuinely hard and not established. Treat the strong version as a hypothesis, not a fact.

The "cortisol detox / cortisol face" wellness distortion. The current pop-culture wave — cortisol blamed for belly fat, "cortisol face," anxiety, and bad skin, with supplements and "cortisol detox" routines sold to fix it — is mostly marketing on top of a kernel of truth. Yes, chronic cortisol excess (as in actual Cushing's) causes central fat and facial changes. No, your puffy face is almost certainly not pathological hypercortisolism, and no supplement "detoxes" a hormone your body tightly autoregulates. The legitimate core — chronic psychological stress dysregulates a real hormonal system with real metabolic consequences — gets buried under a pile of unfalsifiable product claims. Same pattern as every law in this curriculum: a true mechanism, monetized into magic.

The CAR is real but fragile as a measure. The cortisol awakening response is a genuine phenomenon, but it's one of the most methodologically temperamental biomarkers in the field — wildly sensitive to exactly when you take the saliva sample relative to true waking. Many published CAR findings are underpowered or non-replicated. Believe the phenomenon; be skeptical of any single dramatic CAR claim.

08

How to work with it

The HPA axis is not directly willed — you can't decide to lower your cortisol the way you decide to lift your arm. But it is regulable through its inputs, because the top of the cascade (the PVN) integrates appraisal, context, and physiological state. The levers:

  1. Protect the rhythm. The axis is a clock first. Consistent sleep/wake timing, morning light, and not blunting the natural morning peak with a wired-but-tired all-day stress drip — these keep the diurnal cortisol curve healthy. A flat curve is a warning sign; a steep, well-timed one is health. Chronic sleep deprivation directly elevates evening cortisol and flattens the rhythm.
  1. *Make stress responses terminate.** The thing that breaks isn't the surge, it's the shutoff. The goal is not zero stress; it's complete recovery* between stressors. Whatever returns the system to baseline — slow exhale-emphasized breathing (engages the parasympathetic brake), exercise (acutely raises then resolves cortisol, improving regulation over time), genuine downtime, social safety — is "feedback training" for the axis.
  1. Use safety, not just calm. Because the PVN is driven by appraisal (amygdala) and restrained by context (hippocampus, PFC), the axis responds to perceived safety and controllability. Sapolsky's baboon work and decades of human research converge: predictability, control, and social support blunt the cortisol response to the same objective stressor. Reframing a stressor as a challenge rather than a threat measurably changes the hormonal response.
  1. Treat early-life and chronic adversity as a biological input, not a moral failing. A dysregulated axis from chronic stress or early adversity is a tuned set-point, not weakness. It can be re-regulated — the plasticity findings are the hopeful half of the story — but it takes sustained input, not a one-off.
  1. Don't chase cortisol numbers; chase recovery. Outside of diagnosing actual endocrine disease (which is a doctor's job, via proper testing), obsessing over a cortisol reading is the wellness trap. The variable that matters is functional: does your system fire when it should and come all the way back down when the threat passes?
09

How It Connects

the Law of Rhythm — the HPA axis is rhythm made flesh: a pendulum that swings up under threat and must swing back, plus a daily tide (the circadian/ultradian cortisol curve) that the body keeps whether you attend to it or not. Rhythm's lesson — that the swing back is not optional and that health lives in the return, not the peak — is the literal physiology of cortisol's negative feedback. A flattened, arrhythmic cortisol curve is Rhythm broken at the hormonal level. · Amygdala (the threat input that drives the axis) · Hippocampus (the brake and the casualty — both restrains the PVN and is remodeled by chronic cortisol) · Prefrontal cortex (appraisal and top-down regulation) · Allostasis / allostatic load (McEwen's organizing frame) · The autonomic / sympathetic nervous system (the fast arm of stress, paired with the HPA's slow arm) · Neuroplasticity (why a dysregulated axis can be re-regulated) · the Law of Polarity (acute cortisol protects, chronic cortisol harms — same hormone, opposite ends of a dose-and-duration pole) · the Law of Cause and Effect (chronic psychological stress as a real physical cause of downstream disease, via this exact conduit).

The Compression

Cortisol isn't the enemy — a broken shutoff is. A stress response is healthy by how fast it ends, not how hard it fires; chronic stress kills you slowly by eroding the brake, until the alarm and the daily clock both lose their off-switch.

Your Practice
My Depth
My Notes
← Previous
The Vagus Nerve
Next →
The Mirror Neuron System