What Controls Appetite in the Brain?

Ask most people what part of the brain controls appetite and you will get a vague gesture toward the stomach. The honest answer sits about ten centimetres higher and slightly behind the eyes, in a structure roughly the size of an almond. The hypothalamus does not feel hunger so much as compute it — weighing signals from the gut, the fat stores, the bloodstream, and the higher cortex, then issuing a verdict that arrives in consciousness as the urge to eat or the contentment to stop. Understanding appetite control in the brain means understanding that circuit, because every modern weight medication works by tilting it.

In 2000, Michael Schwartz and colleagues at the University of Washington published a review in Nature that gave the field its working map. They described a system in which peripheral hormones report on the body's energy state and a small set of hypothalamic neurons translate those reports into behaviour. Two decades of refinement later, the map has more detail, but its architecture has held. This is a tour of that architecture.

The arcuate nucleus: where the body reports in

At the base of the hypothalamus, pressed against the floor of the third ventricle, lies the arcuate nucleus. Its position is not incidental. Here the blood-brain barrier is unusually permeable, which lets circulating hormones reach neurons that would otherwise be sealed off from the bloodstream. The arcuate is, in effect, the brain's customs house — the place where messengers from the body present their papers.

Two populations of neurons sit in the arcuate, and they want opposite things. This opposition is the heart of homeostatic appetite, the slow accounting system that tries to keep body energy stable over weeks and months. It runs underneath conscious awareness, and it is older, in evolutionary terms, than the desire for anything in particular.

AgRP/NPY neurons: the accelerator

One population co-expresses agouti-related peptide (AgRP) and neuropeptide Y (NPY). These are the orexigenic neurons — the ones that drive eating. When energy is low, when the stomach is empty, when ghrelin is rising, AgRP/NPY neurons fire. Their activity is among the most powerful appetite signals known: in rodents, switching them on artificially provokes voracious feeding within minutes, even in animals that have just eaten to fullness. These are the cells that make hunger feel urgent rather than optional. If you want the longer story of the chemical messengers that switch them on, we cover it in hunger hormones explained.

POMC/CART neurons: the brake

The opposing population expresses pro-opiomelanocortin (POMC) and cocaine- and amphetamine-regulated transcript (CART). These are the anorexigenic neurons — they suppress appetite. When energy stores are adequate, when leptin and insulin are reporting plenty, POMC neurons fire and the urge to eat recedes. POMC is a precursor protein, cleaved inside the neuron into smaller active fragments. One of those fragments, α-melanocyte-stimulating hormone, is the molecule that carries the satiety message onward. The interplay between these signals and the sensation of fullness is the subject of our piece on the satiety hormones GLP-1, PYY and CCK.

Crucially, the two populations do not merely coexist — AgRP/NPY neurons actively inhibit POMC neurons, releasing GABA onto them and silencing the brake while pressing the accelerator. Hunger is not the absence of satiety. It is the active suppression of it.

The melanocortin system and the MC4R switch

The α-MSH released by POMC neurons does not act locally. It travels to a second relay, chiefly the paraventricular nucleus of the hypothalamus, where it binds the melanocortin-4 receptor — MC4R. Roger Cone, whose laboratory mapped much of this pathway, described the central melanocortin system in a 2005 review as the point where long-term adipostatic signals and acute hunger cues are integrated into a single output. MC4R is the switch that converts that integration into a change in feeding.

AgRP is the reason this matters so much. AgRP is not simply an appetite stimulant — it is an inverse agonist at MC4R, meaning it actively blocks the receptor and dials its baseline activity down. So the two arcuate populations compete for the same lock: POMC's α-MSH turns MC4R on to suppress appetite, while AgRP turns it off to release it. The melanocortin system is where the accelerator and brake meet a single pedal.

The clinical proof sits in human genetics. MC4R mutations are the commonest known monogenic cause of severe obesity, accounting for a meaningful fraction of childhood-onset cases. Lose the receptor and you lose the satiety signal at its most important junction — a vivid demonstration that this is not an academic circuit but a load-bearing one. The same logic underpins how appetite regulation works as a whole: a chain of signals in which any broken link reshapes eating behaviour.

The brainstem: the other appetite brain

The hypothalamus gets the headlines, but it is not eating alone. Far below it, in the brainstem, the nucleus tractus solitarius (NTS) receives a torrent of information from the gut by way of the vagus nerve. As the stomach stretches and the intestine senses nutrients, vagal afferents fire and the NTS registers it. This is the circuit that ends a meal in real time — the within-meal satiation that tells you the plate is finished before any hormone could have circulated.

The NTS does two things. It can act locally, talking to nearby brainstem nuclei to halt feeding directly; and it can relay upward to the hypothalamus, integrating the moment-to-moment gut report with the longer-term energy accounting of the arcuate. Gregory Morton, Thomas Meek and Schwartz, in a 2014 Nature Reviews Neuroscience synthesis, framed the modern picture as exactly this kind of distributed system — hypothalamus and brainstem in constant dialogue rather than a single command centre. When you wonder why you feel hungry at a particular moment, the answer is usually a vote tallied across both.

How peripheral hormones feed the circuit

The neurons described so far are listeners. What they listen to is a chorus of hormones, each reporting on a different aspect of the body's state.

Leptin, secreted by fat tissue, is the long-term adiposity signal. More fat means more leptin, which acts on the arcuate to stimulate POMC neurons and inhibit AgRP/NPY neurons — the brain's way of sensing how much energy is in the bank. Insulin, from the pancreas, carries a parallel message and acts on overlapping arcuate targets. Schwartz's original model placed leptin and insulin at the centre precisely because they encode stored energy rather than the last meal.

Ghrelin, from the stomach, runs the other way. It is the only major peripheral hormone whose primary action is to stimulate appetite, and it does so by exciting the AgRP/NPY neurons directly — the gut's request for food, written straight onto the accelerator. The satiety hormones GLP-1 (glucagon-like peptide-1) and PYY (peptide YY), released from the intestine after eating, push back: they signal through both the NTS and the arcuate to suppress intake and bring a meal to a close. Homeostatic appetite, in the end, is the running sum of these competing reports.

Homeostatic versus hedonic: the two hungers

All of this describes hunger as bookkeeping — eat when energy is low, stop when it is restored. But anyone who has eaten dessert after a large meal knows that bookkeeping is not the whole story. There is a second appetite, and it does not care whether you are full.

This is hedonic appetite — eating for pleasure and reward rather than need. It runs largely through the mesolimbic dopamine system, the same circuitry implicated in other rewards, where dopamine projects from the ventral tegmental area to the nucleus accumbens. Palatable, energy-dense food activates this system powerfully, generating what researchers distinguish as the "wanting" (incentive motivation) and "liking" (sensory pleasure) of food. Hans-Rudolf Berthoud, reviewing the question in 2011 under the title "who is the boss?", argued that in a modern food environment the hedonic system can routinely override the homeostatic one — which is much of why availability and palatability, not hunger, predict overeating.

The two systems are not separate boxes. AgRP signalling reaches reward circuits, leptin modulates dopamine neurons, and the hypothalamus and mesolimbic system talk constantly. But the distinction is real and clinically useful: you can be homeostatically satiated and hedonically ravenous at the same time. Our deeper look at how medication acts on this reward axis is in how GLP-1 quiets food cravings in the brain, and the wider set of related pieces lives in the appetite regulation hub.

How GLP-1 medications act on these circuits

GLP-1 receptor agonists such as semaglutide and the dual agonist tirzepatide are sometimes described as gut drugs, but their appetite effect is overwhelmingly central. In 2014, Anna Secher and colleagues at Novo Nordisk traced where liraglutide actually goes in the brain. The answer was precise and consequential: the drug reached the arcuate nucleus and was taken up by the very POMC/CART neurons that suppress appetite, directly stimulating them while indirectly inhibiting the AgRP/NPY neurons through GABA signalling.

In other words, these medications push the arcuate balance toward the brake and away from the accelerator — the same fulcrum the body's own leptin and satiety hormones use. They also engage the NTS in the brainstem, reinforcing the within-meal stop signal, and they appear to dampen the hedonic system, which is why people on them so often report that food has simply lost its pull. The full pharmacological picture is laid out in how GLP-1 affects appetite.

The conceptual point is that these drugs did not invent a new way to feel full. They borrowed the brain's existing one. Everything in this article — the arcuate, the melanocortin switch, the brainstem, the reward system — was already there. The medication simply speaks the circuit's native language. For the broader library of evidence-led writing on this, see our appetite and hunger coverage.

Key takeaways

• The hypothalamus — specifically the arcuate nucleus — is the primary brain region that controls appetite, sitting where circulating hormones can reach neurons across a leaky blood-brain barrier.
• Two opposing arcuate neuron populations set hunger: AgRP/NPY neurons drive eating (the accelerator) and POMC/CART neurons suppress it (the brake), with AgRP/NPY actively inhibiting POMC.
• The melanocortin system converges on the MC4R receptor; α-MSH turns it on to curb appetite while AgRP blocks it. MC4R mutations are the commonest monogenic cause of severe human obesity.
• The brainstem's nucleus tractus solitarius reads real-time gut signals via the vagus nerve and ends meals, working in concert with the hypothalamus.
• Leptin and insulin report long-term energy stores; ghrelin drives hunger; GLP-1 and PYY drive satiety — all feed into the arcuate and brainstem circuits.
• Homeostatic appetite (eat for need) is distinct from hedonic appetite (eat for reward, via mesolimbic dopamine), and the reward system can override need in a modern food environment.
• GLP-1 medications act centrally — stimulating arcuate POMC neurons, inhibiting AgRP neurons, engaging the brainstem, and dampening reward — by borrowing the brain's own satiety machinery.