Hunger and Cognition: What Ghrelin Does to the Hippocampus Before Lunch

Introduction

Two people once ate lunch three times in ninety minutes.

They were not competitors and nobody was testing their appetite. They were patients with dense amnesia, studied by Paul Rozin and colleagues in 1998, and they had almost no explicit memory for anything that had happened more than a minute earlier [1]. Ten to thirty minutes after finishing a full meal, they were offered a second one. Both ate it. Then a third was offered, and they usually started that too.

Their stomachs were full. Their blood chemistry knew it. What they had lost was the memory of having eaten, and that turned out to be enough.

This was two patients, three occasions each. A sample that small proves nothing on its own, and the article you are reading will keep saying so. But hold the image anyway, because it points at something the rest of this piece has to explain. If a full stomach were the signal that ends a meal, those two people would have stopped. They did not. Something in the brain was supposed to be keeping score, and in their case it was gone.

That is the first crack in the story most people carry around about hunger and thinking. The story goes like this: your brain runs on glucose, hunger means low glucose, low glucose means poor focus, so eat something and you will think better. It is tidy. It is repeated on almost every page that ranks for this question. It is also, in most of its specifics, wrong or unproven.

What follows is the version the research actually supports, and it is stranger. A hormone made in your stomach reaches the memory centre of your brain and builds synapses there. The same hormone starts rising before you are empty, on a schedule you taught it without noticing. Hunger does not degrade your thinking across the board. It reallocates it, sharply, toward food and away from everything else, and recent work suggests it leaves your non-food judgement surprisingly intact. And the single most quoted fact in this whole area, the one about judges and parole, was rebutted in the same journal in the same year by people who had gone and looked at the court schedule.

The Hormone Nobody Was Looking For

Science sometimes finds an answer years before it finds the question.

In the 1970s and 1980s, researchers built synthetic compounds that made animals release growth hormone. The compounds worked. Nobody knew why. They clearly bound to something, and in 1996 that something was identified as a receptor and given the plain name it still carries: the growth hormone secretagogue receptor, GHS-R. The receptor existed, it did something, and its natural partner was missing. In the trade this is called an orphan receptor, and it is an itch a laboratory cannot leave alone.

Masayasu Kojima and colleagues went looking for the missing molecule and found it somewhere nobody expected. Not in the pituitary. Not in the hypothalamus. In the stomach.

Their 1999 paper in Nature reported a 28-amino-acid peptide, purified from rat stomach, that activated GHS-R and released growth hormone [2]. They called it ghrelin. Within two years it was clear the growth hormone story was almost a side effect. Ghrelin's headline property is that it makes animals eat. Give it to rats and they eat more [3]. Infuse it into people and they eat more too [4]. Delete the receptor in mice and both the appetite effect and the growth hormone effect largely disappear [5].

A stomach. Of all the places to look for the body's only hunger hormone, the organ that fills and empties is obvious in hindsight and was not obvious at all at the time. The field had been searching upward, in the brain, for years.

Ghrelin remains the only well-established hormone in the body that makes you hungry rather than full. Everything else in the gut, the whole crowd of satiety signals like GLP-1, CCK and PYY, pushes in the opposite direction [6]. GLP-1 in particular has become famous in the last few years for exactly that reason: it tells you to stop [7].

That asymmetry is worth pausing on. Evolution built many ways to tell you to stop and essentially one way to tell you to start. Starvation was the problem that needed solving, not abundance.

There is one more oddity worth carrying forward. Ghrelin only works if it is chemically decorated first. An enzyme called ghrelin O-acyltransferase attaches a fatty acid to the third amino acid, and without that attachment the molecule does not activate its receptor properly. Mice engineered without the enzyme cope badly with severe food restriction [8]. The body went to real trouble to build this signal. That is usually a hint that the signal matters.

So we have a stomach hormone that causes hunger. The obvious next question is when it fires.

Your Stomach Is Not Watching Your Stomach

Here is where the folk model starts to come apart.

If ghrelin were a fuel gauge, it would rise as your stomach emptied and keep rising until you ate. That is not what it does. In 2001 David Cummings and colleagues sampled plasma ghrelin in people across a normal day and found something much more interesting: the hormone surged shortly before each meal and dropped within about an hour of eating [9]. Before. Not during the emptiest stretch. Before.

A gauge does not anticipate. A clock does.

This is the single fact that most changes how you should read your own mid-morning. It reframes the sensation from a shortage into an appointment.

Julie Frecka and Richard Mattes tested the implication directly in 2008. If the surge is anticipatory, it should follow your habits rather than your energy balance, and it should move when your habits move. That is roughly what they found: the ghrelin rhythm appeared entrained to habitual meal patterns [10]. Your body learned when you eat and started preparing for it.

Think about what that means for the eleven-forty slump you probably know well. Part of it is not a fuel problem at all. It is a schedule your endocrine system memorised, firing on time.

You can test the idea against your own experience. Most people feel it at roughly the same time each working day, and it softens on the days their schedule falls apart. A fuel gauge would not behave like that.

A randomised controlled trial published in 2026 pushed the timing further, reporting that ghrelin levels predicted how much people later ate as much as three hours ahead of the meal [11]. The signal is a forecast, not a report.

This anticipatory quality also explains why meal timing keeps showing up in metabolic research as something that matters in its own right, independent of what is eaten [12], and why shifting food earlier in the day changes glucose handling and circadian markers even when calories are held constant [13]. Disrupt the feeding rhythm badly enough and the consequences reach into cognitive processes as well [14]. If you want the broader version of that argument, our piece on circadian rhythms and the best time to study covers how much of your daily performance curve is clock rather than effort.

None of this is advice about when to eat. It is a statement about what the hormone is doing, and the honest summary is that ghrelin is less a hunger alarm than a scheduled preparation.

How a Stomach Signal Reaches a Memory

A hormone in your blood cannot affect your memory unless it gets to your brain or hands its message to something that can. This step is where almost every popular article stops thinking, and it is where the interesting biology begins.

Ghrelin does cross the blood-brain barrier, and William Banks and colleagues showed in 2002 that the crossing depends on its exact chemical structure, with transport running in both directions [15]. The acylated form and the unacylated form behave differently. That fatty acid tail is not decoration.

So the barrier is not really a wall here. It is a door with an opinion about who comes through, and the fatty acid tail is the key that fits.

The next question is what the brain has waiting for it. In 1997 Xiao-Ming Guan and colleagues mapped where the ghrelin receptor's messenger RNA appears across brain and body [16]. A fuller mapping in rat and mouse brain followed from Jeffrey Zigman's group in 2006 [17]. The receptor turns up where you would expect, in the hypothalamic circuits that run appetite. It also turns up somewhere you would not.

Receptor maps are dull to read and enormously informative. They tell you which parts of the brain a hormone is even allowed to talk to.

It is in the hippocampus. Dentate gyrus, CA1, CA2 [18].

That is the structure that builds new episodic memories. It is the part of the brain that decides which of today's moments you will still have next year, a job covered in detail in our article on how the hippocampus decides what to remember. And it is studded with receptors for a hormone released by your stomach when you have not eaten.

Nobody designed that. It has to mean something.

Two routes, one destination. The bloodstream route was found first. The nerve route took another fourteen years, and it is the more surprising of the two.

2006: The Year a Stomach Hormone Started Building Synapses

Sabrina Diano, Tamas Horvath and sixteen colleagues published a paper in Nature Neuroscience in 2006 that should be far better known than it is [19].

They showed that circulating ghrelin enters the hippocampus and binds neurons there. They showed that it promotes the formation of dendritic spine synapses, the small physical connections between neurons where learning is stored. They showed it generates long-term potentiation, the strengthening of synaptic connections that is the closest thing neuroscience has to a memory being written. And they showed that mice with more ghrelin signalling did better on spatial learning and memory tasks.

Then they did the thing that turns a correlation into a mechanism. They deleted the ghrelin gene. Those mice had fewer spine synapses in CA1 and performed worse on memory tests. Then they gave the ghrelin back, and both the synapses and the performance recovered rapidly.

Remove it, memory degrades. Restore it, memory returns. That is about as close to causal as this kind of work gets.

Papers of that design are rarer than you would think. Most of what gets reported in this field is a correlation dressed in confident verbs.

Every word of that paragraph describes mice. This matters enormously and it will keep being said in this article, because the gap between a rodent hippocampus and yours is exactly where confident health writing goes wrong.

The mechanism work since then has filled in the how. Ghrelin receptor activation makes dentate gyrus granule cells more excitable and improves short-term spatial memory in rodents [20]. It increases glutamate release at the perforant path synapses feeding the dentate gyrus, oddly by cross-activating dopamine D1 receptors [21]. Even the receptor's baseline activity, without any ghrelin bound at all, shapes hippocampal excitability, long-term potentiation and fear memory [22].

Different laboratories, different techniques, the same direction of effect. That kind of convergence is what makes a finding hard to wave away, and it is also why the rodent caveat below matters so much rather than so little.

There is also a growth effect. Brianne Kent and colleagues raised acyl-ghrelin in rats to levels still inside the normal physiological range and found more newly born neurons in the dentate gyrus, along with better performance on a task that depends on that structure [23]. The adult brain does grow a small number of new neurons in this one region, a process we cover in our article on adult neurogenesis, and a hunger hormone appears to be one of its dials.

Building new neurons is expensive. A body that spends on it when food is short is telling you something about what it thinks the shortage requires.

Now the honest caveat, because a good article carries its own counterexamples. Not every population of ghrelin receptors moves memory. A 2025 study found that ghrelin receptors in the olfactory bulb of male mice suppressed food motivation and increased exploration while leaving spatial memory untouched [24]. The receptor is not a memory switch wherever it appears. Location decides.

The Nerve That Carries the Message

For years the assumption was that ghrelin's effect on the brain was a bloodstream story. Hormone travels in blood, crosses barrier, finds receptor. Simple.

In 2020, Elizabeth Davis and eleven colleagues published a result in Current Biology that complicated it in a useful way [25]. They injected rats' nodose ganglia, the cluster of cell bodies belonging to the sensory fibres of the vagus nerve, with a virus carrying a short hairpin RNA that knocked down the ghrelin receptor. The knockdown was targeted: only the vagal neurons that innervate the gut lost the receptor. The hormone in the blood was untouched. The hippocampus was untouched.

The rats developed feeding and metabolic problems, which was expected. They also developed impaired hippocampus-dependent contextual episodic memory, which was not.

Cut the nerve's ability to hear ghrelin and memory suffers, even though the hormone and the memory structure are both still there. The gut is not shouting at the brain through the blood alone. It is talking up a cable.

Consider how strange that is as an engineering choice. The message could have travelled entirely in the blood. Instead there is a second wire, and cutting the wire matters even when the blood is fine.

This fits a wider picture in which the vagus nerve turns out to be a genuine information channel for cognition rather than a plumbing control line. Chronic dietary patterns that damage cognition appear to work partly through this route [26], and stimulating gastric vagal afferents can shift hippocampal-dependent performance in rodent disease models [27]. If you want the fuller map of that traffic, our article on the gut-brain axis and cognition lays out the routes.

For our purposes the point is narrower and sharper. There are two independent paths from an empty stomach to a memory, and interfering with either one changes what the memory does.

The Arrow Points Both Ways

Everything so far has treated the brain as the receiver. Stomach sends, brain listens. That is how every popular article on this subject frames it, and it is half the picture at most.

Scott Kanoski's laboratory spent a decade demonstrating the other half.

In 2013 they showed that ghrelin signalling in the ventral hippocampus of rats stimulates learned and motivational aspects of feeding [28]. Not reflexive eating. Learned eating. In 2015, in eLife, Ted Hsu and colleagues traced the circuit further: ghrelin acting on the ventral hippocampus drives meal-entrained conditioned appetite, and it does so through neurons that talk directly to orexin-producing cells in the lateral hypothalamus [29]. Later work followed the pathway down into the hindbrain, where ghrelin and orexin interact to increase how much a rat eats in a single sitting [30].

Read that again. The memory structure is issuing instructions about eating.

It took the field roughly fifteen years to take that sentence seriously, and the reason is worth naming. The hippocampus had been filed under memory, and appetite research looked in the hypothalamus. Nobody was reading the other department's papers.

The same group also showed hippocampal ghrelin receptor signalling driving socially learned food preferences in rats, meaning the structure is not just deciding when to eat but what counts as worth eating based on prior experience [31]. Their 2017 review pulls the whole argument together: the hippocampus contributes to food intake control through mnemonic, neuroanatomical and endocrine mechanisms all at once [32].

And now go back to the two amnesic patients.

They were the human demonstration, thirteen years before anyone traced the circuit. Their hippocampal memory was gone, and with it went the thing that normally says the meal is finished [1]. It is two patients, and two patients is not a population. But the effect does not require brain damage to see. Suzanne Higgs showed in healthy people that memory for recent eating influences how much they eat next [33]. Remembering lunch changes the size of your afternoon.

That is the ordinary sequence in neuroscience, inverted. Usually the animal work comes first and the human case confirms it. Here the human case sat in the literature for over a decade, waiting for a mechanism to explain it.

The traffic in this system is circular, not one-way.

That loop has no obvious start. It is a system in which memory and metabolism are the same conversation, which is why hunger feels like a state rather than a sensation. If you have read our article on state-dependent memory, the shape will be familiar: internal states are not background noise to cognition, they are part of the address system that cognition uses.

What Hunger Actually Feels Like

Ask someone why they get irritable when they have not eaten and you will hear about blood sugar. The neuroscience answer is more direct and more unsettling.

Deep in the hypothalamus sits a population of neurons called AgRP neurons. Switch them on in a mouse and it eats, immediately and vigorously, even if it is already full [34]. They are as close to a hunger button as the brain has, and their inputs are now well mapped [35].

In 2015 Nicholas Betley and colleagues asked what activating these neurons feels like from the inside. The way you ask a mouse this question is to see whether it will work to avoid the state. It will. AgRP activity turned out to transmit a negative-valence teaching signal [36]. The animal treats the state as something to be escaped, and learns from it.

Hunger, in other words, is not a neutral readout of energy stores. It is aversive by design. It is built to be unpleasant so that you do something about it.

There is a design logic to that. A signal you could comfortably ignore would not have kept anything alive. The unpleasantness is the feature.

That reframes the everyday experience considerably. When you are snappish at half past eleven, the plausible mechanism is not that your neurons are running out of fuel. It is that a circuit whose entire job is to generate an unpleasant motivating state is doing its job.

There is a further twist, and it is a human one. Jennifer MacCormack and Kristen Lindquist ran studies asking when hunger actually turns into hostility, and found that it depended heavily on context. Hunger produced negative emotion most reliably when the situation invited people to interpret their internal state in emotional terms [37]. Being hungry and unaware of it, in a context that gave people no emotional frame, did not reliably make them unpleasant.

Read that finding carefully, because it is easy to overstate in both directions. Hunger did not simply make people unpleasant, and it did not fail to affect them either.

So hanger is real, and it is also partly constructed. Your body supplies an aversive signal. Your interpretation supplies what it means. Recent theoretical work on interoception makes the same point at a systems level, treating metabolic state and mental state as parts of one regulatory loop rather than as cause and symptom [38].

These circuits are also faster than nutrition. When a mouse simply sees or smells food, arcuate feeding circuits change their activity within seconds, long before a single calorie has been absorbed [39]. The system is running on prediction, not on measurement, which is the same theme as the preprandial ghrelin surge showing up in a different part of the brain.

Seconds. Not the twenty minutes a meal takes to register in your blood. The brain is answering a question about food long before your body has any data.

The Glucose Story, and Why It Is Not Enough

Now for the claim you have read a hundred times: your brain runs on glucose, so when you are hungry your brain is starving.

The first half is broadly right. The brain is a heavy and fussy consumer of glucose, it has its own family of transporters to move it across the barrier and into cells, and disruptions to that transport system cause serious neurological problems [40]. Brain fuel metabolism is genuinely central to how the organ ages, and shortfalls in it are part of the story in Alzheimer's disease [41]. Insulin acts in the brain too, not merely on it, and that signalling matters for hippocampal function [42].

The second half is where it breaks. A healthy adult who skips one lunch is not experiencing a brain fuel shortage. Blood glucose in a person without diabetes is defended within a narrow band by several overlapping systems, and the brain has priority access. Missing a meal at one o'clock does not push a normally fed body outside that band.

This is where the popular version does its real damage, because the first half is true enough to carry the second half along with it.

What does damage the hippocampus through metabolic routes is chronic, not acute. Diet-induced insulin resistance impairs hippocampal synaptic plasticity and cognition in middle-aged rats [43]. Long-term Western-pattern diets are associated with hippocampal dysfunction and cognitive impairment [44]. Sustained obesity relates to measurable differences in cognitive function [45].

Notice the mismatch. The evidence for metabolic harm to memory concerns months and years of dietary pattern. The claim being sold concerns the three hours between breakfast and lunch. They are not the same phenomenon and they should not share a paragraph, let alone a headline.

Timescale is the most commonly dropped variable in health writing. It is also usually the one that decides whether a claim is true.

Real starvation is a third and separate thing again. Ancel Keys and colleagues documented it in extraordinary detail in The Biology of Human Starvation, published by the University of Minnesota Press in 1950, following volunteers through months of semi-starvation. What they recorded was profound: obsessive preoccupation with food, apathy, irritability, social withdrawal, depression. That is a real and terrible cognitive effect of hunger. It is also the effect of half a year of it, and quoting it to explain an eleven o'clock slump is a category error. So is quoting the very real cognitive consequences of chronic childhood malnutrition [46].

Three different things wear the same word. Missing a meal, eating badly for years, and genuine starvation are not points on one line.

Keep those three separated for the rest of this article. Almost every confused claim about hunger and the brain comes from sliding between them.

The Famous Study That Did Not Survive

If you have encountered one piece of research about hunger and thinking, it is almost certainly this one.

In 2011, Shai Danziger, Jonathan Levav and Liora Avnaim-Pesso published an analysis in PNAS of 1,112 parole rulings made by eight Israeli judges over 50 hearing days across ten months [47]. The finding was dramatic. The proportion of favourable rulings started at roughly 0.65 at the beginning of a session, declined steadily toward nearly zero as the session went on, and jumped back to roughly 0.65 immediately after a food break.

It travelled fast, and it travelled stripped of qualifications. Judges are hungry, therefore judges are harsh. It appears in bestselling books, corporate training decks and productivity blogs to this day.

You have probably heard it. It is one of those findings that arrives already sounding true, which is exactly the property that should slow a reader down.

Later the same year, in the same journal, Keren Weinshall-Margel and John Shapard published an objection with a devastatingly simple premise: they went and looked at how the hearings were actually scheduled [48].

Case ordering was not random. The boards worked through all the cases from one prison before taking a break, then started a different prison afterwards. And crucially, whether a prisoner had a lawyer mattered enormously and was not evenly distributed across the session. Unrepresented prisoners made up roughly a third of cases and were granted parole at roughly 15 percent, against roughly 35 percent for represented prisoners. The authors examined the original data, added 12 hearing days of their own covering 227 decisions, and interviewed three attorneys, a panel judge and five prison and court staff.

Notice what they did there. They did not run a better statistical model. They went and asked how the room actually worked.

The original authors replied, defending the finding and arguing the confound could not account for the whole effect [49]. That exchange has never been resolved cleanly.

Then in 2016 Andreas Glöckner approached it from a different direction. Rather than argue about the data, he simulated it. He built models in which case ordering worked the way the courts described and no hunger effect existed at all, and asked whether the published pattern could emerge anyway. His conclusion was that the magnitude of the effect had been substantially overestimated, and that plausible scheduling mechanisms could generate a pattern like the reported one without any biology [50].

Where does that leave things? The pattern in the data is real. The cause is unresolved, and has been unresolved for well over a decade. Anyone who tells you that hungry judges deny parole is quoting one side of a live argument and leaving out the other.

That is worth sitting with for a second, because the failure here is not really the original researchers'. They published a striking finding and defended it in the open. The failure is downstream, in a hundred retellings that dropped the rebuttal because it made a worse story.

Willpower Was Never a Fuel Tank

Attached to the hungry judges story is a larger idea that deserves the same scrutiny: that self-control runs on glucose, and that using it depletes a finite supply.

The idea is intuitive. It also does not hold up well.

Robert Kurzban asked the direct physiological question in 2010: does the brain actually consume measurably more glucose during self-control tasks? Working through the energetics, the answer he reached was that the numbers do not support the model [51]. The brain's metabolic rate is high and remarkably stable. Effortful thinking does not visibly drain a tank.

Which is a strange thing to learn about an idea that feels so obviously right. Concentrating for four hours is genuinely tiring. It is simply not tiring in the way that running for four hours is tiring.

Empirical work has pointed the same way. One study found blood sugar levels related to aggression between partners without supporting the glucose model of self-control as the explanation [52].

One study is one study. What follows is not.

Then came the replication. In 2016, Hagger, Chatzisarantis and 62 co-authors ran a preregistered multilab replication of the ego depletion effect, the flagship phenomenon of the entire self-control-as-resource literature. Across laboratories, using an agreed protocol fixed in advance, they did not find the effect at the size the original literature reported [53].

This does not mean effort is imaginary. Anyone who has concentrated hard for four hours knows something real happens. It means the specific mechanism, a glucose reservoir that self-control drains, is not the explanation. The current thinking leans toward motivation and attention allocation rather than fuel.

The distinction matters practically. If depletion is about fuel, you eat. If it is about motivation and attention, the answer is a different one entirely, and the honest position today is that we do not have it nailed down.

So when the next article tells you to eat something because your willpower is running low, the sentence is doing something odd. It is invoking a mechanism the field has largely set down, to explain a feeling that is real, and prescribing an intervention that may work for entirely different reasons.

So Does Hunger Make You Worse at Thinking?

We can finally answer the question the title implies, and the answer has a shape that no popular article gives it.

The best-designed recent attempt comes from Jennifer March and colleagues, published in Scientific Reports in 2026 [54]. Seventy participants completed three different tasks, once hungry and once sated, while an eye-tracker recorded where they looked. One task was food choice. One was intertemporal discounting, the classic smaller-sooner versus larger-later decision. One was social preferences, involving allocations between themselves and others.

Hunger had a large effect on the food task. Hungry participants looked longer at tasty options over healthy ones and chose accordingly. Computational modelling localised the effect precisely: hunger changed how attention fed into the accumulation of evidence for a choice.

In the two non-food tasks, attention and choice were essentially unaffected.

Hold on to that asymmetry. It is a much narrower claim than the one you usually meet, and a much more believable one.

That is a domain-specific effect, and it contradicts the blanket claim. Hunger did not make people impulsive in general, impatient in general or selfish in general. It made them extremely interested in food.

This dovetails with a well-established finding about attention: hungry or not, the strength with which food cues grab you predicts eating behaviour and weight change over time [55]. Hunger turns that dial up. Ghrelin is part of how it does so, acting on midbrain dopamine neurons that assign motivational value [56] and shifting activity in human brain regions that govern appetitive behaviour [57]. The reach of this system into general reward circuitry is broad enough that central ghrelin signalling is required for normal alcohol reward in mice [58]. Our article on dopamine and learning covers what that valuation machinery does when it is not being pulled around by a hormone.

The picture is not settled, and the article would be dishonest to pretend otherwise.

Pulling the other way is a 2014 study by Denise de Ridder and colleagues with the wonderful title "Always Gamble on an Empty Stomach" [59]. Across three studies with 30, 50 and 45 participants, hungry people performed better on the Iowa Gambling Task, choosing more often from the advantageous decks, and showed a greater appreciation of delayed larger rewards without taking more risk on a separate risk task. That is the opposite of the folk prediction.

Pulling in a third direction, a 2026 human imaging study looked for effects of unacylated ghrelin on risk-taking and neural response and reported no credible evidence for them [60]. And other 2025 work reports decision-making scores correlating with how susceptible a person is to hunger, mediated by a metabolic signalling factor rather than by body fat [61].

Four studies, four different shapes of answer. The domain-specific reading is the most parsimonious synthesis available right now. It is not a verdict.

Which is what an open question looks like from the inside. It is not a failure of the field. It is a field that has not finished.

Read down the second column and the headline of this article assembles itself. There is no single arrow. There is a reallocation.

What Happens When You Actually Skip Lunch

Enough mechanism. What do the human trials say when someone simply does not eat?

Less than you might hope, and that is informative in itself.

When a phenomenon is large, the trials agree quickly. When they keep coming back small and scattered, that is usually the answer rather than a delay in reaching one.

A 2024 study looking at short-term fasting and cognitive ability found effects that were modest rather than dramatic [62]. A 2025 review of continuous calorie restriction and fasting on cognition in adults without eating disorders reached a similarly restrained conclusion [63]. A randomised controlled trial published in 2026 tracked people adapting to intermittent fasting and found cognitive performance simply stable throughout [64].

Stable. Not sharpened, not wrecked. That is a real result and it deserves to be reported as one, even though it makes a bad headline.

Nobody writes an article about a null result, which is precisely why you have not read about this one.

Where changes do show up, they are often tradeoffs rather than gains or losses. Three weeks of a ketogenic diet in undergraduate students produced responses that were faster but less accurate [65]. Speed bought at the cost of precision is not an improvement, and it is not a decline either. It is a different operating point.

Other nutritional manipulations land in the same modest territory. Ketone ester supplementation shows measurable but limited cognitive effects [66]. An umbrella review of intermittent fasting on metabolic and cognitive health in adults with obesity gathers a literature that is broad but not decisive [67]. Even the composition of a single meal can shift postprandial hunger, satiety and attention in measurable ways [68], and a controlled trial found that what accompanies a high-carbohydrate breakfast changes both metabolic and cognitive outcomes afterwards [69].

The pattern across all of it is small effects, high variability between people, and a strong dependence on what exactly is being measured. That is what an honest literature looks like when a phenomenon is real but modest.

None of the above is a recommendation about your eating. It describes what researchers measured in volunteers under controlled conditions. Anything involving your own health, your own eating patterns, or any medical condition belongs with a clinician and not with an article.

Where It Genuinely Matters

There is one place where the link between hunger and cognition is not a matter of a mildly interesting mid-morning dip, and it deserves to be stated plainly.

Benton and Parker's 1998 work sits near the origin of the modern research on breakfast, blood glucose and cognition [70]. Two decades later, Katie Adolphus, Clare Lawton and Louise Dye reviewed what the field had accumulated on breakfast, behaviour and academic performance in children and adolescents [71]. Their review is worth reading precisely because of how carefully it hedges. There are associations. There are also serious confounds: socioeconomic status, habitual eating patterns, school-level differences, and the simple problem that children who eat breakfast differ from children who do not in many ways that have nothing to do with breakfast.

Nearly thirty years later the question is still not cleanly answered, which tells you how hard it is to study something that correlates with everything else in a child's life.

The honest position is that a child arriving at school having eaten is doing better on average, and that we cannot cleanly separate how much of that is the food. Broader dietary quality in children and adolescents shows associations with cognitive function too, with the same interpretive difficulties [72].

That is not a dodge. It is what the evidence supports, and writing anything firmer would be doing the exact thing this article has spent three sections criticising other people for.

Food insecurity is a different order of problem from a skipped meal, and chronic undernutrition in early life has consequences that persist [46]. Nothing in the mechanism sections above should be read as making that smaller.

The clinical edge matters as well. Ghrelin signalling is being examined in Alzheimer's and Parkinson's disease [73], in diabetes-related cognitive impairment [74], and in the relationship between metabolic state and mood [75]. Anorexia nervosa has become a place where the metabolic regulation of synaptic plasticity is studied directly, because the disorder forces the question [76]. These are research directions. They are not treatments, and nothing here should be read as suggesting otherwise.

The State You Are In Is Not Only About Food

One more thread, because it changes how you should read your own eleven o'clock.

Ghrelin is not driven by food alone. Sleep moves it. Restrict sleep in healthy young men and ghrelin rises alongside self-reported hunger [77]. Population data links short sleep duration to elevated ghrelin [78]. A 2026 randomised clinical trial in adolescents found sleep restriction shifting both macronutrient intake and reported hunger [79], and the two-way relationship between ghrelin and sleep now has its own review literature [80].

Stress moves it too. In mice, ghrelin appears to defend against the depressive consequences of chronic stress rather than to cause them [81], which sits oddly alongside the popular framing of stress hormones as uniformly destructive. Bruce McEwen spent a career arguing that these mediators are protective and damaging depending on context and duration [82], an argument our article on stress, cortisol and memory follows in detail.

Protective and damaging at once, depending on how long it lasts and what else is happening. That description fits almost every mechanism in this article.

Then there is the finding that is hardest to shake. In a randomised trial, making people feel lower in subjective socioeconomic status raised their ghrelin [83]. Not their diet. Their sense of where they stood relative to others.

Sit with that for a moment. A social comparison, held entirely in your head, moved a hormone made in your stomach.

Metabolic and psychiatric measures keep turning up entangled in this way. Fasting ghrelin appears to mediate part of the relationship between obesity and depressive symptoms [84], and appetite hormones show up alongside working memory dysfunction in adolescents with affective disorders [85].

So when you feel that mid-morning fog, the inputs include when you habitually eat, how you slept, how stressed you are, and how you feel about your position in the world. Calling it hunger is a simplification that happens to be convenient.

Why Evolution Would Build It This Way

Step back far enough and the strange part stops being strange.

Consider an animal that has not eaten. What does it most need? Not calm reflection. It needs to find food, and finding food is a memory problem. Where was the fruiting tree. Which route led to water. What did the place look like that had grubs under the bark last season. An animal whose spatial memory sharpened when its stomach emptied would out-forage one whose memory faded.

Under that reading, the hippocampal effects are not a quirk. They are the point. The hormone that says food is needed also improves the machinery for finding it. The attention narrowing that March and colleagues measured is the same logic operating at a shorter timescale: hunger makes food-relevant information more salient because food-relevant information is what solves the problem.

It also explains why the effect would be domain-specific rather than general. A hungry animal does not need to be better at everything. It needs to be better at one thing, urgently.

This arrangement is very old. In fruit flies, hunger state switches a memory circuit through an AMPK-dependent feedback loop [86]. In another invertebrate system, octopamine signalling drives hunger-enhanced olfactory learning [87]. Animals separated from us by hundreds of millions of years of evolution independently link an empty gut to sharper learning about where food is.

The mammalian version simply runs on different molecules. The hypothalamic circuitry that manages it is now mapped in considerable detail [88], and the same architecture that made this useful for a foraging animal is what makes it awkward in a supermarket [89]. A system tuned to make food salient when food is scarce behaves differently when food is three metres away and engineered to be irresistible.

Look at the shape of that list. The first decade is chemistry. The second is mechanism. The third is the field discovering that the arrow it drew was pointing the wrong way, and that some of its most famous psychology did not replicate.

Three decades, and the most recent entries are the ones subtracting from what the earlier ones claimed. That is science working, and it reads like science failing, which is why so little of it reaches the reader.

What the Evidence Does and Does Not License

Here is what can be said with reasonable confidence.

Ghrelin comes mainly from your stomach, needs a fatty acid attached to work, and makes people eat more when it is given to them. It rises before your habitual meals rather than in proportion to how empty you are. It reaches the hippocampus by two routes, through the bloodstream and up the vagus nerve.

In rodents it builds spine synapses, generates long-term potentiation, supports the birth of new dentate neurons, and improves spatial memory, with knockout-and-rescue evidence behind that claim. The hippocampus, in turn, helps decide when a meal starts and when it is over, in rats through orexin pathways and in humans through the memory of having eaten. Hunger neurons are aversive by design.

And sleep, stress and social comparison all move the same hormone.

Here is what cannot be said.

That hunger reliably degrades your general thinking. The best-controlled recent evidence says the effect is confined to the food domain.

That judges deny parole because they are hungry. That claim has two published rebuttals and no clean resolution. That willpower is a glucose tank that hunger drains. The flagship effect for that model failed a 64-author replication.

That skipping a meal starves your brain. In a healthy adult it does not. That any of the rodent synaptic findings translate directly into a memory benefit for you. Nobody has shown that, and one human ghrelin imaging study came back explicitly null.

And here is the part worth keeping.

Your relationship to food is not a separate system from your relationship to memory. They share hardware. The structure that stores your afternoon is the same structure being asked when the afternoon should include lunch, and the hormone in that conversation was made in your gut before your brain ever heard about it. Whatever you feel at half past eleven, it is not your neurons running out of fuel. It is a very old system, built for a world with less food and more walking, doing exactly what it evolved to do in a building where the kitchen is down the hall.

If that leaves you curious about which other everyday chemicals push the same machinery around, our piece on how caffeine affects learning takes the same evidence-first approach to a molecule you probably use more deliberately.