Brain Temperature: Why Heat Slows Thinking Before You Notice It

Introduction

In July 2016, a heat wave settled over Boston and 44 university students woke up every morning to a cognitive test on their phones.

They were healthy. They were young, average age twenty. Half of them lived in dormitories with air conditioning and half lived in older buildings without it, which is the only reason the study worked. For twelve days they took a Stroop test and an arithmetic test before getting out of bed [1].

The students without air conditioning were 13.4 percent slower on the Stroop test. Their arithmetic reaction time was 13.3 percent slower. Their output fell too, by about ten percent on one task and six percent on the other.

The indoor temperature in those rooms averaged 26.3 degrees Celsius.

Twenty-six degrees. That is a warm room. It is not a dangerous room, not a fever, not an emergency. Nobody in that building was ill. They just could not think as fast, and there is nothing in the experience of sitting in a 26 degree room that tells you that you are ten percent worse at something.

That is the whole problem with heat and cognition. The impairment arrives before the alarm does.

None of this is exotic. It is the most ordinary environmental variable there is, and it is the one nobody schedules around.

This article is about why. It runs through what a human brain actually measures on a thermometer, which turns out not to be the number almost everyone carries in their head. It covers what happens inside a skull when the air gets hot, which system fails first, and the one experiment that should change what you do about it. It also covers a thirty year argument that scientists are still having about whether the human brain can cool itself at all.

Some of this is settled. Some of it is genuinely contested, and where it is, this article names the people on both sides rather than picking a side for you.

The Number You Have Is the Wrong Number

Ask almost anyone how warm the human body is and you get 37 degrees Celsius. Ask how warm the brain is and most people give the same answer, or shrug and add half a degree.

In 2022, a team led by Nina Rzechorzek published the first proper map of temperature inside healthy human brains, and the answer was nothing like that [2]. They used magnetic resonance spectroscopy on 40 healthy adults, 20 men and 20 women, aged 20 to 40. They scanned them at different times of day.

The mean brain temperature was 38.5 degrees Celsius. In the same people, at the same time, oral temperature averaged 36.0.

Read those two numbers together for a second. The gap is about two and a half degrees. If you took an oral reading of 38.5 in a clinic you would be told you had a fever. That is the ordinary, healthy, resting temperature of the organ inside your skull.

That is not a subtle correction. It is the difference between what you would call healthy and what you would call ill, applied to the same measurement in the same person on the same afternoon.

It gets stranger. Individual brain regions in those healthy volunteers ranged from 36.1 all the way up to 40.9 degrees. Parts of a normal brain routinely sit above 40, which is a temperature nobody would tolerate as a whole-body reading.

So the first thing to give up is the idea of a single reading. Asking for the temperature of a brain is like asking for the temperature of a house. It depends which room, and when.

The brain also has a clock. Global brain temperature in those volunteers moved by 0.57 degrees across a single day, and deep regions moved by 0.86. It is coolest at night. That rhythm is not decoration. In the second half of the same study, the researchers looked at 114 patients with traumatic brain injury and found that the patients whose brains had lost the daily rhythm were the ones who died. A missing temperature cycle predicted mortality better than the raw temperature did.

Two more findings from that dataset are worth carrying forward, because they change who this article is about. Brain temperature ran 0.36 degrees higher in women in the luteal phase of the menstrual cycle compared with follicular-phase women and men. And deep brain temperature climbed roughly 0.6 degrees across twenty years of age.

So there is no single brain temperature. Later work using the same family of techniques found that even at rest the temperature fluctuates dynamically, and that the gap between brain and body is not a fixed offset you can look up [3].

What there is instead is a distribution that shifts with the region, the hour, the age, and the cycle. This is the point where the topic stops being trivia. If you want to know what heat does to your thinking, you first have to accept that the baseline you were working from was wrong.

What does this mean? It means the brain does not have a comfortable margin of two or three degrees below some dangerous ceiling. It is already running hot. The interesting question is not how it survives being warm. It is how it stays as stable as it does, and what happens on the day that stability gives way.

Why Nobody Knew Until Recently

It is worth asking why a fact this basic took until 2022 to settle. The answer is not that nobody cared. It is that measuring the temperature inside a living human head is genuinely hard.

For most of the twentieth century, the only reliable way to know a brain's temperature was to put a probe in it, which restricted the entire evidence base to neurosurgical patients and people in intensive care. Those are not healthy brains. Building a picture of normal physiology from them is like describing normal traffic by watching a motorway after a crash.

The workaround came from chemistry. Certain molecules resonate at frequencies that shift predictably with temperature, so a magnetic resonance scanner tuned to read those shifts can calculate temperature without touching anything. Karaszewski and colleagues demonstrated it in acute ischaemic stroke patients in 2006 [4], and the technique has been refined steadily since [5].

Other approaches were tried and are still in use. Diffusion-based thermometry reads temperature from the movement of water in cerebrospinal fluid [6]. Near-infrared spectroscopy attempts it optically [7]. Diffusion-weighted methods have been applied to mild traumatic brain injury [8].

Each has its own error bars, which is exactly why comparing published brain temperatures across methods is a trap.

Biophysical modelling now runs alongside the scanners. Researchers build a thermal model of a head, predict what the temperature field should be, and check it against whole-brain thermometry to see where the model breaks [9]. Those bioheat models have improved, though the field is honest that they are not finished [10]. More recent work combines several non-invasive monitoring streams to predict a temperature map rather than measure one directly [11].

The reason this history matters is that a lot of what circulates about brain temperature was written before 2022. Reviews from that earlier period are careful and useful, and they still put core brain temperature in the region of 36 to 37.5 degrees [12].

They were reading the instruments they had. The instruments got better.

Where All That Heat Comes From

Before going further it helps to know where the heat is coming from in the first place. The answer is unglamorous and it explains most of the rest.

A brain is about two percent of your body weight and takes something in the region of a fifth of your energy budget. Almost all of the energy a neuron uses ends up as heat, because that is what happens to energy in a cell that is not doing mechanical work.

So the brain is a small, well-insulated furnace sitting inside a bone box.

That framing explains the puzzle in the previous section. The brain is warmer than the rest of you because it produces more heat per gram than the rest of you, and it is wrapped in skull and scalp that slow the escape of that heat. The blood arriving from the body is cooler than the tissue it flows into, which means blood is not just a delivery service for oxygen. It is the cooling loop [13].

Temperature is not incidental to that machinery either. Ion channels open and close at rates that depend on temperature, enzymes work faster when warm until they do not, and the release of neurotransmitters across a synapse is a physical process with a thermal profile. Everything in the way neurons actually pass signals to each other has a temperature term in it somewhere.

Every part of that loop is a temperature variable. Cerebral blood flow carries heat out. Cerebrospinal fluid moves heat around. The meninges, the skull and the scalp each contribute a layer of thermal resistance, and the scalp has a high density of sweat glands, which is why a hot head sweats first and hardest.

There is a newer proposal worth mentioning as a proposal rather than a finding. Some researchers have argued that lactate shuttling between brain cells acts as an allostatic mechanism for handling heat, moving metabolic load around rather than simply generating it in place [14].

It is an interesting idea. It is not settled, and this article does not treat it as though it were.

Understanding the loop matters for the rest of the argument. If blood flow is how the brain sheds heat, then anything that reduces blood flow to the brain reduces its ability to cool itself. Hold that thought, because heat does exactly that.

Does the Brain Cool Itself? Thirty Years of Argument

Up to now this article has described measurements. This section describes an argument, and it is one where the honest answer is that nobody has won. It is also the place where most popular writing quietly picks a side without telling you.

The idea is called selective brain cooling. In its strong form it says the body has a dedicated mechanism for keeping the brain cooler than the rest of the core, so that when you overheat the brain is protected first. It is an attractive idea. It would explain a lot.

For some animals it is not in doubt. Sheep, goats, gazelles, camels and many other hoofed mammals have a structure called the carotid rete, a dense network of small arteries that sits in a pool of venous blood cooled by the nasal passages. Warm arterial blood heading for the brain runs through cool venous blood coming back from the nose, and heat crosses between them before the blood arrives. It is a counter-current heat exchanger, and it works [15]. Comparative anatomists have traced its evolution across artiodactyls in detail [16] [17]. In sheep it becomes stronger when the animal is dehydrated, which reads as a water-conservation strategy rather than a brain-protection one [18].

Notice what that mechanism is actually for. In sheep it gets stronger when the animal is short of water, which is a clue about its real job. A brain kept cool sends a weaker overheating signal to the rest of the body, so the animal sweats less and conserves water. That is a desert survival trick. It is not a system for protecting thought, and the two get run together constantly in popular writing.

Humans do not have a carotid rete. A 2023 review states it plainly: the rete mirabile does not exist in humans [19].

So the argument in humans has always been about whether something else does the job. Michel Cabanac spent years arguing that it does. In 1993 he published a paper with the memorable title asking whether selective brain cooling in humans was fancy or fact, and came down on the side of fact [20]. His case rested on heat loss through the upper airways and on emissary veins carrying cooled blood from the face and scalp inward. He and colleagues documented nasal mucosal vasodilatation during passive heating, which would increase that heat exchange [21] [22].

The counter-argument came from two directions at once, and it was brutal.

The first line of attack was on the measurements. Shane Maloney, Andrea Fuller, Duncan Mitchell and colleagues showed in 2001 that rectal temperature measurement itself produces artifactual evidence of selective brain cooling [23]. The rectum responds slowly. Compare a slow-responding reference against a faster-responding brain probe and you generate an apparent brain-cooler-than-body signal out of nothing but measurement lag. Eckhart Simon made a parallel argument in 2007 about tympanic temperature, concluding that ear measurements are simply not suited to indicating selective brain cooling in humans [24].

Much of the human evidence had been built on those two measures.

Stop on that for a second, because it is a bleak kind of finding. It does not say the effect is small. It says the instrument that produced the effect was capable of producing it out of nothing. Two decades of human measurements were suddenly evidence about probe response times rather than about brains. That is the sort of correction that quietly deletes a literature.

The second line of attack was on the animals. The same group went looking for selective brain cooling in free-ranging animals under real heat and could not find it. Not in zebras in their natural habitat [25]. Not in unrestrained baboons exposed to heat [26]. Liang Zhu approached it from theory instead and calculated how much heat conduction and counter-current exchange could plausibly contribute in a human head, and the answer was: not much [27].

Those two absences are worth taking seriously. Zebras and baboons are large mammals living in genuinely hot places with every reason to protect their brains. If a selective cooling mechanism existed anywhere it should be visible there, in the field, under real heat. It was not.

And then there is the most awkward observation of all. When a human brain gets hot from the inside, during fever, a 1998 study looked for a specific protective response and found none [28].

The title says it: no specific brain protection against thermal stress in fever.

None of this means the human head has no cooling. It obviously does. Your scalp sweats, your face flushes, blood carries heat away, and you lose heat with every breath. Clinical researchers are actively working on ways to cool a brain selectively for medical reasons, precisely because the body does not do it well enough on its own [29].

What is contested is whether those ordinary routes add up to a preferential system that protects the brain ahead of the body. Cabanac said yes. Simon, Maloney, Fuller, Mitchell and Zhu said the evidence for it is an artifact of how it was measured.

Where this leaves us. Do not believe anyone who tells you your brain has a special cooling system that keeps it safe. Do not believe anyone who tells you it has nothing either. The honest position is that humans cool their heads with the same general machinery they cool everything else, and whether that amounts to something selective has been argued about for three decades without resolution.

The Brain Holds Its Temperature Until It Cannot

So the brain is warm, variable, and cooled by ordinary machinery with no special protection. The obvious next question is what happens to all of that when the air gets hot. The answer is stranger than a simple rise.

If you want to understand why heat catches people out, this is the study to know.

In 2024, a team led by Xiang Ren Tan put 11 people through magnetic resonance imaging while driving their rectal temperature up to 39.5 degrees, either by exercise or by immersing them in warm water [30]. They measured brain temperature, cerebral perfusion and task-related brain activity throughout.

At rest, motor cortex temperature sat at 37.3 degrees and stayed there. Rectal temperature moved underneath it. Brain temperature did not follow.

That is the finding. At rest, the brain regulates its own temperature tightly and does not simply track the body. Then, as genuine hyperthermia developed, brain temperature climbed and dovetailed with rising core temperature. The independence broke.

The rest of the results describe what breaks with it. Bilateral motor cortical activity was suppressed during high-intensity tasks once core temperature reached 38.5 degrees, which the authors read as reduced central motor drive. Grey matter perfusion fell, and so did regional perfusion in sensorimotor cortex.

Put the two halves together and you have the shape of the whole problem. The system defends its set point, and defends it, and gives you no progressive signal that it is working harder. Then the defence is overwhelmed and several things degrade at once.

A gradual decline you can feel coming. This is not that.

This is also why the everyday version of the question is harder than it looks. You are not trying to notice a temperature. You are trying to notice the moment a regulatory system that was succeeding stops succeeding, from the inside, using the organ that is being affected. If you want a sense of how badly-placed you are to judge your own cognitive state, the same problem shows up in how people misjudge their own best study times across the day.

What Actually Breaks, and In What Order

Knowing that the defence eventually fails is not the same as knowing what fails with it.

"Heat is bad for your brain" is true and nearly useless. The interesting result is that heat does not degrade thinking uniformly. It takes specific systems and leaves others alone, and knowing which is which is what makes the finding actionable.

In 2010, Nadia Gaoua and colleagues heated 16 people and tested them [31]. Three attention tests: choice reaction time, match-to-sample visual search, rapid visual processing. Two memory tests: pattern recognition memory and spatial span. The hot condition brought core temperature to 38.6 degrees, against 37.1 in the control.

Working memory was impaired. Attentional processing was not altered at all.

That dissociation is the most useful single fact in this literature. It maps directly onto the subjective experience that people describe on hot afternoons, which is not "I cannot concentrate" so much as "I keep losing the thread". You can still track what is in front of you. What you lose is the ability to hold several things at once and manipulate them, which is what the relationship between attention and memory is built on.

An earlier study from the same group with the same sample size found that two hours at 50 degrees impaired short-term memory and also degraded the transmission of motor drive at the spinal and neuromuscular level [32]. Heat is not only doing something to your thoughts. It is doing something to the wiring underneath them.

The picture got more precise in 2018. Gaoua's group ran a smaller study, 10 participants, using EEG to watch frontal theta power while people did an easy version and a hard version of a planning task [33].

Core temperature in the hot condition reached 39.5 degrees. Theta power rose with heat and rose with task difficulty, which is what you would expect if heat were acting as an extra cognitive load. Then on the hardest version the heat effect disappeared, at the same moment performance fell apart. The authors read that as saturation: the system had run out of resources to allocate, so the marker of effort stopped rising because effort could no longer be increased.

Two things about that table deserve saying out loud. First, these are small studies. Ten to sixteen people is normal for this kind of work, because heating a person to 39.5 degrees under medical supervision is expensive and unpleasant, but it means every individual number carries real uncertainty. Second, the last two rows disagree with the rest, and this article comes back to them shortly.

The broader reviews have been saying for two decades that complexity is the variable that matters. Hancock and Vasmatzidis summarised the state of knowledge in 2003 and put task complexity at the centre of it [34]. A later focused review across environmental stressors reached the same conclusion [35].

Simple, well-practised, automatic tasks hold up. Anything requiring you to keep several items in play and reason over them does not.

The Mechanism: Breathing, Carbon Dioxide, and Blood

Everything so far has described what breaks. This section is about how, and the chain has a middle step that most accounts skip. It is your breathing, and it is the part that makes the rest make sense.

Start with what heat does to circulation. To lose heat, you send blood to the skin. That blood has to come from somewhere, and the redistribution is systemic [36].

Meanwhile, getting hot makes you breathe faster. Hyperventilation blows off carbon dioxide, and arterial carbon dioxide is the single strongest chemical regulator of how wide your cerebral blood vessels sit. Low carbon dioxide constricts them.

So the brain gets less blood, for two reasons at once, at the exact moment its own metabolism is running warm.

The measurements bear this out. Lars Nybo and colleagues found middle cerebral artery blood velocity reduced with hyperthermia during prolonged exercise [37], and followed it with direct work on cerebral blood flow and metabolism under the same conditions [38]. Fujii and colleagues went after the carbon dioxide link specifically, using chemoreflex manipulations in resting heated humans to show how hyperthermic hyperventilation drives the change in cerebral blood velocity [39]. Cerebral oxygenation falls along with the flow [40].

There is a parallel effect on the motor side that the sports science literature has chased for years. Nybo and Nielsen described central fatigue under hyperthermia in 2001: the reduction in performance was happening in the nervous system rather than in the muscle [41]. Nybo later reviewed brain temperature and exercise performance directly [42]. More recent work has separated the thermal cost from the mental one, finding that elevated core temperature and mental fatigue impair exercise capacity through partly independent routes in highly trained athletes [43], and has begun tracking cognitive neuroelectrical activity during the preparation phase before exercise under hyperthermia rather than during it [44]. The general physiology of all of this is set out at length in two modern reviews that are worth knowing about if you want the full account [45] [46].

Now the honest part. Everything above is a correlation with a plausible causal story attached. Blood flow falls, cognition falls, and the temptation is to say the first caused the second. Proving that requires breaking the two apart, and researchers are doing exactly that right now.

One approach is to hold carbon dioxide constant while raising temperature, which separates the thermal effect from the ventilation effect. A 2026 study did this with moderate isocapnic hyperthermia and looked at dynamic cerebral autoregulation [47]. Another approach is to raise cerebral blood flow pharmacologically during combined stress and ask whether cognition comes back with it [48]. A third separates hyperthermia from hypoxia using EEG [49].

Those experiments are in progress. The reduction in cerebral blood flow is real and not in dispute. Whether it is the cause of the cognitive change, or a companion to it, is still open. Anyone who tells you the mechanism is fully worked out is ahead of the evidence.

The Experiment That Should Change What You Do

Everything so far describes a problem. This section is about the obvious solution, and why it does not work. If you remember one experiment from this article, make it this one.

If your brain is too warm and your head is the closest part of you to it, cooling your head should help. Cold cloth on the forehead, cold water on the face, a fan aimed at your head. It is what everybody reaches for and it feels like it works.

In 2017, Manabu Shibasaki and colleagues tested it properly [50]. Fifteen men, average age 21, in water-perfused suits. They ran a Go/No-go task while EEG recorded event-related potentials, which separate two things that behaviour alone confuses: the act of executing a response, and the act of withholding one. The Go-P300 component tracks execution. The No-go-P300 tracks inhibition.

Passive heat raised oesophageal temperature by 1.30 degrees. Cerebral perfusion fell, thermal comfort fell, and both P300 components shrank. So far, exactly as predicted.

Then they cooled the subjects' faces and heads.

Cerebral perfusion recovered. Thermal comfort recovered. The subjects felt fine. And the P300 amplitudes stayed suppressed. Both of them.

Then they cooled the whole body. The Go component returned to baseline. The No-go component did not.

Read that sequence again, because there are two separate results in it and both matter. Local head cooling restored the things you can feel and measure easily, comfort and blood flow, without restoring the cognitive markers at all. And even full-body cooling back to normal temperature left response inhibition impaired.

Feeling cool again is not thinking clearly again. And of the two functions, the one that recovers last is the ability to stop yourself doing something.

That last point deserves a moment. Inhibitory control is one of the jobs handled by the prefrontal cortex and the executive systems built on it, and it is what keeps you from sending the message, taking the corner too fast, snapping at someone, or committing to the first answer that comes to mind. If it is the slowest function to come back after heat, then the window in which you are most likely to make a bad decision extends past the window in which you feel hot.

Other work on head and neck cooling has found real physiological and cognitive effects, so this is not a simple story of local cooling being useless [51]. The clinical literature has invested heavily in cooling brains from the outside and through the nose, and the honest summary of that work is that shifting the temperature of an adult human brain by external means is difficult [52] [53].

A skull is good insulation. That is its job.

What Shibasaki's result rules out is the assumption that comfort is a proxy for capability. It is not.

The Studies That Disagree

It is also worth being clear about how strong the evidence in the previous section really is. Fifteen men in one laboratory is a finding, not a law.

An article that only reported the results supporting its argument would not be worth reading. Two recent studies point the other way and they belong in the body, not in a footnote.

In 2026, Schilder and colleagues put military personnel through virtual simulations under passive heat stress [54]. Decision making was affected. Situational awareness and executive functioning were not.

The same year, Coehoorn and colleagues tested firefighters after rapid heat stress exposure and found cognitive performance preserved [55].

Neither result is a fluke and neither should be waved away. There are several plausible reasons they differ from the rest of the literature, and the honest thing is to present them as candidate explanations rather than conclusions.

The first is duration. Rapid exposure is not the same stimulus as an hour at 50 degrees, and the studies that find the largest effects generally heat people for longer.

The second is training. Firefighters and soldiers are habitually heat-exposed populations, and heat acclimation changes the physiological response substantially [56].

An acclimated person reaching the same air temperature is not in the same internal state.

The third is the task, which brings us back to Gaoua. If impairment scales with complexity, then two studies using different tasks can honestly reach different conclusions about the same temperature. Situational awareness in a familiar simulation is a well-practised skill for a soldier. Holding an unfamiliar sequence in working memory is not.

A 2026 study looking across three consecutive days of simulated occupational heat stress describes itself as exploratory, which is the right label for work in this area at the moment [57]. Other recent work has examined heat acclimation against simple and complex cognitive tasks specifically [58], cognition during self-paced cycling in the heat [59], and affective processes and risk-taking under heat stress [60].

One finding that sits slightly apart from the temperature debate but matters practically: direct solar radiation falling on the head impairs motor-cognitive performance beyond what ambient temperature alone predicts [61]. Sun on your head is its own variable. A hat is not a superstition.

Ten Million Students

There is a question hanging over everything so far, and it is a fair one. Does any of this show up outside a laboratory where people are deliberately heated to 39 degrees?

Laboratory studies with sixteen people establish mechanism. They cannot tell you whether any of it matters at the scale of a life. For that you need a very large number of people and something real being measured about them.

In 2020, Park, Goodman, Hurwitz and Smith published exactly that in the American Economic Journal [62]. They took roughly 10 million students who had retaken the PSAT, which gives you the same student measured twice, and matched their scores against the temperatures of the school year in between.

Without air conditioning, a school year one degree Fahrenheit hotter reduced that year's learning by about one percent.

One percent of a year sounds small until you notice it accumulates, and until you notice what it correlates with. Hot school days disproportionately affected minority students, and the authors calculate that this accounts for roughly five percent of the racial achievement gap in the United States. The pattern in the data was consistent with air conditioning infrastructure largely offsetting the effect, which reframes the whole finding: this is a building problem before it is a biology problem.

Weekend and summer temperatures did far less, which is the detail that makes the causal story credible. Heat is interfering with the hours when learning happens.

The Boston dormitory study from the opening of this article is the small, precise companion to that enormous one [1]. Forty-four students, twelve days, an indoor mean of 26.3 degrees against 21.4 in the air-conditioned buildings, and reaction times 13 percent slower. Same conclusion, from an entirely different direction, in healthy young adults at temperatures nobody would call extreme.

Other work has found associations between extreme heat exposure and lower learning, general cognitive ability and memory in United States children [63], and between high temperature and cognitive function in a large Chinese cohort [64]. That second one is cross-sectional, so it shows association and cannot establish direction. Systematic reviews have looked at ambient temperature and children's health more broadly [65].

The occupational literature runs in parallel and reaches the same place from the workplace side. Tord Kjellstrom and colleagues have argued for years that heat is a first-order issue for human performance and occupational health as the climate warms [66] [67]. Recent field work has measured thermophysiology and cognitive performance in live-line electrical workers in high heat and humidity [68], which is a setting where a lapse in inhibitory control has consequences.

At the far end of the scale sits the physical limit. Sherwood and Huber calculated in 2010 that there is a wet-bulb temperature above which a human cannot shed metabolic heat at all, regardless of behaviour or acclimation [69].

That is a different subject from thinking clearly. It is worth knowing it exists.

And then there is what heat does at population scale during a crisis. Excess mortality during the 2003 French heat wave was documented in detail [70]. Mental health presentations rise during heat waves in temperate cities [71]. Vulnerability tracks neighbourhood microclimate [72], and the public health literature was flagging extreme heat events as a growing problem well before the current decade [73] [74].

The Night Shift

One more pathway deserves its own section, because it explains why a heat wave feels cumulative rather than daily. This second route has nothing to do with what happens while you are hot.

Sleep is thermally regulated. Falling asleep requires your core temperature to drop, and that drop is driven by dumping heat through your skin. A room that is too warm interferes with that process directly, and the effects of thermal environment on sleep and circadian rhythm are well described [75]. A systematic review of heat-exposed workers found consistent damage to sleep quality [76], and researchers working in low-resource urban settings have started mapping the pathways linking heat, sleep and mental health together [77].

Then the next day arrives and you are worse at everything, for reasons that have nothing to do with the temperature of the room you are now sitting in. If you want the detail on how much of your learning depends on the night after it, the case is set out in what the brain is doing while you sleep.

There is a neat symmetry here that comes from animal work, and it should be read as animal work rather than as a human finding. Studies in cats showed brain temperature falling across the wake-sleep cycle and rising again during REM sleep, driven by systemic haemodynamic changes [78] [79].

Sleep is when a brain runs coolest. A hot night takes that away.

So the cost of a heat wave is not one hit. It is a daytime impairment stacked on a night of degraded sleep, repeated, and the second effect follows you into rooms where the first one has stopped.

Who Feels It First

Everything so far has treated the reader as a generic adult. That is a simplification, and for some people it is the wrong one.

Heat is not distributed evenly across people, and neither is the ability to handle it.

Ageing changes thermoregulation directly. Older adults sweat less at a given thermal load, redistribute blood less effectively, and reach a higher core temperature for the same exposure [80]. The cardiovascular consequences during heat waves have been reviewed in detail [81], and a recent systematic review pulled together the evidence on heat tolerance and thermoregulatory vulnerability in older adults specifically [82].

Recall from the Rzechorzek data that deep brain temperature rises about 0.6 degrees across twenty years of age [2]. An older brain starts from a warmer baseline and has less capacity to shed additional heat. Those two facts compound.

Fitness and habitual activity move things in the other direction. Recent work has looked at how age, cardiorespiratory fitness and regular physical activity affect physiological strain and cognitive performance across a six-hour extreme heat exposure [83], and at whether habitual physical activity protects cognition after passive heat exposure [84].

Hydration is its own axis and it is easy to get wrong in both directions. Dehydration impairs aerobic performance through mechanisms that are partly separable from heat itself [85], and fluid ingestion after heat or exercise-induced dehydration measurably affects cognitive function [86]. Acid-base state turns out to matter too, which is the kind of detail that shows how many variables are moving at once in these studies [87].

One clarification worth making, because the two get confused constantly. Fever is not the same thing as environmental heat. Fever is a regulated change: the body raises its own set point deliberately as part of an immune response, and the mechanisms are specific and well described [88].

Environmental hyperthermia is a failure to hold the set point you already had. They feel different, they are driven by different machinery, and evidence about one does not transfer to the other.

Brain temperature has also started showing up as a clinical variable in its own right. It rises with focal oedema in temporal lobe epilepsy [89], and increased along with free water after mild COVID-19 infection [90]. Relationships between brain and body temperature after ischaemic stroke have been examined at length [91], and the brain's thermal response has been proposed as a neuroimaging biomarker of cerebrovascular impairment [92]. Fifteen years ago none of that was measurable.

What Actually Helps

Which leaves the practical question. Given all of that, what does the evidence actually support? This is a science article and not an advice column, so what follows describes the state of the evidence rather than handing you a set of instructions.

The intervention with the strongest evidence behind it is also the least interesting: change the air temperature. The PSAT study's own conclusion was that air conditioning infrastructure largely offset the effect [62]. The dormitory study compared two groups whose only meaningful difference was a building system [1].

Neither of those is a physiological trick. They are engineering.

There is something slightly deflating about that. A subject this full of magnetic resonance spectroscopy and evoked potentials arrives at a recommendation about ventilation and thermostats. But it is the honest ranking of the evidence, and the size of the effect is the reason. A building system changed outcomes across ten million students. No physiological intervention in this literature has been tested at anything close to that scale.

Acclimation genuinely works, and it works on the physiology rather than on the perception. Repeated heat exposure improves plasma volume, sweating response and cardiovascular stability [56] [93], an effect documented in humans as far back as the early 1990s [94]. Whether it protects cognition specifically is a newer and less settled question [58].

That gap matters more than it sounds. Acclimation makes you handle heat better in every way that can be measured from the outside. Your heart rate settles. You sweat sooner and more efficiently. What nobody has established is whether the part of you that holds a plan in mind acclimates on the same schedule as the part of you that sweats. Those could easily come apart, and the two 2026 studies of trained groups are the first real hint that they might not.

Combined cooling interventions during activity in the heat have been tested and show effects on cognition [95], and cooling garments have been trialled in real occupational settings, including nurses working in full protective equipment [96]. Those results are promising, and they are about maintaining performance in people who cannot leave the hot environment.

Read that framing carefully, because it is a different goal from the one most readers have. A cooling vest is designed for a firefighter who cannot leave the building. It is not designed to make an ordinary warm afternoon feel like a cool one, and the evidence for it does not transfer to that use.

And then there is the thing that does not work as advertised. Cooling your face and head restores comfort and cerebral perfusion without restoring the cognitive markers [50].

That is worth restating because it is the most counterintuitive result in this whole article. The remedy that feels most effective is the one that has been most clearly shown to fix the feeling rather than the function.

The practical implication is about scheduling rather than technique. If complex work degrades and simple work holds, and if the recovery of inhibitory control lags behind the recovery of comfort, then the thing to move is the task rather than the person. Hard thinking wants the cool part of the day. This is the same logic behind matching demanding study to the hours when your body is set up for it, and it interacts with everything else that determines your state, including how stress hormones shape what you retain and what caffeine does and does not do for learning.

One thing to be careful about. Heat stroke is a medical emergency and it is not what this article is about. If someone is confused, disoriented or collapses in the heat, that needs emergency care immediately, not a discussion about working memory.

What Is Still Unknown

The gaps in this field are large, and pretending otherwise would be dishonest.

Whether humans have any selective brain cooling mechanism at all remains argued about by researchers who have been arguing about it for thirty years [20] [24]. The measurement critiques from Maloney, Fuller and Simon were serious enough that a substantial part of the older human evidence base has to be treated as unreliable.

Working out how much of the older evidence survives that critique is not a tidy job. Some of it was never dependent on the disputed measures. Some of it clearly was. Nobody has gone back and sorted the pile paper by paper, which means the field carries an unknown amount of unsound material forward by default.

Whether reduced cerebral blood flow causes the cognitive impairment, or merely accompanies it, is being tested right now [47] [48]. Those results will matter, because the answer determines whether interventions should target blood flow or target temperature.

It is unusual to be able to say that about a physiology question. Usually the open questions are about detail. Here the open question is directional: two plausible mechanisms, opposite implications for what you would do about it, and an answer expected within a few years.

Almost every threshold quoted in this literature comes from a study of ten to sixteen people, and several of those samples were all male, including Shibasaki's fifteen [50]. Given that the Rzechorzek data found brain temperature varying with menstrual cycle phase [2], the shortage of women in the heat-and-cognition literature is a real gap and not a technicality.

The thresholds themselves need one more caution, and it is the most important sentence in this section. Nearly every number you will see quoted about cognitive decline starting at 38 or 38.5 degrees refers to core body temperature measured in a laboratory, not to brain temperature. Only a handful of studies have measured a living human brain directly while testing cognition. Treating a core temperature threshold as a brain temperature threshold is the single easiest way to get this subject wrong, and it happens constantly.

Finally, almost none of the everyday range has been studied properly. The laboratory work heats people to 38.5 or 39.5 degrees core, which requires deliberate effort. The temperatures most people actually live in are far below that. A 2026 pilot study using smartwatches and momentary sampling found that thermal discomfort at ordinary, unremarkable temperatures was enough to worsen performance on an N-back task [97].

It is a pilot, with everything that implies. It also points at the question that matters most to the largest number of people, and it is nearly unstudied.

The clearest way to picture the whole system is not as a thermostat with a wide safety margin. It is a small furnace, well insulated, kept steady by the same blood supply that heat is busy diverting elsewhere, running a daily rhythm you cannot feel, in an organ that has no dedicated way to cool itself and no reliable way to tell you it is struggling. It works remarkably well. That is precisely why the failure is quiet.