A third of people taste something when you warm the tongue

A metal disc about the size of a fingernail sits against the tip of the tongue. Sixty four square millimetres of computer-controlled Peltier element, cooled to 15 degrees, then warmed back up at one degree a second. Nothing has been put in the mouth. No sugar, no salt, no acid, nothing dissolved in anything at all. A large minority of the people this is done to will tell you they taste something.

For most of the twentieth century that result was not supposed to be possible, and the reason nobody went looking is better than an oversight. Something like half the neurons in mammalian taste pathways respond to temperature, and since temperature had never been shown to produce a taste, the working assumption was that the gustatory system somehow cancelled the thermal signal on its way up.¹ The signal was known to be there. What nobody expected was that it arrives as flavour.

Twenty four people, one cold tongue at a time

In February 2000 Cruz and Barry Green published two pages in Nature reporting that heating or cooling small areas of the tongue can in fact cause sensations of taste.¹ Direction turned out to matter. Warming the anterior edge of the tongue from a cold temperature evoked sweetness; cooling the same region evoked sourness, saltiness, or both. The back of the tongue also produced thermal tastes, on a different nerve, with a different relationship to temperature.

There were 24 subjects. Twenty one of them reported at least one taste quality, and 19 reported two or more at one site or another along the front edge.² Hold them to a stricter standard, rating the sensation above weak on a scale and producing it again on a repeat trial, and about half the panel still qualified.

That word about is doing real work, and more studies have not made it easier. Published prevalence for thermal tasting runs from 21 to 50 per cent, with a further fifth to three fifths of participants left unclassified, and a 2026 systematic review of 69 papers concludes the spread is methodological rather than biological.² Four choices move the number: how wide the temperature window is, which taste words the panel is offered, how intense the sensation has to be to count, and whether it has to happen twice. Tighten those and the estimate drops to 20 per cent.³ The largest pooled sample assembled, 12 separate recruitment drives, found 254 thermal tasters among 781 people, or roughly a third.⁴

That pooled sample also put a number on the asymmetry Cruz and Green had described in words. Sweet thermal tasters were nine times more likely than the rest to get their sensation while the tongue was being warmed. Sour thermal tasters were eight times more likely to get theirs while it was being cooled, both at p below 0.0001.⁴ Twenty years and ten times the sample, and the shape of the finding did not move.

The tongue map is wrong, and the tongue is still not uniform

The diagram with sweet at the tip and bitter at the back has been dead since 1974, when Collings measured taste thresholds across the tongue and soft palate and found every quality detectable everywhere it could be detected at all, separated by modest differences in threshold and nothing like the tidy zones.⁵ The map survives on placemats because it is easy to draw.

The joke is that regional differences do exist. They are thermal ones, and they do not match the picture. The front edge of the tongue turns warming into sweetness; the back of the tongue, on a different cranial nerve, does something else entirely.¹ The real geography of the tongue was found by the experiment nobody thought to run.

A channel that opens when it gets warm

In December 2005 a group led by Talavera reported in Nature that TRPM5, an ion channel sitting downstream of the receptors for sweet, umami and bitter, is heat-activated, and steeply so: inward currents rise sharply between 15 and 35 degrees.⁶ That range is nothing exotic: it runs from a cold rinse up to body temperature, which is to say through everything a mouthful passes on its way in.

The clean part of that paper is the mouse. Warm the tongue from 15 to 35 degrees in a wild-type animal and the gustatory nerve response to sweet compounds increases markedly. Do it in a mouse with no TRPM5 and the enhancement is gone.⁶ The authors said plainly what they thought it explained: sweetness rising with temperature in humans, and thermal taste.

Then somebody blocked the door

The obvious objection is that the brain might simply be mislabelling a temperature it has no better word for. In 2020 Nachtigal and Green tested exactly that, and the design is the reason this note exists. They dropped the thermode and used water: sweetness could be evoked with nothing but deionised water heated from 20 to 35 degrees in the mouth.⁷ Static temperatures did much less. Heating from a cool start is what mattered.

Then they gave people 8 millimolar lactisole, an inverse agonist that shuts down the sweet receptor TAS1R2 and TAS1R3, and the thermal sweetness disappeared.⁷ So it is not a phantom the brain assembles somewhere upstairs. It is the sweet receptor itself, opened by warmth instead of by sugar, reporting the only thing it knows how to report.

The same paper is careful about what it did not find. Only a subset of participants got the sensation, it varied between trials and between sessions for the people who did, and heating dynamically did not make actual sucrose taste any sweeter than holding it at 35 degrees.⁷ The door opens. Not everything that walks through it is large.

Whose temperature, yours or the drink's

Here is the finding that changes how you think about a cup, and it is nearly forty years old. In 1987 Green and Frankmann cooled the tongue and the solution together from 36 degrees down to 28 and to 20, and measured what happened. Sweetness of sucrose fell. Bitterness of caffeine fell. Saltiness of sodium chloride and sourness of citric acid were untouched. And the critical variable was the temperature of the tongue rather than the temperature of the liquid.⁸

The following year they separated the two properly and found glucose and fructose behaving the same way, with cooling the tongue costing more sweetness than cooling the solution did. Aspartame split the difference. Saccharin did not care about temperature at all.⁹ From that gap between the sugars and saccharin they argued there had to be more than one mechanism carrying sweet taste, seventeen years before anyone tied TRPM5 to heat.

They then spent a paper explaining why. Mild cooling, 37 degrees down to 21, does not reduce the initial sweetness of sucrose at all. What it does is make the sweetness fade faster once it has arrived. Go properly cold, down to 5 or 12 degrees, and the intensity itself drops.¹⁰ Both effects vary by sweetener, which is a polite way of saying the tidy version of this story does not survive contact with the data.

One popular line collapses completely. Iced coffee is not more bitter for being cold. Cooling reduced the bitterness of caffeine in 1987,⁸ and in mouse chorda tympani recordings run from 23 to 43 degrees, bitter stimuli were the ones only weakly affected by temperature while sweet and umami tracked TRPM5.¹¹ If cold coffee tastes different to you, the honest suspects are dilution, what came out of the grounds, and the aroma you are no longer getting up the back of the nose.

The people who taste more of everything

Thermal tasters are not just people with a party trick. In 2004 Green and George found they rated ordinary chemical stimuli more intense than non-thermal-tasters, often by better than two to one, and not selectively: sucrose, saccharin, salt, citric acid, quinine, monosodium glutamate and PROP, across the front of the tongue, the back, the soft palate and the whole mouth.¹² They also rated vanillin stronger, particularly smelled retronasally, which puts the difference outside taste altogether.

The tempting explanation is that they have more equipment. They do not. Thermal taster status shows no relationship to fungiform papillae density in either study that has looked, it is independent of PROP status, and it is not associated with the TAS2R38 variants that drive PROP reactivity.²³ A purely peripheral account has very little left to stand on, and both the 2004 paper and the 2020 pooled analysis land on central processing instead.¹²⁴

It follows them into a glass. Across beer, cider and wine, thermal tasters have consistently rated bitterness, sourness, sweetness, astringency, carbonation and overall flavour intensity higher.² Whether any of that changes what they enjoy is a different question with a duller answer: differences in liking appeared only where ethanol was high, in dry red wine and spirits, and not in cider, white wine or beer. The review's own conclusion is that liking cannot be explained by thermal taster status alone.

How hot the cup actually is

All of this converges on a practical number that almost nobody gets right. Hot drinks are frequently served between 71 and 85 degrees. The preferred drinking temperature of coffee, measured across 300 people, is 60 degrees give or take 8.3. And when Brown and Diller ran the optimisation properly, weighing scald risk against the warmth people actually want, the answer came out at about 58.¹³

There is a line above that. In June 2016 an IARC working group of 23 scientists classified drinking very hot beverages as probably carcinogenic to humans, Group 2A, and defined very hot in a footnote as any beverage consumed above 65 degrees.¹⁴ The same evaluation moved coffee itself down to Group 3, not classifiable, from the possibly carcinogenic rating it had carried since 1991. As the agency's director put it, it is the temperature rather than the drinks themselves that appears to be responsible. IARC also says outright that the share of oesophageal cancers attributable to hot drinks is not known.¹⁴

Line the figures up and they are almost polite about it. People want about 60. The engineers say 58. The hazard begins at 65. And the cup you are handed is somewhere between 71 and 85, which is above the line by enough that waiting for it is not fussiness.

So the sage and black tea in this index asks you to let it stand, and the reason printed on the page is extraction. There is a second reason nobody writes down. Somewhere in the descent from scalding to drinkable, the cup crosses the range where TRPM5 does its steepest work, and if you are one of the third, your own tongue is quietly adding a sweetness the leaf never put there. You cannot taste which part is the tea.