Eighty in the throat, none on the tongue

Take a spoonful of a fresh, green, badly behaved olive oil and swallow it. For a second or two nothing happens. Then something catches at the back of your throat, builds, and makes you cough. Your lips felt nothing. Your tongue felt nothing. The sensation arrived in one small place, late, and then would not leave for a couple of minutes.

That is not how oral irritation normally works. Chilli goes everywhere. So does mustard, so does cinnamon bark, so does neat spirit: they light up the lips, the tongue, the palate and the throat more or less together. Olive oil skips almost all of it and goes straight for the pharynx. The explanation turned out to be worth a decade of work, and the answer is a count.

Tasting the fractions

The first person to isolate the sensation did it the slow way. In 2003, two years before olive oil became a headline, a group at Unilever pulled the polyphenols out of virgin olive oils, separated them by liquid chromatography, collected the fractions one at a time, and then tasted them. ¹

Almost every fraction was bitter. In their words, "almost all fractions containing polyphenols were described as bitter and astringent". One was different: it "produced a strong burning pungent sensation at the back of the throat". ¹ Bitterness, in other words, is a committee. The throat sting is a single molecule.

The molecule was deacetoxy-ligstroside aglycon. Its near twin, deacetoxy-oleuropein aglycon, differs by one hydroxyl group, and at the same concentration it gave "only a slight burning/numbing sensation, which was perceived more on the tongue". ¹ One hydroxyl, and the sensation moves house.

Olive, sting, aldehyde

Two years later the same molecule arrived in Nature under a different name and for a different reason. Gary Beauchamp of the Monell Chemical Senses Center had tasted a particularly pungent extra virgin oil in Sicily and noticed that the burn in his throat felt like the one he knew from ibuprofen. ³ The paper that followed reported that the resemblance was not a coincidence: the two compounds are "structurally dissimilar" and yet "both these molecules inhibit the same cyclooxygenase enzymes in the prostaglandin-biosynthesis pathway". ²

They named it oleocanthal, and the name is a description rather than a label. Oleo for olive, canth for the sting, and al for aldehyde, which does double duty because the molecule carries two aldehyde groups. ^{5 6} It is a modest thing, C17H20O5, weighing 304 daltons. ⁴

I should be careful with the origin story, because a detail circulates that I could not stand up. Many accounts place the tasting at a specific meeting in the hill town of Erice and give a year. The only source I could open puts it in Sicily and gives no year at all, so Sicily is where this note leaves it. ³

Eighty and zero

The answer to the throat question came in 2011, in a paper whose title is its own abstract: unusual pungency from extra virgin olive oil is attributable to restricted spatial expression of the receptor of oleocanthal. ⁷

First they found the receptor. Oleocanthal selectively activates hTRPA1, the channel that mustard and cinnamaldehyde also work on, with a half-maximal concentration of 2.8 micromolar. Then they did the experiment that turns correlation into cause: sensory neurons taken from mice bred without TRPA1, 187 of them from five animals, are simply not excited by oleocanthal. Those same neurons still responded to capsaicin perfectly well, so nothing was broken in general. ⁷

Then they counted. Staining human tissue for TRPA1, they found roughly 80 positive cells per 0.12 square millimetres in the pharynx. In lingual tissue, at the taste buds, they found 0 per 0.05 square millimetres. ⁷

Eighty in the throat. None on the tongue. There is no mystery left after that: your tongue cannot feel oleocanthal because the receptor for it is not there. The sensation is restricted because the hardware is restricted.

The obvious objection is that oil coats the throat and that the pattern is an artefact of the medium. They tested it. Oleocanthal delivered in water and in corn oil at the same concentration produced the same throat-first pattern, "independently of the medium". ⁷ And in the neatest of the human experiments, they matched olive oil against horseradish so that the two stung the throat with equal intensity, then asked about the front of the mouth: the oil "elicited very little pungency on the anterior tongue, whereas horseradish irritation was strongly sensed on anterior tongue". ⁷ Same burn in the throat, and only the tongue could tell them apart.

The psychophysics had already pointed there. In 2009 a group measured irritation in three places and found it greater in the oropharynx than on the anterior tongue or in the anterior mouth, at p below 0.001, peaking around 15 seconds after exposure and lasting past 180. ⁸ That delay is the reason the cough always seems to arrive after you have swallowed and decided the oil was fine.

The same study turned up something I liked: how strongly a person feels oleocanthal has no relationship to how strongly they feel the prickle of carbonated water. The correlation was essentially nothing. ⁸ Two irritations, two populations, and the friend who cannot drink fizzy water is not necessarily the friend who chokes on your oil.

One loose end, which the paper is honest about. Most electrophilic TRPA1 agonists work by covalently modifying cysteine residues on the channel. Oleocanthal has the reactive chemistry to do that and does not do it, and nobody in that paper says what it does instead. ⁷ Fifteen years on, the best-explained sensation in olive oil still has an unexplained mechanism inside it.

The cough is in the standard

Here is the part I did not expect. What may legally be sold in the European Union as extra virgin is decided, in the end, by people tasting it. Implementing Regulation 2022/2105 says so plainly: the organoleptic characteristics "shall be deemed consistent with the declared category if a tasting panel approved by a Member State confirms the category". ¹¹ The method requires between 8 and 12 tasters. ¹⁰

Open that method and section 4.1 is headed Negative attributes. Section 4.3 is headed Positive attributes, and it contains exactly three entries: fruity, bitter, and pungent. Pungent is defined as a "Biting tactile sensation characteristic of oils produced at the start of the crop year, primarily from olives that are still unripe. It can be perceived throughout the whole of the mouth cavity, particularly in the throat." ¹⁰

That is a standards committee describing oleocanthal, filed under the things that count in an oil's favour. Note also that bitter is classified as a taste and pungent as a tactile sensation, which is chemically correct: the sting is trigeminal, not gustatory, and TRPA1 is not a taste receptor. ¹⁰

What actually demotes an oil is the other list, and it is a vocabulary worth reading once: fusty, musty, winey, rancid, frostbitten. There is one for oil pressed from fruit that the olive fly got to, grubby, and one for oil sealed in a tin too long, cucumber, which comes from 2,6-nonadienal forming inside the container. To be extra virgin, an oil needs a median defect score of exactly zero and a median fruity score above zero. ¹⁰ Nothing on the defect list is peppery. Nearly all of it is some version of stale.

One correction while we are in the legal texts, because a great deal of food writing has not caught up: Regulation 2568/91, the one everybody cites for the acidity limits, was repealed on 23 November 2022. The current instrument is Delegated Regulation 2022/2104, and extra virgin is the category at or below 0.80 grams of free acidity per 100 grams. ¹²

Seventy per cent of what

Which recasts the most repeated statistic in this whole subject. In 2010 the UC Davis Olive Center reported that "69 percent of imported olive oil samples and 10 percent of California olive oil samples labeled as extra virgin olive oil failed to meet the IOC/USDA sensory (organoleptic) standards". ¹⁴ That number has been travelling ever since as proof that most olive oil is fake.

Read the next sentence of the report and the claim falls apart. The failing samples "scored a median of up to 3.5 sensory defects such as rancid, fusty, and musty" and were "classified at the lower grade of 'virgin'". ¹⁴ They were not fake. They were real olive oil that had gone tired, downgraded to a legitimate category that people happily buy.

The study is also smaller and more interested than the retellings suggest. It tested 14 imported brands and five Californian ones, bought in California in 2010. It is marked DRAFT and prints its own warning that the results "should not be used to characterize the quality or authenticity of olive oil currently available in California or elsewhere". And its acknowledgements thank Corto Olive, California Olive Ranch and the California Olive Oil Council "for their financial support of this research". ¹⁴ A finding that imported oil fails and Californian oil passes was paid for by Californian producers, which does not make it wrong and does mean the sentence should never travel alone.

For actual adulteration, the number is different and much duller. The US Food and Drug Administration tested 88 commercial extra virgin samples against the international purity criteria. Three failed. ¹⁵ Sensory quality and purity are separate questions, and the popular fraud story collapses them.

Real prosecuted fraud does exist, and it looks like neither story. In December 2023 Europol's Operation OPSON took 11 arrests in Spain and Italy and seized over 260,000 litres. The scheme was lampante oil, the lowest grade, unfit to eat as it stands, blended and sold as extra virgin. ¹⁶ The money is in misgrading olive oil, not in replacing it.

The one true catastrophe in this literature was not olive oil at all. In 1981 in Spain, "the ingestion of an oil fraudulently sold as olive oil caused an outbreak" of a disease nobody had seen before. Of the 20,000 people affected, "approximately 300 died shortly after the onset of the disease and a larger number developed chronic disease". Twenty years later a multidisciplinary review still could not name the agent: "Whether PAP esters are simply markers of toxicity of oils or have the capability to induce the disease remains to be elucidated." ¹⁷ Four decades on, the poisoning that made olive oil fraud famous has never been fully explained.

How much ibuprofen is in a salad

The pharmacology is real and it is routinely stretched past what it can carry. What the enzyme work established is that at equal concentrations oleocanthal inhibits more cyclooxygenase activity than ibuprofen does: one review by a co-author of the original paper puts it at 41 to 57 per cent against 13 to 18 per cent, both at 25 micromolar. ¹⁸ That is a statement about molecules in a test tube.

Between there and a person, several links are missing. I could not find a single human pharmacokinetic measurement of oleocanthal, and the reviews concede the problem in their own words. Nor is there a clinical trial of the compound. I went through every study on the public register that mentions it, all 16, and every one administers an olive oil, an extract or a supplement characterised by its oleocanthal content. ¹⁹ Nobody has ever given a human being purified oleocanthal for anything.

The only health claim the European Union permits here is narrow, and it is usually misquoted, so it is worth the exact words: "Olive oil polyphenols contribute to the protection of blood lipids from oxidative stress", allowed only for oil containing "at least 5 mg of hydroxytyrosol and its derivatives" per 20 grams, with the consumer told that the effect comes from 20 grams a day. ¹³ That is roughly a tablespoon and a half. There is nothing authorised about inflammation, pain, cancer or memory, and oleocanthal is not named in the claim at all.

The famous trial gets pressed into service here too, and it should not be. PREDIMED was retracted in 2018 after protocol deviations including enrolment of household members without randomisation, then republished with revised estimates. In the republished paper, 7,447 people followed for a median of 4.8 years, the extra virgin olive oil arm had a hazard ratio of 0.69 against controls, which in plain numbers is 3.8 per cent of that arm reaching an endpoint against 4.4 per cent of the controls. ²⁰ It tested a whole dietary pattern against low-fat advice, in high-risk Spaniards aged 55 to 80. It is not evidence about a molecule.

One smaller thing surfaced that I have not seen flagged anywhere. Two separate reviews state that a daily 25 to 50 millilitres of olive oil provides no more than 0.9 milligrams of oleocanthal. ^{5 6} Their own concentration data contradict it. Fifty millilitres is about 45 grams, and at commonly measured concentrations that is several milligrams at least, often tens. To land at 0.9 you would need an oil sitting at the very bottom of the observed range. It reads like a decimal that slipped once and then got copied.

What is actually in the bottle

Because there is no such thing as a typical amount. A systematic review of the measurements found oleocanthal "ranging from as little as 0.2 mg/kg to as high as 498 mg/kg", and put Coratina at 78.2 mg/kg against Taggiasca at 8.3 in the same comparison. ²¹ Three orders of magnitude. The number is a fingerprint of variety, harvest date and freshness rather than a property of olive oil.

And it leaves quickly. Umbrian oils tracked over two years lost roughly 60 per cent of their oleocanthal and 75 per cent of their oleacein after just three months of storage, and the samples kept in the dark were depleted too. ²² Light makes it worse but time does most of the work. In 40 days of ordinary household use, oil in a clear plastic bottle lost 26 per cent of its total phenols against 11 per cent in a bag-in-box. ²³ The half-empty PET bottle next to the hob is the worst case anyone has measured.

Heat is stranger than the folklore. Heated to 240 °C for 90 minutes, oleocanthal proved "heat stable compared with other olive oil phenolics, with a maximum loss of 16%" by chromatography, while its measured biological activity fell by up to 31 per cent. ²⁴ The molecule survives the pan better than the sting does, and the authors read the gap as heating producing something that gets in the way. Frying actual food is the destructive case: ten cycles at 180 °C stripped over 90 per cent of the phenolics when the food was not breaded. ²⁵

Refined olive oil has none of it. In one trial the refined comparator "was confirmed to be devoid of phenolic compounds", against 255 mg/kg in a standard extra virgin and 1,249 in a high-phenolic one. ²⁶ That is the cleanest fact in the whole subject: light olive oil does not sting because the pungency and the pharmacology are the same molecule, and they leave together.

Older than the farming

The oldest evidence of anyone pressing this oil sits under about four metres of water off the Carmel coast. At Kfar Samir, excavators found "crushed olive pits and pulp and few whole pits" packed into "an unlined round pit, dug directly into the clay paleosol, which had unworked pebbles paving at its base", with "several large stone basins found near the pit" that were probably the crushers. The assemblage "was radiocarbon dated to ca. 7,000 BP". ²⁷

The detail that upends the usual telling is in the same paper: "the olives from the Hishuley Carmel and Kfar Samir sites were collected from wild trees". ²⁷ People were making olive oil, at scale, in a purpose-dug pit, from fruit off wild trees, before anyone had domesticated the olive.

Where and when that domestication happened is genuinely unresolved. One camp reads the plastid diversity as pointing to the northern Levant, ²⁸ another to the southern Levant, ²⁹ and a third to several independent events around the Mediterranean. The disagreement is formal enough that two letters in the same journal are titled "Single vs multiple independent olive domestications: the jury is (still) out" and "The jury may be out, but it is important that it deliberates". ^{30 31} Anyone who tells you confidently where the olive was domesticated is ahead of the evidence.

The bitterness came first and it is the same chemistry. Raw olives are inedible because of oleuropein; curing works by taking it apart, hydrolysed by microbial β-glucosidase and then broken down further by esterase into non-bitter fragments including hydroxytyrosol. ³² Cato wrote the water-leaching version down in the second century BCE and it still works: "Bruise the olives before they become black and throw them into water. Change the water often." ³³ Four thousand years after Kfar Samir, and two thousand before the chromatography.

The tree has stayed where it started. The World Heritage property that covers the terraced valleys south of Jerusalem, with their dry stone walls and their grapevines and olives, is registered under the name "Palestine: Land of Olives and Vines". ³⁴

The kindness of egg yolk

There is a footnote to all of this that changes how you taste. In 2021 the Monell group found that food proteins, specifically from egg yolks and whey, "reduce or even eliminate both the throat pungency and bitterness" of pungent oils, in proportion to how much protein is present. ⁹

So the fierce oil that made you cough off the spoon goes quiet the moment it meets a mayonnaise or a yoghurt dressing. That is a kindness at the table and a nuisance in a tasting room, and it is the reason merchants make you drink the stuff neat, warmed in a covered glass, with nothing else in your mouth. It also means the bottle you bought for its ferocity and stopped noticing may not have gone off at all. You may just have started putting it on things.