Research note 012 — 2026-08-31
Cooking Does Not Erase Bitterness — It Edits It
「塩でもむ」 (in translation: "salt-rubbing"), 「茹でこぼす」 (in translation: "parboiling and discarding the water"), 「発酵させる」 (in translation: "fermenting"), 「深く焙煎する」 (in translation: "roasting deeply"), 「砂糖や乳を加える」 (in translation: "adding sugar or milk"). There are many cooking techniques that deal with bitterness. But calling everything that happens there「苦味が抜けた」 (in translation: "the bitterness has been removed") conflatesa decrease in the compound itself, a weaker feeling on the tongue, a different aroma coming to the front, and food becoming easier to eat.
What this page establishes
- One study confirms that simply soaking bitter-gourd flesh in salt (NaCl) does not change its bitterness, though it does significantly lower its sourness.
- Multiple studies confirm that adding thickening or inclusion agents such as gum arabic, CMC, or beta-cyclodextrin to bitter-gourd juice lowers its bitterness, but not every additive works the same way, and CMC's effect is reported to be far smaller than beta-cyclodextrin's.
- For tea, one orthogonal-design study confirms that raising the leaf-to-water ratio, lowering the water temperature, and shortening the steeping time lowers the combined bitterness-and-astringency attribute.
- The same test confirms that adding milk weakens the perceived bitterness of coffee, but the release of aroma compounds itself does not change greatly.
Five kinds of editing
- Move
- Dissolve and diffuse it out into water or brine.
- Change
- Change the makeup of compounds or flavor through heat, fermentation, or enzymatic reactions.
- Layer
- Add sweetness, milk, aroma, or viscosity, changing what stands in the foreground or background of perception.
- Create
- Generate new bitterness, burnt aroma, or hop aroma through roasting or boiling.
- Leave
- Stop at the endpoint a dish needs, without bringing bitterness down to zero.
1. Salt Alone Did Not Work
In a study that soaked pieces of bitter gourd in 3.5% salt water for 60 minutes, the juice's bitterness rating was nearly the same as untreated. Adding gum arabic to the same juice, on the other hand, lowered the bitterness rating. For the latter, the possibility that it changed viscosity or mouthfeel, not just removed the bitter compounds, cannot be ruled out. The ingredient names 「塩」 (in translation: "salt") and 「多糖」 (in translation: "polysaccharide") alone cannot explain the process.
Scope: Bitter gourd juice evaluated by a trained panel of 10 in 1 Malaysian study
Note: A null result contrary to the folk expectation that salt-rubbing reduces bitterness. The same soak significantly reduced sourness rather than bitterness. Single study, single fruit lot.
Show data
ingredient-bitter-gourd —has_perceived_bitterness_unchanged_by→ "Fruit pieces soaked in 3.5% (w/v) NaCl solution at 25 C for 60 minutes before juicing (bitterness 10.6 +/- 1.1 on a 15 cm line scale versus 10.7 +/- 1.7 untreated, p > 0.05)"Scope: Bitter gourd juice evaluated by a trained panel of 10 in 1 Malaysian study
Note: In the same study, bitterness showed a negative correlation with viscosity, so part of the effect may be masking by matrix/mouthfeel rather than removal of the bitter compounds.
Show data
ingredient-bitter-gourd —has_perceived_bitterness_suppressed_by→ "Addition of 5% gum arabic to juice (8.6 +/- 0.6 versus control 10.7 +/- 1.7 on a 15 cm line scale, p < 0.05; 7.3 +/- 1.4 at 15%)"2. Liking Shifts Even When Bitterness Does Not
In a comparison of turnips prepared 4 ways — boiled and puréed, steamed and puréed, roasted, and stir-fried — there was no significant difference in the trained panel's bitterness. Even so, consumer preference shifted. That is because the whole dish moves, including sweetness, umami, aroma, texture, and appearance. This is an example of why 「食べやすくなった」 (in translation: "it became easier to eat") should not be rephrased as 「苦味が減った」 (in translation: "the bitterness decreased").
Scope: A trained panel of 10 compared UK turnips
Note: All samples were perceived as bitter, but the main effect of cooking method on bitterness was p=0.38. Do not generalize this to 『茹でれば必ず苦味が減る』 (in translation: "boiling always reduces bitterness").
Show data
ingredient-turnip-root —has_perceived_bitterness_unchanged_by→ "4 cooking methods: boiled purée, steamed purée, roasted, and stir-fried"Scope: A liking comparison of 4 cooking methods by 74 UK adults
Note: Overall liking and taste liking differed by cooking method, but this is not attributed to bitterness intensity alone.
Show data
ingredient-turnip-root —has_liking_associated_with→ "a sensory profile — including sweetness, umami, aroma, and texture — that varies by cooking method"3. Boiling, and Entrusting Bitterness to the Water
Thailand's Senna siamea is recorded in an ethnobotanical survey as being heated before eating to reduce bitterness. Korea's ssumbagwi greens (씀바귀) are briefly boiled, soaked in water, and then stir-fried or dressed. Both processes are recorded, but the temperature, time, and the strength of bitterness left behind are not measured. What remains here is the cook's own words, still waiting to be asked.
Scope: Pa Tio District, Yasothon Province, Thailand
Note: An example of a cooking-culture practice of reducing bitterness by parboiling and discarding the water.
Show data
organism-senna-siamea —is_documented_as_having→ "Recorded as a wild edible plant under the local name ขี้เหล็ก (Khi Lek). It is heated before cooking to reduce bitterness."Scope: a public encyclopedia from the Academy of Korean Studies
Note: Time, temperature, and residual bitterness are not quantified.
Show data
ingredient-ssumbagwi-greens —is_prepared_by→ "briefly boiled, then soaked in water to remove bitterness before being stir-fried or seasoned"4. In Fermentation, Bitterness Moves, Splits, and Becomes a Different Taste
A study that naturally fermented Nyons black olives in brine identified diffusion from the flesh into the brine and endogenous enzymes as the candidates most involved in debittering this system. Microbial enzymes are also candidates from prior research, but this trial did not identify them as the contribution of a single microbe. In a separate comparison of 5 cultivars, however, even when the amount of bitterness-related phenolics differed, there was no significant difference in the final product's perceived bitterness. Analytical values matter, but they are not the bitterness itself, as tasted.
Scope: Natural fermentation of Nyons black olives in 10% brine for 15 months
Note: Chemical breakdown and microbial enzymes are candidates from prior research, but in this system, neither their contribution ratio nor causation by a single microbe was identified.
Show data
process-olive-debittering —has_debittering_most_likely_due_to→ "In this Nyons system, diffusion from the flesh into the brine and endogenous enzymes"Scope: The final product of naturally fermented olives across 5 cultivars
Note: The amount of bitterness-related phenolics differed among cultivars, but there was no significant difference in the final perceived bitterness rated by an 8-member panel. This does not mean chemical quantity in general is irrelevant.
Show data
compound-oleuropein —does_not_reliably_track→ sensation-bitterIn one 36-day fermentation time series for aged pu-erh tea, bitterness and astringency were low at the endpoint, and a rise in sweetness, a shift to a reddish-brown color, and a change in aroma that the source calls stale aroma were observed at the same time. Because this is not a factorial experiment with a control vat, it cannot conclude the effect of fermentation in general, or the cause of each individual change.
Scope: A 36-day fermentation time series of one sun-dried green maocha from old-tree tea
Note: A comparison between RM and F4. This is a time series from a single fermentation batch, not the causal effect or effect size of fermentation in general.
Show data
process-tea-pile-fermentation —was_observed_with→ "Lower bitterness and astringency than the raw material at the 36-day endpoint"Scope: The same 36-day fermentation, evaluated by a trained panel of 9
Note: Multiple attributes were observed at the same time. Do not convert the correlation of sugars, pigments, and microorganisms into the sole cause of each individual change.
Show data
process-tea-pile-fermentation —was_observed_with→ "A rise in sweetness, a shift to a reddish-brown color, and a change in aroma to what the source called "stale aroma""5. Tea Turns the Amount of Water, Temperature, and Time into Separate Levers
For Tieguanyin, raising the leaf-to-water ratio and lowering the water temperature or steeping time lowered the trained panel's composite attribute "bitterness-astringency". What was measured here is not bitterness alone. Aroma and extracted compounds move together with it. Diluting it is not simply the right answer — it is a choice of what to leave behind.
Scope: An orthogonal regression design and trained panel using Tieguanyin
Note: The evaluated attribute is the composite "bitterness-astringency." It does not claim to have varied bitterness and astringency separately.
Show data
ingredient-tieguanyin-infusion —has_bitter_astringency_affected_by→ "Tea-to-water ratio, water temperature, steeping time, and number of infusions"6. Roasting Does More Than Add Bitterness
In a planned roast of 100% chocolate, a decrease in bitterness corresponded to a rise in roast temperature, within a certain range. For coffee, bitterness and roast aroma were temporally dominant in medium roasts, while bitterness and burnt aroma were dominant in dark roasts. Behind the single axis called「焙煎度」 (in translation: "roast degree"), the order of taste and aroma is being reshuffled.
Scope: 27 roasting treatments across 3 origins, 145 chocolate consumers, gLMS
Note: The direction is the reverse of the intuition that 「焙煎温度が高いほど苦くなる」 (in translation: "the higher the roasting temperature, the more bitter it becomes"). No p-value is given for this effect size.
Show data
process-cocoa-roasting —reduces→ sensation-bitterScope: Comparison of Temporal Dominance of Sensations (TDS) between medium-roast and dark-roast Arabica
Note: For medium roast, bitterness and roasty aroma dominate; for dark roast, bitterness and burnt aroma dominate. TDS is not an intensity curve.
Show data
process-coffee-roasting —can_change→ "The temporal composition of bitterness, roasty aroma, and burnt aroma that dominate during drinking"7. What Milk Changed Was Perception, Before the Analytical Values
Adding milk or an oat drink to coffee at 30% lowered the dominance rate of bitterness on TDS across multiple time periods. But the coffee-derived volatile compounds released retronasally did not decrease uniformly. Milk-derived aroma and mouthfeel are added at the same time, but this study does not separate out the contribution of each factor. At the very least, a uniform decrease in volatile release alone could not explain the change in perception.
Scope: Temporal Dominance of Sensations (TDS) with 30% milk or oat drink added to medium- and dark-roast coffee
Note: The dominance rate of bitterness dropped in several time windows, including the early period. This does not assert intensity reduction, removal of taste substances, or a mechanism attributable solely to fat or protein.
Show data
process-milk-addition —can_reduce_temporal_dominance_of→ sensation-bitterScope: Simultaneous TDS and PTR-ToF-MS measurement in the same coffee test
Note: Adding milk strongly changed perception, but it did not systematically lower the release of coffee-derived volatile compounds.
Show data
process-milk-addition —does_not_reliably_track→ "A perceived decrease in coffee-derived flavor and a uniform decrease in retronasal volatile release"8. Don't Lump Sugar and Fat into the Same 「マスク」 (in translation: "mask")
In chocolate milk, increasing sucrose lowered both bitterness and bitter aftertaste. Yet in the same trial, the difference between 0.1% and 4.5% fat had no significant main effect on bitterness. Rules of thumb such as 「甘味は苦味を抑える」 (in translation: "sweetness suppresses bitterness") and「脂肪は苦味を丸める」 (in translation: "fat rounds off bitterness") need to be kept together with the food and the conditions they came from.
Scope: A trained-panel test that factorialized 2 fat levels and sweetener
Note: Bitterness and bitter aftertaste decreased significantly as sucrose increased. This does not mean that the sugar removed the bitter compounds.
Show data
ingredient-chocolate-milk —has_perceived_bitterness_affected_by→ "Sucrose concentration (0%, 2.5%, 5%)"Scope: The same factorial chocolate-milk test
Note: There was no significant main effect of fat concentration on bitterness. This is not proof that fat in general has no effect on bitterness.
Show data
ingredient-chocolate-milk —has_perceived_bitterness_unchanged_by→ "The main effect of fat at 0.1% versus 4.5%"9. Hops Change Role Depending on When They Go In
Hops added during the boil (kettle hopping) take part in building bitterness through the isomerization of alpha acids, while dry hopping after fermentation tends to leave aroma behind. But dry hopping is not「香りだけ」 (in translation: "aroma only") either. IBU, humulinones, polyphenols, and iso-alpha acids move in different directions, further complicating the relationship between analytical indicators and the bitterness actually drunk.
Scope: A literature synthesis on the timing of hop additions in beer brewing
Note: The background is the isomerization of alpha acids during the boil. This is contextual because it is a literature synthesis in the introduction, not a direct comparison from the main experiment.
Show data
process-boil-hopping —contributes_to→ sensation-bitterScope: The process definition used in a dry-hopping time study
Note: This is not treated as 『香りだけ』 (in translation: "aroma only"). Existing data show that humulinones, polyphenols, IBU, and iso-alpha acids can move as well.
Show data
process-dry-hopping —is_used_as→ "A process that adds hop aroma after fermentation"Scope: Pilot-scale dry hopping of a hop-free ale (0-16 g/L, 0-72 h), and a survey of 15 Pacific Northwest commercial beer brands sampled before and after dry hopping
Note: The 2 halves move in opposite directions: the conventional IBU proxy rises while the compound class it is supposed to stand in for falls. Central counter-evidence against treating IBU as a reading of iso-α-acids.
Show data
process-dry-hopping —increases→ concept-bitterness-unitsIf This Connects to a Bitter Seasoning
Umami-bitter, Sweet-bitter, and 「さわやか苦い」 ("refreshing-bitter") are not simply a formula of adding another taste to bitterness. They are an edit that includes onset, lingering finish, the ordering of aromas, viscosity, temperature, and even the order in which foods are eaten. At NIGAMI LAB, the prototyping log records not only how much bitterness was erased, but what was left, and what was brought to the front.
Honest caveats
- We keep a decrease in a compound from processing, a decrease in perception, and a rise in liking as separate claims.
- We do not reduce the results of fermentation or roasting to the cause of a single compound or a single microbe.
- A null result is not proof that 「絶対に変わらない」 (in translation: "it absolutely never changes"); we keep the sample and the conditions on the card.
- Regional techniques and the stories told about them are passed on as cultural records without numbers, to the next interview or experiment.
Frequently asked
Does salt-rubbing remove bitterness from bitter gourd?
One study that soaked bitter-gourd flesh in 3.5% salt water at 25 degrees for 60 minutes found no significant difference in bitterness intensity, measured on an unstructured scale, between the soaked flesh and untreated flesh. The same treatment did significantly lower sourness, meaning the salt treatment worked on sourness rather than bitterness. Meanwhile, under a different condition in the same study, adding 5% gum arabic significantly lowered the juice's bitterness, so a thickener, not salt, is the one confirmed to work if the goal is to weaken bitter gourd's bitterness. In short, this material does not confirm that salt-rubbing removes bitterness itself.
How can the way tea is brewed be changed to hold down bitterness and astringency?
One study using Tieguanyin confirms that increasing the amount of hot water relative to the tea leaves, lowering the water temperature, and shortening the steeping time significantly lowers the rating of the combined bitterness-and-astringency attribute. However, this study evaluates bitterness and astringency together as a single attribute, and the size of the effect on bitterness alone or astringency alone is not shown separately. The number of extractions is also treated as one of the conditions, but exactly how it worked cannot be read from this material.
Does fermentation eliminate bitterness, or does it change it?
One study on the brine fermentation of black olives concludes that the diffusion of bitter compounds into the brine and the action of the olive's own enzymes were the main mechanisms lowering bitterness. In a study fermenting sun-dried green maocha from old-tree tea for 36 days, bitterness and astringency at the endpoint were significantly lower than in the raw material, but at the same time sweetness rose and the color and aroma also changed — bitterness alone was not the only thing that disappeared. In a separate study comparing 5 olive cultivars, the final phenolic composition differed by cultivar, yet the resulting bitterness intensity showed no significant difference. In other words, this material fits better with the view that fermentation remakes bitterness together with other tastes and aromas, rather than simply eliminating it.
What is not known
- Why CMC and beta-cyclodextrin produce such a large difference in bitterness-reduction effect (such as differences in how readily an inclusion complex forms) has not yet been sufficiently confirmed.
- How the mechanism behind the drop in iso-alpha-acid concentration, seen in more than half of commercial beer brands before and after dry hopping, connects with polyphenol extraction has not yet been confirmed.
- Why the phenolic composition differs across 5 olive cultivars while the final bitterness intensity shows no difference is not confirmed within the scope of this material.
- No material was found examining why fat concentration (0.1% and 4.5%) had no significant effect on chocolate milk's bitterness intensity, or whether the same result holds for other foods.
Related: Salt-rubbing bitter gourd /IBU and bitterness /Hypothesis collection box
How to cite this page
Aoki, M. (NIGAMI LAB) (2026). Cooking Does Not Erase Bitterness. https://nigamilab.com/en/notes/cooking-edits-bitterness/
Each claim card has a stable anchor (#clm-…). The dataset is CC BY 4.0; see the data page. Dataset DOI: 10.5281/zenodo.22739101