日本語版

Research note 020 — 2026-09-01

Can Cooking Create Bitterness?

Talk about bitterness tends to be framed as a story about ingredients: bitter gourd is bitter, coffee is bitter. But tracing what actually happens in kitchens and factories through the sources shows that bitterness is also a product of process. People have made bitterness, removed it, decided it in advance through how something is grown, and hidden it. Laying NIGAMI LAB's processing records back out, the relationship between cooking and bitterness can be organized into the following four quadrants.

What this page establishes

• Roasting is the leading example of a cooking method that "makes" bitterness: multiple studies confirm that in coffee and cacao, new bitter compounds form as roasting conditions change.
• 「加熱すれば苦くなる」 (in translation: "if you heat it, it becomes bitter") does not hold uniformly: one study that compared turnip cooked 4 ways — boiled, steamed, roasted, and stir-fried — found no significant difference in bitterness by cooking method.
• Techniques for reducing bitter compounds through aku removal (soaking in water, boiling, fermenting) are recorded in home cooking around the world, but most do not quantify the rate of bitterness reduction.
• Multiple studies confirm that adding milk or oat drink weakens how bitter coffee is perceived, but the source also reports that this does not necessarily reduce the amount of the bitter compounds themselves (the release of volatile compounds was unchanged).

The four quadrants

Add — fire is a machine for making bitterness

Raw green coffee beans do not yet taste like the coffee we know. At lighter roasts, a group of compounds called quinides forms from chlorogenic acid, contributing to coffee's characteristic bitterness.

As roasting intensifies, the source reports, the quinides themselves break down and a different group of compounds forms instead, bringing a coarser quality of bitterness. In other words, dark-roast bitterness is not light-roast bitterness that has simply increased — its substance has been swapped out.

Furthermore, a compound specific to arabica, when broken down by roasting, is reported to yield breakdown products more bitter than the parent substance. Fire is not only a device for erasing bitterness — it is also a device for creating new substances perceived as bitter at lower concentrations.

In the world of beer too, the heating process of wort boiling converts hop-derived compounds and creates a group of substances that contribute to bitterness.

But 「加熱＝苦くなる」 (in translation: "heating = becoming bitter") is not the rule

Here, one result running in the opposite direction needs to be placed alongside the rest. In cacao, raising the roasting temperature actually lowered the bitterness consumers perceived. Depending on the material and conditions, fire can work on the "reducing" side rather than the "creating" side. The same process of roasting works in opposite directions depending on the material's initial state — and NIGAMI LAB's principle that bitterness is a relationship, not a fixed property of an ingredient, holds up from the process side as well.

In Doi Chang-grown arabica from northern Thailand as well, the relationship between roast level and composition has been tracked.

Remove — debittering techniques recorded around the world

On the opposite side of adding sits a group of removal techniques. Debittering olives, aku removal from fukinoto, soaking lupin beans in the Andes, macerating bitter leaf in Nigeria — each region's own 「抜き方」 (in translation: "way of removing it") is recorded in its own sources.

Closer to home, simply boiling Brassica vegetables reduces glucosinolates. Whether this decrease in content directly translates into a decrease in perceived bitterness, however, is itself recorded as a point that has not yet been settled.

Design — deciding bitterness before it reaches the table

The specification for Radicchio Variegato di Castelfranco records that the process of blanching in the dark after harvest deliberately develops what the specification calls a pleasant bitterness. Rather than adjusting bitterness at the cooking stage, it is built in at the growing stage — a third quadrant, different from both adding and removing.

Suppress — changing perception without removing anything

The last quadrant works in a way distinct from processes that remove the compound. Adding milk to coffee reduces how often bitterness comes to the fore during the time spent drinking it.

And the strangest "process" of all is the experience of continued drinking itself. Separate studies report that repeatedly taking in a bitter compound can lower its bitterness rating at the same concentration, and that vegetables, too, can become more accepted through repeated presentation. The apparatus that suppresses bitterness is built into the drinker as much as into the kitchen.

Add techniques not yet written up — smoking, straw-roasting, and scorched rice

This arrangement still has empty seats. Does smoke from smoking add bitterness? Does the toasty aroma of straw-roasted bonito contain bitter compounds? And in cultures that prize the char at the bottom of the pot as a delicacy in its own right — Japan's okoge, Korea's nurungji, Spain's socarrat, Iran's tahdig — where is the line drawn between toastiness and bitterness? Because no confirmed sources exist yet for these, NIGAMI LAB keeps them not as claims but as questions in thehypothesis box. As sources are found, this table of four quadrants will gain more seats.

Frequently asked

Why does adding milk make coffee's bitterness feel weaker?

Adding milk or oat drink to medium- or dark-roast coffee has been confirmed, using the sensory-evaluation method called TDS, to lower the proportion of time during which bitterness dominates the sensation as it changes over the course of drinking. At the same time, however, measurements of the aroma compounds released show that adding milk does not systematically lower the overall release of coffee-derived volatile compounds. In other words, milk is thought to work by weakening how bitterness is perceived, without necessarily reducing the amount of the bitter compounds themselves.

How have bitter beans and seeds been made edible?

For bitter lupin beans (Lupinus mutabilis), the traditional process of soaking in salt water and then boiling is reported to remove 89-97% of the seed's bitter compounds (alkaloids). Solid-state fermentation with a fungus after soaking and heating is further shown to reduce bitter compounds even more, down to roughly 9% to 43% of the initial value depending on conditions. However, the degree of this reduction varies widely by fungal strain and process conditions, and the review itself states that it 「説明できない」 (in translation: "cannot be explained"). For table olives too, curing and processing method, cultivar, and ripeness are cited as factors that affect the bitter compound (oleuropein), but how much each factor contributes, in terms of magnitude, is not known.

What is not known

• For most aku-removal techniques, no source has been found that quantifies how much the bitter compounds were reduced; they remain only as a record of the process.
• Why the reduction rate in lupin-bean debittering varies so much by fungal strain and process conditions has not yet been confirmed; the review itself states that it 「説明できない」 (in translation: "cannot be explained").
• For the change in chocolate's bitterness with roasting, the original study does not report the p-values or r-values that would show the strength of the correlation with astringency, and the confidence interval for the effect size has not been confirmed.
• The mechanism by which adding milk weakens only how bitterness is perceived, while the release of coffee's aroma compounds stays unchanged, has not yet been sufficiently confirmed.