Topic hub — 2026-10-22
Does fermenting, drying or roasting cacao change its bitterness?
A caveat first: within our search, few studies in which people compared bitterness after changing how cacao is fermented, dried or roasted could be checked down to their conditions. On top of that, in some of the included studies fermentation time or the time drying started differ along with the treatment compared, so the effect of a single step cannot be separated, and comparisons within one processing plant have to be read apart from comparisons across plants. The point of this page lies in that inability to separate. For roasting, a model predicts that unsweetened chocolate is less bitter the stronger the roast. For fermentation days, whether bitterness rose or fell differed between studies and plants. For drying temperature, the hotter drying had the higher mean bitterness, but only some comparisons were significant. Fermentation and drying here are production steps applied to the beans by farms and processing plants, not preparation done at home. This page lines up the included studies together with their conditions: samples, processing, how they were served, and who rated them on what scale. It then separates what was established within one experiment, what differences might explain, and what this comparison alone cannot separate. It is about perceived taste, not advice on choosing chocolate or on health. Differences in cocoa percentage are covered on the cocoa percentage page; this page is limited to differences in processing.
On this page (10 sections)
- What this page establishes
- What the included studies allow us to say
- Bitterness ratings at a glance
- Conditions and results, study by study
- What was established within an experiment
- Candidate explanations for the differences
- What is not known — what this comparison alone cannot separate
- How studies and rows were counted
- What we looked at
- Read in the encyclopedia
What this page establishes
- That in unsweetened chocolate a model predicts lower bitterness at higher roasting temperatures, and that these values are model predictions, not measured means for each treatment.
- That in studies varying fermentation days, whether bitterness rose or fell differed between studies and plants, with test results not reported for some pairs; and that in drying-temperature comparisons the hotter drying had the higher mean bitterness, but only some were significant, and in that pair drying also started at a different time.
- That in some of the included fermentation and drying comparisons, fermentation time, the time drying started and other conditions differ together, so which step mattered cannot be separated, and that plants cannot be compared with each other.
- That a tisane of shells from roasted beans ranked more bitter at higher temperatures, but shells are not beans and are not read together with the bean results.
What the included studies allow us to say
- For roasting, chocolate made from the beans with no added sugar (McClure et al. 2022, 145 consumers) was less bitter the stronger the roast. The authors converted the coefficient for 1 SD of roasting temperature (42 °C) into a value about 1.56 below the mean of 26.9 (the coefficient is on the square-root scale, and there is also an interaction with time). These values are model predictions, not measured means for any treatment. A study of sweetened bitter chocolate (Iida et al. 2018) writes in its discussion that the lower, longer roast had stronger "initial bitterness", but the figure shows no significance mark and its values cannot be read. [rows: McClure 2022 Roasting temperature 1 SD (42 °C) higher vs the mean; Iida 2018 Low-temperature long roast vs high-temperature roast (initial bitterness)]
- Fermentation and drying are steps that cacao producers (farms and processing plants) apply to the beans, not preparation done at home. In studies that varied fermentation days, whether bitterness rose or fell differed between studies and plants. In a study comparing 0 and 6 days of fermentation (Haruna et al. 2024), the 0-day beans were more bitter (bitterness quality scores 2.6 and 6.3, where higher means less objectionable bitterness; significant). In a study comparing 4 and 7 days (Ramírez González et al. 2024), 7 days was more bitter at some plants and 4 days at another, with no test result reported. In the three comparisons of drying temperature (two locations in Streule et al. 2024 and one pair in Streule et al. 2023), the higher temperature always had the higher mean bitterness, but only one difference was significant, and those 80 °C beans had started drying earlier. [rows: Haruna 2024 Fermented 6 days vs 0 days (no pre-drying); Ramírez González 2024 Fermented 7 days vs 4 days (ASA); Ramírez González 2024 Fermented 7 days vs 4 days (CMI); Ramírez González 2024 Fermented 7 days vs 4 days (COC); Streule 2024 Dried at 80 °C vs 60 °C (F); Streule 2024 Dried at 80 °C vs 60 °C (A); Streule 2023 Dried at high vs low temperature (A)]
- The samples and how they were served differ by study (unsweetened 100% chocolate or liquor, sweetened chocolate, cocoa cake, shell infusion), and so do the scales (intensity, quality score, a single bitter-to-sweet scale, ranks), so values cannot be compared across studies. In some of the included studies, fermentation time or the time drying started differ along with the treatment compared, so the effect of a single step cannot be separated. Comparisons within one plant and comparisons across plants have to be read separately, and plants cannot be compared with each other. In a tisane of shells taken from roasted beans (Gumelar et al. 2022), the higher the temperature the higher the bitterness rank, but shells are not beans and are not read together with the bean results. [rows: McClure 2022 Roasting temperature 1 SD (42 °C) higher vs the mean; Gumelar 2022 Shells: 140 °C for 18 min vs 100 °C for 20 min (roasted as whole beans)]
Bitterness ratings at a glance
Human bitterness ratings only. Samples and serving differ by study (unsweetened liquor or chocolate, sweetened chocolate, cocoa cake, shell infusion), so values are not compared across studies. The scales differ too. The "Direction" column shows which sample was more bitter ("... more bitter"). In most rows that is the larger value, but in rows using a quality score (higher means less objectionable bitterness) or an "extremely bitter to extremely sweet" scale a smaller value is more bitter (the unit column of those rows begins with "smaller values mean more bitter"). The nature of each value is given in the unit column, and model predictions and values that cannot be read from a figure are marked as such. The shell tisane row (not beans) and the row comparing origins roasted almost the same way are kept so that they can be told apart. Whether the difference is significant is shown by the test in the same "Direction / test" column. Select a study name to jump to its detailed table. Rows on astringency are in the detailed tables by study.
Results in which people rated bitterness only. Where each value comes from (a table or a reading from a figure) is shown in its row; scales and methods differ between studies, so compare within a row, not across rows.
| Study | Comparison and condition | Measured as | Treated sample | Comparison | Direction / test |
|---|---|---|---|---|---|
| McClure 2022 | Roasting temperature 1 SD (42 °C) higher vs the mean Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time | gLMS (145 people); converted from a model coefficient, not a measured mean for any treatment | 1.56 below the mean of 26.9 (authors' conversion) | 26.9 (mean) | Comparison more bitter Tested, significant |
| McClure 2022 | Lowest vs highest bitterness within the tested range (Ghana) Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time; the region of lowest bitterness lay between 135 and 171 °C, and the highest near raw (0 min, 24 °C) to light roasting (55 min, 64 °C) | gLMS (model predictions, not measured means for any treatment; response surface in Fig. 2 and values in the text) | 22 | 30 | Comparison more bitter No test reported |
| McClure 2022 | Lowest vs highest bitterness within the tested range (Madagascar) Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time; the region of lowest bitterness lay between 135 and 171 °C, and the highest near raw (0 min, 24 °C) to light roasting (55 min, 64 °C) | gLMS (model predictions, not measured means for any treatment; response surface in Fig. 2 and values in the text) | 25.8 | 28.2 | Comparison more bitter No test reported |
| McClure 2022 | Lowest vs highest bitterness within the tested range (Peru) Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time; the region of lowest bitterness lay between 135 and 171 °C, and the highest near raw (0 min, 24 °C) to light roasting (55 min, 64 °C) | gLMS (model predictions, not measured means for any treatment; response surface in Fig. 2 and values in the text) | 27 | 30 | Comparison more bitter No test reported |
| Iida 2018 | Low-temperature long roast vs high-temperature roast (initial bitterness) Bitter chocolates of the same recipe (60% cacao, 40% sugar); Venezuelan beans at a constant 125 °C and Ecuadorian beans at a constant 115 °C for 70–80 min (low), versus up to 147 °C for 45–50 min for both (high); temperature and time differ together | 7-point score (−3 weak to +3 strong); the vertical axis of Fig. 7 is broken and some tick labels are cut off, so no values are used | cannot be read from the figure | cannot be read from the figure | Treated sample more bitter Tested, not significant |
| Iida 2018 | Venezuela (high roast) vs Ghana (control) (strength of bitter aftertaste) Bitter chocolates of the same recipe; roasting was nearly the same, up to 147 °C for 40–50 min for Ghana and up to 147 °C for 45–50 min for Venezuela (high), so mainly the origin differs | 7-point score (−3 weak to +3 strong); the vertical axis of Fig. 7 is broken and some tick labels are cut off, so no values are used | cannot be read from the figure | cannot be read from the figure | Treated sample more bitter Tested, significant |
| Haruna 2024 | Fermented 6 days vs 0 days (no pre-drying) Beans with 0 h of pre-drying fermented spontaneously in heaps; in between, 2 days 4.3 (H) and 4 days 5.3 (EFG) | smaller values mean more bitter; bitterness quality score (0–10; scored inversely to perceived bitterness, so higher means less objectionable bitterness; 7.5 considered optimal) | 6.3 (BCD) | 2.6 (I) | Comparison more bitter Tested, significant |
| Ramírez González 2024 | Fermented 7 days vs 4 days (ASA) Beans taken from the same fermentation box on day 4 and day 7 at each plant, then dried by the plant's own method (as stated in the text); roasting for sensory testing was set by bean size and moisture, with temperature and time not given; bean mass, turning, drying days and clones differ by plant | 0–10 scale; mean calculated by us from the values per box in the supplementary table (in brackets) | 2.67 (3, 3, 2) | 2.00 (2, 2, 2) | Treated sample more bitter No test reported |
| Ramírez González 2024 | Fermented 7 days vs 4 days (CMI) Beans taken from the same fermentation box on day 4 and day 7 at each plant, then dried by the plant's own method (as stated in the text); roasting for sensory testing was set by bean size and moisture, with temperature and time not given; bean mass, turning, drying days and clones differ by plant | 0–10 scale; mean calculated by us from the values per box in the supplementary table (in brackets) | 2.33 (3, 2, 2) | 3.33 (4, 2, 4) | Comparison more bitter No test reported |
| Ramírez González 2024 | Fermented 7 days vs 4 days (COC) Beans taken from the same fermentation box on day 4 and day 7 at each plant, then dried by the plant's own method (as stated in the text); roasting for sensory testing was set by bean size and moisture, with temperature and time not given; bean mass, turning, drying days and clones differ by plant | 0–10 scale; mean calculated by us from the values per box in the supplementary table (in brackets) | 3.0 (3, 3) | 2.5 (3, 2) | Treated sample more bitter No test reported |
| Streule 2024 | Dried at 80 °C vs 60 °C (F) Beans with 0 days of pod storage fermented from the same day and dried at 80 °C (PS0_H) or 60 °C (PS0_L); the 80 °C beans started drying about 8–10 h earlier, and the dryer was loaded differently | 0–10 line scale (mean of 8 panellists) | 5.6 (a) | 4.3 (c) | Treated sample more bitter Tested, significant |
| Streule 2024 | Dried at 80 °C vs 60 °C (A) Beans with 0 days of pod storage fermented from the same day and dried at 80 °C (PS0_H) or 60 °C (PS0_L); the 80 °C beans started drying about 8–10 h earlier, and the dryer was loaded differently | 0–10 line scale (mean of 8 panellists) | 5.6 (ab) | 4.8 (bc) | Treated sample more bitter Tested, not significant |
| Streule 2023 | Dried at high vs low temperature (A) Fermented in jute bags with pre-drying and no turning to day 4 (about 63 h), with only drying temperature varied (JpT vs Jp) | 10 cm VAS (mean of 9 assessors; sample mixed from 2 runs) | 6.6 (ab) | 5.9 (abc) | Treated sample more bitter Tested, not significant |
| Streule 2023 | Fermentation 1 day shorter vs to day 4 (both dried at high temperature, A) Jute bags with pre-drying, no turning and high-temperature drying; the variation fermented to day 3 (about 40 h; JpsT) started drying on day 3, and the one fermented to day 4 (about 63 h; JpT) on day 4 | 10 cm VAS (mean of 9 assessors; sample mixed from 2 runs) | 7.1 (a) | 6.6 (ab) | Treated sample more bitter Tested, not significant |
| Streule 2023 | Short fermentation and high-temperature drying vs box fermentation with turning and low-temperature drying (A) JpsT (jute bag, pre-drying, no turning, to day 3 for about 40 h, high temperature) vs the control Wt (wooden box, turned on days 2 and 3, no pre-drying, to day 4 for about 63 h, low temperature); device, turning, pre-drying, fermentation time and drying temperature all differ together | 10 cm VAS (mean of 9 assessors; sample mixed from 2 runs) | 7.1 (a) | 4.6 (cde) | Treated sample more bitter Tested, significant |
| Ponka 2020 | Natron 1.442 g vs 0 g (per 200 g cocoa cake) Natron mixed into 200 g of the same cocoa cake and heated 10 min over low heat; intermediate amounts: 0.961 g 2.72 (ab), 1.082 g 3.47 (cd), 1.202 g 3.25 (bc), 1.322 g 3.58 (cd) | smaller values mean more bitter; mean on 9 points (1 extremely bitter, 5 neither bitter nor sweet, 9 extremely sweet); larger values mean less bitter | 4.14 (d) | 2.47 (a) | Comparison more bitter Tested, significant |
| Gumelar 2022 | Shells: 140 °C for 18 min vs 100 °C for 20 min (roasted as whole beans) Tisane of shells taken after roasting the beans, not the beans themselves; 120 °C for 20 min scored 37.8 | mean rank for bitterness (Kruskal–Wallis over 6 samples including 3 black teas; larger = stronger) | 54.7 | 17.4 | Treated sample more bitter Tested; significance reported, but which pair differs is unclear |
Conditions and results, study by study
For each study, the conditions for the samples, processing, serving and who rated them, and the results. Each row is one study × one measured result, so different comparisons in the same study are separate rows. Under each row we give where it is in the paper and whether it can be used to judge whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao, with the reason. Rows comparing origins or rating astringency are kept so that it is clear they cannot be used for this. Nothing is left blank: "not given", "not read" and "not applicable" are kept apart.
McClure 2022 sensory only
Optimizing consumer acceptability of 100% chocolate through roasting treatments and effects on bitterness and other important sensory characteristics. Current Research in Food Science — doi:10.1016/j.crfs.2022.01.005
- How deeply checked
- Checked in full text — Full text via Europe PMC (Table 1A, Results 3.1) Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-21.
- Samples
- 100% chocolate from beans of Madagascar, Ghana and Peru (no added sugar or other ingredients); cocoa content not varied
- Processing and cooking
- Roasting: 8 conditions plus replicated centre points applied to 3 origins, 27 treatments in all (including an unroasted control; 24–171 °C, 0–80 min)
- How it was served or eaten
- Discs of about 0.3 g
- Who rated, on what scale
- 145 consumers recruited (brief training on day 1); each sample rated by 104 people on average, 148 ratings; gLMS; mixed-effects model on square-root values
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Roasting temperature 1 SD (42 °C) higher vs the mean Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time | 1.56 below the mean of 26.9 (authors' conversion) | 26.9 (mean) | gLMS (145 people); converted from a model coefficient, not a measured mean for any treatment | Comparison more bitter Tested, significant |
| In the mixed-effects model the temperature coefficient was significant (−0.15290 on the square-root scale, p≤0.05); the time coefficient (−0.02853) was not significant, and the time × temperature interaction (−0.06640) was significant, so the decrease depends on roasting time; 1 SD of temperature is 42 °C Where: Table 1A; first Results section For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Roasting temperature and time varied for the same beans and 145 consumers rated bitterness, but the value is converted from a model coefficient and the chocolate has no added sugar | |||||
| Bitterness intensity (sensory) | Lowest vs highest bitterness within the tested range (Ghana) Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time; the region of lowest bitterness lay between 135 and 171 °C, and the highest near raw (0 min, 24 °C) to light roasting (55 min, 64 °C) | 22 | 30 | gLMS (model predictions, not measured means for any treatment; response surface in Fig. 2 and values in the text) | Comparison more bitter No test reported |
| No test compares the two ends of the range; in the model behind the predictions, the roasting temperature term and the time × temperature term were significant (see the temperature row of the same study) Where: first Results section; Fig. 2A For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Bitterness at the two ends of the roasting range, but model predictions rather than measured means for any treatment | |||||
| Bitterness intensity (sensory) | Lowest vs highest bitterness within the tested range (Madagascar) Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time; the region of lowest bitterness lay between 135 and 171 °C, and the highest near raw (0 min, 24 °C) to light roasting (55 min, 64 °C) | 25.8 | 28.2 | gLMS (model predictions, not measured means for any treatment; response surface in Fig. 2 and values in the text) | Comparison more bitter No test reported |
| No test compares the two ends of the range; in the model behind the predictions, the roasting temperature term and the time × temperature term were significant (see the temperature row of the same study) Where: first Results section; Fig. 2B For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Bitterness at the two ends of the roasting range, but model predictions rather than measured means for any treatment | |||||
| Bitterness intensity (sensory) | Lowest vs highest bitterness within the tested range (Peru) Fermented and dried beans from Madagascar, Ghana and Peru roasted within 24–171 °C and 0–80 min and made into 100% chocolate with no added sugar; the cocoa-percentage page cites the same study as samples whose cocoa content was not varied, while this page reads it as a comparison of roasting temperature and time; the region of lowest bitterness lay between 135 and 171 °C, and the highest near raw (0 min, 24 °C) to light roasting (55 min, 64 °C) | 27 | 30 | gLMS (model predictions, not measured means for any treatment; response surface in Fig. 2 and values in the text) | Comparison more bitter No test reported |
| No test compares the two ends of the range; in the model behind the predictions, the roasting temperature term and the time × temperature term were significant (see the temperature row of the same study) Where: first Results section; Fig. 2C For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Bitterness at the two ends of the roasting range, but model predictions rather than measured means for any treatment | |||||
Iida 2018 sensory only
カカオ豆のロースト条件とビターチョコレート嗜好の関連性. 日本官能評価学会誌 22(2):98-107 — doi:10.9763/jjsse.22.98
- How deeply checked
- Checked in full text — Full text of the J-STAGE PDF, 10 pages (section 2-2, Table 3, Fig. 7, section 3-2, section 4 discussion); the vertical axis of Fig. 7 is broken and some tick labels are cut off, so no values are read Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-22.
- Samples
- 5 bitter chocolates from beans of Ghana (control), Venezuela and Ecuador, all 60% cacao (cocoa butter plus cocoa mass), 40% sugar and a trace of vanilla flavouring (made by the Lotte Central Laboratory)
- Processing and cooking
- Ghana: up to 147 °C for 40–50 min. Venezuela and Ecuador each roasted 2 ways: high (up to 147 °C for 45–50 min) and low (constant 125 °C for Venezuela, constant 115 °C for Ecuador, 70–80 min); the low roasts were also longer
- How it was served or eaten
- Pieces of 1.9 × 3.8 × 0.8 cm at room temperature (20 °C); 3 bites per group of attributes, taste rated by licking
- Who rated, on what scale
- 202 people in all: 193 food-science students of Japan Women's University and 9 trained panellists (all women, aged 18–22); 12 attributes including "initial bitterness" and "strength of bitter aftertaste" on 7 points from −3 to +3 (+3 strong); one-way ANOVA and Tukey test (p<0.05)
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Low-temperature long roast vs high-temperature roast (initial bitterness) Bitter chocolates of the same recipe (60% cacao, 40% sugar); Venezuelan beans at a constant 125 °C and Ecuadorian beans at a constant 115 °C for 70–80 min (low), versus up to 147 °C for 45–50 min for both (high); temperature and time differ together | cannot be read from the figure | cannot be read from the figure | 7-point score (−3 weak to +3 strong); the vertical axis of Fig. 7 is broken and some tick labels are cut off, so no values are used | Treated sample more bitter Tested, not significant |
| Fig. 7 shows no significance mark for initial bitterness (marks are given only for pairs at p<0.05; one-way ANOVA and Tukey test); the direction follows the discussion (for both Venezuelan and Ecuadorian beans, the low roast had stronger "initial bitterness") Where: Fig. 7; Table 3; section 4 (discussion) For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Roasting varied for beans of the same origin and people rated bitterness, but temperature and time differ together, values cannot be read from the figure, and the direction comes from the discussion | |||||
| Bitterness intensity (sensory) | Venezuela (high roast) vs Ghana (control) (strength of bitter aftertaste) Bitter chocolates of the same recipe; roasting was nearly the same, up to 147 °C for 40–50 min for Ghana and up to 147 °C for 45–50 min for Venezuela (high), so mainly the origin differs | cannot be read from the figure | cannot be read from the figure | 7-point score (−3 weak to +3 strong); the vertical axis of Fig. 7 is broken and some tick labels are cut off, so no values are used | Treated sample more bitter Tested, significant |
| Fig. 7 marks the pair (p<0.05), and section 3-2 says Venezuela (high roast) scored higher than the control for strength of bitter aftertaste; the same figure also marks Venezuela (high) vs Ecuador (high); no pair of high and low roasts of the same origin is marked Where: Fig. 7; section 3-2; Table 3 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Not usable — Roasting was nearly the same, so this mainly compares origins | |||||
Haruna 2024 sensory only
Effects of Predrying and Spontaneous Fermentation Treatments on Nib Acidification, Fermentation Quality, and Flavour Attributes of Ghanaian Cocoa (Theobroma cacao) Beans. International Journal of Food Science 2024:5198607 — doi:10.1155/2024/5198607
- How deeply checked
- Checked in full text — Full text via PMC (sections 2.1, 2.2.3 and 2.2.4, section 3.3.1, Table 6) Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-22.
- Samples
- Beans of mixed hybrids from the experimental plots of the Cocoa Research Institute of Ghana (CRIG), October 2022
- Processing and cooking
- Beans pre-dried in the sun for 0, 2, 4, 6 or 8 h, then about 60 kg heaped on plantain leaves and fermented spontaneously for 0, 2, 4 or 6 days (turned every 48 h; 3 replicates each); sun-dried to 7.5% moisture or below and roasted at 118 °C for 23 min
- How it was served or eaten
- Nibs milled into liquor; about 1 g melted at 35 °C and served
- Who rated, on what scale
- 7 experienced trained panellists (4 women, 3 men); quality scores 0–10 (0 extremely bad, 10 excellent); negative attributes such as bitterness scored inversely to perceived intensity, with 7.5 considered optimal for bitterness; ANOVA and LSD (5%)
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Fermented 6 days vs 0 days (no pre-drying) Beans with 0 h of pre-drying fermented spontaneously in heaps; in between, 2 days 4.3 (H) and 4 days 5.3 (EFG) | 6.3 (BCD) | 2.6 (I) | smaller values mean more bitter; bitterness quality score (0–10; scored inversely to perceived bitterness, so higher means less objectionable bitterness; 7.5 considered optimal) | Comparison more bitter Tested, significant |
| The letters in Table 6 do not overlap (ANOVA and LSD, p<0.05); a quality score, not bitterness intensity itself; in "Direction", "higher" refers to the more bitter sample, which here is the opposite of the values (0 days was more bitter) Where: Table 6; section 3.3.1 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Fermentation days varied for the same beans and 7 trained panellists rated them, but as a quality score rather than bitterness intensity | |||||
Ramírez González 2024 sensory only
Fermentation and its effect on the physicochemical and sensory attributes of cocoa beans in the Colombian Amazon. PLOS ONE 19(10):e0306680 — doi:10.1371/journal.pone.0306680
- How deeply checked
- Checked in full text — Full text via Europe PMC (sections 2.1, 2.2, 2.3.5 and 2.4, section 3.5, Fig. 3) and the sensory sheet of supplementary table S1 (values per box); means calculated by us Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-22.
- Samples
- Beans from 3 processing plants (ASA, CMI, COC) in Caquetá, Colombia; ASA and CMI use mixed clones such as CCN51, ICS1, ICS95, TSH565 and FEAR5; clones not given for COC
- Processing and cooking
- Fermented in wooden boxes, with beans taken from the same box on day 4 and day 7; dried by each plant's own method (bean mass, turning and drying days differ by plant); roasting curve set by bean size and moisture, with temperature and time not given
- How it was served or eaten
- Cocoa paste (liquor) refined in a melangeur from the roasted beans; amount and temperature not given
- Who rated, on what scale
- 5 trained panellists; international cocoa evaluation scale 0–10; 3 fermentation boxes per plant (2 rows each for COC in the supplementary table); no test result reported for the sensory values (the methods describe a mixed model and Fisher's LSD)
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Fermented 7 days vs 4 days (ASA) Beans taken from the same fermentation box on day 4 and day 7 at each plant, then dried by the plant's own method (as stated in the text); roasting for sensory testing was set by bean size and moisture, with temperature and time not given; bean mass, turning, drying days and clones differ by plant | 2.67 (3, 3, 2) | 2.00 (2, 2, 2) | 0–10 scale; mean calculated by us from the values per box in the supplementary table (in brackets) | Treated sample more bitter No test reported |
| No test result is reported for the sensory values; the bitterness direction matches section 3.5 ("bitterness and astringency increased at ASA and COC and decreased at CMI") Where: Supplementary table S1 (Sensory); section 3.5; Fig. 3 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Fermentation days compared for beans from the same boxes, but with no test, means calculated by us from the supplementary table, and 5 panellists | |||||
| Bitterness intensity (sensory) | Fermented 7 days vs 4 days (CMI) Beans taken from the same fermentation box on day 4 and day 7 at each plant, then dried by the plant's own method (as stated in the text); roasting for sensory testing was set by bean size and moisture, with temperature and time not given; bean mass, turning, drying days and clones differ by plant | 2.33 (3, 2, 2) | 3.33 (4, 2, 4) | 0–10 scale; mean calculated by us from the values per box in the supplementary table (in brackets) | Comparison more bitter No test reported |
| No test result is reported for the sensory values; the bitterness direction matches section 3.5 ("bitterness and astringency increased at ASA and COC and decreased at CMI"). For astringency the supplementary table gives 3.00 on both day 4 and day 7, which does not match the text saying astringency fell at CMI Where: Supplementary table S1 (Sensory); section 3.5; Fig. 3 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Fermentation days compared for beans from the same boxes, but with no test, means calculated by us from the supplementary table, and 5 panellists | |||||
| Bitterness intensity (sensory) | Fermented 7 days vs 4 days (COC) Beans taken from the same fermentation box on day 4 and day 7 at each plant, then dried by the plant's own method (as stated in the text); roasting for sensory testing was set by bean size and moisture, with temperature and time not given; bean mass, turning, drying days and clones differ by plant | 3.0 (3, 3) | 2.5 (3, 2) | 0–10 scale; mean calculated by us from the values per box in the supplementary table (in brackets) | Treated sample more bitter No test reported |
| No test result is reported for the sensory values; the bitterness direction matches section 3.5 ("bitterness and astringency increased at ASA and COC and decreased at CMI"). The text says beans came from the same boxes, but for COC the supplementary table has replicates 1 and 3 on day 4 and replicates 1 and 2 on day 7, which do not match each other or the 3 replicates stated in the text; means were calculated from the 2 rows available each day, and 1 day-4 row has the same values in every attribute as 1 day-7 row of CMI Where: Supplementary table S1 (Sensory); section 3.5; Fig. 3 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Fermentation days compared for beans from the same boxes, but with no test, means calculated by us from the supplementary table, and 5 panellists | |||||
Streule 2024 sensory only
Effect of Pod Storage and Drying Temperature on Fermentation Dynamics and Final Bean Quality of Cacao Nacional in Ecuador. Foods 13:1536 — doi:10.3390/foods13101536
- How deeply checked
- Checked in full text — Full text via Europe PMC (sections 2.1, 2.2.9 and 2.3, section 3.9, Table 5, section 4.2) Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-22.
- Samples
- Cacao Nacional from Ecuador at 2 locations, Manabí (A) and Guayas (F), each run 3 times independently (September to November 2021)
- Processing and cooking
- Kept in plastic bags for the first night, then pre-dried the next day and fermented in jute bags; beans after 0, 3 or 5 days of pod storage dried at 60 °C, and beans with 0 days of pod storage dried at 80 °C; the 80 °C beans started drying about 8–10 h earlier and in some runs filled 1/3 of the dryer's effective area (runs 2 and 3)
- How it was served or eaten
- Cocoa liquor from beans of the 3 runs mixed; served at 45 °C
- Who rated, on what scale
- 8 trained panellists of ZHAW (10 training sessions); 10 cm line scale (0 not perceivable to 10 very intense), 3 replicates; two-way ANOVA (assessors random) and Fisher's LSD (p<0.05)
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Dried at 80 °C vs 60 °C (F) Beans with 0 days of pod storage fermented from the same day and dried at 80 °C (PS0_H) or 60 °C (PS0_L); the 80 °C beans started drying about 8–10 h earlier, and the dryer was loaded differently | 5.6 (a) | 4.3 (c) | 0–10 line scale (mean of 8 panellists) | Treated sample more bitter Tested, significant |
| The letters do not overlap (two-way ANOVA and Fisher's LSD, p<0.05); the authors note that the 80 °C beans started drying earlier, so the difference in fermentation time may have mattered more than drying temperature, and that the relation between drying temperature or duration and these attributes remains unclear Where: Table 5; section 3.9; section 4.2 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Drying temperature varied and 8 trained panellists rated bitterness, but the 80 °C beans started drying about 8–10 h earlier and the dryer was loaded differently | |||||
| Bitterness intensity (sensory) | Dried at 80 °C vs 60 °C (A) Beans with 0 days of pod storage fermented from the same day and dried at 80 °C (PS0_H) or 60 °C (PS0_L); the 80 °C beans started drying about 8–10 h earlier, and the dryer was loaded differently | 5.6 (ab) | 4.8 (bc) | 0–10 line scale (mean of 8 panellists) | Treated sample more bitter Tested, not significant |
| They share the letter b, so no significant difference was detected; the abstract says there were significant differences between 80 °C and 60 °C with the former more bitter and astringent, but for bitterness at A, Table 5 shows no significant difference Where: Table 5; section 3.9; section 4.2 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Drying temperature varied and 8 trained panellists rated bitterness, but the 80 °C beans started drying about 8–10 h earlier and the dryer was loaded differently | |||||
| Astringency (sensory) | Dried at 80 °C vs 60 °C (A, astringency) Beans with 0 days of pod storage fermented from the same day and dried at 80 °C (PS0_H) or 60 °C (PS0_L); the 80 °C beans started drying about 8–10 h earlier, and the dryer was loaded differently | 5.8 (a) | 4.9 (cd) | 0–10 line scale (mean of 8 panellists) | Treated sample higher Tested, significant |
| Astringency, not bitterness; the letters do not overlap; also significant at F, 5.8 (ab) and 4.3 (d) Where: Table 5; section 3.9; section 4.2 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Not usable — Astringency, not bitterness | |||||
Streule 2023 sensory only
Variations in Ecuadorian Cocoa Fermentation and Drying at Two Locations: Implications for Quality and Sensory. Foods 13(1):137 — doi:10.3390/foods13010137
- How deeply checked
- Checked in full text — Full text via PMC (sections 2.1, 2.3.6 and 2.4, section 3.6, Tables 1, 3 and 6) Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-22.
- Samples
- Cacao Nacional from Ecuador at 2 locations, Manabí (A) and Los Ríos (E) (April to August 2018); about 50 kg lots, 9 process variations
- Processing and cooking
- Combinations of fermentation device (wooden box, jute bag, plastic bag), pre-drying, turning, fermentation time (to day 4, about 63 h; to day 3, about 40 h) and drying temperature (high or low); at A the mean maximum drying temperature was 94.5 °C for the low and 96.3 °C for the high group, and the mean temperature 59.0 °C and 73.2 °C
- How it was served or eaten
- Cocoa liquor from beans of 2 runs mixed (runs 1 and 2 at A, runs 3 and 4 at E); prepared as in an earlier paper, serving temperature not given in this paper
- Who rated, on what scale
- 9 assessors of the Lindt & Sprüngli cocoa panel (selected and trained under ISO 8586); 10 cm VAS in 3 separate sessions; mixed ANOVA (assessors random) and Tukey HSD; in Table 6 the same letter means no significant difference (footnote p<0.001)
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Dried at high vs low temperature (A) Fermented in jute bags with pre-drying and no turning to day 4 (about 63 h), with only drying temperature varied (JpT vs Jp) | 6.6 (ab) | 5.9 (abc) | 10 cm VAS (mean of 9 assessors; sample mixed from 2 runs) | Treated sample more bitter Tested, not significant |
| They share letters, so no significant difference was detected (mixed ANOVA and Tukey HSD); Table 3 gives drying temperatures only for the groups of high and low variations Where: Table 6; Tables 1 and 3; section 2.1 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Only drying temperature varied and 9 trained assessors rated bitterness, but samples were mixed from 2 runs and temperatures are given only for groups of variations | |||||
| Bitterness intensity (sensory) | Fermentation 1 day shorter vs to day 4 (both dried at high temperature, A) Jute bags with pre-drying, no turning and high-temperature drying; the variation fermented to day 3 (about 40 h; JpsT) started drying on day 3, and the one fermented to day 4 (about 63 h; JpT) on day 4 | 7.1 (a) | 6.6 (ab) | 10 cm VAS (mean of 9 assessors; sample mixed from 2 runs) | Treated sample more bitter Tested, not significant |
| They share the letter a, so no significant difference was detected (mixed ANOVA and Tukey HSD) Where: Table 6; Tables 1 and 3; section 2.1 For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — A comparison of shorter fermentation, but the day drying started also differs and samples were mixed from 2 runs | |||||
| Bitterness intensity (sensory) | Short fermentation and high-temperature drying vs box fermentation with turning and low-temperature drying (A) JpsT (jute bag, pre-drying, no turning, to day 3 for about 40 h, high temperature) vs the control Wt (wooden box, turned on days 2 and 3, no pre-drying, to day 4 for about 63 h, low temperature); device, turning, pre-drying, fermentation time and drying temperature all differ together | 7.1 (a) | 4.6 (cde) | 10 cm VAS (mean of 9 assessors; sample mixed from 2 runs) | Treated sample more bitter Tested, significant |
| The letters do not overlap (mixed ANOVA and Tukey HSD); the abstract says that fermenting in plastic bags without pre-drying or turning, or in jute bags for a shorter time, gave low maximum fermentation temperatures and bitter, astringent liquor Where: Table 6; Tables 1 and 3; section 2.1; abstract For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — The difference is significant, but device, turning, pre-drying, fermentation time and drying temperature all differ together, so the effect of any one step cannot be separated | |||||
Ponka 2020 sensory only
The reduction of cocoa cake bitterness using natron and its effects on chocolate nutritive value. Food Science & Nutrition 8(7):3425-3434 — doi:10.1002/fsn3.1624
- How deeply checked
- Checked in full text — Full text via PMC (materials and methods, first results section, Table 4) Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-22.
- Samples
- Mixed cocoa beans collected in a Cameroonian village (Fegmimbang), sun-dried by farmers (fermentation not mentioned)
- Processing and cooking
- 5 kg of beans roasted in a small pan over low heat for 30–40 min, cracked, winnowed, ground, mixed with 5 L of water and heated, and the oil removed; the remainder (cocoa cake), 200 g at a time, mixed with 0, 0.961, 1.082, 1.202, 1.322 or 1.442 g natron (described by the authors as a mixture of salts such as sodium carbonate, sodium sulfate and sodium chloride) and heated 10 min over low heat; 1.202 g is the amount the farmers use
- How it was served or eaten
- Natron-treated cocoa cake tasted (amount and temperature not given)
- Who rated, on what scale
- 35 panellists (training not described); 9 points (1 extremely bitter, 5 neither bitter nor sweet, 9 extremely sweet); one-way ANOVA and Duncan test (p<0.05)
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Natron 1.442 g vs 0 g (per 200 g cocoa cake) Natron mixed into 200 g of the same cocoa cake and heated 10 min over low heat; intermediate amounts: 0.961 g 2.72 (ab), 1.082 g 3.47 (cd), 1.202 g 3.25 (bc), 1.322 g 3.58 (cd) | 4.14 (d) | 2.47 (a) | smaller values mean more bitter; mean on 9 points (1 extremely bitter, 5 neither bitter nor sweet, 9 extremely sweet); larger values mean less bitter | Comparison more bitter Tested, significant |
| The letters do not overlap (one-way ANOVA and Duncan test, p<0.05); the values are not in order of amount everywhere (1.082 g higher than 1.202 g); the single bitter-to-sweet scale cannot separate weaker bitterness from perceived sweetness; in "Direction", "higher" refers to the more bitter sample, which here is the opposite of the values (0 g was more bitter) Where: Table 4; first results section For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Limited — Alkaline salts added to the same cocoa cake and rated by 35 people, but this is a farmers' technique, not industrial alkalization, and the scale runs from bitter to sweet | |||||
Gumelar 2022 sensory only
Antioxidant Content of Tisane of Cocoa Bean Shells as Affected by Roasting Temperatures. Pelita Perkebunan 38(3):200–210 — doi:10.22302/iccri.jur.pelitaperkebunan.v38i3.524
- How deeply checked
- Checked in full text — Full PDF, 11 pages (methods for roasting, deshelling and sensory testing; Table 3); Table 3 checked as a page image Checked by: LLM (one extracted, another checked against the source; no human check), 2026-10-22.
- Samples
- Shells (not beans) taken by deshelling after roasting 0.5 kg of Trinitario beans from an experimental station of the Indonesian Coffee and Cocoa Research Institute (ICCRI); 3 black teas for comparison
- Processing and cooking
- Whole beans roasted at 100 °C for 20 min, 120 °C for 20 min or 140 °C for 18 min before the shells were separated
- How it was served or eaten
- 2 g shells in 150 mL boiling water for 5 min, filtered; 25 mL served without sugar or milk
- Who rated, on what scale
- 15 people selected from 20 untrained people by basic taste recognition (bitterness: caffeine 0.25% and 0.5%); intensity rated, reported as Kruskal–Wallis mean ranks (Table 3); no pairwise comparison between roasting temperatures
| Measured | Comparison and condition | Treated sample | Comparison | Unit | Direction / test |
|---|---|---|---|---|---|
| Bitterness intensity (sensory) | Shells: 140 °C for 18 min vs 100 °C for 20 min (roasted as whole beans) Tisane of shells taken after roasting the beans, not the beans themselves; 120 °C for 20 min scored 37.8 | 54.7 | 17.4 | mean rank for bitterness (Kruskal–Wallis over 6 samples including 3 black teas; larger = stronger) | Treated sample more bitter Tested; significance reported, but which pair differs is unclear |
| The text reports bitterness as significant across the 6 samples (p<0.05), but there is no pairwise comparison between roasting temperatures; the black teas scored 58.5, 62.3 and 42.3 Where: Table 3; results section on sensory attributes For whether processing conditions such as fermentation, drying or roasting change the bitterness people perceive in cacao (liquor or chocolate made from the beans): Not usable — A shell tisane, not the bitterness of beans or chocolate | |||||
What was established within an experiment
Only results observed in comparisons within one study. Results for beans and for shells are kept apart.
- When only roasting temperature and time varied for beans of the same 3 origins, bitterness fell with stronger roasting, and temperature mattered more than time (the temperature coefficient was significant, the time coefficient was not, and the time × temperature interaction was significant). Across the tested range the model predicted 30 to 22 for Ghana, 28.2 to 25.8 for Madagascar and 30 to 27 for Peru. The region of lowest bitterness was 135–171 °C. [rows: McClure 2022 Roasting temperature 1 SD (42 °C) higher vs the mean; McClure 2022 Lowest vs highest bitterness within the tested range (Ghana); McClure 2022 Lowest vs highest bitterness within the tested range (Madagascar); McClure 2022 Lowest vs highest bitterness within the tested range (Peru)]
- When beans without pre-drying were fermented spontaneously for 0, 2, 4 or 6 days, the bitterness quality score rose 2.6, 4.3, 5.3 and 6.3 (higher means less objectionable bitterness), with a significant difference between 0 and 6 days. [rows: Haruna 2024 Fermented 6 days vs 0 days (no pre-drying)]
- For beans taken on day 4 and day 7 at each plant (the text says from the same boxes, but for COC the replicate numbers in the supplementary table do not match), bitterness (means calculated by us) rose from 2.00 to 2.67 at ASA and from 2.5 to 3.0 at COC, and fell from 3.33 to 2.33 at CMI. No test result is reported for the sensory values. [rows: Ramírez González 2024 Fermented 7 days vs 4 days (ASA); Ramírez González 2024 Fermented 7 days vs 4 days (CMI); Ramírez González 2024 Fermented 7 days vs 4 days (COC)]
- When beans fermented from the same day were dried at 60 °C and 80 °C, bitterness values were higher at 80 °C (F: 4.3 and 5.6, significant; A: 4.8 and 5.6, not significant). Astringency was significantly higher at A as well (4.9 and 5.8). The 80 °C beans had started drying about 8–10 h earlier. [rows: Streule 2024 Dried at 80 °C vs 60 °C (F); Streule 2024 Dried at 80 °C vs 60 °C (A); Streule 2024 Dried at 80 °C vs 60 °C (A, astringency)]
- For beans fermented the same way (jute bag, pre-drying, no turning), raising only the drying temperature changed bitterness from 5.9 to 6.6 with no significant difference; shortening fermentation by 1 day as well changed it from 6.6 to 7.1, again with no significant difference. The control fermented in a box with turning and dried at low temperature (4.6) and the variation with shorter fermentation and high-temperature drying (7.1) differed significantly, but device, turning, pre-drying, fermentation time and drying temperature all differ together. [rows: Streule 2023 Dried at high vs low temperature (A); Streule 2023 Fermentation 1 day shorter vs to day 4 (both dried at high temperature, A); Streule 2023 Short fermentation and high-temperature drying vs box fermentation with turning and low-temperature drying (A)]
- Adding natron to the same cocoa cake raised the score from 2.47 at 0 g to 4.14 at 1.442 g (1 extremely bitter, 5 neither bitter nor sweet, 9 extremely sweet), a significant shift towards less bitterness. Intermediate amounts were not all in order (1.082 g 3.47, 1.202 g 3.25). [rows: Ponka 2020 Natron 1.442 g vs 0 g (per 200 g cocoa cake)]
- When whole beans were roasted at 100 °C for 20 min, 120 °C for 20 min or 140 °C for 18 min before shelling, the mean bitterness ranks of the shell tisane were 17.4, 37.8 and 54.7, higher at higher temperature. The test covers all 6 samples including black teas; there is no pairwise comparison between roasting temperatures. [rows: Gumelar 2022 Shells: 140 °C for 18 min vs 100 °C for 20 min (roasted as whole beans)]
Candidate explanations for the differences
Conditions that differ between studies and could explain the differences in bitterness results. Several conditions differ at once, so which one mattered is unknown. These are candidates, not causes.
- As an observation, Haruna et al. 2024 describe liquor from 0-day beans as unpleasantly bitter and extremely astringent (a description of the sensory result). As the authors' interpretations, Streule et al. 2024 regard the stronger bitterness and astringency of the 80 °C beans as consistent with weaker fermentation because drying started earlier, and Ramírez González et al. 2024 link the plants where bitterness and astringency increased to possible over-fermentation. Building on these, this page offers a candidate: the split in direction across fermentation and drying conditions may reflect how far fermentation progressed. This is this page's inference and has not been confirmed. [rows: Haruna 2024 Fermented 6 days vs 0 days (no pre-drying); Streule 2024 Dried at 80 °C vs 60 °C (F); Ramírez González 2024 Fermented 7 days vs 4 days (ASA)]
- For roasting, McClure et al. and Iida et al. share the direction that the hotter roast was less bitter (initial bitterness in Iida et al.). However, McClure et al. give model predictions for chocolate with no added sugar with temperature and time varied over a wide range, whereas Iida et al. compared roasts differing in both temperature and time in chocolate with 40% sugar, with no significance mark in the figure. Because samples, design and significance differ, this is not read as an effect of the same size. In the shell tisane, bitterness rank rose with temperature, but shells are a different material from beans and are treated separately. [rows: McClure 2022 Roasting temperature 1 SD (42 °C) higher vs the mean; Iida 2018 Low-temperature long roast vs high-temperature roast (initial bitterness); Gumelar 2022 Shells: 140 °C for 18 min vs 100 °C for 20 min (roasted as whole beans)]
What is not known — what this comparison alone cannot separate
- In some of the included studies, other conditions differ along with the treatment compared, so the effect of a single step cannot be separated. The 80 °C beans of Streule et al. 2024 started drying earlier and the dryer was loaded differently. In Streule et al. 2023 the shorter fermentation also started drying on a different day, and the comparison with the box-fermented control differs in many conditions at once. Ramírez González et al. 2024 compare beans from the same boxes within a plant, but across plants bean mass, turning, drying days and clones differ, so plants cannot be compared with each other. The samples rated were liquors mixed from several runs or values per box at each plant, so whether the same direction holds when the treatment is repeated at another time is unknown. [rows: Streule 2024 Dried at 80 °C vs 60 °C (F); Streule 2023 Fermentation 1 day shorter vs to day 4 (both dried at high temperature, A); Streule 2023 Short fermentation and high-temperature drying vs box fermentation with turning and low-temperature drying (A); Ramírez González 2024 Fermented 7 days vs 4 days (CMI)]
- The nature of the values differs by study: McClure et al. give model predictions rather than measured means for each treatment; Haruna et al. give quality scores rather than bitterness intensity (higher means less objectionable bitterness); Ponka et al. use a single bitter-to-sweet scale; Gumelar et al. give ranks; Ramírez González et al. give means calculated by us from a supplementary table; and the values of Iida et al. cannot be read from the figure. Values are not compared across studies. [rows: McClure 2022 Lowest vs highest bitterness within the tested range (Ghana); Haruna 2024 Fermented 6 days vs 0 days (no pre-drying); Ponka 2020 Natron 1.442 g vs 0 g (per 200 g cocoa cake); Gumelar 2022 Shells: 140 °C for 18 min vs 100 °C for 20 min (roasted as whole beans); Ramírez González 2024 Fermented 7 days vs 4 days (ASA); Iida 2018 Low-temperature long roast vs high-temperature roast (initial bitterness)]
- The fermentation, drying and roasting in this table are steps applied to the beans by cacao producers, not ways of handling bought chocolate or cocoa at home. The natron study is a Cameroonian farmers' technique, not a comparison of industrial alkalization (Dutch processing). Differences in cocoa percentage are covered on a separate page. [rows: McClure 2022 Roasting temperature 1 SD (42 °C) higher vs the mean; Ponka 2020 Natron 1.442 g vs 0 g (per 200 g cocoa cake)]
How studies and rows were counted
The number of evidence rows is not the number of studies. Table rows, ledger evidence rows, papers and research groups are counted separately.
- Sources in the table: 8, of which 8 have results on people's bitterness ratings in this table; for the others, only results on something else (compounds, smell) are shown, or they are used only for context.
- Rows in the table (study × measured result): 18. Rows on perceived bitterness: 17.
- Evidence rows in the lab's ledger these rows rest on: 0. One paper can supply several rows. Rows without a ledger evidence row come from other tables of the same papers, or from sources newly read for this question.
- Research groups among the studies with bitterness results in this table, as grouped in the ledger: 7 (reports from the same lab count as 1; co-authors and affiliations can overlap, so this is not a count of independent replications).
What we looked at
The sources examined for this question and why each was included or excluded. We searched Europe PMC and J-STAGE for studies in which people rated bitterness after cacao fermentation, drying, roasting, alkalization or conching conditions were varied, and added the sources already in the lab's ledger. Studies whose full text could not be reached and studies read only as abstracts are marked as such. The search described here is a record, not a claim that nothing was missed.
| Source | Decision | Why |
|---|---|---|
| Optimizing consumer acceptability of 100% chocolate through roasting treatments and effects on bitterness and other important sensory characteristics (2022) doi | Included in the table | Roasting temperature and time varied for beans of 3 origins; 145 consumers rated the bitterness of chocolate with no added sugar; read in full; also used on the cocoa-percentage page |
| カカオ豆のロースト条件とビターチョコレート嗜好の関連性 (2018) doi | Included in the table | High and low roasts of beans of the same origin rated in bitter chocolate by 202 people; figure values cannot be read, so only significance marks and the text are used; read in full |
| Effects of Predrying and Spontaneous Fermentation Treatments on Nib Acidification, Fermentation Quality, and Flavour Attributes of Ghanaian Cocoa (Theobroma cacao) Beans (2024) doi | Included in the table | Pre-drying and fermentation days varied; 7 trained panellists rated the liquor with quality scores; read in full |
| Fermentation and its effect on the physicochemical and sensory attributes of cocoa beans in the Colombian Amazon (2024) doi | Included in the table | Beans from day 4 and day 7 of fermentation at 3 plants rated by 5 trained panellists; values calculated from the supplementary table, no statistical test results reported for the sensory comparisons; read in full |
| Effect of Pod Storage and Drying Temperature on Fermentation Dynamics and Final Bean Quality of Cacao Nacional in Ecuador (2024) doi | Included in the table | Drying temperature (60 °C and 80 °C) and pod storage varied; 8 trained panellists rated the liquor; read in full |
| Variations in Ecuadorian Cocoa Fermentation and Drying at Two Locations: Implications for Quality and Sensory (2023) doi | Included in the table | Fermentation device, pre-drying, turning, fermentation time and drying temperature varied; a 9-member panel rated the liquor; read in full |
| The reduction of cocoa cake bitterness using natron and its effects on chocolate nutritive value (2020) doi | Included in the table | Amount of natron added to cocoa cake varied; rated by 35 people; not industrial alkalization; read in full |
| Antioxidant Content of Tisane of Cocoa Bean Shells as Affected by Roasting Temperatures (2022) doi | Included in the table | Tisane of shells from beans roasted at different temperatures rated by 15 people; not beans, so kept apart from the bean rows; read in full |
| The impact of controlled fermentation temperature on chemical composition and sensory properties of cacao (2025) doi | Read as context (not in the table) | Fermentation temperature was controlled and varied; sensory profiles by treatment and genotype are in Fig. 4 (radar charts), but exact values and a test of bitterness between treatments could not be taken; correlations of bitterness with compounds are in Fig. 5 |
| High temperature and nib acidification during cacao-controlled fermentation improve cadmium transfer from nibs to testa and the liquor's flavor (2024) doi | Read as context (not in the table) | The fermentation trial behind Calvo et al. 2025; flavour was rated for intensity on 0–10 and profiles are shown in Fig. 7, but exact values and tests of bitterness between treatments were not taken; the reported p values concern the frequency of indicator attributes (supplements not read) |
| The impact of yeast starter cultures on the microbial communities and volatile compounds in cocoa fermentation and the resulting sensory attributes of chocolate (2015) doi | Read as context (not in the table) | Compared yeast-starter and spontaneous fermentation, but bitterness is a TDS dominance rate, not an intensity comparison; the figure has TDS significance-level lines, but these are not a test of bitterness intensity between treatments; panel size and test details unconfirmed because the full text could not be obtained |
| Comparison of the drying behavior of fermented cocoa (Theobroma cacao L.) beans dried in a cocoa house, greenhouse and mechanical oven (2017) | Read as context (not in the table) | Compared drying places and methods (cocoa house, greenhouse, 40 °C oven), but bitterness is only described as showing little difference, with no values or panel size to take; a conference report, peer review not confirmed |
| Influences of Depulping, Pod Storage and Fermentation Time on Fermentation Dynamics and Quality of Ghanaian Cocoa (2024) doi | Read as context (not in the table) | Pod storage, depulping and fermentation time varied, but sensory data are consensus values for 5 selected samples with no test; beans fermented 3–4 days were described as strongly bitter and astringent in screening but were not characterised sensorially |
| Exploring the Impact of Fermentation Time and Climate on Quality of Cocoa Bean-Derived Chocolate: Sensorial Profile and Volatilome Analysis (2024) doi | Read as context (not in the table) | Fermentation time and climate varied, but bitterness values are in supplementary figures, which were not read |
| Microwave Roasting as an Alternative to Convection Roasting: Sensory Analysis and Physical Characterization of Dark Chocolate (2023) doi | Read as context (not in the table) | Microwave and convection roasting compared in 70% chocolate using consumer check-all-that-apply (CATA); bitterness was among the most cited terms for both, with no significant difference in taste terms; not an intensity comparison, so not tabled |
| Chemical composition and flavor of Ecuadorian cocoa liquor (2002) doi | Read as context (not in the table) | Abstract only; correlations between polyphenols and bitterness or astringency in liquor, not a comparison of fermentation, drying or roasting conditions |
| Linking chemical composition and sensory attributes in cocoa liquors: a multivariate profiling strategy (2026) doi | Read as context (not in the table) | Abstract only; relation of compounds to bitterness and astringency in liquors from different regions, not a comparison of processing |
| Comparison of the Effects of Conching Parameters on the Contents of Three Dominant Flavan3-ols, Rheological Properties and Sensory Quality in Chocolate Milk Mass Based on Liquor from Unroasted Cocoa Beans (2021) doi | Excluded | Conching temperature and time varied in a milk-chocolate-type mass from unroasted-bean liquor, but the rating is a taste quality score with no bitterness item |
| Cocoa Shell Infusion: A Promising Application for Added-Value Beverages Based on Cocoa’s Production Coproducts (2023) doi | Excluded | Shell infusions compared by particle size and temperature, not by fermentation, drying or roasting conditions |
| Effects of Particle Size and Extraction Methods on Cocoa Bean Shell Functional Beverage (2019) doi | Excluded | Shell beverages compared by particle size and extraction method; no bitterness intensity item |
| Bitter and astringent characteristics of alkalized cocoa powder: analysis using UPLC-Q-TOF-MS/MS and molecular simulation (2025) doi | Excluded | Abstract only; bitter compounds of cocoa before and after alkalization studied by instruments and computation, and no human comparison of bitterness between treatments can be confirmed from the abstract; full text and the 2026 erratum not checked |
| Structures, sensory activity, and dose/response functions of 2,5-diketopiperazines in roasted cocoa nibs (Theobroma cacao) (2005) doi | Excluded | Abstract only; thresholds of bitter compounds in roasted nibs, with no comparison of roasting degree |
| ビーントゥバーの製法による認知度と嗜好性について (2022) | Excluded | A one-page conference abstract; sweetened chocolates from different roasting temperatures were tasted by people involved in the study, and more people found the hotter roasts bitter, but the number of people and any test are not given |
| 発酵条件の違いがカカオ豆成分やチョコレート風味にあたえる影響の解析 (2025) | Excluded | A one-page conference abstract (72nd meeting of the Japanese Society for Food Science and Technology); fermentation starters and other conditions varied and bitterness was rated, but no bitterness result is described |
| The influence of three different types and dosage of alkaline on the inherent properties in cocoa powder (2020) doi | Full text not reachable | Full text could not be obtained from Europe PMC on 2026-10-22 (empty body), and the publisher page was not tried; whether bitterness was compared has not been checked |
| Volatile organic compounds and sensory profile of dark chocolates made with cocoa beans fermented with Pichia kudriavzevii and Hanseniaspora thailandica (2022) doi | Full text not reachable | Full text could not be obtained from Europe PMC on 2026-10-22 (empty body), and the publisher page was not tried; whether bitterness was compared has not been checked |
Read in the encyclopedia
- Cocoa fermentation — fermenting the pulp and beans before drying
- Cocoa bean drying — reducing bean moisture after fermentation
- Cocoa roasting — heating beans or nibs, defined by time and temperature
- Cocoa bean shell — the material of the shell tisane included here
- Cacao bitterness and astringency — the cacao hub
- Is chocolate with a higher cocoa percentage more bitter? — a comparison of cocoa percentage
Source ledger
Each study's source can be reached from its section above. All of the lab's sources are listed in the source ledger.
How to cite this page
Aoki, M. (NIGAMI LAB) (2026). Does processing cacao change its bitterness?. https://nigamilab.com/en/topics/cocoa-processing-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