Indian food uses many spices because the cuisine's flavour logic depends on them: an IIT Jodhpur analysis of 2,543 recipes found that Indian dishes combine ingredients that share fewer aroma compounds than chance would predict, and spices are the main reason. The popular explanation, that hot countries use spices to kill food-borne germs, does not survive a 2021 study of 33,750 recipes from 70 cuisines, which found no link with temperature once shared history and geography were accounted for.
- 2,543 recipes from eight regional cuisines all showed negative food pairing: unlike flavours, combined on purpose.
- Spices were the most strongly contrasting category in seven of the eight cuisines.
- Red chilli is the single biggest contrast driver; milk and butter pull the other way.
- 33,750 recipes from 70 cuisines found no temperature link once shared history was corrected (p = 0.844).
- 53% of the variation in spice use is explained by which cuisines are neighbours or relatives.
So the honest answer has two halves:
- The mechanism is well measured. Spices let Indian cooks build dishes out of contrasting flavours rather than matching ones.
- The origin is much less certain. The evidence now points more to culture, availability and copying between neighbouring cuisines than to a hidden public-health strategy.
This post goes through the numbers behind both halves.
How many spices does a typical Indian recipe use?
A typical Indian recipe has between 3 and 12 ingredients, and spices are the most frequently used category in it. That is the size range Anupam Jain, Rakhi N K and Ganesh Bagler found when they mined 2,543 recipes from eight regional cuisines on TarlaDalal.com in December 2014, mapping 588 ingredient names down to 194 ingredients with known flavour profiles.
Ganesh Bagler trained in physics and computer science and worked in computational biology, including a postdoc at the Max Planck Institute for Molecular Genetics in Berlin, before returning to India in 2010. He has said his move into food happened by accident, while teaching a class, and was not planned at all.
The trigger was a 2011 paper that tested chef Heston Blumenthal's food-pairing idea on world recipes. At IIT Jodhpur, Bagler and his students took the late Tarla Dalal's recipes, 2,543 of them, sorted the ingredients into categories and asked what happened when one was swapped. Swapping spinach for fenugreek in palak paneer barely moved the flavour profile; swapping turmeric for cinnamon changed the dish's essence. The 2015 preprint was picked up by MIT Technology Review, and Bagler later called spices the "molecular fulcrum" of Indian food.
The National (Tania Bhattacharya): The Indian academic making the world look at flavours and food in a fresh way (Apr 2020); Food Innovation Quarterly: interview with Ganesh BaglerSpice hiding inside blends
Spice use also hides inside compound ingredients. The team counted ready-made blends as single ingredients and listed how often each appeared, which shows how much spice is pre-packed into the base of Indian cooking.
| Derived ingredient | What it contains (as mapped in the study) | Recipes using it |
|---|---|---|
| Ginger garlic paste | Ginger, garlic | 166 |
| Coriander cumin seed powder | Coriander, cumin | 87 |
| Chaat masala | Asafoetida, mango, black salt, cayenne, garlic, ginger, sesame, mustard, turmeric, coriander, bay leaf, star anise, fennel | 86 |
| Sambar powder | Pigeon pea, coriander, chickpea, cumin, black pepper, cayenne, ginger, fenugreek, turmeric | 22 |
| Chole masala | Cayenne, garlic, ginger, sesame, mustard, turmeric, coriander, bay leaf, star anise, fennel | 15 |
| Rasam powder | Cayenne, pigeon pea, cumin, coriander, black pepper, curry leaf | 11 |
| Tandoori masala | Garlic, ginger, clove, nutmeg, mace, cumin, coriander, fenugreek, cinnamon, cardamom, black pepper | 8 |
One spoon of chaat masala, in other words, puts a dozen ingredients into a dish. The study's dataset breaks down unevenly by region: Punjabi 1,013 recipes (152 ingredients), South Indian 474, Jain 447, Gujarati 392, Mughlai 179, Bengali 156, Maharashtrian 130 and Rajasthani 126 (78 ingredients).
See the data as a table
| Recipes | Distinct ingredients | |
|---|---|---|
| Punjabi | 1013 | 152 |
| South Indian | 474 | 114 |
| Jain | 447 | 138 |
| Gujarati | 392 | 112 |
| Mughlai | 179 | 105 |
| Bengali | 156 | 102 |
| Maharashtrian | 130 | 93 |
| Rajasthani | 126 | 78 |
Source: Jain, Rakhi and Bagler 2015, PLOS ONE, Table 1 · Recipes from TarlaDalal.com, collected December 2014.

What is food pairing, and why is Indian food different?
Food pairing is the idea that two ingredients taste good together if they share volatile flavour compounds, and Indian cuisine does the opposite: its ingredient pairs share fewer compounds than random recipes would. The idea comes from chefs who noticed that ingredients with overlapping aroma molecules often work together.
The 2011 flavour network
In 2011 Yong-Yeol Ahn, Albert-László Barabási and colleagues built a flavour network linking ingredients by shared compounds. North American and Western European recipes tend to combine ingredients that share many compounds, while East Asian recipes tend to avoid it. India was not in that analysis. Bagler's group filled the gap.
How negative pairing is measured
- Pair up. For every pair of ingredients in a recipe, count the flavour compounds they share.
- Average. Average across the recipe, then across the cuisine (the score Ns).
- Randomise. Build thousands of fake cuisines with the same recipe sizes and compute the same score.
- Compare. If real recipes share fewer compounds than random ones (ΔNs below zero), the cuisine shows negative food pairing.
2,543 recipes from eight regional cuisines were compared with four randomised controls. All eight showed significantly negative food pairing, from Mughlai (ΔNs = −0.758, Z = −10.2) to Maharashtrian (ΔNs = −4.523, Z = −52.0), and the effect held across recipe sizes. Only controls that preserved how often each ingredient is used reproduced the pattern; preserving category mix alone did not. Spices were the most significantly negative category in every cuisine except Mughlai. The limit: one recipe website, flavour data for only 194 of 339 ingredients, and shared compounds are a proxy for taste, not a taste test.
Read the paper (free)Which spices push Indian food towards contrast?
Cayenne (red chilli) is the single biggest driver of negative food pairing in Indian cooking, followed by green bell pepper, coriander, garam masala and tamarind.
The earlier, whole-cuisine version of the IIT Jodhpur work, posted on arXiv in 2015, ranked ingredients by how much their presence shifts a cuisine's average flavour sharing (a score called χ). Nine of the ten strongest negative contributors were spices or spice-like ingredients. The positive side was dominated by dairy and staples.
See the data as a table
| χ contribution | |
|---|---|
| Milk | 0.336 |
| Butter | 0.315 |
| Bread | 0.113 |
| Rice | 0.087 |
| Cottage cheese | 0.074 |
| Corn | 0.071 |
| Cinnamon | -0.044 |
| Clove | -0.046 |
| Ginger | -0.047 |
| Tamarind | -0.059 |
| Garam masala | -0.067 |
| Coriander | -0.078 |
| Green bell pepper | -0.134 |
| Cayenne | -0.139 |
Source: Jain, Rakhi N K and Bagler, Spices form the basis of food pairing in Indian cuisine, arXiv:1502.03815, Table S3 · Positive values push recipes towards shared flavour compounds; negative values push them towards contrast. Cayenne appeared in 1,179 recipes.
Read the chart as a balance. A dish built on milk, butter and rice drifts towards flavour harmony, the Western pattern. Add red chilli, coriander and garam masala and it swings hard towards contrast.
Most Indian savoury dishes carry several of the contrast-side ingredients at once, which is why they taste layered rather than uniform.
The Mughlai exception
Mughlai food is the telling exception. It had the largest recipes, a bimodal size distribution and the weakest negative pairing, and spices were not its most negative category.
Its heavy use of cream, ghee, nuts and yoghurt pulls it back towards the dairy side of the chart, which fits its origin in a royal kitchen with different raw materials.

Do hot countries use spices to kill germs?
The germ theory is the most famous explanation, and the best recent evidence says it does not hold up. Its best-known version is a 1998 review by J. Billing and P. W. Sherman in the Quarterly Review of Biology, and it was tested with proper controls only in 2021.
The 1998 case for germs
Billing and Sherman analysed 4,578 meat-based recipes from 93 traditional cookbooks in 36 countries, tracking 43 spices. In hotter countries:
- more recipes contained spices;
- recipes used more of them;
- the most strongly antibacterial spices were used more often.
They concluded that flavour is the immediate reason people use spices but that the deeper reason is protection from food-borne pathogens.
The flaw: Galton's problem
The flaw is statistical, and it has a name: Galton's problem. Neighbouring and related cultures share recipes, climates and many other traits, so they are not independent data points. Lindell Bromham's group at the Australian National University rebuilt the test with methods from evolutionary biology that correct for shared ancestry and proximity.
The team combined five datasets into 33,750 recipes from 70 cuisines using 93 spices, including the eight Indian regional cuisines (2,761 recipes, 45 spices). Proximity and relatedness explained 53% of the variation in mean spice use; after correcting for them, temperature had no significant effect (β = 0.03, p = 0.844, Bayes factor for no effect 5.9), and historical pathogen load was not supported either (p = 0.053). Spice use did track GDP per capita, life expectancy and even road deaths, which predicted spice use better than food-borne illness. The limit: recipe collections are not measurements of what people actually eat, and the result shows association, not cause.
Read the paper (open copy)Bromham's point is not that spices have no antimicrobial effect; many do in a dish. It is that the global pattern in spice use is better explained by a bundle of socio-economic factors that travel together than by an evolved defence against spoilage.
Lower-income regions use more spices per recipe. They also have more food-borne illness, more road deaths and lower life expectancy, and the authors note that none of these causes the others.
Road deaths predicted spice use better than food-borne illness did. A correlation that strong, with no causal story, is a warning against tidy explanations.
A cancer-prevention version of the idea has also been tested. A 2023 study in Evolution, Medicine, and Public Health compared spice consumption with age-standardised gastrointestinal cancer rates across 36 countries and found almost no relationship.
Which explanations survive the data?
The data support a mechanism (spices create contrast) and an inheritance pattern (neighbouring cuisines copy each other), but not a single functional purpose. The table sets out the main hypotheses against what the studies found.
| Hypothesis | Claim | Key evidence | Verdict on current data |
|---|---|---|---|
| Antimicrobial | Hot climates adopted spices to stop spoilage | Billing and Sherman 1998 found a temperature link in 36 countries; Bromham 2021 found it vanishes after correcting for shared history (p = 0.844) | Weak |
| Flavour contrast | Indian cooking is built on ingredients with unlike flavours | Negative pairing in all 8 regional cuisines; spices the leading negative category (Jain et al. 2015) | Strong as a description of how it works |
| Cultural copying | Recipes evolve by copying and small changes from neighbours | Jain et al.'s copy-mutate model reproduced Indian ingredient use and negative pairing; Bromham found proximity and relatedness explain 53% of spice variation | Supported |
| Availability | Cuisines use the spices that grow nearby | Bromham tested plant and crop diversity and spice-plant ranges and found no significant effect on spice use | Not supported globally, though trade history matters for India |
| Socio-economic | Spice use tracks poverty and health outcomes | GDP per capita predicts spice use as well as food-borne illness does (Bayes factor 1.8) | An association, with no clear causal story |
| Medicinal or cancer prevention | Spices persisted because they kept people healthy | No link with GI cancer rates in 36 countries (Dujon et al. 2023) | Not supported at population level |
So why does India in particular use so many spices?
Put the findings together and a plausible account emerges: India's cuisines inherited a large spice vocabulary from each other and from centuries of trade, and that vocabulary became the grammar of the food.
- Copying explains why neighbouring regions share it.
- Contrast-building explains why it keeps working on the plate.
Trade is in the ingredient list
Black pepper, cardamom and cinnamon are Old World spices with long histories on the Malabar coast, while the chilli that now tops the χ ranking is a New World plant. A cuisine that could absorb a foreign fruit so completely that it became its single most important flavour driver is clearly one that runs on spice. The full story is in the history of chilli in India.
Cheap, local and traded at scale
The economics still point the same way. Chilli was 27% of the value of India's spice exports in 2025-26 and cumin 12%, according to the Spices Board, and India exported 17.34 lakh tonnes of spices that year. A country that grows and trades spices at that scale has cheap access to them, which keeps them in everyday cooking rather than reserving them for feasts.
What the data does not show
What I would not claim is that grandmothers designed masalas as medicine. Bagler's group itself suggests the negative pairing may carry the signature of a health-centred food tradition, but that is a hypothesis in the discussion section, not a result.
The measured findings are narrower and more useful: spices are what make Indian flavour combinations contrast, and the blends that bundle them, such as garam masala and sambar powder, are how cooks reproduce that contrast reliably. If you are unsure what separates two of those blends, read garam masala vs curry powder.

Can the data predict a good new spice combination?
Partly: the same flavour databases now drive recipe-generation tools, though taste tests remain the bottleneck. Bagler moved to IIIT-Delhi, where his Complex Systems Lab built FlavorDB and RecipeDB.
A PhD thesis from the lab, submitted in August 2025, describes RecipeDB2, which links recipes, ingredients and flavour molecules, and uses language models for recipe generation and flavour analysis.
For more on the food science behind this, the Rasa world page collects the rest of the series.
What's new
- IIIT-Delhi thesis on computational gastronomy builds RecipeDB2 linking recipes to flavour molecules IIIT-Delhi repository, Aug 2025
- Spice exports reach 17.34 lakh tonnes worth ₹39,140 crore in 2025-26, with chilli the largest item Spices Board India, 2026
- Bagler's lab releases an AI recipe tool trained on recipes from 74 countries ThePrint, Aug 2024
Questions people ask
Is Indian food spicy because of the hot climate?
Not on current evidence. Temperature does correlate with spice use, but a 2021 study of 70 cuisines found the link disappears once you account for the fact that neighbouring and related cultures share recipes and climates.
Why does Indian food taste so complex?
Because it pairs ingredients with unlike flavour compounds. In 2,543 recipes, Indian dishes shared fewer compounds per ingredient pair than randomised recipes, and spices such as chilli, coriander and garam masala drove that contrast.
Do spices really kill bacteria in food?
Some do in laboratory tests, and Billing and Sherman's 1998 review listed many with antibacterial activity. That does not mean cuisines adopted them for that reason, and spices are not a substitute for cooking and storing food safely.
Which spice matters most in Indian cooking?
By frequency and by effect on flavour pairing, red chilli. It appeared in 1,179 of the recipes in the IIT Jodhpur dataset and had the strongest negative pairing score.
Sources
- Jain A., Rakhi N K, Bagler G., Analysis of Food Pairing in Regional Cuisines of India, PLOS ONE (2015)
- Jain A., Rakhi N K, Bagler G., Spices form the basis of food pairing in Indian cuisine, arXiv:1502.03815 (2015)
- Ahn Y.-Y. et al., Flavor network and the principles of food pairing, arXiv:1111.6074 (Scientific Reports) (2011)
- Bromham L. et al., There is little evidence that spicy food in hot countries is an adaptation to reducing infection risk, Nature Human Behaviour 5 (2021)
- Billing J., Sherman P.W., Antimicrobial functions of spices: why some like it hot, Quarterly Review of Biology 73(1) (1998)
- Dujon A.M. et al., No evidence that spice consumption is a cancer prevention mechanism in human populations, Evolution, Medicine, and Public Health 11 (2023)
- Spices Board India: export performance 2025-26 (2026)
- Spices Board India: item-wise export shares (2026)
- IIIT-Delhi repository: Mansi Goel, PhD thesis on computational gastronomy (2025)
- ThePrint (PTI): Prof, scholar develop AI tool that thinks like a chef (2024)
- The National (Tania Bhattacharya): The Indian academic making the world look at flavours and food in a fresh way (2020) (Bagler's background, the Tarla Dalal recipes and the swap examples in the story)
- Food Innovation Quarterly: interview with Ganesh Bagler (c. 2023) (his own account of the unplanned start)