Thursday 19 January 2012

The mathematics of flavour


We travel to enjoy new experiences and in particular the cuisines of other cultures. No two cuisines are exactly alike, even when the countries have a similar ethnic heritage. Some differences are obvious – eating lamb rather than beef or rice rather than bread, for example. Even when a cuisine’s ingredients are familiar, we can be bemused by the flavour combinations that the locals take for granted. In Japan, a very popular snack combines pretzels and chocolate; in the USA, pumpkin can be found inside sweet pies, and in parts of both Scotland and New Zealand, chocolate bars are sometimes covered in batter and deep-fried. But sometimes the animal and plant species consumed are totally foreign, presenting a challenge for all but the most adventurous. Exactly what is it that defines a cuisine and distinguishes one cuisine from another?

In the early 1980s, Elizabeth Rozin published “The Flavour Principle Cookbook” [1] in which the characteristic dishes of different cuisines were described in terms of unique combinations of specific ingredients used across a variety of foods within a culture. For example, a characteristic combination of ingredients in Japanese cooking is soy sauce, mirin (sweet rice wine) and dashi (a stock made from flakes of the bonito fish). While Korea is geographically close to Japan, its flavour principles are very different, with the intense flavours of garlic, chilli, sesame and soy dominating many dishes. In Moroccan cooking, cinnamon, cumin, tumeric, ginger, paprika, coriander, saffron, anise, cardamom and other spices are combined into Ras El Hanout, a universal blend that renders the chicken, beef or lamb that it flavours characteristically Moroccan. 

Together with her husband Paul Rozin, a noted researcher in the psychology of food, Elizabeth Rozin pointed out that these combinations of different ingredients provide a characteristic flavour that unites foods within a culture, and identifies them as originating from that culture [2]. Such a characteristic and preferred combination of flavourings may also assist in ensuring that a culture can be flexible in the range of foods that it consumes, thus maintaining dietary variety and an adequate intake of nutrients. A problem with incorporating new foods into diets lies in our tendency to reject those foods that are unfamiliar, a phenomenon known as food neophobia. Since food neophobia is largely based on fear of a food tasting unpleasant, using a familiar combination of flavours is a way to address this problem. This explains how whole cultures can successfully incorporate new foods, if circumstances suddenly change. If your staple meat is chicken, which you fry with vegetable oil, garlic, chilli and lemongrass, becoming vegetarian is not the only option if chicken is unavailable for any reason. The flavour principle – that is, frying in this particular combination of spices – allows a newly chickenless culture to start to consider other sources of animal protein without worrying if it will be rejected [3].

Our understanding of flavour principles has been largely anecdotal, gleaned as much from the information provided by chefs and books of recipes as anything else. In particular, there has been little understanding of the underlying principles that distinguish cultures on the basis of enjoying particular flavour combination and avoiding others? So, are ethnic flavour combinations merely the result of availability of particular ingredients or is there something more fundamental underlying flavour choices? 

It is often assumed – without any real evidence - that ingredients with similar flavours are more likely to be combined because of their compatibility than ingredients with contrasting flavours. Now, for the first time, quantitative research has addressed this question. Yong-Yeol Ahn of the USA’s Northeastern University, together with colleagues from Harvard, Indiana (USA) and Cambridge (UK) universities, recently published the results of a study [4] in which a rather complex mathematical method was used to explore the links between a set of 381 ingredients used is a variety of the world’s cuisines and the more than a thousand chemical compounds that give these ingredients their characteristic flavour. The basis of their analysis was the fact that many ingredients share one or more such compounds and so the number of shared compounds can be used as a measure of the similarity of different ingredients to one another. For example, the chemical compound 4-methylpenanoic acid contributes to the flavour of both tomatoes and Parmesan cheese.

The researchers asked if different cuisines were more or less likely in recipes to use ingredients that are linked strongly within this network of flavour similarities. To answer this question, they used databases containing more than 50,000 recipes representing a range of the world’s cuisines including those from North America, Western and Southern Europe, Latin America, and East Asia.

They found that cuisines from North America and Western Europe cuisines show a strong tendency towards recipes using ingredients that share flavour compounds. In contrast, both East Asian and Southern European cuisines are characterised by recipes with ingredients using distinctly different flavour compounds. Latin American cuisine tended to fall into the middle ground. This finding seems very consistent with ideas from both Japanese and Chinese cuisines in which the balance of distinctly different flavour qualities is emphasised. 

Digging into this unusually rich data set further, the researchers were able to discover that cultural differences in this “shared compound effect” did not involve the majority of ingredients used by the different cuisines. Rather, each cuisine was dominated by a small set of key ingredients that were very frequently used. They noted, for example, that around 75% of all North American recipes contained 13 key ingredients, with this list dominated by dairy products, eggs and wheat. East Asian cuisines on the other hand rely heavily on plant derivatives such as soy sauce, sesame oil, and rice and ginger. An important consequence of this study was therefore to provide empirical support for the Rozins’ “flavour principles”. 

Because of the ways in which Ahn and colleagues studied the inter-relationships of ingredient combinations across cultures, they were also able to investigate the similarity of flavour principles between different cultures. While we can all do this on an anecdotal basis – we know that German and Dutch dishes are more similar than German and Turkish dishes, for example – their data analysis provides a more scientific basis on which to make such judgments. Hence, they found, as expected, that North American and Western European cuisines were closely related, but also observed that in terms of signature ingredient combinations, Southern European cuisine is much closer to Latin American than it is to Western European cuisine.

Another intriguing finding from these data was to do with the complexity of different cuisines. Bookshops these days are filled with cookbooks containing supposedly authentic national, regional and sub-regional cuisines from all over the world. These give the impression both that many “authentic” dishes require sometimes dozens of different ingredients, and also that there are large cultural differences in cuisine complexity. We do not think it is surprising that French recipes tend to have many ingredients, more than a typical Brazilian recipe, for example.  But the final word on this comes from the flavour study. In their analysis of the huge number of recipes, they noted variation in the number of ingredients per recipe but in fact it is typically only a handful of “authentic” ingredients that distinguish a stew from a cassoulet or a curry.



1.      Rozin, E., Ethnic Cuisine: The Flavor-Principle Cookbook.1983, Brattleboro: The Stephen Greene Press.
2.      Rozin, E. and P. Rozin, Culinary themes and variations. Natural History, 1981. 90(2): p. 6-14.
3.      A more detailed account of flavour principles and food neophobia can be found in Taste Matters. Why we like the foods we do. 2012, London: Reaktion Books.
4.      Ahn, Y.-Y., et al., Flavor network and the principles of food pairing Science Reports, 2011. 1(196): p. 1-7.