Tikalon Header Blog Logo

Flavor Networks

December 2, 2011

My longtime employer nicely provided a fitness center for its employees. Mid-day, I would exercise on a treadmill while watching television. Since daytime television is uniformly awful, I eventually settled on watching the Food Network. My exercise time coincided with a show about desserts hosted by Debbi Fields of the famous Mrs. Fields cookies.

In watching that show, I saw that most of the dessert recipes contained a set of common ingredients, such as flour, sugar, butter, oil, milk and eggs. To make a new recipe, you just needed to vary the proportions of these and add a pinch of some other ingredients, such as cinnamon, vanilla, or chocolate.

If I had been adventuresome, I would have written a computer program to randomly generate dessert recipes, baked them up, and seen how they tasted. It would have been a Monte Carlo calculation, with ingredient distributions derived from existing recipes. Alas, all this eating would have been contrary to the idea of my being at the fitness center in the first place.

No scientific idea is lost forever. It's generally rediscovered by another scientist. An eclectic group of scientists from the Department of Physics, Northeastern University (Boston, MA), the Dana-Farber Cancer Institute, Harvard University (Boston, MA), the School of Informatics and Computing, Indiana University (Bloomington, IN), and the Cavendish Laboratory, Cambridge University (Cambridge, UK) has recently published a paper on the arXiv Preprint Server that analyzes the connections between food ingredients in western and non-western foods.

The number of potential food recipes is huge. If a typical recipe has just eight ingredients, there would be about 1015 possible combinations using the 300 most common ingredients. The authors point out that this could still be an underestimate, since one authoritative source lists 1,000 ingredients,[2] and there could be ten ingredients per recipe, for a total of 1023 combinations. Obviously, humans are exploiting just a small fraction of the "culinary space," although most of these combinations would be quite unpalatable.

The authors find that western cuisines use ingredient pairs that share many flavor compounds. This finding supports what's called the "food pairing" hypothesis that some foods seem to naturally go together. A strange example of this is the discovery that chocolate and cauliflower taste good together.[3] East Asian cuisines, however, avoid such flavor pairing.

So, what ingredients are used most often? The following table shows the most popular food ingredients for North American and East Asian cuisine.

Contributing ingredients to North American and East Asian cuisine. The numbers are the relative contributions of each ingredient to the total cuisine, as defined in Ref. 1. These are ranked in order of greatest use from top to bottom.[1]
North AmericanΧiEast AsianΧi
milk0.529rice0.294
butter0.511red bean0.152
cocoa0.377milk0.055
vanilla0.239green tea0.041
cream0.154butter0.041
cream cheese0.154peanut0.038
egg0.151mung bean0.036
peanut butter0.136egg0.033
strawberry0.106brown rice0.031
cheddar cheese0.098nut0.024
orange0.095mushroom0.022
lemon0.095orange0.016
coffee0.085soybean0.015
cranberry0.07cinnamon0.014
lime0.065enokiatke0.013
tomato-0.168beef-0.2498
white wine-0.0556ginger-0.1032
beef-0.0544pork-0.0987
onion-0.0524cayenne-0.0686
chicken-0.0498chicken-0.0662
tamarind-0.0427onion-0.0541
vinegar-0.0396fish-0.0458
pepper-0.0356bell pepper-0.0414
pork-0.0332sesame seed-0.041
celery-0.0329black pepper-0.0409
bell pepper-0.0306shrimp-0.0408
red wine-0.0271shiitake-0.0329
black pepper-0.0248garlic-0.0302
parsley-0.0217carrot-0.0261
Parmesan cheese-0.0197tomato-0.0246

Network-oriented papers always have some very pretty figures, such as the one shown below. See the caption of the original figure for an explanation.[1] Network of flavors

This figure shows the connections between food ingredients. No, you're not expected to read the tiny characters. Click on the image for a more readable version, and see the caption of the original figure for an explanation. (via arXiv Preprint Server, Fig. S4 of Ref. 1).


References:

  1. Yong-Yeol Ahn, Sebastian E. Ahnert, James P. Bagrow, Albert-László Barabási, "Flavor network and the principles of food pairing," arXiv Preprint Server, November 25, 2011.
  2. O. Kinouchi, R.W. Diez-Garcia, A.J. Holanda, P. Zambianchi and A.C. Roque, "The non-equilibrium nature of culinary evolution," New Journal of Physics, vol. 10, Document No. 073020 (July, 2008).
  3. Martin Lersch, "Flavor Pairing," Khymos Blog

Permanent Link to this article

Linked Keywords: Treadmill; television; Food Network; dessert; Debbi Fields; Mrs. Fields; recipe; ingredients; flour; sugar; butter; oil; milk; egg; cinnamon; vanilla; chocolate; computer program; randomness; random; Monte Carlo calculation; scientific; scientist; Department of Physics; Northeastern University (Boston, MA); Dana-Farber Cancer Institute; Harvard University (Boston, MA); School of Informatics and Computing; Indiana University (Bloomington, IN); Cavendish Laboratory; Cambridge University (Cambridge, UK); arXiv Preprint Server; western; non-western; western cuisine; flavor; compound; food pairing; chocolate; cauliflower; East Asian cuisine; milk; rice; butter; red bean; cocoa; vanilla; green tea; cream<; cream cheese; peanut; egg; mung bean; peanut butter; strawberry; brown rice; cheddar cheese; nut; orange; mushroom; lemon; coffee; soybean; cranberry; cinnamon; lime; enokiatke; tomato; beef; white wine; ginger; pork; onion; cayenne; chicken; tamarind; vinegar; fish; pepper; bell pepper; sesame seed; celery; shrimp; red wine; shiitake; garlic; parsley; carrot; Parmigiano-Reggiano cheese; tomato.