Fitting food for physics
Jul 10, 2008 2 comments
Culinary xenophobes take note: your favourite national dish is unlikely to be pushed off the menu by exotic recipes from afar. That’s the conclusion of physicists in Brazil who have applied statistical methods to cookery books from different countries and different eras in the name of interdisciplinary science. The study, published today in New Journal of Physics, also suggests that the average number of ingredients per recipe is similar between countries (about 7–10) and has remained constant over time.
The outcome may surprise foodies in the UK, where the Indian-themed dish “chicken tikka masala” reportedly has displaced the humble “fish ‘n’ chips” as the nation’s favourite dish. Then again, their analysis could be a bit overseasoned.
“Having observed that culinary ingredients and recipes constitute a bipartite network, just like actors and films do, we realized they can be modelled using tools from mathematics and physics,” team-leader Antonio Roque of the University of Sao Paulo told physicsworld.com. “Besides, food is an unusual topic among the physics community but one of high social and cultural importance.”
Roque and co-workers collected data from four classic cookbooks: Pleyn Delit, a collection of medieval recipes; the UK’s New Penguin Cookery Book; Larousse Gastronomique from France; and Dona Benta from Brazil. For each they ranked ingredients according to how often they appeared, and then plotted the number of recipes in which each ingredient appears as a function of decreasing rank (New J Phys 10 073020).
The distributions for each book had long “tails”, reflecting the fact that there are fewer recipes containing low-rank ingredients, and the team was able to fit each one using the same power-law. This suggests that cuisines evolve independently of culture, although with just one book per country and no error bars present it is difficult to judge just how similar this evolution really is.
The researchers also found the same power-law behaviour for three different editions of Dona Benta from 1946, 1969 and 2004, hinting that the structure of Brazilian recipes is not only similar to those in France and the UK but has not changed in the last 50 years — despite the shift from a regional to a more globalized consumer profile during that period.
The traditional culinary ingredients and recipes of a given culture are strongly resistant to replacement pressures from external culturesAntonio Roque, University of Sao Paulo
Clearly it is satisfying for physicists to be able to describe complex behaviour in terms of simple formulae, but at the end of the day these are just statistical fits. Such “scale-invariant” (i.e. power-law) behaviour pops up in countless social systems, such as linguistics, not to mention numerous physical situations such as coastlines and drainage networks. The conclusion of such studies is often a statement of the obvious.
What makes the cookbook analysis interesting is that the Brazilian team was able to reproduce the universal power-law behaviour using a “copy-mutate” algorithm, which describes culinary evolution by a branching process similar to that in biology. The model suggests that idiosyncratic ingredients, for example the edible plant chayote in Central and South American diets, are preserved in cultures just like certain genetic patterns persist across generations. The classic French pot-au-feu and other stews are good examples of “mother recipes” that generate several daughter-recipes by a copy-mutate mechanism.
“Because the system does not forget the initial conditions, the traditional culinary ingredients and recipes of a given culture are strongly resistant to replacement pressures from external cultures,” says Roque. “The model may help biologists model speciation, for example by making an analogy between recipes/ingredients and species/phenotypes.”
It is very difficult to define ingredients — is sugar an ingredient, for example?Hervé This, College de France
Hervé This, a physical chemist at the College de France in Paris who cofounded the science of molecular gastronomy, thinks the idea and methods of the work are good but finds its conclusions hard to swallow. “It is very difficult to define ingredients – is sugar an ingredient, for example, and what about the carrots used to make the stock for a recipe?” He also says we should recognize cookery books as literature, not necessarily as reliable gastronomic databases. That’s as clear as consommé to anyone who has attempted to prepare an evening meal from the Larousse.
The team now plans to consider other cookery books and to apply its evolutionary algorithm to different problems.
About the author
Matthew Chalmers is a writer based in Bristol, UK