Why some foods go well together and some not!
Sebastian Ahnert gave an interesting talk on the Flavor Network 2.0 in Barabasi Lab. Back in 2011, Sebastian and his colleagues published a paper in Scientific Reports of Nature on scientific aspect of culinary practice. The paper examines the hypothesis of whether there are any general patterns governing proper combination of ingredients. In other words, is there any universal rule which determines the tastiness of two foods combining with each other?
To begin with, a bipartite network of foods and flavor compounds was developed which from a weighted food-food network was derived. Since the obtained network was too dense, the backbone of the network was extracted to have a better visualization of the network.
In order to test the hypothesis, a reasonable number of recipes have been pulled altogether to derive a ingredient-ingredient network from, assuming ingredient pairs appearing in a recipe means those ingredients go well together. Comparing two ingredient-ingredient networks - one developed based on shared flavor compounds, the other based on recipes - shows there is a tendency for ingredients in a recipe to share flavor compounds which is more significant for certain cuisines.
While Western cuisines seem to have a tendency to use ingredient pairs with many common flavor compounds, East Asian cuisines tend to avoid compound sharing ingredients. The paper enumerated several reasons for observed disagreement: 1) Asian cuisines may rely more on the effect of ingredient combinations rather than compound sharing; 2) umami plays an important role in Asian cuisines which is not quite determinant in Western cuisines; 3) recipes are usually problematic data sets; 4) there are cultural biases in the compound-ingredient data. Any of these reasons may undermine the observed contrast between Western and Asian cuisines.
As a new component added to this on-going project, Sebastian started developing Flavor Network 2.0 by collecting new data. While the first phase of the study was a binary version of associations, he started incorporating flavor compound concentration data into the picture. Flavor compounds have different concentrations in the food space. They also possess different thresholds to be perceived by human body. Using concentration and threshold value, he calculated the Odor Activity Value (OAV) as a measure for quantifying the perceivablity. Taking the OAV into account, he showed that the observed generic pattern is even more significant. Finally, he developed a bipartite network for olfactory receptors-flavor compounds, leading to a tripartite network of ingredients, compounds, and receptors. Sebastian is working on testing whether there is a general rule for ingredients whose compounds share common receptors.
If you would like to know more about Sebastian's research, visit his website above, or follow him on Twitter!