Copenhagen MG seminar: Flavor pairing (part 2)

Wender Bredie presented results from experiments designed to test the flavor pairing hypothesis

A topic that I was particularily excited to hear about at the molecular gastronomy seminar in Copenhagen was flavor pairing. Since Heston Blumenthal presented his white chocolate and caviar combination based on amines in 2002 and Francois Benzi of Firmenich the pork liver-jasmine combination based on indole the idea has been further elaborated by Bernard Lahousse and Lieven De Couvreur who launched the foodpairing website and by me in the TGRWT food blogging event. Despite the interest and fascination it is fair to say the flavor pairing is still controversial – see for instance the discussion with in particular Jorge Ruiz. What is clearly lacking in the field is a more stringent scientific approach (as well as someone with time, interest, a sensory panel and the money to finance the activities…). It was therefore great to hear that sensory science professor Wender Bredie together with PhD student Ditte Hartvig actually set out to test the flavor pairing hypothesis formulated as: if major volatiles are shared between two foods it may very well be that they go well together. To achieve this they used a sensory panel to assess the odor of food pairs mixed and unmixed. Bredie proposed that a hyper addition of odor intensities would perhaps be the holy grail of flavor pairing – that is if the intensity of the mixed odors would be more than the sum of the unmixed intensities. Or even better: if there would be a hyper additive effect on pleasantness

A finding in their initial study with 8 pure volatiles smelled alone and in pairs was that the total odor intensity only increased when combining up to 2-3 components. This is in accordance with Fechner’s law which states that it takes a larger difference in physical stimuli to give the same rise difference in perceived intensity. The pleasentness of the individual components decreased when mixed. In a further study they prepared 53 binary mixtures from 19 food odors. The recorded responses were: intensity, pleasantness, complexity, “harmonic” and novelty. They found no real hyper additive effects for intensity – in 91% of the cases the mixture was perceived to be more intense than one compound perceived alone, but less intense than the other (= compromise region of hypo-addition). But turning to pleasantness there were a couple of interesting findings which were presented as the good, the bad and the ugly combinations, based on the pleasantness score from the sensory panel:

The good
lemon peel + butter
cinnamon + apple
ginger + apple
cinnamon + peanut butter
malt + peanut butter
cacao + malt
lemon peel + strawberry

The bad
ginger + blue cheese
caviar + apple
basil + caviar
vanilla + blue cheese
anis seed + basil
basil + green tea
anis seed + caviar

The ugly
cinnamon + blue cheese
anis seed + garlic
cacao + garlic
cinnamon + garlic
malt + blue cheese
caviar + blue cheese
malt + caviar

To test the flavor pairing hypothesis they then took the pairings and analyzed the overlap of volatiles using data from the VCF database. They found absolutely no correlation between the hedonic score of the food pairings and the overlap of volatile flavors! But knowing that many (if not most) of the volatiles compounds found in foods do not contribute to the aroma this is not very surprising. A further interesting test was to evaluate pleasantness as a function of complexity, but the conclusion in short was that one cannot use complexity to predict hedonic response.

Wender concluded that flavor pairing should be the subject of further and more elaborate studies, for instance at lower concentrations (in the so-called hyper additive region where the intensity increases exponentially with increasing concentration). Another approach could be to study sensory (dis)similarity rather than chemical similarity. One could also explore the arousal potential in pairs of dissimilar but liked flavours. He also suggested that one should keep on searching for additive hedonic responses. And he finished his presentation with the following quote:

“Flavour pairing is like painting in the dark… you do not really know what you are creating, but at daylight you may discover that you have moved art beyond imagination”.

I’m really happy that the topic has been brought into the scientific community – and I’m very much looking forward to see the results published. At the same time is has also become very clear to me that the term flavor pairing needs some clarification. I’ll discuss that in a separate post.


  1. The whole concept of flavour pairing reminds me very much of astronomy, meticulous scientific observations turned into a system of belief. And the disciples are very reluctant to test the theories. If there were any truth in the hypothesis a lot of famous classic flavour combinations would consist of ingredients similar aromatic profiles. They do not.

    • Lisa: I guess you mean astrology? 🙂 I can see your point, but as I’ve emphasized in many of my previous posts – this is and will always remain a niche tool to discover a couple of surprising combinations that work quite well, the reason for that being that a similar aroma. Most of what we do in the kitchen however is not based on this kind of flavor pairing!

  2. To me , the advancement of flavour pairing is an inherently good thing but at the same time, won’t it ever so slightly ruin some of the fun of the kitchen? Maybe that would only be the case if it expanded from being a niche tool.

  3. Has Bredie published anything about his study yet? I’d be interested to see the results. In particular, if pleasantness is related to intensity of individual components, then we know that mixture supression causes individual intensities to be reduced in mixtures. This is a known, well studied psychological effect. Therefore it is no surprise that hedonics of individual components goes down in mixtures.

    A hypo additivity effect was also previously observed at in a study done by Harry Lawless for the US Army. He looked at barter systems for multiple food items. Soldiers were trading “desert bars” for individual and multiple MRE components. Using a “desert bar scale”, Harry found that soldiers would discount combinations of components (1 for 3, 2 for 5, etc.) and proposed that perhaps liking isn’t so well represented on traditional 9 point scales. Some clever chefs intuitively know they get more money from the individual for a la carte items than meal combinations (app/entree/dessert combinations) (although the economics is more complex if your tasting menu helps get people to come in the first place)

    It is good to see more people looking at flavor pairings. I have been working with the Army on developing new methodologies for deciding what field ration menu items to combine together to create more likable combinations as meals (not just pairs). After all, how often do we just pair two flavors together? Isn’t there often much more going on than that? The problem is in combinations – if we have 25 flavors to choose from, there are millions of flavor combinations between 1-8 components. Any sort of analysis of all possible combinations is a previously intractable problem 🙂

Leave a Reply

Your email address will not be published. Required fields are marked *

The reCAPTCHA verification period has expired. Please reload the page.