by Kristen Rasmussen
All proper scientific evaluation requires objectivity and sensory science is no exception. This is difficult because our senses are, in fact, very subjective. Think about the last time you tried a new food – maybe it was a tropical fruit on vacation or an unfamiliar pseudo-grain. Whether or not your palate accepted that novel food depends on many factors—some that can be measured chemically and physically, such as taste, smell, and touch, but other factors that are harder to quantify, like culture and past experiences, also play a role.
The projects conducted at Nordic Food Lab always include an element of evaluating and/or attempting to modulate specific flavour profiles. As a culinary-minded nutritionist, I like to point out that beyond the benefits of enjoying food, no food is nutritious unless it is eaten. I firmly believe that we should enjoy all food that passes our lips. For this reason I would argue that any research involving food, even research without a culinary focus, should similarly consider and appreciate the importance of flavour. But how does one measure something as nuanced as flavour?
Because sensory experience can be so personal and subjective, the ‘gold standard’ of sensory testing involves panels trained to be as objective as possible in a particular descriptive analysis test. However, this process is notoriously costly, time-prohibitive, and difficult to analyze, leading to a search for other testing methods that are faster and cheaper, while still producing robust and repeatable results. During my stint as visiting researcher at NFL, I spent the majority of my time conducting a sensory study to analyse the validity of one such alternative process. The purpose of the study was to capture the sensory differences in a large set of spice blends and pastes using two fast sensory methods, Napping® and Ultra-Flash Profiling (UFP). The results could then be analyzed to determine the adequacy of the sensory evaluation methods.
The sample set of 29 aromatic blends
‘Aromatic blend’ samples were chosen for the study in order to include a variety of traditional blends from around the world in addition to several samples from NFL’s work, such as Juniper Ant Paste and Peaso. Consequently, the samples represent, though incompletely, the large differences that exist between mixes of different flavourful ingredients from a range of cultures and approaches to ‘aromatic blends’.
Preparing the aromatic spice blends
Napping® is a sensory method where participants are presented all samples at one time and asked to arrange them on X,Y coordinates of a sheet, placing samples closer together that are more similar, and farther apart that are less similar (Figure 2). The beauty of Napping® is that it requires minimal training and is fast, versatile, and holistic, meaning that participants consider all characteristics. By comparing samples holistically, we get a much better representation of the true sensory experience, rather than just looking at one factor such as ‘sweetness’ or ‘spiciness’ as is done in many other methods.
As we wanted more information than just coordinates, the Napping® exercise was combined with UFP, where participants added sensory descriptors they found appropriate to describe each sample (Figure 3). To reduce bias, no prompts were given and subjects could write down anything that came to mind. As you might imagine, we received a wide variety of descriptors, from “garlic” and “umami” to “sexy”, “tastes good with pork buns”, and “reminds me of snowfall in Christmas”. Although these sample descriptions were highly varied, by grouping descriptors with the same or very similar meanings, a consistent set of descriptive words was formulated and summed over respondents be used for further statistical analysis.
Subject in action
A total of 26 subjects including chefs, students majoring in a food-related field, and other food professionals participated in the study. Due to our large sample size and sample complexity we decided that the study would benefit from individuals experienced in tasting food, as they are able to identify and name the flavours in the mixes more quickly and accurately than novices and in principle would take longer to fatigue.
A sample of napping boards after the experiment
The study included 29 different aromatic blends at the same time (imagine tasting 29 different wines at a wine tasting, without the alcohol of course), which, as we mentioned, is quite a large number of samples. Through a statistical tool called Principal Component Analysis, we were able to determine the variances between sample placements on the grid within the group as a whole. This method showed us that our subjects largely placed samples containing similar ingredients near each other, such as a pickling blend near juniper ant paste and BBQ chipotle near mole negro. Additionally, a majority of the words and phrases used to describe the aromatic blends are associated with samples that are near others described with similar words and phrases. For example, “fishy” is near “anchovy” and “Mexico” is next to “chili pepper”. These results may not appear to be dramatic—but what was exciting is that we could obtain such overlaps with an untrained panel and so many complex samples.
Figure 1. Score plot from PCA, Principal components 1 and 2. Map of 29 spice blends, showing the interrelationship between samples.
Figure 2. Correlation loading plot from PCA, Principal Components 1 and 2. Map of respondents’ positioning (small dots, labels omitted for brevity and clarity of the figure) and descriptors (triangles) used by respondents for sensory properties of the spice blends. For clarity only important descriptors are labelled.
As with any sensory study, our methods are not perfect—one subject’s definition of “salty” may differ from another’s, for example. However, our results with like-samples and descriptions appearing in the same vicinity demonstrate that combining Napping® and UFP can serve as a more practical approach to the time-consuming descriptive analysis completed by trained panels. Additionally, some limitations of the study include our large number of samples and complete freedom for descriptive words—we believe that more limitations on these parameters might diminish fatigue and thus yield even more reliable results.
One fact of gastronomy that makes sensory science so complex and interesting is that every food, even every ingredient, contains a wide variety of tastes and aromas that combine in different ways to create flavour. Our research was descriptive in nature rather than affective, but if you were to use the same 29 spice blends in an acceptance test, the results would be much more wide-ranging, as humans have such diverse preferences. This concept will be explored on from a scientific and culinary perspective in the next following post, Calibrating Flavour Part II:Exploring formulas for deliciousness.
Ares, G., Deliza, R., Barreiro, C., Gimenez, A., & Gámbaro, A. (2010). Comparison of two sensory profiling techniques based on consumer perception. Food Quality and Preference 21 (4), 417–426.
Dehlholm, C., Brockhoff, P. B., Meinert, L., Aaslyng, M. D., & Bredie, W. L. P. (2012). Rapid sensory methods – Comparison of free multiple sorting, partial napping, napping, flash profiling and conventional profiling. Food Quality and Preference , 26 (2), 267–277.
Frøst, M.B., Giacalone, D., & Rasmussen, K. K. (2014). Alternative methods of sensory testing: working with chefs, culinary professionals and brew masters. In J. Delarue, J. Ben Lawlor, & M. Rogeaux (Eds.), Rapid Sensory Profiling Techniques and Related Methods – Applications in New Product Development and Consumer Research (1st ed., pp. 363–382). Cambridge: Woodhead Publishing. doi:10.1533/9781782422587.3.363