Sensory Evaluation 1 (Wine): Measuring the Indescribable

(Last Updated On: March 27, 2017)

wine_glasses qpad 200pxScience Fact:  Sensory Evaluation goes beyond tasting or smelling or hearing: it implies scoring, grading, in fact, measurement.

The senses are inherently personal.  For example, it’s been shown that different people taste foods and see colors quite differently.  Therefore scientists have had to become very clever to objectively judge, that is measure, the subjective experience of the senses.  Moreover, experiences like taste and smell are largely indescribable – in Western countries, people use different terms to describe the same taste or smell, there’s no common vocabulary.  (Things can be different in other cultures – for example, the Janai people of Southeast Asia use very consistent words to describe different smells.)

Today’s blog discusses Wine Tasting, a deep subject that has justified centuries of study and a library of books; the following blog, to be posted next Tuesday, covers Concert Hall Measurement, a field that borrows from wine tasting but involves additional complications of its own.

A previous blog pointed out how, when tasting wine, almost everyone is misled by factors other than sight, smell and taste – the price of the bottle, the name on the label, the wine’s reputation, even the décor and comfort of the taster’s immediate environment.  And studies have shown that even wine professionals are unable to give consistent evaluations of the same wine on two different occasions.

In many cases, it really doesn’t matter whether our senses are easily fooled.  If you host a wine tasting party, your primary goal may be to entertain your guests, not to add to the world’s storehouse of knowledge.  If you’re a winemaker or wine dealer, your goal may be to educate consumers who will then want to buy more of the wines you have poured.  And the organizers of wine competitions have many goals including tourism and publicity for the participating wineries.  You could say that as long as the total experience fulfils the goals of the organizer, the wine tasting has done its job.

What about wine tasting that really hopes to learn something about the wine, something that doesn’t depend on the extraneous factors?  A winemaker may want to improve his product so that it will win the hearts of consumers and garner awards.  And researchers, especially those associated with agricultural universities, hunger for the real knowledge hiding beneath all the subjective factors.

That brings us to the scientific approach to wine tasting, which we will call Sensory Evaluation.  Serious winemakers and researchers share a systematic approach to wine tasting, one that involves quite a few steps.  Since the winemaker is trying to increase the world’s knowledge in this area the winemaker is (himself or herself) a researcher, so we may refer to the study leader as a researcher in either case.  Opinions differ, but I would offer the following ten steps in a serious wine study:

1.  Goals.  Setting clear goals is an early step.  Are you trying to tweak your wine to appeal to a particular market segment, people of certain demographics?  Do you want to cultivate trend-setters who might influence many folks to buy your product?  Do you want to create a wine that will appeal to professional tasters and capture influential prizes, or at least a high rating from Wine Spectator?  For the moment, let’s choose the third of these options and see where it leads us.

2.  Study Design.  The researcher must then lay out the design of the study: how tasters and wines will be selected, how the measurement process will be managed, how the data will be analyzed and especially, the expected results or outputs of the study.  For example, the sponsor might want to bring his or her wine closer to some specific ideal, or some particular other wine; or simply to characterize the variations and similarities among the wines in the study group.  The study design will be influenced by the availability of wines, tasters and suitable research space, as well as the budget allocated by the sponsor.

3.  Supertasters.  We’re all familiar with the range of personal preferences in foods – one person can’t abide beets, the next one goes out of her way to eat them.  What’s less obvious is that some people among us are genetically gifted, able to detect and distinguish a wider range of tastes and aromas than the rest of us.  Those who want to scientifically measure wine taste use these “supertasters” as their tasting panels.  Supertasters make up approximately 25% of the population and can be readily identified by two specific tests:
– Supertasters are known to be sensitive to two particular chemicals – PROP and PTC – that to the rest of us have no taste at all.
– Supertasters also have more papillae – taste buds – on their tongues: 65 or more per square centimeter for supertasters, less than 35 for the 25% of “non tasters,” and intermediate values for the rest of us.  It’s easy to make this measurement: stain your tongue with blue food coloring, place a “hole reinforcer” near the tip of your tongue and count the pink taste buds visible in the center hole, peeping up through the field of blue.  40 or more (some say 30) qualifies you as a supertaster.

4.  Field of Inquiry.  The number of wines may not exceed the number of grains of sand on the beach but it’s still an exceedingly large number.  Someone will be paying for the study, the sponsor, and the sponsor’s goals will in part determine the selection of wines to be tasted.  Many decisions are required, such as wine or grape type; retail price range; geographical area of origin; geographical area of distribution; vintage; and number of wines to be tasted.  Choosing some or all of these factors will define a universe of wines for the comparison.

5.  Panel of Judges.  Next we must assemble a panel of supertasters to act as judges.  These must be people who can devote serious time to tasting.  One wine study required each judge to participate for six weeks, tasting every wine several times.  Because of statistical variations between individuals, it’s generally accepted that you need a minimum of eight judges for valid results.  The maximum number of judges is determined by the following factor.

6.  Vocabulary Profiling / Descriptive Analysis.  Even among experienced supertasters, there are considerable differences between the terms they use to describe tastes, aromas and visual appearance.  One judge might call a wine “crisp,” another “acidic,” still another “sharp,” and all three might be referring to the same feature of the wine.  Moreover, the taster who called the wine “crisp” might use the term “sharp” to describe a completely different sensation on the palate.  Thus when a new tasting panel is assembled they need to begin by finding a common language, a process that is called Consensus Vocabulary Profiling or Sensor Descriptive Analysis.  The panel moderator chooses a test substance, perhaps a wine, perhaps a chemical or an herbal extract, and guides the panel to reach consensus agreement as to what words will be used to describe this particular flavor.  This is where the maximum size of the panel enters in – typically, a dozen people is the maximum number that a moderator can keep focused on the subject and not distracted by side conversations.  This process has to be repeated in all three dimensions – visual, palatal and nasal – until the panel members are satisfied that they have agreed on enough words of all three types to describe all the sensations of the wines they are about to taste.

7.  Tasting.  Finally the panel is ready to actually taste the wines.  For visual profiling, the wines will be served in clear glassware.  However, when measuring aroma and taste, black glassware may be used so that the appearance of the wine doesn’t bias the sensory evaluation.  Tasting is “blind” of course, with wines identified only by a random number that reveals nothing about their origin or even the type of grape.  Ideally, the judges are isolated from one another when they taste so they will not be affected by nonverbal cues from other judges.  The most reliable and repeatable measurements make direct comparisons of one wine with another and typically twenty to fifty different wines may be tasted in one sitting.  Each panelist describes all the visual, or aromatic, or palatal sensations of each wine, using the agreed-upon vocabulary, and rates how strongly each factor appears.

8.  Scoring.  Next, the researcher needs to make sense of the data.  Even with all the careful preparation, there will be variation between the descriptors and scores given by the various panelists.  The score sheets are transferred to a data base and mathematically analyzed.  Each wine receives a consensus score on each factor that the judges seem to agree applies to that wine.  For example, if a majority scored a wine as “earthy,” that term would be included in the descriptors for that wine.  Statistical analysis may be used to compute the probability that the scores of the judges are consistent, that they don’t result from random chance.  Scores that are too contradictory may lead to throwing out that factor for the wine, and if too many scores have to be discarded, the researcher may need to go back and conduct additional Vocabulary Profiling, this time with additional test substances.

9.  Clustering.  Assuming that each wine has received a consensus score that seems valid, the researcher will look for “clustering.”  Conceptually, the results are plotted in “feature space,” in which the various vocabulary terms each represent an independent axis, and we look for wines that are “close together.”  Another way to say this is, if two wines have been judged to have the same sensory features, we classify them as similar to each other, and the degree of similarity can be mathematically calculated.

10.  Study Results.  Finally, the researcher can pull together the results in the form anticipated by the sponsor.  The result might take many forms: a profile of a class of wines for a publication or an association; a proprietary report to guide an individual winemaker; an academic publication.  If the study has been well designed and competently executed with carefully selected tasters, the results should be both valid and repeatable – in other words, add to the universe of verifiable knowledge about this most elusive and personal taste experience.

I was frankly amazed at how much there is to this field, and I hope I haven’t worn you out by trying to describe it.  I highly recommend the video of an excellent talk on scientific wine tasting by Prof Hildegarde Heymann of UC Davis.

Science Speculation:  The science of sensory evaluation is  a serious discipline, whether conducted by a winemaker, a wine expert or a university professor.  We can all be grateful that these studies are continually being carried out, whether for economic or academic reasons, because their end result is to improve the wine consumer’s experience.

Yet recall the previous blog (Taste Bias & the $90 Wine).  Just because many factors can affect our sensory evaluation when we pick up a glass of wine, that doesn’t mean we should try to banish those factors.  The point of sipping wine – for me, at least – is to have an enjoyable sensory experience.  We should not resent the fact that our sensory evaluation is colored by price, brand, the coziness of the tasting room, the taste of accompanying foods, the joy of being with our friends.  Rather, we should embrace and absorb all these factors – taste, aroma, sight and all the others – as part of the multidimensional reward that we gain from that simple-yet-complex entity, a glass of wine.

Next Tuesday I will post the second half of the Sensory Evaluation story, but shifted into an entirely different realm: music and the acoustics of concert halls.

Have you attempted a sensory evaluation of wine, for example a blind tasting?  What was the occasion, and what did you take away from the experience?

Drawing Credit: qpad, on


Sensory Evaluation 1 (Wine): Measuring the Indescribable — 10 Comments

  1. Two other control variables that aren’t mentioned, but which I would think are important, are (1) do the results from a single panelist tend to vary over time of day (fresher in the morning?) and (2) how does “fatigue” in tasting get factored out?

    I know there is the advice to “freshen your palate” between tastings but it’s just hard to believe that tasting one particularly memorable (good or bad) wine doesn’t muddy the result if a second tasting occurs within a few minutes afterwards. That could occur on a psychological level, skewing the results because of the recent experience, or because the taste buds themselves are affected and require time to “recover” before they can once again yield a consistent baseline result (if that’s even possible).

    I guess I’m suspicious about how well we understand the senses of taste and smell to be able to attempt a scientific study like this without unintentionally polluting the results given the serial nature of these evaluations. It would be like a chemist or biologist using the same specimen dish for a sequence of tests without thoroughly washing it out (if not sterilizing it) between tests, or like a physicist using an instrument to measure multiple items over time but not being able to recalibrate on a regular basis.

    • In Tuscany, most wine tastings are accompanied by cheese. It is obvious that the coating of the tongue with these fats enhances the taste of specific wines (probably masking the effects of some unpleasant chemicals).


    • Charles, I agree that details like those you mention are important. I know that each judge tastes wines repeatedly and the order of the wines is randomized each time. Thus if there is fatigue, or a lingering effect of one wine on another, that should be revealed by an analysis of the data. I don’t know whether time of day matters, so ideally that should be varied in repeated tastings as well. Perhaps that’s why it takes a number of weeks to conduct one of these scientific tests, unlike more casual wine tastings.