FREE! Subscribe to News Fetch, THE daily wine industry briefing - Click Here


SPONSORED BY:

CLICK TO VIEW VIDEO

epic AD final

Click ad, above, to see how corks with TCA above 0.5 ppt can be removed from production.

Commerce7Click to learn more

Wine wreck-commendations: Genes determine that no two people taste the same wine the same way.

AUTHOR’S NOTE: This is #4 in a series about Netflix and wine

Article #1: How A Netflix-Style Recommender Is Vital to Reversing Wine’s Market Marginalization

Article #2: Why solving the “Paradox of Choice” is a major reason Netflix’s recommender is worth $1 Bil/yr. And why wine fails at that.

Article #3: Reviews and 5-Star ratings are so useless for recommendations that Netflix tossed its prized $1-million algorithm. They’re even worse for wine

Next: How Netflix has a big genetic recommendation advantage over wine + what wine can do about that.


Product recommendation systems generally fail regardless of the product. Consumers tend to see them as inaccurate, useless, and annoying.

Wine recommendations fail even more miserably because of psychology, education, social pressure, tasting environment, misunderstandings of descriptions, and individual sensory sensations. Add to this a lack of experience and the inherent difficulty people have in verbalizing the scents and tastes they experience.

Wine recommendations are especially sabotaged because unique, individual genetic variations in smell and taste mean that no two people experience the same flavors the same way. This holds even for “identical” twins.

This means that all current wine recommendation systems cannot deliver accurate suggestions.

This series previously addressed some of the reasons why Netflix abandoned five-star ratings and user reviews. For its part, Netflix has a far, far easier path to recommendation success because of its as a digital sight and sound phenomenon, rather than wine’s combination smell/taste phenomenon with a physical product.

Significantly, wine’s recommendation predicament has special pitfalls for the currently popular practice of trying to match specific consumers and wines using sensory profiles.

FLAVOR = Smell + Taste

First of all, it’s important to understand that flavor is not the same as taste. Flavor is the experience produced by the dual and simultaneous interaction of your sense of smell and taste.

Taste correctly refers to the sensations produced by receptors on the tongue and mouth. When those sensations are combined with the olfactory scents produced by chewing and swallowing the food, then that combined sensation is flavor.

The complete integration of the two flavor senses is so strong and integrated that some scientists see it as a form of synaesthesia.

How much of flavor is smell? How much is taste?

Contrary to numerous assertions that 75% to 95% of flavor comes from what we smell, even scientists who say there is no precise percentage that can be calculated agree that:

“the pleasure, all the interesting dimensions of what is commonly called taste, the meaty, the floral, the fruity, the herbal, the citrus, the burnt, all derive primarily from the contribution of olfaction.”

Clearly, anyone who has lost their sense of smell during a cold (or a bout with COVID-19) can verify that a substantial portion of their flavor experience has been lost.

In addition to smell and taste, solid scientific data indicates that sight, sound, and other sensory stimuli also play supporting roles in the perception of flavor. This multi-sensory phenomenon further complicates wine descriptions among different people and skews recommendations and sensory profiling.

Screen Shot 2021-02-22 at 7.44.50 PM

189 people studied, 189 different sets of odor-related genes.

A study of genetic taste receptor variations found that perceptions of wine were considerably different depending upon unique personal gene sequence.

The graphic to the left is one of two published in a scientific paper — (“Different noses for different people,“) that illustrates this vast genetic diversity.

According to the study, investigators genotyped 189 people and found that none of them had the same odor receptor pattern.

“Each of the individuals examined had a unique genotypic pattern.”

What’s more, the genes involved with olfactory receptors, have a very high rate of variation/mutation (aka polymorphisms).

This is confirmed by another major study that found DNA sequence polymorphisms were especially extremely high among olfactory genes.

Does the genetic variation actually matter?

Yep. It may be true that “A rose is a rose is a rose,” but

“[N]ew research indicates that might not be the case when it comes to the rose’s scent.

Researchers from the Monell Center and collaborating institutions have found that as much as 30 percent of the large array of human olfactory receptor differs between any two individuals. This substantial variation is in turn reflected by variability in how each person perceives odors. — Variability in olfactory receptors affects human odor perception.

Finally, studies show that even monozygotic twins have small but detectable differences in how they perceive odors. Monozygotic means twins that come from a single egg, a phenomenon that was formerly — and incorrectly — called “identical twins.”

Further, a significant human study — Genetic variation across the human olfactory receptor repertoire alters odor perception — found that changes in the genotypes of even a handful of olfactory receptors altered the perception of the odors presented to a cohort of 332 test subjects.

Previously, a 2017 research project investigating the flavor of polyphenol anti-oxidants in red wine also determined that individual perceptions varied according to the genotypes among 528 test subjects.

SMELL: Trillions of scents

The debate continues about the influence of smell versus taste in determining flavor. Significantly, the human sense of smell that can detect at least 1 trillion different scents. using approximately 400 active olfactory receptors.

Taste, on the other hand, has five or six basic tastes. Adding to a “taste” sensation are accessory receptors that detect heat, mechanical motion, pain, and chemical composition which the brain adds to sensations like  “mouthfeel and minerality” that are often inaccurately considered “taste “sensations.

Adding to the power of the smells is the presence of another 600 or so olfactory “pseudogenes.” that were once thought to be inactive. However, olfactory pseudogenes are getting a second look at potential sensory activity. This parallels the discovery that the “junk DNA” once thought to dominate the overall human genome are active in newly discovered ways.

The amazing number of different odors that can be detected is possible because a single odor can activate combinations of multiple receptors or just a single one.

By contrast, human vision has only four receptors: three for color-vision cones which detect red, blue and green. The fourth is the rod which functions in low light and basically sees shades of gray.

Taste matters

Scientists and philosophers have long debated what sensations qualify as “basic” tastes. Screen Shot 2021-02-07 at 10.39.30 AM

Six, not five basic tastes?

Current scientific work indicates that human basic tastes may actually number six with fat now considered as a 2015 addition to the original five. (See: Oleogustus: The Unique Taste of Fat)

  • Bitter
  • Sweet
  • Sour
  • Salty
  • Umami
  • Fat

Tastes are processed by three main types of receptors and variations It’s important to note that scientific study of taste receptors lags behind that of smell and still has many unanswered questions. Some sources estimate that taste is controlled by fewer than 100 genes.

Regardless of the number of genetic variations in taste receptors, the fact remains that these variations further increase the enormous genetic variability of the human population. This makes it nearly impossibly rare for any two people to experience the same flavor the same way.

While flavor studies focusing on smell and taste in wine drinkers are tiny in number, this study of genetic taste receptor variations found that perceptions of wine were considerably different depending upon gene sequences.

What about supertasters?

Much has been made in wine circles about “supertasters” who are exceptionally sensitive to bitter tastes.

According to “A review of the associations between single nucleotide polymorphisms in taste receptors, eating behaviors, and health,” this is a consequence of a genetic variation (polymorphism):

“Polymorphisms in the bitter taste receptor T2R38 have been shown to influence taste for brassica vegetables. Individuals that  intensely taste the bitterness of brassica vegetables (“supertasters”) may avoid vegetable consumption and compensate by increasing their consumption of sweet and fatty foods, which may increase risk for chronic disease.”

The study also noted (citations omitted and one clarification in [brackets]) that:

“The underlying cause of this relatively heightened oral response is not known, although it is thought to be due to both an increased number of fungiform papillae [taste buds] and the T2R38 phenotype; however, those factors do not completely explain the supertasting phenomenon as it may be attributed to other orosensory phenotypes.

“Among adults, the T2R38 genotype has been linked to avoidance of alcoholic beverages , increased prevalence of colon cancer through inadequate vegetable consumption, avoidance of cigarette smoking, and a preference for sweetness….”

The full text and details are available in the study which is free to access.

There’s more to taste than just taste … mouthfeel and more

In addition to the traditional notions of taste and smell, the experience of flavor — whether it be wine or a cheeseburger — exceeds the boundaries of the olfaction and basic taste sensations.

Neglected in most discussions of the sense of taste is the oral trigeminal system which encompasses an entirely different set of sensations from the basic tastes. These include; intensity, heat/cold, pain, texture, and astringency,

And because they are genetically determined, the sensations serviced by trigeminal nerves further complicate the issue of recommendations and determining individual taste preferences.

Consider “mouthfeel” which is frequently referenced as a quality of wine perception and often described in tasting notes and reviews.

Mouthfeel is determined largely (but not entirely) by the viscosity of the wine. Viscosity is sensed by mechanoreceptors which provide a sense of touch in the mouth that senses size, firmness, and texture of food in the mouth (oral cavity).

Viscosity in the mouthfeel of a wine or other liquid, is actually a delicate measurement of the oral sheer stress (a measure of flow) as it passes over the receptors. These receptors are especially prevalent on the tip of the tongue.

In addition, there are pain receptors that protect against harm associated with extreme heat, cold, and tissue damage. They also work in tandem with taste and smell receptors to distinguish thermal damage from pungency (hot spicy peppers) and cold (menthol).

As for mouthfeel:

“Astringency mouthfeel perception, which is a dry puckering sensation, has been linked to a loss of lubrication from the interaction of salivary proteins with astringent molecules such polyphenols in wine and tea” Oral processing, texture and mouthfeel: From rheology to tribology and beyond

Other viscosity researchers found that increasing viscosity decreased the perception of sweetness. That confirmed a previous study finding that perceived sweetness decreases with viscosity.

While those studies were not conducted with wine, the conclusion to be drawn is that the receptors are genetically based, and will vary according to the individual. Those variations further confound the wildly varying combination of smell and taste genes. This further confirms the irrelevancy of flavor recommendations based on comparisons of one individual taster to any other.

What does this mean for wine recommendations, especially sensory profile matching?

Psychology, education, life experience, vocabulary, tasting environment, and other factors continually sabotage the usefulness of wine descriptions as an accurate way to help people choose wines they will like.

For this reason, profiling wines has been tried for some time now as a logical way to finesse the biases and inaccuracies inherent in individual interpretation of words.

The profile concept theoretically asserts that if an expert evaluates a wine according to a set of objective sensory flavor standards (tannin, fruit, intensity, acidity, oak, etc.) it creates an objective profile for that wine.

Following this theory, consumers can evaluate the same wines and create their own objective personal profiles. As a result, theory indicates that personal profiles can then be compared with reference wines profiled by experts (or others) to develop a match that identifies wines the consumer will like.

Epic fail

It’s a logical, but fatally flawed hypothesis that fails when tested in the real world.

The example described below, is based on SmartTaste, a pioneering data-based, sensory profiling recommendation engine.

SmartTaste was the first implementation of an automatic, online data-based system designed to match people’s preference profiles with given wine profiles created by professional critics.

Multiple tests of profiling uncover fatal flaws

SmartTaste and its later variant, SavvyTaste, failed because of inconsistencies between reference profiles and personal profiles. Indeed, both reference and personal profiles were found inconsistent in additional profiles of the same wine by the same taster conducted at different times and places.

Screen Shot 2021-02-18 at 7.08.48 AM

There are numerous sensory profiling systems that are based on the same principles as SmartTaste and SavvyTaste and fail for the same reasons.

Screen Shot 2021-02-22 at 9.36.57 AM

Screen Shot 2021-02-22 at 9.33.44 AM

Screen Shot 2021-02-22 at 9.34.12 AM

SmartTaste began with a professional tasting panel — usually five or six — assembled by the former president of Wine Spectator. Tastings were done completely blind.

The professional tasters created the reference profile and a numerical rating for each wine on a 100-point scale.

Website visitors were then invited to use the same scale to describe and rate wines they had tasted.

Using that data, the SmartTaste system then created a profile of wines that consumers liked and tried to match it to the expert wine profiles.

Inconsistent performance on reference profile

SmartTaste failed the consistency test.

Interviews with consumers and a closer look at the expert rating process revealed why.

First of all, even the expert taster profiles and ratings varied. In fact, individual tasters were substantially different on the exact levels of tannins, complexity, and other organoleptic measures. They also disagreed on ratings, frequently by more than 5 points on the 100-point scale. Sometimes the variations were 10 points or more.

Disagreements were resolved by a consensus discussion. This meant that the profiles were composites lacking in precision.

Disagreements among professional tasters are well documented:

Genetics = Poor recommendations

Subsequent interviews with consumers indicated that the recommendations were inconsistent but neither better nor worse than the wines they chose previously based on reading published reviews of wines. We then invited consumers to create their own profiles of the wines they had purchased. None were exactly the same and none matched any of the individual professional tasters.

In the end, SmartTaste and SavvyTaste were discontinued because of the inevitable individual rating and profile inconsistencies caused by genetics along with environmental, educational, experiential, psychological, and other individually uncontrollable factors.

Adjustable flavor sliders were no help

Another current “old is new again” system in attempting to match consumers and wine is to use a slider that consumers can use to try and match their own preferences.

Screen Shot 2021-02-22 at 9.33.58 AM

In 1996, SmartTaste created individual flavor histograms for each varietal rated by each user.

That was graphically represented as an interactive version of the SmartTaste system where users could adjust the tasting categories to better suit their own individual taste.

Screen Shot 2021-02-22 at 9.18.10 AM

The flavor slider did not work for two reasons:

  1. All but the most sophisticated consumers did not understand how to calibrate the flavor categories to their taste perception experience
  2. Flavors are experienced as an integrated perception, a momentary sensation, and not as a matrix of discrete scents and tastes. Moving a slider on “body” (also interpreted by some as mouthfeel) would also affect the perception of sweetness (as we saw above). Slider movement on that single characteristic would result in genetically individual modifications of perceptions that would also affect acidity, fruit. Changing one in a slider would, in reality, change the entire perception.

And, once again, genetics comes into play. The individual flavor genome will determine how a specific person will perceive the interplay among taste and smell components.

Perception capture a future direction?

Given the inherent problems of trying to create an accurate, computationally stable profile of a wine that manifests itself as a unique internal sensory experience of a unique individual, a possible path forward may lie in methods that can accurately capture an individual’s unique perception of specific wines (or other taste/scent/flavor products) in a manner that can be accurately, reliably, and consistently captured.


Further reading: