Review
Recent development of taste sensors
2023 Volume 29 Issue 2 Pages 87-99
Details
2023 Volume 29 Issue 2 Pages 87-99
The first taste sensor was developed 30 years ago, and its use has become widespread in the food and pharmaceutical industries. Numerous efforts have been made to improve taste sensor technology since that time. Now, over 600 taste sensors are used around the world to quantify and visualize the taste of food. In this article, we first explain the mechanisms underlying the operation of taste sensors and how they respond to the five basic sensations of taste. A taste map of beers that presents their taste in a visual manner will be introduced as a typical application of the taste sensors used in the food industry. We then discuss the ability of the taste sensor to detect taste interactions, such as bitterness-suppressing effects as well as attempts to produce tasty low-calorie foods. Recent attempts to visualize personal preferences and the dynamic changes in the taste of food during chewing using the taste sensor are also explained. The environment surrounding taste sensors and the taste sensors themselves have progressed considerably, with new food services appearing constantly. We are entering an era where we can increase our enjoyment of eating.
Palatability is highly subjective. Is there universality in palatability? For example, is it possible to segment palatability according to ethnic group, culture, location, and age? Is it possible to quantify and visualize the taste of food in a manner that is obvious to all? Among the five senses, eyesight, hearing, and touch are called physical senses because they result from the reception of physical stimuli. On the other hand, taste and smell are referred to as chemical senses because they result from the reception of chemical stimuli. Palatability is a complicated concept because it is a perception that results from both physical and chemical senses.
When you drink orange-colored sugar water, you may think that it is orange juice. However, you cannot taste the juice when you have caught a cold, finding that what you have experienced as taste is actually smell. These examples show that “taste”, which is considered to be sensed by the tongue, is actually a sensory quantity recognized using all five senses. In fact, the “taste” casually talked about in everyday life, described in words and imagined in our minds, is not the “taste” sensed by the tongue, but taste as recognized using all five senses. However, the term “taste” in this article refers specifically to the “taste” sensed by the tongue.
As a result, palatability reflects a large amount of information based on our five senses, preferences, cultures, and promotional messages. Is it possible to present this perception in a scientific manner and convey it to others? Many people take photos of their meals with their cellphones while traveling and send them to their friends and family. Visual information can certainly be recorded, reproduced, and conveyed to others. Of course, such visual information is only the simplest approximation (first-order approximation) of the environment or of particular objects both natural and artificial.
The science and technology related to the five senses attempt to approximate the world of nature or objects. A television is a device that receives signals and broadcasts images of objects with sound. Since it was launched over 50 years ago, the television has advanced from being black and white to colored. Televisions now have a large screen with high definition, are digital, and are capable of 3D display and 4K resolution. In other words, the level of fidelity of the approximation of objects has increased.
Taste and smell are senses resulting from the reception of low-molecular-weight compounds. Fig. 1 shows taste and smell in terms of detection thresholds, the number of receptors, and the presence/absence of basic elements (basic taste qualities and basic odors) (Bachmanov et al., 2014; Beidler, 1971; Pfaffmann, 1959). While taste is sensed at high concentrations above 1 ppm, smell is sensed at very low concentrations at the level of ppb and ppt levels. While sour and bitter substances are sensed at concentrations above 1 ppm, sweet substances are sensed at concentrations above 1 000 ppm.
World of taste and smell.
In humans, the number of types of receptors for taste differs markedly from that dedicated to smell; for example, there are only 30 receptors for taste, but as many as 400 receptors for smell. In addition, while there are basic taste qualities (sourness, sweetness, bitterness, umami, and saltiness) for taste, there are no basic odors for smell, which may be related to the different numbers of receptors for taste and smell. In addition to the five basic taste qualities, there are the additional taste qualities of astringency and pungency.
A taste sensor, or electronic tongue (e-tongue), is a device that can be used to quantify the taste sensed by humans as a result of chewing food in the mouth. It also quantifies and discriminates the chemical characteristics of liquid samples. A number of review papers have been published on the first taste sensor that was developed in 1989 (Kobayashi et al., 2010; Tahara and Toko, 2013; Toko, 2000; Toko et al., 2016; Toko et al., 2021; Wu et al., 2020a), and other types of e-tongues (Ciosek and Wróblewski, 2007; Riul Jr and Correa, 2021; Riul Jr et al., 2010; Sharma et al., 2015; Vlasov et al., 2005; Woertz et al., 2011a; Woertz et al., 2011b). The mechanism of the taste sensor developed in Japan and its applications are introduced in this article.
This taste sensor is recognized as an e-tongue with global selectivity (Riul Jr et al., 2010; Anand et al., 2007), which means that it has the ability to decompose the characteristics of a chemical substance into each type of taste quality, and then to quantify the taste. On the other hand, an e-tongue proposed in 1995 was a sensor that was used to analyze solutions using arrays of nonspecific chemical sensors and pattern recognition (Vlasov et al., 2005; Ghasemi-Varnamkhasti et al., 2010; Winquist, 2008).
A membrane composed of a lipid, a plasticizer, and polyvinyl chloride (PVC) polymer (lipid/polymer membrane) serves as the receptor for taste substances in taste sensors. Taste is quantified (digitized) based on the responses of potential outputs from several receptive membranes. The basic taste qualities can be quantified by a simple proportional calculation, because the intensity of the response of the sensor changes logarithmically. The TS-5000Z taste-sensing system was developed and marketed by Intelligent Sensor Technology, Inc. More than 600 units of the taste sensing system have been used around the world (Fig. 2). The taste-sensing system provided the first-ever “taste scale” in the world and has become easy to use because it has been nearly 30 years since its initial release and various improvements have been made to the instrument over time. It has proved to be a powerful tool for marketing, for developing new food products, and for ensuring food quality.
Taste sensor (taste sensing system TS-5000Z manufactured by Intelligent Sensor Technology, Inc.).
Fig. 3 shows a schematic diagram of the lipid/polymer membrane of the C00 sensor, which is composed of tetradodecylammonium bromide (TDAB), 2-nitrophenyl octyl ether (NPOE), and PVC, used for measuring the bitterness of food compounds (e.g., iso-α acid). The membrane is characterized by having the hydrophilic group of the TDAB molecule (i.e., the part of the molecule that is attracted to water) directed toward the water phase and a hydrophobic group (i.e., the part of a molecule that repels water) oriented towards the PVC due to hydrophobic interaction inside the membrane. This surface structure is the key component for receiving taste substances. Note that this hydrophilic property of the surface layer containing lipid molecules is the same as that of biomembranes.
Schematic diagram of receptive membrane of bitterness sensor C00 for foods.
This structure of the bitterness membrane in C00 was confirmed by X-ray photon spectroscopy (XPS) and gas cluster ion beam time-of-flight secondary ion mass spectrometry (GCIB-TOF-SIMS) (Yatabe et al., 2015). A preconditioning process is necessary for the lipid/polymer membrane to respond to taste substances (Harada et al., 2016). The membrane is typically immersed in a solution containing several kinds of taste substances for a specified period of time, usually from one to several days. After this preconditioning period, the lipid concentration was found to be higher in the surface region than in the bulk region of the membrane. The plasticizer and PVC have no electric charge, but the amphipathic lipid molecules do have charge. As a result of thermodynamic stability, the hydrophilic components adjust themselves so as to face the aqueous solution.
Fig. 4 shows the measurement procedure. First, the potential in the reference solution corresponding to saliva, which is nearly tasteless, is measured. Here, the obtained potential is denoted as V r. Next, the potential in the sample solution (V s) is measured. The difference between the two potentials (Vs-Vr) is the normal potential response called the relative value. Then, the sensor electrode is washed lightly and then dipped into the reference solution again. At this time, the membrane has not returned to the initial state because the taste substances having hydrophobicity, such as bitterness, astringency, and umami, are still adsorbed on the membrane. As a result, the potential generated (V r’) differs from that in the initial reference solution (V r). The difference between this potential and the initial potential (V r’ −V r) indicates the change in the membrane potential caused by the adsorption of chemicals onto the membrane (called the CPA value). This value corresponds to the aftertaste perceived by humans and depends on the amount of taste substances adsorbed onto the membrane and the state of the charge of the membrane (Toko et al., 2014). The aftertaste of the five basic taste qualities can be quantified by measuring the CPA value. In particular, the quantification of koku (rich taste), the aftertaste mainly resulting from umami substances, is valued by the food industry engaged in the production of broth and soup, as it agrees well with the sensory evaluation by humans (Doi, 2011). As the final step in the measurement of the sample solution, the membrane is washed thoroughly in a specific cleansing solution in order to desorb the adsorbed taste substances and to return the membrane to the initial state. Generally, this procedure (i.e., rotation) is repeated three to five times to quantify the taste of the sample solution.
Measurement procedure: relative and CPA values.
Fig. 5 shows the responses of the taste sensor to the five basic taste qualities (Toko et al., 2021). In addition to the five basic taste qualities, astringency can also be measured. BT0, CA0, AAE, CT0, and GL1 are the receptive membranes that have been developed to measure bitterness, sourness, umami, saltiness, and sweetness, respectively. The CPA value as well as responses from the bitterness sensor BT0 are used because the aftertaste of bitterness is also measured.
Responses of receptive membrane of sensor to five basic taste qualities. From Toko et al. (2021), reproduced with permission of IOP Publishing Limited through PLSclear.
As shown in Fig. 5, the responses are proportional to the logarithm of the concentration in some areas. This feature agrees well with the well-known Weber–Fechner law in biological systems, which states that the intensity of a sensation is proportional to the logarithm of the stimulus (Beidler, 1971; Pfaffmann, 1959). The response to each taste quality is the response of the lipid membrane sensor that is dedicated to the taste quality. The threshold of each human taste quality is as follows: bitterness is several μM, sourness is about 0.1 mM, saltiness is 1 to 10 mM, sweetness is 3 to 30 mM, and umami is about 1 mM (Indow, 1966; Herrmann, 1972; Yamaguchi, 1987). The thresholds shown in Fig. 5 are near those of the four taste qualities, except for sweetness, for which the threshold of the taste sensor is around 100 mM, which is higher than that of humans.
The bitterness sensor BT0 hardly responds to sweetness, sourness, saltiness, umami, or astringency, but responds strongly to quinine, cetirizine, hydroxyzine, and bromhexine, which have a bitter taste. Also, the response values of the bitterness sensor BT0 to various bitter substances corresponds closely with the sensory values perceived by humans (Kobayashi et al., 2010). The sensor output is large for loperamide, which tastes very bitter to humans and very small for ambroxol, which does not taste very bitter to humans. Namely, the bitterness sensor BT0 responds strongly to bitter chemicals and responds weakly to less bitter chemicals. Thus, it can quantify the bitterness detected by humans.
Similarly, membranes and measurement procedures with high selectivity have been established for other taste qualities. Lipid membranes responding only to specific taste qualities were developed by finely adjusting the proportions of the lipid and plasticizer; in other words, by striking a balance between electrical charges and hydrophobicity. For example, the lipid content is decreased and hence hydrophobicity is increased in the bitterness sensor BT0. On the contrary, hydrophilicity is increased in the saltiness sensor CT0 by increasing the content of the charged lipid to facilitate more electrostatic interactions with ions. Modifiers such as aromatic carboxylic acids (gallic acids, trimellitic acids, etc.) are added to the lipid/polymer membrane of the sweetness sensor GL1 so that the sensor responds to sweet substances, such as sucrose and glucose (Yasuura et al., 2015; Ye et al., 2022). In this case, the membrane is washed in the alkaline cleansing solution (shown in Fig. 4), which contains high concentrations of metal ions such as K+ and Na+. Therefore, the electrostatic attraction occurs between the metal ions and the carboxy group of the modifiers, increasing V r in the reference solution to a positive value. In the subsequent measurement of sucrose solution, complex formation between sucrose and metal ions occurs after detaching from the carboxy group, decreasing V s in the negative direction.
Each receptive membrane of the taste sensor specifically and selectively responds to each taste quality. This property is called global selectivity, as mentioned above (Riul Jr et al., 2010; Sharma et al., 2015; Toko et al., 2016) because, compared with conventional chemical biosensors, which are highly selective for chemicals, the taste sensor is not selective for individual chemicals but for the taste qualities into which the chemicals are grouped. Just like conventional chemical biosensors, which are selective for chemicals, antigen-antibody reactions in a living body are highly selective and specific for antigens, and enzyme catalysis is also selective for chemicals. Chemicals (substrates) are bound in a one-to-one manner with receptors. This property is used in glucose sensors such as blood sugar level sensors. On the other hand, the taste sensor is not selective for chemicals, but for taste qualities, and this property reproduces the taste mechanism. It is well known that the receptors in humans can bind to different chemicals of the same taste quality simultaneously.
Taste sensors have not only been used to quantify and evaluate the quality of the taste of liquid food and beverages (Kobayashi et al., 2010; Tahara and Toko, 2013; Toko, 2000; Toko et al., 2016; Toko et al., 2021; Wu et al., 2020a; Doi, 2011; Hayashi et al., 2020; Iiyama et al., 2000; Iiyama et al., 1996; Toko et al., 1995; Fujimoto et al., 2021; Hayashi et al., 2006; Guo et al., 2016; Akitomi et al., 2013), such as coffee, beer, wine, green tea, Japanese sake, shochu (distilled spirit), juice, milk, yogurt, mineral water, broth, soup, seasoning, and soy sauce, but also solid food, such as rice, bread, meat, fish and seafood, jiaozi (dumplings), vegetables, and fruit. Solid food is mixed with water and ground using a mixer, and the prepared liquid sample is subjected to measurement.
Fig. 6 shows a taste map of world beers, which was evaluated using bitterness C00 and sourness CA0 sensors. The vertical axis indicates the bitterness intensity, which represents the “malt taste”. The horizontal axis indicates the sourness intensity, which represents the “dry taste”. The scales for sourness and bitterness are indicated on the taste map. One scale unit on the axis indicates the minimum difference between two taste strengths that humans can detect. A difference of one scale unit is set to a 1.2-fold concentration of citric acid (sourness) and sodium chloride (saltiness). If the difference between two taste strengths is two or three scale units, then the difference is easy to detect; this is the “taste scale” based on the taste sensor.
Taste map of world beers.
We can see that Sapporo Yebisu (Japan) and Bavaria (Netherlands) beers have a strong bitter taste, whereas Edelweiss (Austria), Inedit (Spain), and Weissbier (Germany) beers have very mild bitterness, while Chang (Thai), Asahi Super Dry (Japan), and Castlemaine's XXXX (Australia) beers are sour. In this way, we can visualize the taste.
The taste sensor has made it possible to perform assessments that were not possible using conventional analytical instruments. One such assessment is the quantification of the interactions between taste substances or taste qualities. For example, sweetness increases and bitterness decreases when sugar is added to bitter coffee. Some medicinal products that are bitter by nature—as in the old saying “Good medicine tastes bitter”––are sugarcoated to make them easier to swallow. This is an application of the bitterness-suppressing effect. The suppression of bitterness is a critical issue in the pharmaceutical industry, and several approaches have been attempted to achieve this aim. The most common approach is the addition of sweet substances; for example, pediatric syrup is manufactured with this approach. In addition, bitterness inhibitors that are formulated using phospholipids are marketed and taste sensors are used to detect their bitterness-suppressing effect (Terashita and Wakabayashi, 2013; Harada et al., 2010; Woertz et al, 2011b).
A comparison of the results of sensory tests with those of taste sensor measurements was performed when a commercially available bitterness inhibitor was added to quinine, a bitter substance (Takagi et al., 2001). The bitterness score (τ scale) decreased when the bitterness inhibitor was added in both the sensory tests and the taste sensor measurement. The sensor detected little bitterness when at least 0.5% bitterness inhibitor was added to 0.1 mM quinine, which agreed with the results of sensory tests. It shows that the taste sensor can reproduce and quantify taste interactions. Less bitter medicinal products for children are developed by making use of these achievements (Woertz et al., 2013). Bitterness sensors are used at sites where medicinal products that require the bitterness-suppressing effect are developed.
Next, let us discuss what happens when artificial sweeteners are added to coffee (Wu et al., 2016; Wu et al., 2020b). Although bitterness is suppressed in this case, no studies have demonstrated that this bitterness-suppressing effect takes place on the tongue. We therefore considered that this phenomenon takes place in the brain and carried out a study to evaluate bitterness. Quinine, a bitter substance, and three types of artificial sweetener, acesulfame potassium, saccharin sodium, and aspartame, were used in the experiment. First, a scale for bitterness detected by humans (in sensory tests) was created using the bitterness sensor. In other words, the intensity of bitterness detected by humans was represented by the output of the bitterness sensor. Then, the intensity of sweetness detected by humans was represented by the output of the sensor for artificial sweeteners. The scales for bitterness and sweetness were created by operating these two sensors. The CPA values of both the sensors were used.
It should be noted that the responses of these sensors are mutually independent. Namely, the bitterness sensor does not respond to the sweetness of artificial sweeteners and the sensor for artificial sweeteners does not respond to bitterness. As a result, two independent scales were created. The CPA value represents the detection of only adsorptive substances and does not represent the detection of electrolytic taste qualities, such as saltiness and sourness. The CPA value is highly selective for adsorptive taste qualities. This property of the CPA value facilitated the development of two mutually independent sensors, contributing to the construction of the two scales.
Then, sensory tests were performed by adding the three above-mentioned artificial sweeteners to quinine. The results of the sensory tests show that the bitterness decreased when the artificial sweeteners were added. Our goal was to quantify the decrease in bitterness using the scales for bitterness and sweetness. As shown in Fig. 7 (Wu et al., 2016), the suppression of bitterness by aspartame was quantified with high accuracy (R2 > 0.9) using a linear sum equation of the two scales. The horizontal axis in Fig. 7 shows the bitterness score obtained by the sensory tests and the vertical axis shows the bitterness score estimated using the linear sum of the scales created using the taste sensor. The bitterness score obtained by the sensory tests decreased with increasing concentration of aspartame (from top right to bottom left in the figure), indicating that the taste sensor evaluated the bitterness score accurately.
Derivation of estimation equation for bitterness-suppressing effect. Reprinted from Wu et al. (2016) with permission from Elsevier.
This achievement is of great significance because it demonstrates the ability of the taste sensor to quantify not only the “taste detected by the tongue”, but also the “taste perceived by the brain”. The taste perceived by the brain can be quantified using a linear sum equation that represents the interactions of taste qualities in the brain on the basis of the results of the taste sensor measurement. In addition, the latest achievement demonstrates the ability of the taste sensor to detect the saltiness enhancement effect, whereby the saltiness perceived by humans is maintained at the same level with a lower salt content by adding specific substances to salt (Nakatani et al., 2019). The taste sensor can objectively quantify the saltiness perceived by humans, such as “strong saltiness with a low salt content” and “the same saltiness with a salt content lower than that in conventional food”. The taste sensor enables food to be developed that ensures an active and healthy life in an aging society. Recently, with the progress of artificial intelligence (AI), more complex flavors will be represented by leveraging the ability of the taste sensor to quantify the taste detected by the tongue.
Taste sensors have also been used to develop diabetes-friendly food. Chateraise Co., Ltd. has developed low-carbohydrate dorayaki (a sandwich of small pancakes filled with bean jam) with 86 % less sugar and 48% fewer calories than conventional dorayaki i). Low-carbohydrate Western-style confectioneries, such as cakes and puddings, already exist. However, the development of Japanese-style confectioneries has been considered difficult because the alternative ingredients used for bean jam deteriorate in taste when the amount of sugar is markedly reduced. To solve this problem, the taste of the trial low-carbohydrate dorayaki was compared to that of the conventional dorayaki using the taste sensor. On the basis of the obtained data, attempts were made to improve the taste of the trial product by adding different ingredients and examining the similarity of the taste profile to that obtained from the conventional dorayaki. For example, powdered green tea was used to add umami, which was lacking in the trial product. Dietary fiber and flavoring ingredients were used instead of high-carbohydrate wheat flour. Such product development is generally a trial-and-error process. Healthy human product testers might have some resistance to tasting trial products that are not palatable, and this is why the taste sensor was used in the development of the low-carbohydrate dorayaki with 86% less sugar.
Taste sensors are not only used for the development of new products, but also for marketing. Let us take coffee production as an example. It is not easy to consistently offer coffee with a desirable taste because coffee bean crops and their price vary considerably from year to year. The product design has been performed by people with a trained sensitivity who are called blenders. However, a product design approach using taste sensors can shorten the development period and thus allow both the taste and price to be optimized, which cannot be achieved even by excellent blenders. The coffee served on Japanese Airlines flights was developed using a taste sensor (Ishiwaki, 2013).
First, the target taste profile, such as very strong bitterness and moderately strong sourness, is specified. Next, the taste of each coffee raw material is measured using the taste sensor. Then, an optimization calculation is carried out with the superposition of the obtained taste profiles. The inventory status and purchase price, such as which coffee beans can be purchased at a low price, are also considered in the optimization calculation. Coffee blending used to be a trial-and-error process, but it can now be performed objectively on the basis of a database of quantified tastes. Consumers can also be shown objectively measured taste profiles to give them an idea of what a product tastes like.
The quantification of taste enables the selection of only necessary coffee beans, leading to improved purchasing and manufacturing efficiencies. The development period is also shortened because candidate coffee beans can be selected from the database to achieve the target taste profile. It also enables product manufacturers to keep the price low without changing the established taste profile of a given product.
Fig. 8 shows the case of reproducing the soup of a famous ramen restaurant in Kyoto where customers often have to wait in line (Wako and Tajima, 2012). The octagon indicated by the thin lines and the circles (○) at “0” on the axes show the target taste profile of the ramen broth. The numbers indicated on the axes are the taste scores obtained using the taste sensor, similar to those shown in Fig. 6. As a first step, a trial ramen broth was made, which was not as palatable as the target ramen broth. The broken lines and solid squares (■) indicate the results of the taste sensor measurements for the trial ramen broth, showing that koku (rich taste), the aftertaste resulting from umami, was lacking. Having identified the problem, koku was increased by changing the ingredients and the cooking procedure. Finally, a soup with the taste profile indicated by the bold lines and filled solid (●), which was almost the same as that of the broth of the famous ramen restaurant, was achieved. The developed ramen broth was as palatable as the broth of the famous ramen restaurant and this product was soon put into practical use. Using the taste sensor, it is possible to visually display how palatable a food is. As a result, a food product with a desired taste can be produced within a short period of time.
Food development in cooperation with famous ramen restaurant. From Wako and Tajima (2012).
As explained above, if taste, an attribute of food, can be quantified and visualized using the taste sensor, then is it possible to predict personal taste preferences by leveraging the ability of the taste sensor? The Taste & Aroma Strategic Research Institute Co., Ltd. has made it possible with the following procedure (Koyanagi, 2022). A questionnaire consisting of 15 questions that can be answered on a smartphone was prepared. Each question required the comparison of two items that were selected from the response patterns obtained by measuring food using the taste sensor. For example, the intensity of bitterness differs considerably between black coffee and coffee with milk. In this case, the question and choices were as follows: “How do you usually take your coffee? Black coffee/Other”.
The 15 questions were answered using a smartphone. Scores were assigned to the choices of questions in advance. A personal “preference vector” of the respondent, (for example: sourness, sweetness, bitterness, umami, and saltiness = 4, 2, 5, 2, and 3, respectively) was obtained when the answers were summed up. The same number of preference vectors as the number of respondents was obtained. Next, characteristic and representative preference vectors (preference patterns) were obtained using a cluster analysis of the preference vectors of the respondents. While any number of preference patterns can be obtained, a few dozen patterns were typically obtained from about 30 000 respondents.
Fig. 9 shows the procedure and the obtained preference patterns. For example, the pattern of preference for bitterness was characteristic of middle-aged men, while the patterns of preference for umami and sweetness were relatively common in younger people and women.
Prediction of personal taste preferences using questionnaire answered on smartphone (provided by Taste & Aroma Strategic Research Institute Co., Ltd.).
Then, choices that respondents would prefer were predicted for several food products on the basis of the above results, and a survey was conducted to see if the prediction was correct. The rate of matching answers was approximately 80 %. This result indicates that personal taste preferences can be predicted using the taste sensor, i.e., the taste sensor can not only be used for “measuring products”, but also for “knowing people”.
Some applications of the taste sensor were explained above. The taste sensor is commercially available and used around the world. However, it measures the taste of food in its static state. In the case of medicinal tablets, the change in taste as they disintegrate has been dynamically measured (Uchida et al., 2013). Nevertheless, there have been no reports on measurement of the dynamic changes in the taste of foods such as sweets, jiaozi (dumplings), hamburger steak, or sushi, which have a more complicated taste profile.
In the previous study (Toko, 2022), we visualized the changes in the taste of food during chewing using a model food that has a three-layer structure, i.e., a salty gel, mashed potato, and a wafer. The taste sensor measurement showed that salty substances were gradually released from the salty gel with an increase in the number of repetitions of application of a mashing force meant to simulate chewing (30 repetitions in total) in the case of a single ingredient (top left of Fig. 10). The numbers 0, 1, 5, 15, and 30 shown below indicate the number of repetitions of the mashing. The release of taste substances from the mashed potato (bottom left of Fig. 10) was more rapid than that from the salty gel. Saltiness, umami, and sweetness were dominant in the mashed potato, while bitterness was only perceived in the wafer (top right). The changes in the taste of the model food (bottom right) therefore mainly reflected the changes in the taste of the salty gel and mashed potato.
Radar charts representing changes in five taste qualities during chewing. From Toko (2022).
Two kinds of sensory test were also performed on the model food using the temporal dominance of sensations (TDS) method (Pineau et al., 2009). Three components, i.e., mashed potato, salty gel, and wafer, were selected in one sensory test. The model food was swallowed for about 5 s. The result showed the wafer was the dominant component, followed by the salty gel, in the initial stage of chewing. Then, the mashed potato was the dominant component in the middle and final stages. In the last part of the final stage, the wafer began to be perceived. Another sensory test related to eight attributes, i.e., sweetness, umami, sourness, bitterness, saltiness, astringency, smell, and texture. In the initial stage, the texture was found to be dominant, followed by saltiness. In the middle stage, umami was dominant. In the final stage, smell, texture, and bitterness were dominant. By comparing the results of the above two sensory tests, we concluded that texture and bitterness were dominant in the wafer, saltiness was dominant in the salty gel, and umami was dominant in the mashed potato.
We obtained good agreement between the results of the taste sensor measurement and those of the sensory tests using the TDS method: the saltiness of the salty gel and mashed potato was perceived in the initial and middle stages of chewing of the model food. Subsequently, the umami of the mashed potato was perceived from the middle stage onwards. As for the wafer, its taste was weak, but the texture was the main attribute perceived by the panelists. In this way, we were able to quantify the changes in the taste of a model food perceived during chewing. Visualizing the release of taste substances along with the disintegration of food in the mouth using the taste sensor will contribute significantly to food design.
The necessity and convenience of the taste sensor will increase in this era of information technology (IT) and AI. Taste can be recorded and reproduced anytime and anywhere. Recently, a new business has emerged (Kamiya and Sakashita, 2022; Kanno et al., 2019) that allows consumers to search for food that matches their preference by reading the bar code of food, converting it to a quick response (QR) code in which their personal preference for food and the taste data obtained using the taste sensor are stored, and showing the food that matches their preference on a tablet at a store (Fig. 11).
Food search service in IT era (provided by Cultivate Japan Co., Ltd.).
The Taste & Aroma Strategic Research Institute Co., Ltd. owns the taste data of over 100 000 commercial food and beverages obtained using the taste sensor. The Institute also offers a service of comprehensively analyzing scientific data, such as taste data, and market data, such as point of sales (POS) data, and providing information on the taste preference trends of consumers. Government offices also use the services to discover little-known local foods and ingredients and demonstrate the characteristic flavors scientifically as new branding strategies for supporting the development of rural areas.
It is vital to visually quantify the taste profiles and qualities of local food products to promote the branding of local food in the era of globalization. We have a natural desire to eat food that suits our preferences even if the feeling of what is palatable varies among individuals. Consumers feel more confident in buying food if its taste profile is indicated because they know they are buying food that suits their preferences. As a result, their satisfaction will increase. The delivery of this information is important because consumers are attracted to promotional messages based on objective facts. The taste sensor enables the delivery of such information about the taste profile of food.
Sensors for other senses, namely, smell, eyesight, hearing, and touch, may also be required to evaluate the palatability and quality of food. In the information era, the sharing of databases created using such sensors and a database created by human sensory tests will lead to the development of new products based on the results of the analysis of POS data coupled to taste and smell data, the development of three-dimensional (3D) food printers, and the improved productivity of high-value-added food. The sharing of these databases also enables the matching of supply and demand, the passing down of the skills of artisans and their transition into explicit knowledge, the seamless sharing of value information from production sites to consumers, and the formation of new food export markets.
Recently, Itochu Corporation, the Taste & Aroma Strategic Research Institute Co., Ltd., and WingArc1st Inc. jointly launched a digital transformation (DX) support service “FOODATA” for the planning and development of food products (Fig. 12) ii). FOODATA is a data analysis tool that can combine “product data” on the taste, nutrition, and ingredients of food and “human data” on the behaviors and preferences of consumers, such as identification point-of-sale (ID-POS) data, awareness, and reviews, and allow visualization of the analytical findings on a dashboard. This service provides food companies with an environment for the efficient validation of ideas by offering solutions to three major challenges in the process of product planning and development: 1) providing evidence to back up “intuitions and experiences”, 2) shortening the time required for analysis, and 3) reducing the cost of data acquisition. In this way, the service supports the enhancement of product planning and development capabilities of food companies.
DX support service “FOODATA” for planning and development of food productsii).
This article focused mainly on the applications of the taste sensor and also referred to the development of recent applications. The environment surrounding the taste sensor as well as the taste sensor itself has progressed considerably. In this way, it can be said that we are entering an era where we can increase our enjoyment of eating.
Acknowledgements This work was partly supported by the Cabinet Office of the Government of Japan, the Moonshot Research and Development Program for Agriculture, Forestry and Fisheries (Funding Agency: Bio-oriented Technology Research Advancement Institution; Grant Number JPJ009237).
Conflict of interest The author holds stocks in Intelligent Sensor Technology, Inc. and Taste & Aroma Strategic Research Institute Co., Ltd.
