The 4 Requirements for Evaluating taste Qualities
Taste and smell are among the 5 human senses, the most difficult to evaluate. The senses of sight, hearing, and touch are perceived only through well defined physical properties. For this reason, the development of sensors replicating those senses happened much earlier. When it comes to evaluating taste and smell, the perception mechanisms are much more sophisticated. The number of receptors and the number of active molecules involved is the source of greater complexity. We have already defined in a previous article what is the electronic tongue and what is its working principle. The electronic tongue is an analytical solution capable of evaluating taste. Similarly to taste perception in humans, there is an interaction between its sensors and the taste active factors found in the sample. The sensors translate this information from the sample into a signal that feeds the electronic tongue artificial brain. Based on those readings and the electronic tongue artificial neural network, those electrical signals are converted into taste data. You then not only know, what are the different tastes present in the sample but also, what are their intensities. To make sure the final result is pertinent, the initial reading done by the taste sensor needs to be as accurate as possible.
At New Food Innovation, we are agents in the UK for Japanese company Insent (Intelligent Sensor Technology). We offer a range of demonstration and analytical services as well as advice and support. For this reason, the present article focuses on the taste sensors used on the latest version of their electronic tongue the e-tongue TS-5000Z.
What is a taste sensor?
The design of the sensor is unique to Insent and relies on an artificial lipid-based membrane. This one consists of a combination of lipids, plasticizers, and PVC (polyvinyl chloride). Its exact composition and the proportion used will change for each sensor. This membrane is about 200μm thick and is attached to the rest of the sensor. Through changes in the formulation, they could optimize the properties of the membranes, and their ability to sense the different tastes. As for human taste perception, each sensor responds to a range of compounds based on their electric charge density and their hydrophobicity. Those advanced taste sensors were improved and re-engineered so each of them (almost) solely responds to a given taste. The following table, adapted from the publication of Toko et al. 2016 [1], gives the list of all the exploitable taste information taste sensors can give. First, there is an initial evaluation of the taste, which is the initial taste information. After a brief rinsing, mimicking the action of swallowing, some taste compounds still interact with the membrane. This interaction generates a long-lasting residual taste, the aftertaste. Taste sensors are thus able to evaluate the 5 tastes, their aftertastes, as well as astringency. This is also the most accurate way to get readings of some specific bitter substances that are then, better distinguished
How does a sensor read taste?
The measurement system works with the evaluation of electrical potentials. Each sensor has a native electrical potential associated. During measurement, interaction with taste compounds affects this electrical potential. This response is dependent on the taste sensor and the taste substance. This change in potential of the lipid-polymer membrane reaches an Ag/AgCl electrode through a highly conductive solution. Each electrode then connects to an amplifier and the recorded signal sent to the computer. As a final step, the computer converts each signal into exploitable data.
Why should you trust taste sensors?
The ultimate goal of the taste sensing device is to establish an international standard for measuring taste. Indeed, there is to date no internationally exploitable way to measure tastes. Some anecdotic exceptions exist. Yet, they are not accepted or used globally across all industries. For example, the International Bittering Units scale, or IBU scale, gives an estimation of the bitterness of beers. Most commonly, it uses spectrophotometry to get an evaluation of the iso-alpha acid content of the beer, which is the main bitter compound extracted from hop during brewing. From this measurement, we can calculate an approximate bitterness value on the IBU scale. This method might be well accepted within breweries but doesn’t take into account all bitter molecules found in the product. Also, it only applies to beers and can’t transpose to other products. Thus, there is a real need for an efficient alternative. The electronic tongue does fill that gap. It has advanced taste sensors that sense taste qualities and provide users with an objective taste evaluation of any sample.
To be trustable and simulate human perception, four requirements have to be fulfilled by the sensors :
Global selectivity to all compounds responsible for a given taste
Match human taste perception threshold
High correlation with human sensory scores, with similar taste rating
Detection of interaction and report taste-enhancing, taste masking effect.
Taste sensors have a perception threshold that matches the human tongue
The threshold is the smallest concentration required of a taste substance to be perceivable. All the challenge is to identify the threshold for each taste for humans and ensure that the taste sensors have the same.
Why does it matter?
A sensor can read appropriately the taste of a sample and give a quantitative reading. If the reference point isn’t the same, a coffee might score 1 in bitterness in sensory analysis and 4 in an electronic tongue evaluation. Both are the same sample, but both scales don't have the same 0. It results in a significant shift that we need to avoid.
The threshold for each taste is different, mainly as a consequence of human evolution. The body requires high quantities of salts and carbohydrates or sugars. Sweet foods, containing sucrose, glucose, and other sugars are recognized and identified as a source of energy for the body. The cations contained in the salty foods maintain the electrolyte balance, which is also an absolute necessity. The threshold for taste recognition is high for those essential substances to promote an adequate intake. For example, we start tasting salt (NaCl) from 0.58g/L and sucrose from 6.85g/L. Umami is also a useful taste to recognize to detect sources of nutrients. Indeed, its perception is triggered by certain amino acids and proteins. Sourness, which comes from acids (mainly acetic, lactic, citric acids) is an indicator of food decomposition or contamination by certain microorganisms. Perceived as a signal of decay, the sensitivity to those compounds is higher. The associated threshold is, for example for citric acid, around 0.38g/L. Bitter molecules are often found in poisonous foods. A bitter taste can be a warning of toxicity and helps to prevent their consumption. Thus, a very low threshold for those substances is an advantage to detect as soon as possible a potential danger. The threshold concentration for quinine is 0.0026 g/L. For strychnine, a poison extracted from the strychnine seeds (and often exploited in detective novels) shows its bitterness for a concentration as low as 0.00003 g/L [1,2].
Sensors and humans are on the same starting point, sharing the same threshold for taste perception. Next step is to ensure they evaluate changes in taste intensities the same way.
Taste sensors show a global selectivity
There are only a limited number of taste receptors in the mouth. On the other side, there are millions of taste substances responsible for the five main tastes. This implies that a wide range of molecules can stimulate each taste receptor and produce a response. Post brain processing, hundreds of interactions lead to only five tastes. This complicated progress has to be simulated on the taste sensors. They must respond consistently to a range of compounds, as would the human tongue do.
All the molecules and ions detectable by the taste buds can be plotted on a chart based on their charge (ionic strength) and their hydrophilicity (polarity). Almost all the elements responsible for a given taste end up in the same area of the chart. They have chemical similarities and generate the same taste, as seen on the following graph:
The taste sensors show global selectivity, as would the tongue do. This is the ability to respond in similar ways to a range of different stimulating molecules sharing chemical properties. Salts such as NaCl are hydrophilic so they are easily hydrated in an aqueous solution and have a strong ionic charge once dissolved. They are hardly adsorbed by the hydrophobic part of lipid molecules contained in the sensor membrane. Sour substances such as acetic acid have similar behavior toward lipids but have a much lower ionic charge. Bitter and astringent materials are slightly soluble in aqueous solutions as they are more hydrophobic. This hydrophobicity also helps the adsorption on the tongue, causing a lasting aftertaste. Sensors have been conceived, tested, and selected for their global selectivity. The following trial was done with the sensor in development to sense astringency.
The sensor does respond to the range of samples assessed. In the meantime, tests were also done to make sure of the specificity of the sensors. Indeed, each taste probe needs to be responsible for a single taste. Insent research team and their academic partners managed to achieve this goal almost entirely. Only a few exceptions are identified and automatically taken into consideration during data processing and taste score calculations ensuring the reliability of the results.
Taste sensor results correlate with sensory scores
Global selectivity work was carried out for numerous compounds and all the taste sensors. This work also confirmed that the taste sensors output correlate with the human sensory score. This means that there is a link between the results of both entities. The correlation coefficient, which is an indicator of the strength of the relation between them is very high for all tastes. So, when a human finds that one sample is two times more bitter than another, the sensors will also give the same conclusion.
The last step is to make sure that those sensors do not only correlate with sensory scores but give the same values. The information collected from the taste sensors must be converted into a unit that can easily be understood and is as close as possible to a sensory scoring. The achievement of this conversion uses 3 principles:
The smallest detectable increase for the gustatory sense is about 20%
The relationship between a gustatory stimulus and the corresponding perceived intensity is logarithmic.
The reading of taste sensors is proportional to the concentration of taste active compounds
The first 2 being based on the Weber-Feshner law, as demonstrated by studies carried out in the 40s [3,4].
To create this taste scale, arbitrarily, two rules were set as a starting point:
The first unit (value 1) corresponds to the threshold of taste perception
Any change of 1 unit on that scale corresponds to a perceivable difference in taste
From there, we can conclude that an increase of 1 unit corresponds to an increase in 20% in tastant concentration. If we take saltiness as an example, there will be 1 unit difference between a sample A containing 1g/L and a sample B containing 1.2g/L. Then, we can measure those two samples with the taste sensors. We can define that 1 unit corresponds to the difference between those two readings. Repeating this for a range of concentrations permits us to define precisely the scale. In the meantime, we are making sure that the taste score given by the e-tongue corresponds to scores given during sensory evaluations. This is an example with the sensory score of astringency and the corresponding values on the taste scale calculated using the sensor outputs.
Taste sensors can detect interactions
Taste substances interact with each other. The taste intensity increases or decreases depending on the types of taste substances involved. The taste sensors are able to detect interactions between taste substances. This property is especially important for pharmaceutical applications where bitterness masking plays a big role to increase palatability. Even if they are less common, the e-tongue can also measure synergetic, or taste-enhancing effects. Once again, very similar results were obtained when comparing sensory scoring and e-tongue scoring.
With great care in the design and evaluation of their taste sensors and an advanced definition of their taste scale, Insent managed to produce an analytical tool that is capable of measuring the tastes of any sample. It has major advantages of being objective with results very close to a sensory appreciation. Even if sensory scores can change due to social, cultural, or physiological factors, taste evaluation using the electronic tongue can be taken as a reference point that can provide a measure of taste as an international standard.
Florian Woisel
Sources:
[1] Toko, K., Tahara, Y., Habara, M., Kobayashi, Y., Ikezaki, H., & Nakamoto, T. (2016). Taste sensor: electronic tongue with global selectivity. Essentials of machine olfaction and taste, 87-174.
[2] Purves, D., Augustine, G. J., Fitzpatrick, D., Katz, L. C., Lamantia, A. S., McNamara, J. O., & Williams, S. M. (2001). Taste perception in humans. Neuroscience, 2nd Edition. Sunderland MA: Sinauer Associates.
[3] Lewis, D. R. (1948). Psychological scales of taste. The Journal of psychology, 26(2), 437-446.
[4] C. Pfaffmann, in, J. Field, (ed.) Handbook of Physiology, Neurophysiology,
Am. Physiol. Soc., Washington, DC, 507–533 (1959).
Other notable source : Toko, K. (Ed.). (2013). Biochemical sensors: mimicking gustatory and olfactory senses. CRC Press.