Panel training on odour and aroma perception for sensory analysis

The core of food flavour is formed not by the five basic taste types – sweet, sour, bitter, salty and umami – which are perceived on the tongue via the taste organ, but instead by the large number of substances and sensations that are perceived via the olfactory organ. These include both volatile odours and flavouring substances dissolved in the food (Legrum W., 2015, p. 26ff). In sensory analysis, the odour and aroma forms part of the spectrum to be analysed, alongside appearance, texture and basic taste. This Expert Knowledge Report aims to provide an insight into the training of odour and aroma perception in sensory analysis. This is because the perception of olfactory sensory impressions is a skill that panel members within sensory analysis need to train regularly. Although the complexity of the aroma perception (orthonasal and retronasal smelling) is explained, the focus of this Expert Knowledge Report lies on the training of the orthonasal smelling, i.e. the perception of volatile flavour compounds in the form of odours.

1. Introduction

Odour affects our life in many situations. Alongside the emotional and social aspect of odours, e.g. their influence on likes and dislikes, positively stimulating odours and aromas also serve to quicken the appetite, while off-aromas and odours help consumers to recognise dangers, for instance in spoiled foods. The different stimuli (totality of all gustatory, olfactory, haptic and trigeminal stimuli) perceived via the tongue and in the oral cavity during eating account for a relatively small part of the flavour. One central aspect is the odour of the dish, in other words the volatile compounds of its aroma. Before the first bite the nose tests how the dish smells and arouses various associations and emotions in the brain. Complementing this, chewing dissolves the soluble flavour compounds and allows the perception of the aroma via the mouth-throat-nose connection which finally leads to the product enjoyment in its full complexity. Through their sense of smell, humans can perceive, differentiate and recognise a large number of different odours. However, odour perception and recognition depend on experiences gathered in connection with various odours and on odour training.

Sensory food tests are an important component for ensuring high product quality. Sensory analysis comprises the testing of appearance, taste, odour, consistency and texture. In order to obtain the most informative results possible and a high degree of “measuring” precision in human sensory analyses, the tasks are carried out with a trained panel. Especially when smelling via the nose (olfactory sensory impression), untrained analysts find it difficult to recognise fragrances and odours perceived and to describe these in words.
This is attributable to the fact that in the case of odour we have not just five qualities of perception, as in taste, but instead it is possible to distinguish between a vast number of different olfactory impressions. These are caused by thousands of perceptible odour substances, which moreover can occur in any different combination. Classifying these odour impressions in odour groups at least is difficult and needs to be trained in order to be able to build on an odour memory.

2. Sensory physiology of the sense of smell

The sense of smell, also called the olfactory sense, serves to perceive aromas. Here we distinguish between perception of the odour of volatile substances by inhalation through the nose (direct smelling, orthonasal perception) and perception of aroma substances that are dissolved in the food and reach the olfactory epithelium in the nose via the mouth-throat-nose connection during exhalation (retronasal perception).

2.1 The olfactory and the nasal trigeminal system

The actual olfactory system in the narrower sense of the word and the nasal trigeminal system are involved in olfactory perception (perception of odour). These are two physiologically different systems. The olfactory system is localised in the upper area of the two nasal cavities in what we call the olfactory zone (Regio olfactoria) or olfactory mucosa. This covers about 5 cm² in each nasal cavity. Here there are around ten million olfactory receptor cells, each of which expresses only one of the around 350 different odour receptors. The nerve fibres leading out from these olfactory receptor cells grow bundled through the ethmoid bone (a spongy bone of the neurocranium) into the cranial cavity to the olfactory bulb (Bulbus Olfactorius), which can be considered an “offshoot” of the brain and from which the central-nervous processing of the odour stimuli begins (Fig. 1). (https://de.wikipedia.org/wiki/Olfaktorische_Wahrnehmung, Legrum W., 2015, S. 5ff.) Outside the Regio olfactoria, volatile substances can be perceived in the nasal mucosa through the nasal trigeminal system. Here the facial nerve “Nervus Trigeminus” transmits sensations such as burning, pungent, acrid, cooling in the nasal cavity (nasal trigeminal, volatile substances) and the sensations hot or sharp, tingling, cooling in the oral cavity (oral-trigeminal, volatile and non-volatile substances).

By contrast with the actual olfactory system, these sensations – except for the perception of hot and cold – are not perceived via specifi c sensory cells, but instead via free ends of nerve cells of the trigeminus nerve (Fig. 2). Many volatile substances trigger a sense of smell both in the olfactory system and in the nasal trigeminal system. These include volatile acids (e.g. acetic acid, butyric acid …), volatile pungent substances (including isothiocyanate from mustard, propanethial- S-oxide from onions, mucosa-irritating chemicals) and (pseudo-) cooling substances (menthol, eucalyptus oil). Carbonic acid/carbon dioxide on the other hand is an example of a substance that can only be perceived purely trigeminally (and not via olfactory paths in the narrower sense), and vanilla is perceived exclusively via olfactory paths. When humans inhale through the nose, the air fl ows through the nasal atrium, the nasal cavity and the various nasal passages until the odour substances interact with the olfactory receptor cells in the olfactory epithelium in the olfactory zone (Fig. 1). In sensory analysis a certain smelling technique is applied (frequent light sniffi ng), so that both nostrils can be activated at the same time and consequently the sensitive odour zones are reached better by the respiratory air (Derndorfer, p. 20ff.)

The volatile odour substances reach the Regio olfactoria as described before via the orthonasal perception track directly on inhalation or smelling of the food when it is taken into the mouth.

(Individual) flavouring substances are chemical compounds that can be allocated to different substance categories or flavour categories depending on their molecular structure. These include for instance aldehydes and ketones, alcohols, acids, esters (and lactones), terpenes, pyrazines, phenylpropanoids, phenols and various sulphur compounds. Another way of classifying them is on the basis of their origin, e.g. as metabolic products, or formed by microbial fermentation in the foods (constitutive flavouring substances), as well as by the influence of heat (process flavourings). (Legrum, W., p. 103 ff.)

3.3.1 Constitutively formed flavouring substances

The constitutive or native flavouring substances are essentially described as those that occur in fruit, vegetables, spices and herbs through the plant metabolism or that are formed during harvesting, storage, drying or (e.g. enzymatically) after chopping/grinding/hacking. They can occur both as flavouring substances experienced as pleasant, or as (untypical) flavouring defects. Constitutively formed flavouring substances can be grouped on the basis of their structures. Some of the substances groups are set out below:

• Carbonyls (aldehydes and ketones)
• Esters
• Lactones
• Sulphur compounds
• Pyrazines
• Terpenes
• Phenylpropanoids

The first group, the carbonyl compounds, is made up of the degradation products of fatty acids formed microbiologically or enzymatically, the aldehydes and ketones. They result from autoxidation of unsaturated fatty acids with the involvement of the enzymes lipoxygenase and hydroperoxide lyase. In this way a large number of carbonyls result in fruit and vegetables from oleic, linoleic and linolenic acid. These display a low odour threshold and an enormous variety. Carbonyl compounds are formed increasingly under the introduction of oxygen, e.g. by cutting up fruits, which also promotes release of the lipoxygenases from cell compartments and development of the flavour.
Certain carbonyl compounds are described with descriptors that are more typical of off-aromas, such as rancid, fishy (Z-4-heptenal), metallic (trans-4,5-epoxy-2-(E)-decenal), cardboard-like (E-2-nonenal) or old. 1-octen-3-ol and 1-octen-3-one on the other hand are perceptible with a mushroom-like, earthy aroma already at low recognition thresholds (< 1 ppb). Acetaldehyde is one of the best-known representatives of the aldehydes and is important for the fresh, fruity note in orange juices. The odour of the pure substance is described as artificially fruity and/or pungent. Further known representatives are the short-chain aldehydes hexanal and (Z)-3-hexanal. They are formed by degradation of the linolenic acid and are responsible for green odour notes. These and further known flavouring substances from the group of carbonyls can be seen in Table 2.
Esters belong to the second substance group. They are formed from organic acids and alcohols with removal of water. The functional group formed from the reaction of the carboxyl group of a carbonic acid (-COOH) and the hydroxyl group of an alcohol (-OH) is characteristic of esters and is described by the general form R – COO– R´. In many fruits esters belong to the determining flavouring substances. They are formed in cellular manner in the plant from various alcohols and from activated (fatty) acids (Acyl-CoA). Apart from a few exceptions, the odour thresholds of the esters are frequently not particularly low (Legrum, W. p.105ff.). A well-known ester is the acetic acid hexyl ester (=hexyl acetate) which contributes substantially to the flavour of apples. A further well-known ester is the butyric acid ethyl ester (ethyl butyrate) that is co-responsible for the flavour of pineapples (see Table 3).
In chemical terms, lactones are intramolecular esters of hydroxycarboxylic acids (hydroxy fatty acids) and form the third group. Like the open-chain esters they occur in many types of fruit. Oleic and linoleic acid, which are specifically oxidised and converted into hydroxy acids, serve as the source substances for lactone formation. Through the β-oxidation, rings of different sizes form after the shortening of the fatty acids. Most frequently five-membered rings, the γ-(gamma) lactones, or six-membered rings, the δ-(delta) lactones result. Examples of this include the γ-undecalactone (lactone of γ-hydroxy undecanoic acid) and the δ-dodecalactone, which smell fruity, like peach or coconut respectively (see Table 4).
Sulphur compounds play a crucial role as flavouring substances, above all in vegetables, and form the fourth substance group.
Mustard oil glycosides (glucosinolates) are compounds containing sulphur and nitrogen that are formed above all by cruciferous plants (Brassicaceae). These include the cabbage varieties (e.g. cauliflower, Brussels sprouts, white cabbage, broccoli), rucola, garden cress, radish and horseradish. The mustard oil glycosides as undecomposed compounds are mainly responsible for the bitter taste of these vegetables. During the disintegration process the mustard oil glucosides are, however, partly split by released enzymes (thioglucosidase, myrosinase) into sugar (glucose) and the corresponding mustard oil (isothiocyanate).
These mustard oils are either non-volatile and are perceived orally-trigeminally as spicy, or they are volatile and then a pungent impression is added via the nasal trigeminal avenue. For example black mustard contains the glucosinolate Sinigrin, which is hydrolysed during chopping, chewing or boiling the mustard seeds to allyl mustard oil (allyl isothiocyanate). Amino acids containing sulphur and their degradation products characterise the flavour of plants in the leek family (Allicaceae), which include leek, garlic and onions. For example, after chopping garlic the amino acid alliin containing sulphur is degraded by the enzyme alliinase to form allylsulfenic acid, from which among other products diallyl disulfide smelling of garlic results via the intermediate stage of allicin. 

In onions, the tear-triggering propanethial oxide first develops from isoalliin through alliinase, which is further degraded to form dipropyl disulfide. (See Table 5) In addition there are sulfurous flavouring substances with extremely low recognition thresholds, such as the 4-methoxy- 2-methyl-2-butanethiol, which as blackcurrant can already be perceived as of 0.02 ng/l, but in higher concentrations then smells of cat urine.
Pyrazines also occur constitutively above all in vegetable varieties. A characteristic of this fifth substance group, which belongs to the heteroaromatics, is a ring scaffold with two nitrogen atoms. Pyrazines are responsible for the characteristic, green, vegetative odour notes in sweet peppers, chillies and carrots (see Table 6). Pyrazines play a major role as roasting flavours when foods are heated (see processively formed flavouring substances).

The Maillard reaction, a non-enzymatic browning reaction, is the most important of these. The source substances of this reaction are amino acids and reducing sugars (chiefly glucose, fructose, maltose and lactose), that react with each other under the influence of heat (as of around 120°C). A large number of reactions take place at the same time and after each other here, leading to browning and flavour formation. The material diversity of the resulting flavouring substances is very large, for example furanones, thiazoles, pyrroles and pyridines occur. Examples of these are shown in Table 9.

In the case of Strecker degradation of amino acids, a further important reaction in the formation of process flavourings, amino acids are desamined and decarboxylated under oxidative conditions. The products α-amino ketones and Strecker-aldehydes result from this, including for example the flavouring substances ethanal (acetaldehyde), 3-methylbutanal and methional (see Table 10).
Caramelisation is the third type of reaction leading to the formation of process aromas. Here monosaccharides or disaccharides (e.g. glucose, fructose, sucrose) are heated dry or under acid conditions until they dehydrate while browning. Flavour-active substances resulting from this are furane, pyrone and pyrane derivatives. Caramel is a typical product obtained by caramelising (see Table 11).
The volatile, flavour-active compounds are leading components of a flavour. It is not important whether they are formed constitutively or processively. They lend the food its characteristic flavour.

Aroma and flavour wheels are standardised systems for sensory classification of flavouring substances, mainly with a reference to individual food groups. They aim to support the testers and analysts in perceiving, recognising and describing flavours. For this purpose overarching flavour groups (e.g. fruity) are linked with various sub-descriptors beneath them for specific flavours in this group (e.g. unripe apple). In this way they help a panel to find a common language regarding the verbal description of flavours. Odour references for the most important of these descriptors should be made available for training. Aroma and flavour wheels are available in standardised form for different product groups in the food sector, e.g. wine, bread, cheese, beer, coffee or tea. However they can also be tailored individually to the language of a panel for a product group. Recognised aroma and flavour wheels of a product group can be considered as standard for describing and identifying aromas and flavours and serve for comparability of sensory descriptions, thus making sensory analyses objective. The structure of the wheels, generally consisting of three concentric circles of sensory descriptors and references, follows a largely uniform principle.

The following overview lists individual flavours that are suitable for odour training. They are based on proposals of DIN EN ISO 8586:2014-05 and in addition include key flavours and off-flavours occurring in many food groups taken from the DLG Basic Vocabulary.

Food flavours
The training substances listed in the following Table 12 with their respective odour characteristics fall under the category of food flavours for training purposes. Each flavouring substance possesses specific properties. These include e.g. their occurrence in foods and the description of the odour impression.

Off-flavours
Possible substances in the category of off-flavours that frequently give foods the impression of quality defects are listed with their odour characteristics in Table 13.

4.2. Forms of offering and presenting odour substances

The offering of aromatic substances depends on the nature of the flavour training session or test. Generally, the intensity of the odour references should lie well above the odour identification threshold of the analysts. However, the intensity should not be chosen too high, as with some flavouring substances the aroma impression changes with increasing intensity and can influence other analysts sitting close by.

As already mentioned, building up an odour memory is relevant above all for members of a descriptive panel, as is the ability to describe unknown odours in words. The odour memory enables analysts to recognise important aromas reliably, to describe them verbally and to allocate odour samples (see Matching Test).

5.2.1 Odour recognition test with standardised or non-standardised flavourings

The odour recognition test is used to determine whether an analyst can allocate a description correctly to odour samples (supported test). For screening, an odour recognition test is frequently conducted with standardised flavours. Here standardised odours are presented to the analysts at the same time in the form of isolated individual aromas for sniffing and odour recognition. The number of samples should not exceed seven to nine in order not to over-stress the olfactory system. The samples are placed together on the table and tasted in parallel. After sniffing them, the analysts mark the appropriate aroma description on a standardised test sheet which contains the relevant odour attributes. (See Fig. 8)

Odour identification tests can be carried out not only with defined individual substances or defined substance mixtures, but also with non-standardised flavours. Here the non-standardised flavours are generally produced with spices or foods or parts of foods in which the flavours to be identified occur not in isolated form, but integrated in a matrix. In some cases this type of odour recognition is easier for the persons being tested, as e.g. the complex mixture of all typical flavour-active substances is present in the food and not just one key flavouring substance.

5.2.2 Matching-Test

A further, relatively simple variant of the identification test is the matching test, of which there are many variants. For instance the test persons can first be presented with various samples of odour references for training and “familiarisation”. They are then given a series of coded samples that have to be allocated to the odour references in the original set. In a further simple variant, the analysts are given six to ten coded samples, of which two to three in each case contain the same flavouring substance. The analysts first have to sort and group these and then if required describe the odour.

5.2.3 Descriptive test

The descriptive test is used to practice verbal descriptions of unknown odours. By contrast with the recognition test (see 5.2.1, Fig. 8), no odour attributes are provided for selection. A descriptive test is also a component of a profiling test. Precise recognition of the chemical substances presented is not so important here, but instead the most objectifiable description possible of the sensory impression, so that in this case the method of the “simple descriptive test” can generally be used. Here the power to express sensory perceptions is required in order to develop a descriptive vocabulary for describing the flavour profile of the product samples presented. This type of test also allows a large number of other variations. For instance the analysts can be given an aroma/flavour wheel as an aid, or else lists of attributes with possible descriptors can be made available.

In practice it may be desirable to detect off-flavours as sensitively as possible with the help of a panel. In this case it is interesting to determine the odour sensitivity with respect to selected substances in the form of stimulus/perception and recognition thresholds. For intensity assessment the analyst must be able to reliably recognise slight differences in intensity. This can be achieved for instance with the help of discrimination tests and by determining the difference threshold.

5.3.1 Odour sensitivity

The stimulus threshold and the recognition threshold of an analyst for an odour substance can be determined in order to describe the odour sensitivity. These threshold values can vary widely depending on the medium used.

Accordingly several powers of ten can lie between the recognition thresholds of a fl avouring substance in air, water and beer. There are also differences between orthonasal (sniffi ng of a solution) and retronasal (tasting) perception of a fl avouring substance.
To determine the recognition threshold, a test substance is presented in rising concentrations for sniffi ng or tasting. Re-tasting is not allowed. This test fi rst fi nds out as of what concentration an odour impression is perceived (stimulus/ perception threshold). Then with rising concentration, the analysts smell and taste further and in each case state when a difference in concentration is perceived. Provided that the odour is identifi ed clearly, the analyst notes what odour this is (recognition threshold) and tests the sequence up to the end. Experience has shown that when this test is repeated there is frequently a risk of the analysts guessing and not concentrating on the actual test. Alternatively the stimulus or recognition threshold can be additionally secured or verifi ed with a discrimination test, such as the Duo Test or Triangle Test. Recommendations on the individual concentrations of the solutions with the identifi cation thresholds and evaluation options are listed in DIN 10959. (See Fig. 10)

5.3.2 Recognition of differences in intensity

Alongside the qualitative recognition of odours, recognition of intensity differences in standardised odours is also important within the context of screening and monitoring analysts. This ability is a prerequisite for assessing the intensity of flavours.
In principle all discrimination tests (difference or comparison tests) can be used in order to examine the ability to recognise differences in intensity. (For further information on this contact sensorik@DLG.org).
Pairwise comparison tests and triangle tests with a standardised aroma in two concentrations are particularly suitable. If more than two concentrations are to be compared, the ranking test can be used. In this test at least three samples of a fl avouring substance with different concentration levels are produced and handed out together to the panellists for testing. The test task consists of arranging the samples in the sequence of rising concentration of odour intensity perceived (ranking) and noting this intensity sequence using the sample numbers (see Fig. 11). The flavouring substances selected for the difference test should be important for the product that is subsequently to be tested (see Table 12). Further information can be taken from DIN ISO 8587: 2010-08 Sensory Analysis – Test Methods – Ranking Test.

The intensity of an odour substance is evaluated with the help of a scale prescribed by the test leader. Both ordinal scales (verbal description of the extent of the attribute, for instance low – high (e.g. Fig. 11) and metric scales (extent of the attribute defi ned in fi gures, e.g. 0 to 5, see Fig. 12) are applied. Using the intensity scale is the subject of an analyst training session. Intensity evaluations are part of the profile tests and the tests of building products or materials entering into contact with food for unwanted odour substances. In addition to a verbal description of the intensity, reference materials for certain intensity stages or a whole series of intensity stages with a selected aroma are helpful.

Product-specific training is carried out after the intensity training for individual flavouring substances. Depending on the objective and the possible assignments of the panel, different product variants are offered for testing here. It should be noted that ideally the test sequence is evaluated in triple repetition, as a final selection of the chosen panellists is possible by looking through the data gained in this way and comparing the panel performance characteristics.

Those analysts who display a high standard deviation between the repetitions in the test results or who cannot discriminate well between the products are not suitable for a quantitative panel. Panellists who display only a low level of diffi culty in this connection can be qualifi ed appropriately via further product-specific training units.