Introduction

Wine mouthfeel is one of the most important characteristics for wine appreciation, along with appearance and flavour (Araujo et al., 2021; Hopfer & Heymann, 2014; Sáenz-Navajas et al., 2016). While wine researchers have largely invested efforts in disentangling the physicochemical and psychological underpinnings of wine aroma, much less attention has been paid to wine mouthfeel. In this regard, in his book, Rigaux (2015) states that current wine tasting focuses mainly on the olfactory and aromatic attributes of wines, leaving aside wine mouthfeel. The author calls for wine tasting in which wine appreciation focuses on mouthfeel, consistency, suppleness or viscosity to infer wine quality. Scientific studies that relate wine appreciation to wine sensory properties tend to be limited to measuring a few mouthfeel-related terms, such as overall astringency, drying, hot sensation, body and viscosity (Hopfer & Heymann, 2014; Paissoni et al., 2023), which contrasts with the large range of aroma terms used to characterise wines (Noble et al., 1987). Interestingly, an increasing number of studies have been published on this topic in the last decade (Araujo et al., 2021; Paissoni et al., 2020; Rinaldi et al., 2020; Niimi et al., 2017; Sáenz-Navajas et al., 2017; Vidal et al., 2015). In their pioneering work, Gawel et al. (2000) published a red wine mouthfeel wheel created by wine experts. The authors provided semantic definitions for each of the terms and defined a limited number of them by giving handfeel touch examples when a consensual verbal definition could not be achieved. Despite the importance of this work for the study and evaluation of wine texture, the use in scientific publications of the mouthfeel vocabulary generated in this work is limited, which could be due to a certain lack of clarity in the terminology, as well as the necessity for extensive panel training. Two main categories appear in the wheel: astringency and non-astringency terms. Within the non-astringency category, six more specific categories are included (flavour, acidity, weight, texture, heat and irritation), which refer to several sensory modalities (aroma, taste and mouthfeel) and could lead to misunderstandings. The astringency category is made up of seven specific terms: particulate, surface smoothness, complex, drying, dynamic, harsh and unripe, whose definitions are somewhat confusing, as some terms are linked to hedonic properties. For example, the complex category is defined as “a positive hedonic grouping consisting of an amalgam of pleasing astringency sensations, flavour and balanced acidity”. Similarly, the terms harsh and unripe are associated with negative hedonic properties. In addition, some of the terms included in the astringency category are related to other sensory modalities that differ from mouthfeel, such as aroma for rich or flesh, or tastes for hard (bitterness), green (acidity), supple (acidity) or sappy (acid, bitter), which leads to multidimensional terms. There are also terms that are defined on the basis of the level of intensity of a given attribute. These include aggressive (excessive astringency), abrasive (excessive astringency of a strongly roughing nature), soft (light and finely textured astringency) or supple (low to moderate astringency). Lawless and Civille (2013) have already highlighted that terms related to mouthfeel should be reviewed and clarified in order to broaden their application in formal descriptive analysis. In response to this concern, Paissoni et al. (2023) reviewed recent publications focused on understanding and describing mouthfeel. They concluded that the scientific community has not yet defined a common vocabulary to communicate oral sensory perceptions, calling for the homogenisation of definitions and references, and consequently of the use of mouthfeel descriptors. The authors conclude by stating: “When we judge a wine, we should all speak the same sensory language”. To achieve this goal, we should start by understanding the concept of mouthfeel, which requires the examination of the processes linked to this concept, namely sensation and perception.

Sensation, in general, can be defined as the input resulting from the activation of receptors in response to a given stimulus (Coren, 2003). The sensation of mouthfeel is based on inputs from mechanoreceptors present in the oral surface that respond to mechanical events, mainly stress and strain caused by indentation and the stretching of tissues (Foegeding et al., 2015). Studies of skin receptors in the human hand (mainly the palmar side of the hand and fingers) provide a basis for understanding human oral receptors, as both surfaces are innervated by very similar nerve fibres (Di Stefano & Spence, 2022). The only known difference between facial and extra-facial afferents are those associated with a class of receptors (Pacinian corpuscles) that have not been observed in most oral tissues (Trulsson & Johansson, 2002). It is important to note that a particular mouthfeel is not the result of the activation of a single mechanoreceptor, but of a combination of signals activated in a receptive field or population of receptive fields (Foegeding et al., 2015). The analogy between oral cavity and fingertip sensation has also been reported for mouthfeel, with the simultaneous activation of multiple mechanoreceptors rather than a single one (Weber et al., 2013).

Once a stimulus has interacted with the receptors, and the signals have been generated and integrated to form the corresponding sensation, it must be interpreted by the individual before a perception can be formed (Coren, 2003). To interpret the sensation, several cognitive processes take place in an attempt to extract useful information from the signals elicited by the stimulus. Among these, cross-modal integration between two or more different sensory modalities plays an important role as demonstrated by the McGurk illusion (McGurk & McDonald, 1976), for example. In this illusion, the visual information received from observing a person speaking alters the auditory perception of the sound, due to the incongruence between the visual and auditory inputs (seeing “ga” and hearing “ba” give rise to the perception of “da”). This phenomenon of people connecting inputs from different sensory modalities is also known as sensation transfer (Pramudya et al., 2020). Recently, Pramudya et al. (2024) provided empirical evidence of the sensory transfer between handfeel touch and mouth irritation mediated by capsaicin. Sandpaper material was found to be strongly associated with mouthfeel irritation, whereas materials such as silicone, sponge or stainless steel showed weaker associations. Other authors report cross-sensory influences of handfeel touch cues on oral somatosensory perception (Pramudya & Seo, 2019). For example, Piqueras-Fiszman and Spence (2012) report an increase in crunchiness and hardness of biscuits presented in a container with a rough sandpaper compared to the same biscuit presented in a smooth-coated container.

Given the importance of mouthfeel in overall red wine quality perception, and considering the existence of a sensory transfer between handfeel touch and oral somatosensory perception, it seems plausible to develop a methodology that gives a complete picture of wine mouthfeel properties based on handfeel perception. Many psychological studies have been devoted to the investigation of handfeel perception (e.g., Picard et al., 2003). Most of these studies were based on the sorting of everyday surface textures (e.g., wood, sand paper and fabrics) coupled with multidimensional scaling to explore the handfeel perceptual space. Three main dimensions of texture emerged from these studies: roughness/smoothness, hardness/softness and coldness/warmness. However, Sakamoto and Watanabe (2017) suggest that the previous studies may have been reductive due to the limited number of handfeel dimensions represented. To overcome this shortcoming and to cover all the basic categories within the handfeel perceptual space, they developed a lexicon of touch perception as a guide for collecting materials. They claim that the generation of a detailed and reliable vocabulary is important for meaningful descriptions of perceptual spaces, and thus their strategy for collecting materials illustrating different perceptions of handfeel was based on the sensory vocabulary generated at the outset. In the present study, a comparable approach was adopted for the development of reference handfeel touch materials, which were created to illustrate the mouthfeel of red wine. This methodology was based on the hypothesis that cross-sensory associations exist between handfeel touch sensations and wine mouthfeel perception. The study built on the renewed interest in the collaboration between the arts and sciences. This transdisciplinary art-science initiative predates the twentieth century, when art and science were clearly linked. A clear example is the Renaissance period, when art and science went hand in hand, the clearest example being the works of Leonardo Da Vinci (Clark et al., 2020).

In this context, the objective of the present study is to develop handfeel touch references capable of describing the mouthfeel properties of red wine through a transdisciplinary approach. We first compiled a large number of mouthfeel terms used to describe red wine by different Spanish wine stakeholders, including people working in the wine sector, such as technical experts (i.e., winemakers) and people involved in wine distribution and sales, as well as regular consumers not working in the wine sector. Reference standards were then designed to illustrate the terms developed in the first step following an art-based strategy, working on the hypothesis that different mouthfeel terms generated by red wines can be illustrated by touch materials. Finally, the ability of the handfeel touch reference materials to discriminate between red wines was validated and a sample of references selected as a working tool for further studies.

Materials and methods

Figure 1 shows the workflow of the study with the preliminary step involving two tasks (Task P.1 and Task P.2) and two main tasks (Phase 1, Phase 2).

1. Preliminary step

1.1. Generation of mouthfeel terms

A list of 44 mouthfeel-related terms (Table 1) was generated by winemakers, wine sellers and consumers in a pre-test using a long-term memory-based task (see Appendix A for information about this task). Synonyms in the list were grouped using an online task. In this task, 25 native Spanish speakers were provided with the 44 terms, and were asked to associate terms that had similar meanings to each other.

The number of times the same two terms were associated with each other was counted. When the same two terms were associated to make a pair by at least 35 % of the participants, they were considered to have a similar meaning. The following pairs of words were thereby identified as having similar meaning: dynamic/agile; watery/light; fine/delicate; soft/velvety; aggressive/sharp; burning/warm; rough/coarse; creamy/fleshy; lubricant/unctuous; viscous/doughy; thick/dense; mouthcoating/full; ample/with volume; firm/robust. One of the terms in each pair was selected and the other discarded. Thus by the end of this task the initial list of 44 was reduced to 25 terms.

1.2. Design of handfeel touch reference materials

The reduced list of 25 terms (marked in bold in Table 1) was given to three groups of Master students of the design school IADE escuela de diseño (Madrid, Spain), each comprising two or three participants. Each group generated an independent set of handfeel touch reference materials and the three sets were compared in order to evaluate the generalisation of the procedure.

The creative process carried out to design the references is illustrated in Figure S1 of Appendix B.

The design students were instructed to find different materials that represented each of the 25 terms in the list. The resulting three sets (coded A-C) of handfeel touch reference materials each contained 25 (sets A and C) and 24 (set B; as set B did not contain a reference material for the term “hard”) items. Each reference material was associated with one of the five following distinct patterns of exploratory hand movements: i) grab (C for coger in Spanish), ii) experiment (E), iii) touch (T), iv) push (P), and v) rub (F for frotar in Spanish). Each reference material was coded using the letter of the set (A-C), followed by the code letter of the exploratory hand movement (C, E, T, P, F) and then a number from 1 to 10 (Figure 2). For example, reference AP1 belonged to set A and participants had to push to explore it. Images of the references (Figures S2 and S4) and their descriptions (Tables S1 and S3) are provided in Appendix B.

Each reference material was then placed in a homogenised container, which was big enough for a hand to be introduced and explore the item inside. These containers were designed in such a way that the reference materials could be touched blind in order to prevent any visual interaction with them, as they were of different colours and shapes. Figure 2 shows the set-up and procedure for the exploration of the reference materials.

2. Validation of handfeel touch reference materials for the description of red wine mouthfeel

2.1. Selection of six wines with different mouthfeel properties

2.1.1. Procedure

First, 23 commercial wines from different origins (France, Spain, Argentina, Germany and Peru), varieties (Gamay, Malbec, Mencía, Garnacha Tinta, Merlot, Cabernet-Sauvignon, Syrah, Tempranillo Tinto, Moristel, Pinot noir, Garnacha Tintorera, Monstrell, Tinta de Toro, Bobal, Prieto Picudo and Tannat), and vintages (2015-2021) were selected (Table 2). These wines were chosen in consultation with wine experts for being representative of different mouthfeel properties. Samples were purchased from local wine stores and wineries. Appendix C contains further information regarding the conventional oenological parameters of the studied wines.

Second, a sensory profile was carried out to select a subset of six wines from the initial set of 23 wines. The wines were characterised in February 2022 by 31 Oenology students (48 % males and 52 % women, average age = 26, range of age = 21-63) in their 3rd and 4th year of a bachelor’s or master’s degree (Universidad de La Rioja) following the RATA methodology. A list of 20 mouthfeel terms previously developed by Sáenz-Navajas et al. (2017) to describe mouthfeel properties of red wines was used. Fifteen mL of samples were served in dark wine glasses labelled with 3-digit random codes and covered with petri dishes. The order of presentation of the 23 samples was different for each participant. Nose clips were worn when evaluating the samples to avoid aroma interaction and to help participants focus on the taste and mouthfeel properties of wines. The participants were asked to taste the wine samples and to select the terms they associated with each sample from the RATA list, and then to rate their intensity using a 7-point scale (where 1 = very low and 7 = very high). The order of presentation of the terms was randomised and different for each participant.

The task was carried out in a ventilated and air-conditioned tasting room (at approximately 20 ºC) and the wine samples served at room temperature. After each wine had been tasted and evaluated, the participants followed a mandatory rinsing protocol with mineral water and pectin (1 g/L) before tasting the next sample. Every six samples, a 5-min pause was imposed to minimise the carry over effects.

2.1.2. Data analysis

Data were analysed using two-way ANOVAs for each of the 20 terms and considering the participants as random and the wines as fixed factors. A normalised principal component analysis (PCA) with varimax rotation was then carried out with the mean intensity scores of the significant attributes (N = 13). A hierarchical cluster analysis (HCA) with the Ward criteria was finally applied to the scores of the wines on the four varimax-rotated dimensions with an eigenvalue greater than 1. The six wines that differed the most were selected for the next step (see appendix D for selection criteria and sensory characterisation of wines).

2.2. Description of the red wines using the handfeel touch references

2.2.1. Procedure

Based on the results of Section 2.1., the six wines were described after the participants had first assessed a warm-up sample, which was then discarded. The six selected wines (marked with an asterisk in Table 2) deemed to capture maximal variability in terms of mouthfeel among the sample set selected. They were made from 6 different grapevine varieties: Tannat, Garnacha Tinta, Prieto Picudo, Merlot, Moristel and Pinot noir, and were diverse in origin: Peru (South America), Germany and three different regions in North Spain: Huesca (close to Pyrenees), Zaragoza (North East) and León (North West) (see Appendix C, Table S1).

Wine samples of 15 mL were presented to each participant in dark glasses coded with three-digit codes in a monadic and randomised order. The task was carried out in a ventilated, air-conditioned tasting room (approximately 20 ºC) and the wine was served at room temperature.

Thirty students in their second and third years of the bachelor’s or master’s degree in Oenology and Viticulture at the Universidad de La Rioja (Spain) (50 % male and 50 % female, average age = 26, age range = 21-63) participated in six sessions over three different days in March 2022. A 30 min pause between two sessions held on the same day was imposed. In the first three sessions, the seven samples were described using the three sets of handfeel touch references (A-C) without nose clips, and in the next three sessions the same procedure was carried out with nose clips. The order of presentation of the reference sets was randomised among participants. In each session, the participants were first familiarised with the handfeel touch references. They were instructed to interact with each of the 24-25 references following the exploring procedure indicated next to the reference (Figure 2) and group the references according to their perceived similarity. After familiarisation, they were presented with the seven red wines and asked to taste each sample and to describe it by selecting all the handfeel references that they considered appropriate to describe the sample following a Check-all-that-Apply (CATA) methodology (Figure 3). A rinsing protocol, consisting of three rinsing steps: 1) water, 2) pectin (1 g/L), and 3) water, was imposed between samples.

2.2.2. Data Analysis

2.2.2.1 Validation of the reference sets

To evaluate the overall efficiency of the three sets of reference materials in discriminating between the wine samples, a first analysis was carried out independently of the tasting conditions (i.e., with and without noseclip). For each set, the number of times each handfeel touch reference was used to describe each wine was counted. Three frequency tables with six rows (six wines) and 25 columns (references) for sets A and C, and 24 colums for set B were created.

A Cochran-Q test was carried out on each handfeel touch reference material in each set in order to evaluate its discrimination capacity. Further, a chi-squared test was carried out on each set in order to evaluate its overall ability to discriminate between the six evaluated wines. For this analysis, only references with a significant cochran-Q value were used. Data from both conditions (with and without nose clips) were considered as replicated responses. In both tests, alpha was set to 5 %.

In order to visualise the projection of wines based on the CATA responses of the three sets and to compare them with the RATA responses obtained in the verbalisation phase, a Multi Factor Analysis (MFA) on mixed data (frequency data for CATA and continuous data for RATA) was performed. Only references used by at least 15 % of participants were considered for this analysis.

2.2.2.2 Evaluation of the effect of the nose clip

In order to evaluate the effect of nose clips on CATA descriptions, the number of times each handfeel touch reference material was used to describe each wine was counted separately for each reference material set and each of the tasting conditions. Six frequency tables with six rows (six wines) and 25 columns (handfeel reference materials) for sets A and C (with and without nose clip), and 24 for set B (with and without nose clip) were generated.

A Cochran-Q test was calculated for each reference within each set to find significant differences between the two conditions (with and without nose clips). An overall independent chi-squared test was then carried out on each set and on each of the conditions using the reference materials for which Cochran-Q was significant.

To visualise the effect of using a nose clip an MFA for frequency data was performed. The frequency tables obtained for the two tasting conditions were concatenated to form three matrices with twelve rows each and 25 columns (reference materials) for set A and C, and 24 for set B. Each wine sample is thus represented twice in these matrices, once in the “without nose clip” (noted WNC) and once in the “with nose clip” (NC) conditions.

3. Selection of references for the final touch reference set

3.1. Procedure

Twenty-five individuals (44 % males and 56 % women, average age = 36, age range = 24-58), all belonging to the ICVV, participated in a sorting task. They were presented with the total 74 handfeel reference materials from all three sets simultaneously. Participants were asked to group references based on their handfeel similarity, building as many groups as they wanted and leaving out any reference materials that did not belong to any of the groups. No time restriction was imposed. Once they had finished the task, they recorded the results on a paper ballot.

3.2. Data Analysis

The sorting results of each participant were entered in an individual similarity matrix (reference materials × reference materials) using a code: 1 stood for two reference materials in the same group and 0 for two reference materials each in a different group. All the individual matrices (i.e., of all the participants) were then summed; the resulting co-occurrence matrix represents the overall similarity matrix in which the larger numbers indicate higher similarity between references.

4. Ethical approval

Ethical approval for the involvement of human subjects in this study was granted by CSIC Research Ethics Committee on 17/11/2021 with Reference number 180/2021.

Results

1. Phase 1

1.1. Validation of handfeel touch references for the description of red wine mouthfeel

The Cochran-Q test results (Table 3) show that 11 and 8 of the 25 reference materials in sets A and C respectively, and 9 of the 24 materials in set B significantly discriminated between the six wines, suggesting that participants perceived differences in the sensory characteristics of the evaluated wines. The Chi-squared test performed on the significant descriptors of each set showed that the three sets had the same overall capacity to differentiate the samples, with values ranging from 83.78 for set A to 133.5 for set C (p < 0.0001 in all three cases); this suggests that the CATA strategy using handfeel touch reference materials was efficient, regardless of the set employed.

Figure 4 shows the first two dimensions (67 % of the variance) of the MFA performed on the CATA (sets A, B and C) and the RATA data. Figure 4a shows that sets B and C and RATA contribute mostly to the first dimension, whereas set A contributes more to the second dimension (54 %). The proximity of sets B and C to each other indicates that the wine samples were described similarly by these two reference sets. Moreover, the proximity of these two sets to RATA also suggests a similarity with the verbal descriptions (employed in the RATA methodology) given by the experts.

The difference between set A and the other two sets was mostly due to samples Mer_Hu_18, Mor1_Hu_16, Gar_Za_18, and PRI_LE_19 (Figure 4b). This can be explained by the fact that more significant references were identified in the Crochran-Q tests (11 for set A versus 9 and 8 for sets B and C). In addition, the RV coefficients confirm the notably high similarity between the results for RATA and set B and for (to a lesser extent) RATA and set C: 0.94 for RATA-set B and 0.87 for RATA-set C. Meanwhile, the RV coefficients for the other combination of components were: 0.89 for sets B-C, followed by slightly lower values for sets C-A (0.77), then RATA-set A (0.69) and sets B-A (0.66). These significant RV coefficients confirm the correlation between handfeel touch reference materials and the mouthfeel RATA descriptions of the same wines.

Figure 4c shows the correlation circle of all three sets of handfeel touch reference materials and the 13 significant descriptors used in the vocabulary-based RATA approach. Only reference materials that had been associated with at least one wine by at least 50 % of the participants facilitated the interpretation of the results. Table 4 summarises the descriptions of the wines.

The MER_HU_18 wine, made from the Merlot variety from Huesca (Spain), was mainly characterised by the touch references AF5, AA4, AF2, BF3, BF4, BE2, BA5, CF2, CF10, CF4, CA7 and CF3, along with the RATA terms drying, sticky, grainy, coarse, burning, prickling, and persistent (Figures 4b and 4c). In the first component, this sample and its descriptors are opposite the wines MOR1_HU_16, made with Moristel from Huesca (Spain), and PIN_GER_19, produced with Pinot noir from Germany (Figure 4b). The Moristel wine was mainly characterised by the touch references AA4, BE3, BF5, CA7, CE6, and CT1 and the descriptors silky, watery, prickling, and puckering, while the Pinot noir was associated with the reference materials AF9, BA3, BF3, CA3, and CE3 and the term oily; both these wines shared the attributes silky, watery, and prickling.

For the second component, wines made from Garnacha Tinta (GAR_ZA_18) and Prieto Picudo (PRI_LE_19) are plotted in the lower part of the graph (Figure 4b), and are opposite to the Merlot and Pinot noir wines described above. Garnacha Tinta was mainly described using the terms sticky, burning and prickly together with the reference materials AE3, AF2, AF9, AF8, BF4, CA7, and CF4, meanwhile, Prieto Picudo was also found to be sticky and was associated with the reference materials AF2 and CF4, but it was also described as grainy and silky, being characterised by the reference materials BE2, BF3, BA4, BF9, BE5, CF10, tyand CA3.

1.2. Effect of using nose clips on wine descriptions

Figure 5 shows the overall projection (from the perspective of all three reference sets together) of the wine samples on the first two dimensions (43 % of variance) of the MFA. Each wine is represented twice on this map: once for the “without nose clip” (WNC) conditions and once for the “with nose clip” (NC) conditions. Regardless of whether they had been evaluated using or not using a nose clip, the samples are very close to each other on the map. These results were corroborated by those of the chi-square test for both conditions: they were not found to be significant for sets A (Χ² = 0.289, p = 0.438) and C (Χ² = 0.168, p = 0.682), but were significant for set B (Χ² = 9.267, p = 0.010); however, the latter results only took into account the three reference materials that were found to be significant in the Crochran-Q test (BA4, BF1, BE2). Interestingly, only BE3 appeared to significantly differ among the wines studied (see Table 4). Overall, these results suggest overall that the use of nose clips does not induce important differences when using handfeel touch reference materials to describe the palate properties of wines.

2. Phase 2: selection of references for the final handfeel touch reference set

The last step of the study involved the selection of a final reference material set that could be used to identify maximal differences between wines. Given that all three reference sets enabled differentiation of the samples, and that set A in particular showed different levels of differentiation to sets B and C, the three sample sets were pooled (74 references), and a sorting task was carried out in order to identify the reference materials that differed the most from each other. Figure 6 shows the HCA calculated on the first four dimensions (kruskal stress = 0.199) of the MDS coordinates.

The HCA dendrogram (Figure 6) proved to be a useful tool for identifying the redundant handfeel touch reference materials, and thus could be discarded, and the ones that differed the most from each other. The lower in the dendrogram two references are merged, the more similar they are (i.e., any two references that were similar to each other were merged in the lowest part of the dendrogram). The question being at which point two different references can be considered as redundant. To address this question we partitioned the dendrogram at different heights. The resulting grouping of the references was evaluated by three experimenters whose task was to select the best partition (i.e., a partition yielding maximal differences in terms of reference materials). The partition which provided 33 groups of references was consensually selected to be the optimum (Table 4).

The most representative reference material of each group was then selected by the three experimenters to form the final handfeel touch reference material set (underlined and in bold in Table 5). The criteria used for the selection of the most representative references within a group were: i) they must have obtained a significant Cochran-Q in the previous task, and ii) they can be easily reproduced. Therefore, references with a significant Cochran-Q were prioritised over non-significant references. Moreover, references that were easier to replicate were also selected as representative. Table 4 shows the final list of reference groups. Ninteen (marked with an asterisk in Table 3) of the 33 groups of references with marked handfeel differences contained at least one reference that had obtained a significant Cochran-Q in Phase 1. While the final reference set of touch references can be formed by these 19 references, the 14 remaining groups of references were not totally discarded as they may have been able to explain the differences among the other wine sets that differed to those of Phase 1.

Discussion

The present results corroborate the initial hypothesis of the existence of an association between handfeel touch sensations and mouthfeel elicited by red wine. This constitutes a significant finding, as it confirms that sensory transfer can be used for product description and marketing. It also adds to the substantial body of literature examining sensory associations and sensory transfer (Di Stefano & Spence, 2022), which has demonstrated the impact of handfeel touch evoked by extrinsic cues (primarily in the form of packaging or container designs) associated with food or beverage products on consumer experiences and product perception (Piqueras-Fiszman & Spence, 2012). Similarly, the literature shows the influence of handfeel touch on flavour perception, encompassing aroma, taste (Di Stefano & Spence, 2022), and trigeminal sensations (Pramudya et al., 2024). In contrast with the limited research on sensory correspondence between handfeel touch and mouthfeel investigated in the field of product description, as is the case with the present work.

Our results confirm that it is possible to discriminate between red wine mouthfeel properties by means of handfeel touch reference materials using a CATA-based approach. They can thus be used to provide practical guidelines for developing effective reference materials for describing wine perception and exploring the mouthfeel perceptual space of wines. We worked under the hypothesis that the sensations perceived with the hand by touching are similar to the mouthfeel elicited in the buccal cavity by red wine, which is based on the fact that both the buccal and hand surfaces contain similar mechanoreceptors (Foegending et al., 2015). The first important step was the generation of handfeel touch reference materials by first developing an extensive vocabulary related to red wine mouthfeel - involving more than 170 participants - in order to explore as broad a perceptual mouthfeel space as possible. The collection of touch materials was based on sensory vocabulary suggested by Sakamoto and Watanabe (2017). An initial non-redundant list of 25 mouthfeel-related terms was generated (Table 1), and provided to designers that developed three sets of 24-25 handfeel touch reference materials (Figure 2). The capacity of these materials to describe the mouthfeel space of very different red wines by means of the generated touch references was evaluated relative to a vocabulary-based approach (i.e., RATA) carried out by wine experts (Figure 4). The results show that a higher variability in mouthfeel characterisation can be attributed to one of the sets of handfeel touch reference materials (Table 3). These results confirm that it is possible to use handfeel touch reference materials for wine mouthfeel descriptions, as well as to identify maximal variability in mouthfeel.

Another methodological outcome of the study was that, when using handfeel touch reference materials to characterise the mouthfeel properties of red wine, wearing a nose clip did not markedly influence the perception of mouthfeel, or at the very least, its description (Figure 5). Thus, when wearing a noseclip, participants appeared to rely primarily on mouthfeel sensations to establish relationships between handfeel and wine sensations perceived in the mouth. This indicates that the utilisation of nose clips, which can be onerous, can be circumvented when describing the palate properties of wines through handfeel touch reference materials. However, we considered it essential to corroborate these findings, given the potential for cross-sensory interactions between touch and aroma to influence the description of mouthfeel in wines when nose clips are not worn, as has been previously reported. Therefore, six red wines were described in triplicate under two types of conditions: with (NC) and without (WNC) nose clips, and by means of three sets of touch reference materials produced in the first step of the study (Figure 5). In former conditions no aroma stimulus could reach the receptors of the tasters, and in the latter, aroma stimuli were present. The three distinct sets (sets A-C) of handfeel touch reference materials were used to facilitate the generalisation of the findings. The absence of effect associated with the use or not of nose clips could be due to the participants giving all their attention to in-mouth properties, as they were explicitly instructed to do. Similar outcomes have been observed in previous studies on descriptions of the palate properties of red wines, with no or limited effect of nose clip use being reported (Sáenz-Navajas et al., 2020).

Conclusions

Our study confirms that sensory associations exist between handfeel tactile sensations and red wine mouthfeel, and it explores these associations further by using a variety of handfeel touch reference materials. We propose an original guiding principle based on preliminary vocabulary-based strategies for the development of handfeel touch reference materials, which can then be integrated into a multidisciplinary strategy that incorporates artistic elements. In contrast to classical strategies, our proposal focuses on the development of handfeel touch references that enable the complete characterisation of the mouthfeel properties of red wines without the use of vocabulary, in order to avoid any potential misinterpretations or idiosyncrasies that may arise from the use of language. This work has significant implications for wine industry with regard to the communication of sensory perceptions. It addresses the limitations of classical vocabulary-based approaches, which are characterised by two main drawbacks: the inefficiency of identifying independent dimensions to fully characterise the sensory space of red wine mouthfeel (Sáenz-Navajas et al., 2020), and the fact that interpreting idiosyncratic vocabulary is challenging and panel-dependent (Gawel et al., 2000). The approach proposed here has proved to be very promising; however, its application is presently limited by the difficulties associated with reproducing handfeel touch reference materials, which could impede its general use in wine industry when describing mouthfeel. It would therefore be beneficial for future research to focus on selecting materials that can be applied universally and extrapolated to other contexts. In addition, it would be useful to carry out research to assess the potential of applying our strategy in a variety of contexts (education, marketing and technical support, among others) and in the absence of the vocabulary of wine professionals and consumers, a fact that needs to be proven and opens up an interesting field of research.

Funding

This research was funded by the Spanish Ministry of Science and Innovation, the Spanish Research Agency and FEDER with grant number PID2021-126031OB-C22, and by CSIC through the JAE Programme (JAE INTRO 2021 grant for RBM).