*3.1. Sensory Responses to the Wine Sample in Di*ff*erent Environments*

Results of the analysis of variance (ANOVA) for the different sensory parameters are shown in Table 1 (acceptability and intensity). For the acceptability responses (floral aroma, fruity aroma, sweetness, acidity, mouthfeel astringency, body, aftertaste, and overall liking), none of the main effects (environment nor order) were significant (*p* ≥ 0.05) in the ANOVA models. The interaction (environment\*order) effect was only significant (*p* < 0.05) for the fruity aroma attribute on the model of acceptability. For the intensity parameters (floral aroma, sweetness, acidity, and astringency), the type of environment effect was only significant (*p* < 0.05) for the floral aroma and astringency parameters. The interaction effect (environment\*order) was only significant (*p* < 0.05) for the acidity and astringency intensity parameters. The order main effect was not a significant factor (*p* ≥ 0.05) for neither the acceptability nor the intensity parameters; therefore, the means of the two served samples could be pooled for the *post-hoc* means comparison analysis.


**Table 1.** ANOVA <sup>1</sup> table for the acceptability and intensity parameters of the wine by environment samples.

<sup>1</sup> ANOVA = analysis of variance (five types of environments (traditional booths, bright-restaurant, dark-restaurant, bright-VR, and dark-VR) and two positional orders. Liking scores were based on a nine-point hedonic scale (1 = dislike extremely, 9 = like extremely). Intensity scores were based on a 15-point Likert scale (1 = absent, 15 = very strong). <sup>2</sup> *F* value, mean square/mean square error. Effects were considered significant when the probability Pr > *F* was less than 0.05 (bolded and italicized probabilities). <sup>3</sup> The environment effect was crossed with the replicate effect in a two-way factorial design (type of environment by order) using participants as blocks.

Order 0.02 0.89 2.41 0.12 0.47 0.49 0.03 0.86 Environment\*Order <sup>3</sup> 1.42 0.23 1.11 0.35 2.93 *0.02* 5.08 <*0.01*

Table 2 shows the mean acceptability scores of the red wine sample in each environmental condition (traditional booths, bright-restaurant, dark-restaurant, bright-VR, and dark-VR). For the aroma acceptability attributes, the floral aroma scores for all the environments were not significantly different (*p* ≥ 0.05) among them, ranging from 5.65 (dark-VR) to 5.94 (dark-restaurant). On the other hand, the acceptability score of the fruity aroma of the wine sample in the dark-restaurant environment was significantly higher (*p* < 0.05) compared to that of the wine sample in the traditional booths (5.99 vs. 5.65, respectively). No significant differences (*p* ≥ 0.05) were found among the environments in the taste acceptability parameters (sweetness, acidity, astringency, body, and aftertaste) and overall liking of the wine sample. The mean intensity scores of the wine sample in each environment for the floral aroma, sweetness, acidity, and astringency are also shown in Table 2. The real dark-restaurant environment had a significantly (*p* < 0.05) higher floral aroma intensity score compared to those values of the traditional booths and the VR environments (bright and dark; 8.61 vs. 7.45–7.96). On the other hand, the real bright-restaurant had a significantly (*p* < 0.05) higher floral aroma intensity score compared to the value of the traditional booths (8.14 vs. 7.45, respectively), but was not significantly (*p* ≥ 0.05) different compared to the VR environments (bright and dark). Opposite results were observed for the astringency intensity attribute, in which the real dark-restaurant environment had a significantly (*p* < 0.05) lower score compared to that of the traditional booths and the dark-VR environment (7.57 vs. 8.34–8.39, respectively; Table 2). For the sweetness and acidity intensity attributes, no significant differences (*p* ≥ 0.05) were found among the environments.


**Table 2.** Acceptability and intensity mean values of the wine sample in each environment <sup>1</sup> .

<sup>1</sup> Five environments were tested (traditional booths, bright-restaurant, dark-restaurant, bright-VR, and dark-VR). Means and standard deviations of 53 participants. <sup>2</sup> Liking scores were based on a nine-point hedonic scale (1 = dislike extremely, 9 = like extremely). Intensity scores were based on a 15-point Likert scale (1 = absent, 15 = very strong). a–c Means with different superscripts in each column within each attribute indicate significant differences (*p* < 0.05) by the Tukey studentized range Honest Significant Difference (HSD) test.

The frequency distribution (%) of participants' responses for the intensities of sweetness, acidity, astringency, and body of the wine sample in each environment using the just-about-right (JAR) scale is shown in Figure 2. In general, the wine sample was considered to be "just-about-right" (49%–59%) and "too little" (37%–50%) in the sweetness for all tested environments in this study. Moreover, participants

considered the wine sample to be "just-about-right" (52%–66%) and "too much" (28%–38%) in acidity and astringency for all environments. For the body of the wine sample, participants rated this attribute as "just-about-right" (61%–71%) for all the environments (Figure 2). The penalty analysis using the JAR data is shown in Table 3. In general, the wine sample for all the environments was considered to be "too little" in sweetness (*mean drop* = 1.25–1.99; *p* < 0.05) except for the dark-VR environment, in which the mean drop was not significant in the overall liking (0.40; *p* ≥ 0.05).

**Table 3.** Penalty analysis results for the sweetness, acidity, astringency, and body of the wine sample in different environments <sup>1</sup> .


<sup>1</sup> Booths = traditional sensory booths, Bright-real = bright restaurant real environment, Dark-real = dark restaurant real environment, Bright-VR = bright restaurant VR environment, and Dark-VR = dark restaurant VR environment. Values represent the mean drops using the nine-point hedonic scale. Mean drops were considered significant when the probability Pr > *F* was less than 0.05 (bolded and italicized values).

Participants penalized the wine sample in all the environments for being "too much" in acidity (*mean drop* = 1.17–1.94) and astringency (1.02–1.94), but they did not penalize the body attribute (*p* ≥ 0.05; Table 3). Moreover, the purchase intent values of the wine samples in all the environments were not significantly (*p* ≥ 0.05) different (42%–45%; data not shown).

=

## *3.2. Emotions and Multivariate Analysis of the Wine Sample in Di*ff*erent Environments 3.2. Emotions and Multivariate Analysis of the Wine Sample in Different Environments* Figure 3a shows the corresponding analysis of the emotional terms of the CATA question related

Figure 3a shows the corresponding analysis of the emotional terms of the CATA question related to the wine sample in each environment. The principal component one (PC1) and principal component two (PC2) accounted for 26.41% and 36.65%, respectively, which explained 63.03% of the total data variability. The wine sample was only associated with the emotions "polite" and "calm" under the traditional booths environment. Under the real bright-restaurant environment, participants' emotions toward the wine sample were associated with "interested", "secure", "friendly", and "loving". to the wine sample in each environment. The principal component one (PC1) and principal component two (PC2) accounted for 26.41% and 36.65%, respectively, which explained 63.03% of the total data variability. The wine sample was only associated with the emotions "polite" and "calm" under the traditional booths environment. Under the real bright‐restaurant environment, participants' emotions toward the wine sample were associated with "interested", "secure", "friendly", and "loving."

*Foods* **2020**, *9*, x FOR PEER REVIEW 11 of 17

**Figure 3.** (**a**) Correspondence analysis of the emotion terms forthe wine sample in each environment <sup>1</sup> and (**b**) principal coordinate analysis of the emotion terms with the overall liking score. <sup>1</sup> Booths = traditional sensory booths, Bright‐real = bright restaurant real environment, Dark‐real = dark restaurant real environment, Bright‐VR = bright restaurant VR environment, and Dark‐VR = dark restaurant VR environment. **Figure 3.** (**a**) Correspondence analysis of the emotion terms for the wine sample in each environment <sup>1</sup> and (**b**) principal coordinate analysis of the emotion terms with the overall liking score. <sup>1</sup> Booths = traditional sensory booths, Bright-real = bright restaurant real environment, Dark-real = dark restaurant real environment, Bright-VR = bright restaurant VR environment, and Dark-VR = dark restaurant VR environment.

On the other hand, participants only elicited emotions such as "free" and "enthusiastic" towards the wine sample under the bright-VR environment. The emotions "glad" and "enthusiastic" were associated with the wine under the real dark-restaurant environment. Conversely, "nostalgic", "daring", and "disgusted" were associated with the wine under the dark-VR environment. The principal coordinate analysis of the emotion terms concerning the overall liking of the wine sample in different environments

is shown in Figure 3b. In general, overall liking was positively associated with the emotion terms "secure", "free", "interested", "good", and "friendly." Moreover, the overall liking of the samples was negatively associated with "daring", "affectionate", "eager", "adventurous", and "wild."

The principal component analysis (PCA) biplot shows the acceptability and intensity parameter vectors associated with the five environmental conditions (traditional booths, bright-restaurant, dark-restaurant, bright-VR, and dark-VR) in which the wine sample was tasted (Figure 4). Considering all acceptability and intensity sensory parameters, the PC1 and PC2 accounted for 36.93% and 31.77% of the biplot, respectively, explaining 68.7% of the total data variability. The fruit aroma liking and floral aroma intensity (*factor loading* = 0.89–0.91; data not shown) vectors contributed largely to the discrimination of the environments in the PC1. *Foods* **2020**, *9*, x FOR PEER REVIEW 13 of 17

**Figure 4.** Principal component analysis (PCA) biplot visualizing treatments <sup>1</sup> (wine sample in each environment), acceptability (liking; vectors in red color), and intensity attributes (vectors in green color). <sup>1</sup> Booths = traditional sensory booths, Bright‐real = bright restaurant real environment, Dark‐ real = dark restaurant real environment, Bright‐VR = bright restaurant VR environment, and Dark‐VR **Figure 4.** Principal component analysis (PCA) biplot visualizing treatments <sup>1</sup> (wine sample in each environment), acceptability (liking; vectors in red color), and intensity attributes (vectors in green color). <sup>1</sup> Booths = traditional sensory booths, Bright-real = bright restaurant real environment, Dark-real = dark restaurant real environment, Bright-VR = bright restaurant VR environment, and Dark-VR = dark restaurant VR environment.

= dark restaurant VR environment. **4. Discussion** *4.1. Sensory Responses to the Wine Sample in Different Environments* This study showed that the type of environment (traditional booths, bright‐restaurant, dark‐ restaurant, bright‐VR, and dark‐VR) had a marginal effect on the sensory acceptability (floral aroma, sweetness, acidity, astringency, body, aftertaste, and overall liking) of the Cabernet Sauvignon wine sample (Tables 1 and 2). Only the real dark‐restaurant environment had a higher acceptability score for the fruity aroma attribute compared to that of the traditional booths (5.99 vs. 5.65, respectively; Table 2). On the other hand, the type of environment significantly affected the intensity perception On the other hand, the overall liking vector (*factor loadings* = 0.96) contributed largely to the discrimination of the samples in the PC2. According to the PCA, the acidity, sweetness, and aftertaste liking scores were positively associated with overall liking. On the other hand, the liking of the floral aroma was positively associated with the intensity of the floral aroma, but it was negatively associated with the intensity of astringency. The sweetness intensity was positively associated with the liking of the fruity aroma and astringency, but it was negatively associated with the intensity of the acidity. Moreover, the liking of the body was positively associated with the liking of the aftertaste, but it was negatively associated with the liking of the floral aroma. The real dark-restaurant and bright-VR environments were related to higher floral aroma intensity and liking. The dark-VR environment was related to a higher overall liking score, and the traditional booths environment was related to a higher intensity of acidity (Figure 4).

#### (bright and dark) had a significantly higher floral aroma compared to that of the wine tasted in the **4. Discussion**

#### booths (8.14–8.61 vs. 7.45, respectively; Table 2). Conversely, the wine tasted in the real dark environment had a significantly lower astringency compared to that of the booths and the dark‐VR *4.1. Sensory Responses to the Wine Sample in Di*ff*erent Environments*

environment (7.57 vs. 8.34–8.39, respectively; Table 2). Virtual reality technology can provide consumers with simulated scenarios that are close to real environments [30]. The present research showed that changing the environment had a significant This study showed that the type of environment (traditional booths, bright-restaurant, dark-restaurant, bright-VR, and dark-VR) had a marginal effect on the sensory acceptability (floral aroma, sweetness, acidity, astringency, body, aftertaste, and overall liking) of the Cabernet Sauvignon

of the floral aroma and astringency of the wine sample. The wine tasted in both real environments

differences between the real and VR in the dark environments for the perception of floral and astringency. Environments may affect consumers' expectations and experiences of products because their decisions can be unconsciously changed by several extrinsic factors [26]. Ryu and Jang [31] tested different contextual factors such as lighting, facility aesthetics, ambiance music, dining equipment, and employees' interactions on consumers dining experiences. They found that consumers showed positive emotions to simple environmental changes such as the type of music played and the layout of the dining environment. Moreover, the lighting conditions are very important for the sensory evaluation of foods. Bschaden et al. [32] found that the lighting conditions of the testing environment can affect the saltiness perception of tomato soups. Moreover, consumers

wine sample (Tables 1 and 2). Only the real dark-restaurant environment had a higher acceptability score for the fruity aroma attribute compared to that of the traditional booths (5.99 vs. 5.65, respectively; Table 2). On the other hand, the type of environment significantly affected the intensity perception of the floral aroma and astringency of the wine sample. The wine tasted in both real environments (bright and dark) had a significantly higher floral aroma compared to that of the wine tasted in the booths (8.14–8.61 vs. 7.45, respectively; Table 2). Conversely, the wine tasted in the real dark environment had a significantly lower astringency compared to that of the booths and the dark-VR environment (7.57 vs. 8.34–8.39, respectively; Table 2).

Virtual reality technology can provide consumers with simulated scenarios that are close to real environments [30]. The present research showed that changing the environment had a significant effect on perception, but that effect was marginal for acceptability. The overall liking of the wine sample in the real environments was similar compared to that of the VR environments. The same effect occurred for the perception of sweetness and acidity; however, there were significant differences between the real and VR in the dark environments for the perception of floral and astringency. Environments may affect consumers' expectations and experiences of products because their decisions can be unconsciously changed by several extrinsic factors [26]. Ryu and Jang [31] tested different contextual factors such as lighting, facility aesthetics, ambiance music, dining equipment, and employees' interactions on consumers dining experiences. They found that consumers showed positive emotions to simple environmental changes such as the type of music played and the layout of the dining environment. Moreover, the lighting conditions are very important for the sensory evaluation of foods. Bschaden et al. [32] found that the lighting conditions of the testing environment can affect the saltiness perception of tomato soups. Moreover, consumers tend to choose less healthy food options when the ambient lighting is dim [33]. In the present study, the dark environment might have been the most adequate contextual surrounding for consumers to taste the wine sample, as the perception of fruity and floral aroma had positive effects on consumers. In a similar study, Hersleth et al. [34] found that the wine tasting experience in a reception type of room was significantly improved compared to the tasting of the wines in traditional booths. With the development of more efficient virtual reality technologies, more sensory stimuli can be tested with different contextual situations. The virtual reality technology might potentially replace the use of physical environments in the future, becoming an important tool for sensory evaluation [35].
