*2.6. Consumer Sensory Evaluation and Biometrics*

sample?

Check all emojis that depict how you feel when tasting this sample

Check all emotions that depict how you feel when tasting this sample

Overall liking (rated

*2.7. Statistical analysis* 

**3. Results** 

*3.1. Physicochemical results* 

A sensory session was carried out in Monterrey, NL, Mexico, which is the state with the highest alcoholic drinks consumption with beer as the leader [36,37]. The session was conducted with *N* = 61 beer consumers (frequency > three times a month; 54% males; 46% females) between 18 and 51 years old (mean age 25.6 ± 6.9 years). Participants were recruited via email and asked to participate in a graduate research project from the Department of Bioengineering, School on Engineering and Sciences of Tecnológico de Monterrey, Campus Monterrey, Mexico (Ethics ID: CSERDBT-0002). According to the Power analysis conducted using the Power and Sample Size Calculator from the SigmaXL ver. 8.15 software (SigmaXL Inc., Kitchener, ON, Canada), the number of participants was sufficient to find significant differences (1-β = 0.98) among the beer samples. The session was conducted at SensoLab Solutions SC, a sensory and consumer science laboratory center, located at the Technology Transfer and Innovation Center of Tecnológico de Monterrey, Mexico. The laboratory was equipped with eight individual sensory booths with uniform lighting. Each booth had an Android® (Google, Mountain View, CA, USA) Samsung Galaxy Tab 4 tablet (Samsung, Seoul, South Korea) displaying the Bio-Sensory application (App; The University of Melbourne, Parkville, Vic, Australia). The App was able to present the questionnaire (Table 2) and record videos from the participants while tasting the beer samples to further analyze their emotional responses [29]. Samples (30 mL) were served at refrigeration temperature (4 ◦C), and water was used as palate cleanser before and between each sample. To assess the visual descriptors of the beers, a video showing the pouring of the sample using the RoboBEER (The University of Melbourne, Parkville, Vic, Australia) was displayed in the App to avoid bias from the variability due to the pouring method and glass effects [22]. As shown in Table 2, two overall liking ratings were obtained at the start and end of the tasting to verify if there is a bias on this descriptor based on the evaluation of specific attributes.

**Table 2.** Questionnaire presented in the Bio-Sensory application.


ionate/Disgusted/Free/Friendly/Happy /Adventurous/Guilty/Nostalgic/Calm/ Pleasant/Satisfied/Secure/Surprised/W orried\*

, xiii) flushed , xiv) rage , xv) smirk , and xvi) disappointed [39].

at the end of the test) Dislike extremely—Like extremely 15-cm non-structured scale \*Emotion-terms obtained from EsSense Profile® [38].

Videos were analyzed using an application developed based on the Affectiva software development kit (SDK; Affectiva, Boston, MA, USA). This application uses the histogram of the oriented gradient to detect and track the micro- and macro-movements of face features and is able to evaluate all videos in batch. Furthermore, it is capable of assessing facial expressions using support vector machine algorithms to translate them into emotions such as i) contempt, ii) disgust, iii) sadness, iv) surprise, v) joy, vi) valence, vii) engagement, and viii) attention, as well as emojis related to facial expressions such as ix) smiley , x) relaxed , xi) winking face , xii) stuck out tongue

All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, Pennsylvania, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively.

Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−1, respectively), and significantly different from the other samples. On the other hand, LK was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH =

2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17).

Check all that apply (CATA)

Check all that apply (CATA)


**3. Results** 

*3.1. Physicochemical results* 

*Foods* **2019** *8*, x FOR PEER REVIEW 6 of 19

#### , xiii) flushed , xiv) rage , xv) smirk , and xvi) disappointed [39]. *2.7. Statistical analysis 2.7. Statistical analysis 2.7. Statistical analysis 2.7. Statistical analysis 2.7. analysis 2.7. Statistical Analysis*

*2.7. Statistical analysis*  All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, Pennsylvania, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, Pennsylvania, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively. All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, Pennsylvania, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively. All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, Pennsylvania, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively. test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, Pennsylvania, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively. All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, Pennsylvania, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively. All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test using Minitab 17.2.1 Inc., A correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, WA, (self-reported), biometric assessed data on (PCA), multiple factor analysis (MFA) with customized code written in R2019b (Mathworks, Natick, (Addinsoft Inc., USA), respectively. All data were analyzed through ANOVA and least significant differences (LSD) as a post-hoc test (α = 0.05) using Minitab 17.2.1 (Minitab Inc., State College, PA, USA). A linear correlation analysis was conducted for alcohol and hordenine values using Microsoft Excel (Microsoft, Redmond, WA, USA). Chemical, sensory (self-reported), and biometric responses were assessed using multivariate data analysis based on principal components analysis (PCA), and multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively.

#### multiple factor analysis (MFA) with a customized code written in Matlab® R2019b (Mathworks, Inc., Natick, MA, USA) and XLSTAT ver. 2020.1.1 (Addinsoft Inc., New York, NY, USA), respectively. **3. Results**

**3. Results** 

**3. Results** 

**3. Results** 

**3. Results** 

**3. Results** 

#### *3.1. Physicochemical results 3.1. Physicochemical results 3.1. Physicochemical results 3.1. Physicochemical results 3.1. Physicochemical results 3.1. Physicochemical results 3.1. Physicochemical Results*

**3. Results**  *3.1. Physicochemical results*  Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−1, respectively), and significantly different from the other samples. On the other hand, LK Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−1, respectively), and significantly different from the other samples. On the other hand, LK was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH = Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−1, respectively), and significantly different from the other samples. On the other hand, LK was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH = Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−1, respectively), and significantly different from the other samples. On the other hand, LK was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH = Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−1, respectively), and significantly different from the other samples. On the other hand, LK was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH = 2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17). Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−1, respectively), and significantly different from the other samples. On the other hand, LK was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH = Table mean values and the ANOVA for selected parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z the lowest for (26.58) darkest while C the (59.36). yellow (YI) for all other samples, being lowest Spontaneous fermentation and the in density (1.02 1.03 <sup>−</sup> significantly the other the was most viscous by Z and mPa with as viscous mPa s). hand, the fermentation acidic = 2.94, 0.32; pH = 3.17, TA = 0.41), while Z , (xv) smirk Videos to facial well All Chemical, (MFA) Table for being s), hand, , and (xvi) disappointed parameters. significant Z (59.36). was 0.41), while [39]. Table 3 shows the mean values and results from the ANOVA for selected physicochemical parameters. There were significant differences (*p* < 0.05) between samples for all parameters. Sample Z had the lowest mean value for L\* (26.58) as this is the darkest beer, while C had the highest value (59.36). Similarly, the yellow index (YI) was higher for Z (200.40) than all other samples, C being the lowest (15.69). Spontaneous fermentation beers (LK and LF) were the highest in density (1.02 and 1.03 g mL−<sup>1</sup> , respectively), and significantly different from the other samples. On the other hand, LK was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH = 2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17).

was the most viscous (2.16 mPa s), followed by Z and H (1.80 mPa s), with C as the least viscous (1.48 mPa s). On the other hand, the spontaneous fermentation samples were the most acidic (LF: pH = 2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17). 2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17). 2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17). 2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17). 2.94, TA = 0.32; LK: pH = 3.17, TA = 0.41), while Z was the least acidic (pH = 4.42, TA = 0.17). Figure 1 shows the means and ANOVA results of the total sugars, bitterness (Figure 1a), iso-alpha acids, and hordenine (Figure 1b). The spontaneous fermentation beers had significantly higher (*p* < 0.05) total sugar content (LF: 31.23 mg mL−<sup>1</sup> ; LK: 27.53 mg mL−<sup>1</sup> ) than the samples from other types of fermentation; for C, the sugar concentration was non-detectable with the chromatographic conditions used. Sample Z was the highest in both bitterness (34.98 IBU) and total iso alpha-acids (21.41 mg L−<sup>1</sup> ), while LF was the least bitter (bitterness: 5.08 IBU; total iso-alpha acids: 0.60 mg L−<sup>1</sup> ). On the other hand, the top fermentation beers (Z and L) had the highest concentrations of hordenine

(Z: 4.24 mg L−<sup>1</sup> ; L: 3.22 mg L−<sup>1</sup> ), while spontaneous fermentation sample LF had the lowest content (0.98 mg L−<sup>1</sup> ).


**Table 3.** Physicochemical characterization of commercial beers.

Abbreviations: CIELAB color parameters (L\*: lightness, a\*: red/green, b\*: blue/yellow), YI: yellowness index. † Values represent the mean ± standard error (nTitratable Acidity = 2, nColor, Density, Viscosity, pH = 3). Abbreviations of samples may be found in Table 1. Different letters within a column indicate that values are significantly different according to the least significant difference test (LSD; *p* <0.05). *Foods* **2019** *8*, x FOR PEER REVIEW8 of 19

**Figure 1.** Chemical characterization of commercial beers, including (**a**) total sugars (mg mL−1), bitterness (IBU), (**b**) hordenine (mg L−1), total Iso-α-acid concentration (mg L−1). Different letters above bars denote significant differences between beer samples, for the same chemical parameter, according to the least significant difference test (LSD; *p* < 0.05). \* Total sugars not detected in beer C. All values are the mean ± SE (error bars) of independent determinations. *n* = 3, hordenine, and bitterness; *n* = 2, total sugars, and total-α-acids. Abbreviations of samples may be found in Table 1. **Figure 1.** Chemical characterization of commercial beers, including (**a**) total sugars (mg mL−<sup>1</sup> ), bitterness (IBU), (**b**) hordenine (mg L−<sup>1</sup> ), total Iso-α-acid concentration (mg L−<sup>1</sup> ). Different letters above bars denote significant differences between beer samples, for the same chemical parameter, according to the least significant difference test (LSD; *p* < 0.05). \* Total sugars not detected in beer C. All values are the mean ± SE (error bars) of independent determinations. *n* = 3, hordenine, and bitterness; *n* = 2, total sugars, and total-α-acids. Abbreviations of samples may be found in Table 1.

Table 4 shows that the simple sugars from the spontaneous fermentation samples (LF and LK) were mainly composed of glucose (LF: 14.32 mg mL−1; LK: 13.91 mg mL−1), followed by fructose (LF: 13.51 mg mL−1; LK: 12.56 mg mL−1), and maltose (LF: 3.40 mg mL−1; LK: 1.06 mg mL−1). Sample H had higher values of maltose (0.79 mg mL−1) than glucose (0.60 mg mL−1) and fructose (0.50 mg mL−1), while L was higher in fructose (2.04 mg mL−1) than glucose (1.87 mg mL−1) and did not contain maltose. Spontaneous fermentation beers were the highest in salt concentration (LK and LF: 0.10%), while C was the lowest (0.05%). A similar trend was found for TDS with LF and LK; although being significantly different, both presented the highest values (LF: 1226 ppm; LK: 1148 ppm), while C had the lowest with 658 ppm. Top fermentation beers showed the highest alcohol content (Z: 9.47%; L: 6.68%), while spontaneous fermentation samples had the lowest (LF: 2.53%; LK: 3.53%). A similar trend was found for the content of trans-Isocohumulone and trans-Isohumulone parameters with Z Table 4 shows that the simple sugars from the spontaneous fermentation samples (LF and LK) were mainly composed of glucose (LF: 14.32 mg mL−<sup>1</sup> ; LK: 13.91 mg mL−<sup>1</sup> ), followed by fructose (LF: 13.51 mg mL−<sup>1</sup> ; LK: 12.56 mg mL−<sup>1</sup> ), and maltose (LF: 3.40 mg mL−<sup>1</sup> ; LK: 1.06 mg mL−<sup>1</sup> ). Sample H had higher values of maltose (0.79 mg mL−<sup>1</sup> ) than glucose (0.60 mg mL−<sup>1</sup> ) and fructose (0.50 mg mL−<sup>1</sup> ), while L was higher in fructose (2.04 mg mL−<sup>1</sup> ) than glucose (1.87 mg mL−<sup>1</sup> ) and did not contain maltose. Spontaneous fermentation beers were the highest in salt concentration (LK and LF: 0.10%), while C was the lowest (0.05%). A similar trend was found for TDS with LF and LK; although being significantly different, both presented the highest values (LF: 1226 ppm; LK: 1148 ppm), while C had the lowest with 658 ppm. Top fermentation beers showed the highest alcohol content (Z: 9.47%; L: 6.68%), while spontaneous fermentation samples had the lowest (LF: 2.53%; LK: 3.53%). A similar trend was found for the content of trans-Isocohumulone and trans-Isohumulone parameters with Z

being the highest concentration (10.95 mg L−<sup>1</sup> , and 10.46 mg L−<sup>1</sup> , respectively), and LF the lowest (0.22 mg L−<sup>1</sup> , and 0.38 mg L−<sup>1</sup> , respectively).


**Table 4.** Simple sugars, salt, total dissolved solids, ethanol content, and iso-α-acids of commercial beers.

\* Values represent the mean ± standard error (nSimple sugars, Iso-α-acids = 2, nSalt, Total dissolved solids, Ethanol content = 3). ND: Non-detectable. Abbreviations of samples may be found in Table 1. Different letters within a column indicate that values are significantly different according to the least significant difference test (LSD; *p* < 0.05).
