Coffee Drinking and Emotions: Are There Key Sensory Drivers for Emotions? †

beverages

Article Coffee Drinking and Emotions: Are There Key Sensory Drivers for Emotions? †

Natnicha Bhumiratana 1, Mona Wolf 2, Edgar Chambers IV 3 and Koushik Adhikari 4,*

1 Independent Restauranteur, 250/3 Sukhumvit 55/8, Bangkok 10110, Thailand; [email protected] 2 The Wolf Group, 10860 Kenwood Road, Cincinnati, OH 45242, USA; [email protected] 3 Center for Sensory Analysis and Consumer Behavior, Kansas State University, Manhattan, KS 66506, USA; [email protected] 4 Department of Food Science & Technology, University of Georgia, Grifﬁn, GA 30223, USA * Correspondence: [email protected]; Tel.: +1-770-412-4736 † Data for this manuscript was collected at Center for Sensory Analysis and Consumer Behavior, Kansas State University, Manhattan, KS 66506, USA.

Received: 18 December 2018; Accepted: 1 March 2019; Published: 1 April 2019

Abstract: In the past couple of decades the coffee market has exploded, and to remain competitive, it is important to identify the key drivers for consumer acceptance of coffee. This study expanded on the previous emotion study on a population of coffee drinkers in Manhattan, Kansas, USA and focused on identifying the sensory drivers of emotional responses elicited during the coffee drinking experience (CDE). A trained coffee panel performed a descriptive analysis of six coffee samples and identified the key sensory attributes that discriminated each coffee. Utilizing Partial Least Square Regression (PLSR), the descriptive data were then mapped with the emotion data to identify sensory drivers for eliciting the emotional responses. The sensory characteristics of dark roast coffee (roast–aroma and flavor, burnt–aroma and flavor, bitter, and body) might elicit positive-high energy feelings for this population of coffee users. Tobacco (flavor) and cocoa (aroma) may also be responsible for positive emotions (content, good, and pleasant). Citrus and acidity seemed to be negative sensory drivers as they induced the feeling of off-balance. Sensory descriptive data could be useful to describe emotion profiles elicited by coffee drinking, which could help the coffee industry create coffee products for different segments of coffee drinkers.

Keywords: descriptive analysis; emotions; coffee

1. Introduction Human senses are powerful elicitors of emotions and the interactions between the two are rarely debated [1–3]. A number of studies have attempted to define and categorize human emotion, but it is only recently that emotions have been linked to acceptance of food and beverage. Nowadays, there is more awareness that the emotional experiences consumers receive from a product via sensory perception determine acceptability and consumption [4,5]. Therefore, assessment of the emotional responses elicited by the sensorial experience during product consumption is also vital. Several researchers have developed emotion scales to measure the affective feelings evoked by product consumption (EsSense Profile™)[6] or by olfactory stimulations from everyday odors (Geneva Emotion and Odor Scale) [1]. The emotions elicited by the coffee drinking experience were identified in our previous work [7], where they determined a list of 44 emotions suitable for defining the ‘Coffee Drinking Experience (CDE)’ and provided the emotion profiles for each segment of coffee drinkers. The consumers in the study were clustered in six consumer segments. To have a complete understanding of consumers’ perceptions,

Beverages 2019, 5, 27; doi:10.3390/beverages5020027 www.mdpi.com/journal/beverages Beverages 2019, 5, 27 2 of 13 it is important to understand the sensory characteristics that elicit those emotions experienced during coffee consumption. Coffee is well known for its complex sensory characteristics and is often consumed for the sensory experience it provides, in addition to the physiological stimulation of the caffeine [8–11]. Coffee is one of the few food products that utilizes specialized experts (cuppers) to ensure that the sensory properties are up to standard. These experts are not trained sensory assessors but have a lot of experience in determining the quality of coffee for commercial purposes [12]. Trained sensory panels are used extensively for understanding the sensory characteristics of food products including coffee. Bhumiratana et al. [13] conducted a descriptive analysis study on aroma evolution in coffee beans when they are subjected to roasting to different levels of roast. Due to its high complexity, the descriptive sensory panel may need training specifically for coffee, in addition to the usual intensive training program on the sensory characteristics of food and beverage. Many studies support that the amount of training and regular re-training correlate with panelist perception of the sensory attributes and increases the quantification accuracy of attribute intensities [14–17]. Recently, a “living” lexicon for descriptive analysis of brewed coffee was developed by Chambers IV et al. [18] that can be used by sensory researchers for descriptive analysis of coffee. The link between emotion profiles for a food product and its descriptive sensory profile to understand drivers of positive and negative emotions during consumption has been explored to a limited extent by sensory researchers. Interestingly, the two studies the authors found are on dark chocolates but neither used the classic descriptive approach in their respective study [4,19]. Thomson et al. [4] utilized best–worst scaling to gather sensory information on nine dark chocolates from consumers. They also asked the consumers to identify five emotion descriptors from a pre-conceptualized list that come to their mind for each chocolate. The conceptual biplot between the sensory data and the emotions revealed that the sensory differences in products could drive, in part, conceptual differences due to the emotional response. In another study, Jager et al. [19] reported the link between temporal dominance of sensation (TDS; analytical data) and temporal dominance of emotions (TDE; affective data) using unflavored and flavored dark chocolates. The unflavored dark chocolates (70% and 85% cocoa content) characterized by bitter, dry, and cocoa flavor were linked to aggressive, bored, calm, and guilty emotions, while the fruit-flavored dark chocolates (orange and blueberry) that were characterized by crunchy, fruit, and sweet elicited interested, happy, and loving emotions. The fruit-flavored chocolates were liked more by the consumers as was evident through their emotional response as well. The main objective of our study was to identify sensory drivers of emotional response to the experience of coffee drinking. A trained coffee panel performed descriptive analysis on the coffee samples that were used to elicit emotions in coffee drinkers. The sensory data was then utilized to determine the sensory drivers for emotional responses in each segment (cluster) of consumers as reported in Bhumiratana et al. [7].

2. Materials and Methods

2.1. Descriptive Panel The descriptive coffee panel from the Wolf group (Cincinnati, OH, USA) was utilized to evaluate the coffee samples. The panel consisted of six highly-trained members who had completed 120 h of general training and had a minimum of 1200 h of sensory testing of food and beverages. The coffee panelists also completed an additional 120 h of training on various coffee stimuli and key attributes (coffee, robusta, roasted, burnt, earthy, rioy, acidity, bitter, and body). Performance of the panel is evaluated every 3 months in the form of a blind reference sample or samples. This coffee panel has been evaluating coffee products regularly for over 2 years before doing this study. Beverages 2019, 5, 27 3 of 13

2.2. Coffee Samples The six single-serve coffee samples (K-Cup® Keurig, Inc.; Reading, MA, USA) were evaluated by the descriptive sensory panel. These single-serve coffee samples represented the range of roast levels from light to dark. Green Mountain Breakfast Blend represented the light roast. Green Mountain Nantucket Blend represented the blend of medium roasted African and Indonesian beans mixed with some French roast. Green Mountain Sumatra Reserved represented dark roasted organic Sumatra coffee. Tully’s Kona represented the blend including the famous Hawaiian coffee from the Kona Typica varietal, and was classiﬁed as medium roast. Tully’s Italian Roast represented a blend of dark roast. Lastly, Newman’s Own Organic represented a blend of medium and dark roast organic coffee beans. All coffee samples were stored at room temperature (20 ◦C) until testing and were used in the study within six weeks of their delivery.

2.3. Descriptive Sensory Analysis

2.3.1. Sample Preparation and Serving Keurig® Special Edition B60 Brewing System (Keurig®, Inc.; Reading, MA, USA) was used to brew the single-serve K-Cup® coffee samples. The machine was set up and cleaned following the instructions in the user manual. A K-Cup was placed in the K-Cup holder and 157.5 mL of coffee was brewed into a styrofoam cup (Dart J-cup, Dart Container Corp.; Mason, OH, USA). The coffee cups were labeled with 3-digit random numbers prior to serving. After each brewing cycle was completed, the K-Cup was removed and discarded immediately. Each coffee cup was covered with either a saucer (Econo Rim, Syracuse China; Lyncourt, NY, USA) or a plastic lid (Dart Container Corp.; Mason, OH, USA) and was then served immediately to the panelists monadically in random order.

2.3.2. Sample Evaluation A 180 min orientation session was completed to familiarize the descriptive panel with the samples. During orientation, the panel identified and defined aroma, flavor, and texture attributes present in each sample (Table1). Necessary references were determined to anchor and calibrate the intensity measurement on a 0 to 15-point scale with 0.5 increments (0.0 = none; 15.0 = extremely high intensity).

Table 1. The list of aroma, flavor, and texture descriptors identified by the coffee panel. Attributes. Definitions Coffee Amount or strength of Arabica coffee aroma or flavor Degree to which the coffee is roasted; ranges from green/no Roast roast–low–medium–dark–very dark Aromatics associated with blacked/acrid carbohydrates (e.g., burnt toast, Burnt espresso coffee) Aromatic associated with iodine in water; is described as chlorine-like, brassy, metallic, Rioy and chemical Ashy Bark-like lingering aromatics associated with a cold campfire Acidity A sour, sharp, puckering sensation in the mouth caused by acids Characteristic reminiscent of tobacco’s odor and taste, but should not be used for Tobacco burnt tobacco Stale Not fresh, flat, bodied down or reduced; old Citrus Aromatics associated with citrus fruits (e.g., lemon) Brown, sweet, dusty often biter aromatics associated with cocoa beans and powered Cocoa cocoa Bitter The amount of bitter basic taste; (e.g., caffeine solutions) Body (Mouthfeel) Viscosity of the coffee; heaviness on the tongue: thin to thick

Outlined in the following paragraph is the structured tasting protocol: Once the coffee was served, panelists opened the lid and the temperature of the coffee was taken with a digital thermometer (Model T220/38A Latte Thermometer, Comark; Hertfordshire, UK). When the temperature reached Beverages 2019, 5, 27 4 of 13

65.5 ◦C, the lid was replaced, keeping one end slightly opened. The panelists took a sniff to identify aroma descriptors belonging to that particular coffee. Panelists then slurped the sample and gently manipulated it in the mouth for 10–20 s to evaluate flavor and body/mouthfeel attributes. A small amount of sample was swallowed to discern bitterness on the back of the tongue. Afterward samples were expectorated. A 10 min break was taken between each sample, during which buttered bread and distilled water were used as palate cleansers. Buttered bread was prepared by spreading Land O’Lakes Whipped Butter (Whipped Butter Sweet Cream, Salted, 45% less fat, Land O’Lakes, Inc.; Arden Hills, 3 MN, USA) on a 4 cm slice of European Batard bread (Kroger; Cincinnati, OH, USA). During testing, panelists evaluated four samples per 180 min panel session. Samples were served one at a time and tasted individually by each panelist. A group discussion was then initiated by a panel leader to determine attributes present, their strengths, and to identify which references were needed. A new cup of the same sample was then served, along with references. The panel then individually evaluated the sample on ballots. The ratings were collected and written on the board by the panel leader. This was to identify any problem areas and whether other references should be reviewed. The panelists then determined and recorded their final score for the first replication of the sample. The next sample was served after a 10 min break and was evaluated following the same procedure. Two replications were carried out for the entire descriptive study.

2.4. Emotion Data The emotion data for the coffee samples were collected from 94 coffee drinkers (consumers) as reported in Bhumiratana et al. [7]. The consumers were between the ages of 18–70 years and there were 63 female and 31 male participants. In brief, the emotion proﬁle data for each of the six coffees for the 44 terms in CDE (AppendixA) were used. Intensity of emotion elicited by the coffee drinking experience were measured before and during consumption using a 5-point numerical scale (1 = not at all; 2 = slightly; 3 = moderately; 4 = very much; 5 = extremely). The emotion ratings prior to the coffee evaluation were subtracted from the emotion ratings during the evaluation before analyzing the data. Hierarchical cluster analysis (HCA) using Ward’s method was carried out on the overall liking data of the six coffee samples. The emotion proﬁles for each coffee were then separated for the six consumer clusters (AppendixB) and were used in this study. AppendixB also shows the mean liking scores for the coffee samples for each cluster.

2.5. Statistical Analyses Randomized complete block design was used for the descriptive evaluation of the six coffee samples. A two-way Analysis of Variance (ANOVA) using the GLIMMIX procedure (SAS® system version 9.2; SAS institute; Cary, NC, USA) at a 5% level of significance was performed on the data set to determine attributes significant in identifying differences among products. Coffee sample was the fixed effect and panelist was set as a random effect. Post-hoc mean separation was carried out by using Fisher’s Least Significant Difference. Principal component analysis (XLSTAT Sensory 19.01; Addinsoft, NY, USA) with covariance matrix was performed on the sensory descriptors to understand the sensory profile of the coffee samples. To investigate the relationship between the sensory attributes and the emotional responses to the drinking experience, partial least squares repression (XLSTAT Sensory 19.01) was conducted. Sensory drivers associated with the emotional experiences were identified among the 94 coffee users and in each consumer cluster.

3. Results and Discussion

3.1. Descriptive Sensory Data The ANOVA followed by mean separation indicated significant differences among the six coffee samples (p-value < 0.05) as shown in Table2. The coffee descriptive panel differentiated sensory Beverages 2019, 5, 27 5 of 13 elements that were distinctive to each coffee sample. Ashy was identified in Nantucket and Sumatra and was perceived to be more intense in Nantucket (a medium roast). Rioy was detected at the same Beverages 2019, 5, x FOR PEER REVIEW 5 of 13 intensity level in Nantucket, Newman, and Sumatra, but was not present in the other samples. Tobacco appearedintensity in the level Italian in Nantucket, sample, Newman,stale underlined and Sumatra, Newman, but was and not cocoapresentaroma in the wasother unique samples. to Tobacco Kona. appeared in the Italian sample, stale underlined Newman, and cocoa aroma was unique to Kona. Table 2. The mean scores of the descriptive analysis of the coffee samples. Table 2. The mean scores of the descriptive analysis of the coffee samples. Abbreviation Descriptors Breakfast Italian Kona Nantucket Newman Sumatra forAbbreviation Figures Descriptors Breakfast Italian Kona Nantucket Newman Sumatra Aroma for Figures CoffeeAroma CoffeeA 7.79 ab 5.42 c 8.33 a 8.58 a 7.50 ab 8.58 a RoastCoffee RoastACoffeeA 6.927.79b ab 8.505.42a c 8.337.42 ba 8.588.50 a a 7.507.58 ab b 8.588.58 a a BurntRoast BurntARoastA 0.676.92c b 4.508.50a a 7.420.17 bc 8.502.92 a b 7.582.83 b b 8.584.33 a a a a a RioyBurnt RioyABurntA 0.000.67b c 0.004.50b a 0.170.00 bc 2.921.33 b 2.831.58 b 4.331.58 a b b b a b a AshyRioy AshyARioyA 0.000.00 b 0.000.00 b 0.00 b 1.332.75 a 1.580.00 a 1.581.92 a b b a b b b CocoaAshy CocoaAAshyA 0.000.00 b 0.000.00 b 0.002.33 b 2.750.00 a 0.000.00 b 1.920.00 a FlavorCocoa CocoaA 0.00 b 0.00 b 2.33 a 0.00 b 0.00 b 0.00 b c c a a a CoffeeFlavor CoffeeF 8.25 8.00 11.75 10.33 b 11.75 12.50 c a b b a a RoastCoffee RoastFCoffeeF 7.088.25c 10.178.00 c 11.758.54 a 10.338.92 b 11.7510.50 a 12.5010.08 a Burnt BurntF 1.58 d 6.67 b 6.75 b 3.75 c 8.33 a 8.50 a Roast RoastF 7.08 c 10.17 a 8.54 b 8.92 b 10.50 a 10.08 a Rioy RioyF 0.00 b 0.00 b 0.00 b 2.04 a 2.00 a 1.58 a Burnt BurntF 1.58 d 6.67 b 6.75 b 3.75 c 8.33 a 8.50 a Ashy AshyF 0.00 b 0.00 b 0.00 b 2.92 a 0.00 b 2.83 a b b b a a a CitrusRioy CitrusRioyF 4.420.00a 0.000.00b 0.00 b 2.040.00 b 2.000.00 b 1.580.00 b b b b a b a TobaccoAshy TobaccoAshyF 0.000.00b 8.080.00a 0.00 b 2.920.00 b 0.000.00 b 2.830.00 b StaleCitrus StaleCitrus 0.004.42b a 0.000.00b b 0.00 b 0.000.00 b b 0.004.42 b a 0.000.00 b b AcidityTobacco AcidityTobacco 5.920.00a b 4.838.08c a 5.830.00 abb 0.005.92 b a 0.004.92 b c 0.004.92 b c BitterStale BitterStale 3.080.00d b 9.500.00a b 0.008.13 b 0.005.42 b c 4.428.00 a b 0.008.42 b b TextureAcidity Acidity 5.92 a 4.83 c 5.83 ab 5.92 a 4.92 c 4.92 c BodyBitter BodyBitter 6.383.08d d 8.679.50ab a 8.138.33 b 5.427.63 c c 8.009.13 b a 8.428.83 b ab Texture a,b,c Row means with common superscrits are not significantly different at p > 0.05. Body Body 6.38 d 8.67 ab 8.33 b 7.63 c 9.13 a 8.83 ab Principal componenta,b,c Row means analysis with common (PCA) superscrits was performed are not significantly to visualize different the product at p > 0.05. placements on the sensory spacePrincipal based component on the sensory analysis attributes. (PCA) was Figure performed1 illustrates to visualize sensory the product profiles placements of the coffees on the created by thesensory coffee space panel based in the on PCA the biplot.sensory attributes. Figure 1 illustrates sensory profiles of the coffees created by the coffee panel in the PCA biplot.

1.2

AshyA CoffeeA AshyF RioyF RioyA Nantucket CoffeeF Sumatra

Acidity RoastA

BurntA 33%

- 0.0 Stale Newman RoastF BurntF Body PC2 PC2 Citrus Breakfast CocoaA Kona BitterF

Tobacco Italian

-1.2 -1.2 0.0 1.2 PC1 - 46%

FigureFigure 1. Principal 1. Principal component component analysis analysis (PCA) (PCA) biplotbiplot of of the the sensory sensory profiles proﬁles of the of thesix coffees six coffees generated generated by theby coffee the coffee panel. panel.

Beverages 2019, 5, 27 6 of 13

PC1 explained 46% of the data variation and seemed to reflect characteristics generated by roasting. Acidity and citrus anchored the left side of PC1 and described Breakfast. Burnt, roast, bitter flavors and bodyBeveragesanchored 2019, 5, x FOR the PEER right REVIEW sideof PC1 and seemed to characterize Newman and Sumatra.6 ofAcidity 13 , bitter, burnt (flavor and aroma), roast flavor, coffee flavor (except for Italian) and body were influenced PC1 explained 46% of the data variation and seemed to reflect characteristics generated by by degree of roasting. Acidity was more intense in the lighter roasts, while bitter, burnt (flavor and roasting. Acidity and citrus anchored the left side of PC1 and described Breakfast. Burnt, roast, bitter aroma), roast, and coffee flavors, and body increased with degree of roasting. The impact of degree of flavors and body anchored the right side of PC1 and seemed to characterize Newman and Sumatra. roastingAcidity on, aroma bitter, burnt and flavor (flavor in and coffee aroma), has beenroast extensivelyflavor, coffee studiedflavor (except [11,13 for,20 –Italian)22] and and is similarbody were to what was foundinfluenced in this by research.degree of However,roasting. Acidi degreety was of more roasting intense was in not the the lighter only roasts, factor while affecting bitter the, burnt sensory characteristics(flavor andof aroma), coffee. roast PC2, and explained coffee flavors, 33% ofand the body data increased set and with provided degree additionalof roasting. informationThe impact on sensoryof degree elements of roasting for Nantucket, on aroma Kona,and flavor and in Italian. coffee hasCoffee beenaroma extensively and roast studiedaroma [11,13, did20– not22] seemand is to be dependentsimilaron to roast what level. was found The intensities in this research. of these However, aroma attributes degree of forroasting Nantucket was not (medium the only roast) factor were higheraffecting than Newman the sensory (medium-dark characteristics roast). of coffee. The PC2 sensory explained profiles 33% indicated of the data some set sensory and provided attributes mightadditional be independent information of degree on sensory of roasting, elements whichfor Nantucket, confirmed Kona, that and other Italian. factors Coffee might aroma be and influencing roast aroma did not seem to be dependent on roast level. The intensities of these aroma attributes for the sensory characteristics of coffee. The origins of coffee, including growing regions and variety of Nantucket (medium roast) were higher than Newman (medium-dark roast). The sensory profiles bean, evidently have a noticeable impact on the sensory fingerprint of each coffee; this is supported by indicated some sensory attributes might be independent of degree of roasting, which confirmed that numerousother factors studies might [11, 13be, 23influencing–26]. the sensory characteristics of coffee. The origins of coffee, including Thegrowing sensory regions data and from variety the of descriptive bean, evidently panel have was a noticeable then utilized impact in on the the next sensory step fingerprint to identify the sensoryof each drivers coffee; responsible this is supported for the by emotional numerous responsesstudies [11,13, elicited23–26 by]. the coffee drinking experience. The sensory data from the descriptive panel was then utilized in the next step to identify the 3.2. Identifyingsensory drivers Sensory responsible Drivers for the Emotionalemotional responses Experience elicited by the coffee drinking experience. The sensory descriptive data was studied with emotion responses for the same set of coffee 3.2. Identifying Sensory Drivers for the Emotional Experience samples created by 94 coffee drinkers in the study done previously by Bhumiratana et al. [7]. Partial Least SquareThe sensory Regression descriptive (PLSR) data was was used studied to identify with emotion sensory responses drivers for of the the same emotion set of coffee responses samples created by 94 coffee drinkers in the study done previously by Bhumiratana et al. [7]. Partial (Figure2). Coffee aroma, surprisingly, elicited a range of negative emotions (bored, disgusted, annoyed, Least Square Regression (PLSR) was used to identify sensory drivers of the emotion responses (Figure and disappointed) even though it is well known that ‘coffee aroma’ elicites positive feelings, including 2). Coffee aroma, surprisingly, elicited a range of negative emotions (bored, disgusted, annoyed, and alertnessdisappo ofinted the) mental even though state, it and is well is the known driver that of ‘coffee coffee aroma’ consumption elicites positive [11,27]. feelings, This may including be because the definitionalertness of of thecoffee mentalaroma state, used and is by the the driver coffee of panelcoffee consumption and consumers [11,27 could]. This be may different, be because a common the problemdefinition in the offield coffee of consumeraroma used research by the coffeewhen integratingpanel and consumers sensory and could consumer be different, data atogether. commonCoffee aroma,problem by the in definition the field listedof consumer in Table research1, was thewhen aroma integrating of pure sensory Arabica and beans, consumer which data consumers together. may not beCoffee familiar aroma, with by and the might definition have listed led to in aTable negative 1, was perception the aroma [5 ]. of pure Arabica beans, which consumers may not be familiar with and might have led to a negative perception [5].

1.2

Bored Peaceful Disappointed CoffeeA Disgusted Annoyed Wild Grouchy Off balance Soothing Relaxed RioyF RioyA Citrus Warm Breakfast Acidity Stale Guilty CoffeeF Newman AshyA AshyF Nantucket Rested Sumatra Worried Active Awake Empowering 0.0 Nervous Rewarded In control Educated Curious Pleasant Kona Comfortable CocoaA Boosted Special Good BurntF Dim 2 (X: 24%, Y: 24%) Y: 24%,(X: 2 Dim Free Joyful Merry Pleased BurntA RoastF Fun Satisfied Motivated Understanding RoastA Body Balanced Jump start Fulfilled Productive Social Bitter Clear-minded Energetic Content Italian Jolted Tobacco

-1.2 -1.2 0.0 1.2 Dim 1 (X:42%, Y:20%)

FigureFigure 2. Partial 2. Partial least least square square regression regression analysis analysis of descriptivedescriptive sensory sensory and and CDE CDE emotion emotion lexicon lexicon of of the sixthe coffee six coffee samples samples for for 94 94 consumers. consumers.

Beverages 2019, 5, 27 7 of 13

Positive emotions seemed to be driven by cocoa aroma, bitter, tobacco, roast, burnt, and body. Cocoa aroma may elevate good and pleasant emotions, which was consistent with previous studies. King and Meiselman [6] found that among the five food categories evaluated, chocolate was reported to have the highest ratings for 15 of the positive emotions (out of 24 positive emotions on a list of 39 terms). Macht and Mueller [28] reported consumption of chocolate could immediately reduce negative mood state, although the effect was temporary. It is also common knowledge that chocolate and its resemblance usually induces positive feelings in the general population. Tobacco flavor evoked the feelings of jolted and content. Coffee users may initially be surprised (i.e., jolted) by the unfamiliar tobacco attribute that was not commonly found in all coffee (only one coffee sample in this study exhibited this sensory attribute). However, they were accepting of the experience (i.e., content), which indicates that having a tobacco attribute in coffee could potentially enhance the drinking experience for general coffee users. Bitter aroused energetic and productive feelings. Roast and burnt (flavor and aroma), and body texture made consumers feel jump start, satisfied, boosted, and special. On the contrary, citrus, hay-like, and acidity appeared to elicit a feeling of off-balance. Like tobacco, consumers may not be familiar with experiencing these sensory characteristics in coffee and were caught off-guard by them. Unlike tobacco, they may not find these attributes appropriate for coffee, hence the off-balance emotion. Because emotions are context specific [29], it seems that citrus and acidity attributes may not fit well with the concept of coffee, which caused negative feelings to develop. It seems the characteristics of dark roast coffee (roasted, burnt, bitter, and body) elicited positive-high energy feelings. This is likely because there were more participants who preferred darker roasts since coffee preference was not one of the criteria during recruitment. This finding identified tobacco, roasted, burnt, bitter, and body as the sensory drivers for this population of 94 coffee users. Since consumers have varying preferences and are affected differently by sensory stimuli, the 94 coffee users were examined more closely in our previous study [7] through clustering the consumers in six segments. The entire set of 94 coffee drinkers was clustered into six groups based on their acceptability scores for the coffee samples and emotion profiles were generated for each set of consumers. We conducted PLSR analysis on each consumer cluster to determine whether relationships can be drawn between the sensory characteristics and emotions elicited by the perceived attributes for each consumer cluster. For coffee drinkers in Cluster 1 (n = 20), who liked all the coffees [7], the tobacco attribute seemed to elicit social, jump start, and special feelings, while the characteristics of dark roasts (high intensity of roast, burnt, and body/mouthfeel) appeared to make them feel empowering and relaxed (Figure3). Acidity was associated with awake and disgusted and may be a negative attribute for this group. Cluster 2 (n = 17; Figure4) consisted of consumers who dislike Breakfast (classified as light roast) [ 7]. The PLSR map indicated that attributes citrus and acidity elicited negative emotions (e.g., disappointed, disgusted, annoyed), and dark roast characteristics (roast, burnt, bitter, and body) were driving positive emotions (e.g., satisfied, energetic, rewarded, boosted, in control, empowering). This group of coffee drinkers seem to relate the coffee aroma to a grouchy emotion and the tobacco attribute to clear-minded, wild, and good feelings. Cluster 3 (n = 24) was identified to like Nantucket and Breakfast but dislike Sumatra [7]. The PLSR bi-plot (Figure5) illustrated that hay-like, citrus, and acidity brought out positive emotions (e.g., merry, pleasant, understanding, relaxed, rewarded) for this group of coffee drinkers. Empowering and boosted emotions seemed to be induced by coffee flavor, ashy, and rioy, while tobacco elicited feelings of off-balance, jolted, and social. Negative emotions (disappointed and disgusted) were driven by roast, burnt, and body characteristics. Coffee drinkers grouped into Cluster 4 (n = 13) were those that did not prefer any of the six coffees [7]. Nantucket was liked the most by this group, and that might be the reason that peaceful, energetic, pleased, and awake are somewhat encircling this sample in Figure6. Since they did not have any firm preferences, the coffees may have elicited mixed emotions for this group (Figure6), which are not easily discernible. Beverages 2019, 5, 27 8 of 13 Beverages 2019,, 5,, xx FORFOR PEERPEER REVIEWREVIEW 8 of 13

1.2 Cluster 1 CoffeeA Bored AshyA AshyF RioyA Disgusted RioyF CoffeeF Awake Acidity Nantucket Guilty Grouchy Wild Off balance Fun Sumatra Educated Boosted Nervous Curious Worried Rested Active JoyfulJoyful InIn controlcontrol CocoaA Clear--minded Pleased RoastA Citrus Soothing Stale 0.0 Breakfast Kona Newman Energetic Relaxed BurntF Balanced Empowering Body Fulfilled Motivated BurntA RoastF

Annoyed DIM 2 (X: 24%, Y: 21%) Y: 24%, 2(X: DIM DIM 2 (X: 24%, Y: 21%) Y: 24%, 2(X: DIM Disappointed Peaceful Good Productive Understanding Free Bitter Comfortable Warm Satisfied JoltedJolted Pleasant Merry Rewarded Content ItalianItalian Social Special Tobacco JumpJump startstart

-1.2-1.2 -1.2-1.2 0.0 1.2 DIM 1 (X: 45%, Y: 24%)

Figure 3. Partial leastlleast squaresquare regressionregression analysisanalysis of descriptivedescriptive sensorysensory and CDE emotionemotion lexicon lexicon of thethe s sixixix coffeecoffee samplessamples forfor consumerconsumer ClusterCluster 1.1.

1.2 Cluster 2 Good Tobacco ItalianItalian Clear--minded Wild Merry Fun Content Warm Active Bitter Productive Free Energetic Nervous Balanced Energetic JoyfulJoyful Satisfied Rested BurntA RoastF CocoaA Pleased JoltedJolted Body Citrus Breakfast Kona Motivated Social BurntF Comfortable Soothing Rewarded 0.0 Disappointed Newman Stale RoastA Boosted Disgusted Educated InIn controlcontrol Awake Pleasant Relaxed Empowering

Understanding Special Fulfilled DIM 2 (X: 24%, Y: 18%) Y: 24%, 2 (X: DIM DIM 2 (X: 24%, Y: 18%) Y: 24%, 2 (X: DIM Acidity Sumatra JumpJump startstart Worried Curious Annoyed Nantucket CoffeeF AshyA RioyF Bored Guilty Peaceful Off balance AshyF RioyA CoffeeA Grouchy

-1.2-1.2 -1.2-1.2 0.0 1.2 DIM (X: 45%, Y: 40%)

Figure 4. Partial leastlleast squaresquare regressionregression analysisanalysis of descriptivedescriptive sensorysensory and CDE emotionemotion lexicon lexicon of thethe s sixixix coffeecoffee samplessamples forfor consumerconsumer ClusterCluster 2.2..

Beverages 2019, 5, 27 9 of 13

Beverages 20192019,, 55,, xx FORFOR PEERPEER REVIEWREVIEW 99 of 1133

1.21.2 Cluster 3 Social TobaccoTobacco Social Italian Italian Active Guilty JoltedJolted Active Guilty Off balance CuriousCurious FunFun Bitter Pleased Bitter JumpJump startstart Pleased Productive Motivated Productive SpecialSpecial CocoaA Motivated Body Fulfilled CocoaA Annoyed In control Body EducatedEducated Fulfilled Satisfied In control RoastFRoastF BurntF Satisfied Joyful Merry BurntFBurntA Kona Nervous Joyful Merry BurntA Kona Good Citrus 0.0 Disgusted Stale Newman Good BreakfastCitrus 0.0 DisgustedAwake Stale EnergeticEnergetic PleasantBreakfast Awake RoastARoastA BalancedBalanced ComfortablePleasant Worried ComfortableFree Content Worried ClearFree-minded Disappointed Content Clear-minded Acidity DIM 2 (X: 23%, Y: 16%) Y: 23%, 2(X: DIM Peaceful Acidity

DIM 2 (X: 23%, Y: 16%) Y: 23%, 2(X: DIM Bored Warm Bored Peaceful Warm RestedRested SumatraSumatra Understanding SoothingSoothing Rewarded Nantucket RewardedRelaxed CoffeeFCoffeeF Wild Relaxed EmpoweringEmpowering Wild BoostedBoosted Grouchy RioyARioyA RioyF Grouchy RioyF AshyA AshyF CoffeeACoffeeA

-1.2-1.2 -1.2-1.2 0.00.0 1.21.2 DIM 1 (X: 45%, Y: 19%)

FigureFigure 5. Partial 5. Partial least lleasteast square ssquare regression rregressionegression analysis analysis ofof d descriptiveescriptiveescriptive ssensoryensory sensory andand and CDE CDE eemotion emotion lexiconlexicon lexicon of of the sixthethe coffee ssixix ccoffee samples ssamples for for consumer cconsumer Cluster Cluster 3. 3..

1.21.2 Cluster 4 Active CoffeeACoffeeA

Bored Bored InIn controlcontrol Awake Energetic Awake RioyFRioyF Acidity Energetic AshyA RioyARioyA Acidity AshyA AshyF Annoyed Peaceful CoffeeFAnnoyed Peaceful Nantucket PleasedPleased CoffeeF Citrus Nantucket Warm Motivated Citrus BreakfastBreakfast Empowering Warm Motivated Empowering Merry Sumatra Relaxed Merry NewmanSumatra Relaxed Newman StaleStale Soothing Nervous Grouchy 0.0 Wild Soothing Kona Rested 0.0 Wild Kona Fulfilled SatisfiedSatisfied Rested Rewarded Pleasant Educated CocoaAFulfilled Rewarded FunFun Pleasant Educated CocoaA Guilty Boosted CuriousCurious Guilty Balanced Boosted FreeFree Off balance Balanced Disappointed Disgusted Off balance RoastARoastA DIM 2 (X: 25%, Y: DIM 17%) 2(X: Y: 25%, BurntF DIM 2 (X: 25%, Y: DIM 17%) 2(X: Y: 25%, SpecialSpecial BurntABurntA BurntF Good Body RoastFRoastF Good JoltedJolted Body Understanding ClearClear--minded Social Worried Content Jump start Social Comfortable Worried Content Jump start Comfortable Productive Bitter JoyfulJoyful Productive Bitter

ItalianItalian TobaccoTobacco

-1.2-1.2 -1.2-1.2 0.00.0 1.21.2 DIM 1 (X: 43%, Y: 28%)

FigureFigure 6. Partial 6. Partial least lleasteast square ssquare regression rregressionegression analysis analysis ofof d descriptiveescriptiveescriptive ssensoryensory sensory andand and CDE CDE eemotion emotion lexiconlexicon lexicon of of the six coffee samples for consumer Cluster 4. the sixthe coffee six coffee samples samples for for consumer consumer Cluster Cluster 4. 4. Cluster 5 (n = 10) was composed of coffee drinkers who gave a high liking rating for Breakfast ClusterCluster 5 (n 5= (n 10) = 10) was was composed composed of of coffee coffee drinkersdrinkers who who gave gave a ahigh high liking liking rating rating for Breakfast for Breakfast and disliked the dark roasts (Newman, Italian, and Sumatra) [7]. Citrus and acidity were shown to and dislikedand disliked the darkthe dark roasts roasts (Newman, (Newman, Italian, Italian, andand Sumatra) [7]. [7 ].CitrusCitrus andand acidityacidity werewere shown shown to to explain positive emotions (e.g., relaxedrelaxed,, soothingsoothing,, understanding,, peaceful),), andand coffeecoffee aroma explained explain positive emotions (e.g., relaxed, soothing, understanding, peaceful), and coffee aroma explained funfun,, rewarewardedrded,, andand pleased (Figure(Figure 7).7). OnOn thethe otherother hand,hand, coffeecoffee flavorflavor andand rioyrioy appeared to describe fun, rewarded, and pleased (Figure7). On the other hand, coffee flavor and rioy appeared to describe negative emotions, including nervous,, disgusted,, andand annoyed.. CoffeeCoffee drinkersdrinkers inin ClusterCluster 66 ((n = 10) negative emotions, including nervous, disgusted, and annoyed. Coffee drinkers in Cluster 6 (n = 10) were classified as preferring Kona coffee [7]. The PLSR bi-plot (Figure8) reflects that this group of consumers were attracted to the cocoa aroma as most positive emotions (i.e., balanced, productive, fulfilled, Beverages 2019, 5, x FOR PEER REVIEW 10 of 13 Beverages 2019, 5, x FOR PEER REVIEW 10 of 13 Beverages 2019, 5, 27 10 of 13 were classified as preferring Kona coffee [7]. The PLSR bi-plot (Figure 8) reflects that this group of consumerswere classified were as attractedpreferring to Kona the cocoa coffee aroma [7]. The as PLSR most positivebi-plot (Figure emotions 8) reflects (i.e., balanced that this, productive group of, fulfilledawakeconsumers, motivated, awake were, ,motivated and attractedenergetic, and to). theTobaccoenergetic cocoaalso). aroma Tobacco described as also mostgood described positiveand soothing good emotions emotions,and soothing (i.e., whilebalanced emotions,acidity, productiveseemed while, aciditytofulfilled generate seemed, awake mixed, tomotivated generate emotions, andmixed of rewardedenergetic emotions)., free Tobacco of, jolted rewarded also, and , described nervousfree, jolted. ,good and nervousand soothing. emotions, while acidity seemed to generate mixed emotions of rewarded, free, jolted, and nervous. 1.2 Cluster 5 1.2 Cluster 5 Rewarded Satisfied CoffeeA Peaceful RewardedPleased Fun Understanding Satisfied EmpoweringCoffeeA Special Active Peaceful Pleased Fun Awake ActiveBoosted UnderstandingAcidity Pleasant EmpoweringRested ContentSpecial Awake Jump start Acidity Warm Educated Free AshyA Boosted Pleasant Guilty Rested Content JumpProductive start Warm NantucketEducated FreeAshyF AshyARioyF RioyA Citrus Relaxed Soothing Guilty ProductiveAnnoyed Off balanceNantucket AshyF CoffeeFRioyF RioyANervous CitrusBreakfastRelaxed SoothingSocial Good Joyful Annoyed Off balanceIn control CoffeeF DisgustedNervous Breakfast BalancedSocial Good In control Sumatra Merry Joyful Disgusted Balanced Curious Fulfilled Energetic NewmanSumatra StaleMerry 0.0 Kona CocoaA DisappointedCurious ComfortableFulfilled Energetic NewmanClear-mindedStaleWorried 0.0 Kona CocoaA Disappointed Comfortable Clear-minded WorriedMotivated Bored RoastAMotivated DIM 2 (X: 27%, Y: DIM 24%)2(X: Y: 27%, Bored Jolted Wild Grouchy BurntARoastA DIM 2 (X: 27%, Y: DIM 24%)2(X: Y: 27%, Jolted Wild BurntF Grouchy Body RoastF BurntA BurntF Body RoastF Bitter Italian Bitter TobaccoItalian Tobacco

-1.2 -1.2 0.0 1.2 -1.2 -1.2 DIM 1 (X:42%,0.0 Y: 32%) 1.2 DIM 1 (X:42%, Y: 32%)

Figure 7. Partial least square regression analysis of descriptive sensory and CDE emotion lexicon of Figure 7. Partial least square regression analysis of descriptive sensory and CDE emotion lexicon of theFigure six c7.offee Partial samples least sforquare consumer regression Cluster analysis 5. of descriptive sensory and CDE emotion lexicon of the sixsix coffeecoffee samplessamples for consumerconsumer Cluster 5.5.

1.2 1.2 Cluster 6 Comfortable Understanding Cluster 6 Comfortable Pleased Understanding Fulfilled Energetic Satisfied AwakePleased Rested PleasantFulfilled MotivatedEnergetic AwakeBalanced Productive Bitter SatisfiedWarm Rested Motivated Special CocoaAPleasant Fun Clear-minded RelaxedBitter WarmJump start BalancedKona ActiveProductive CocoaA Social Clear-minded Body Relaxed RoastASpecial Joyful Kona Fun Content BurntF Jump startIn control Educated Active Body RoastA Joyful SocialBoosted EmpoweringContent BurntF CoffeeF Merry In control Educated RoastF TobaccoBoosted Empowering CoffeeF Merry Soothing RoastF AshyF AshyA Good ItalianTobacco Guilty BurntA Sumatra Soothing AshyF CuriousAshyA NantucketGood Italian Guilty BurntA SumatraCoffeeA Free 0.0 Curious Nantucket Acidity Jolted CoffeeA Free 0.0 RioyA RioyF Rewarded Jolted Peaceful Acidity Nervous RioyA RioyF Rewarded Peaceful Nervous DIM 2 (X: 20%, Y: DIM 37%) 2(X: Y: 20%, Stale Newman

DIM 2 (X: 20%, Y: DIM 37%) 2(X: Y: 20%, Wild Stale Newman Wild Disappointed Worried Off balance Breakfast Disappointed WorriedBored Citrus Annoyed Off balance Breakfast Bored Annoyed Grouchy Citrus Disgusted Grouchy Disgusted

-1.2 -1.2 0.0 1.2 -1.2 -1.2 DIM 1 (X: 36%,0.0 Y: 27%) 1.2 DIM 1 (X: 36%, Y: 27%)

Figure 8. Partial least square regression analysis of descriptive sensory and CDE emotion lexicon of Figure 8. Partial least square regression analysis of descriptive sensory and CDE emotion lexicon of the six coffee samples for consumer Cluster 6. theFigure six c8.offee Partial samples least sforquare consumer regression Cluster analysis 6. of descriptive sensory and CDE emotion lexicon of the six coffee samples for consumer Cluster 6.

Beverages 2019, 5, 27 11 of 13

This study presented the useful interaction of sensory and emotion data. Using the emotion proﬁles generated by the 44 emotions on the coffee drinking experience lexicon, we were able to identify some sensory drivers for speciﬁc emotions elicited by coffee drinking.

4. Conclusions The PLSR maps indicated that sensory descriptive data might be used to describe emotions proﬁles elicited by coffee drinking. The PLSR maps were used to identify which attributes had an impact on positive or negative emotional responses from various groups of coffee drinkers. In general, coffee aroma, citrus, and acidity elicited negative feelings while cocoa aroma, tobacco, bitter, roast, burnt, and body generated positive emotions. As consumers have differing likes and dislikes, this study also examined each consumer cluster based on their preferences and identiﬁed sensory drivers for the emotions experienced by each cluster. These insights generated by the interaction of sensory and emotion data are valuable to both marketers and product developers by explaining acceptability data and change in consumption or purchase behavior.

Author Contributions: N.B. conceptualized and designed the study with the help of K.A. and E.C.I.; N.B. ran the descriptive analysis under the guidance of M.W.; The data analysis and manuscript preparation were done by N.B. with the help of K.A. and E.C.I. Funding: This research received no external funding. Conﬂicts of Interest: The authors declare no conﬂict of interest.

Appendix A The emotion lexicon for Coffee Drinking Experience (CDE) [7]. Active Disgusted Jolted Relaxed Annoyed Educated Joyful Rested Awake Empowering Jumpstart Rewarded Balanced Energetic Merry Satisﬁed Boosted Free Motivated Social Bored Fulﬁlling Nervous Soothing Clear-minded Fun Off-balance Special Comfortable Good Peaceful Understanding Content Grouchy Pleasant Warm Curious Guilty Pleased Wild Disappointed In-control Productive Worried

Appendix B Mean liking scores on a 9-point Hedonic scale for each consumer cluster and coffee sample [7]. Cluster Breakfast Italian Kona Nantucket Newman Sumatra C1 (n = 20) 7.7 6.9 7.3 7.7 7.5 7.2 C2 (n = 17) 4.4 6.9 6.7 6.1 7.2 6.5 C3 (n = 24) 7.0 6.0 5.8 7.5 5.3 3.7 C4 (n = 13) 4.6 3.5 5.7 6.0 5.5 5.4 C5 (n = 10) 7.1 3.3 3.5 4.1 2.2 2.1 C6 (n = 10) 5.9 6.4 7.0 3.6 5.1 6.1

References

1. Chrea, C.; Grandjean, D.; Delplanque, S.; Cayeux, I.; Le Calve, B.; Aymard, L.; Velazco, M.I.; Sander, D.; Scherer, K. Mapping the semantic space for the subjective experience of emotional responses to odors. Chem. Senses 2009, 34, 49–62. [CrossRef][PubMed] Beverages 2019, 5, 27 12 of 13

2. Porcherot, C.; Delplanque, S.; Raviot-Derrien, S.; Le Calve, B.; Chrea, C.; Gaudreau, N.; Cayeux, I. How do you feel when you smell this? Optimization of a verbal measurement of odor-elicited emotions. Food Qual. Prefer. 2010, 21, 938–947. [CrossRef] 3. Thomson, D.M.H. Sensory Cues for Emotional Responses to Foods & Drinks. International Union of Food Science and Technology (IUFoST), 2006. Available online: http://dx.doi.org/10.1051/IUFoST:20061091 (accessed on 1 December 2018). 4. Thomson, D.M.H.; Crocker, C.; Marketo, C.G. Linking sensory characteristics to emotions: An example using dark chocolate. Food Qual. Prefer. 2010, 21, 1117–1125. [CrossRef] 5. Gibson, E.L. Emotional inﬂuences on food choice: Sensory, physiological and psychological pathways. Physiol. Behav. 2006, 89, 53–61. [CrossRef][PubMed] 6. King, S.C.; Meiselman, H.L. Development of a method to measure consumer emotions associated with foods. Food Qual. Prefer. 2009, 21, 168–177. [CrossRef] 7. Bhumiratana, N.; Adhikari, K.; Chambers, E., IV. The development of an emotion lexicon for the coffee drinking experience. Food Res. Int. 2014, 61, 83–92. [CrossRef] 8. Illy, E. The complexity of coffee. Sci. Am. 2002, 286, 86–91. [CrossRef] 9. Grosch, W. Flavour of coffee: A review. Nahrung 1998, 42, 344–350. [CrossRef] 10. Czerny, M.; Mayer, F.; Grosch, W. Sensory study on the character impact odorants of roasted Arabica coffee. J. Agric. Food Chem. 1999, 47, 695–699. [CrossRef] 11. Illy, A.; Viani, R. Espresso Coffee: The Science of Quality, 2nd ed.; Elsevier Academic Press: London, UK, 2005. 12. Di Donfrancesco, B.; Gutierrez Guzman, N.; Chambers, E., IV. Comparison of results from cupping and descriptive sensory analysis of Colombian brewed coffee. J. Sens. Stud. 2014, 29, 301–311. [CrossRef] 13. Bhumiratana, N.; Adhikari, K.; Chambers, E., IV. Green coffee beans to brewed coffee: Evolution of coffee aroma. LWT-Food Sci. Technol. 2011, 44, 2185–2192. [CrossRef] 14. Chambers, D.H.; Allison, A.; Chambers, E., IV. Training effects on performance of descriptive panelists. J. Sens. Stud. 2004, 19, 486–499. [CrossRef] 15. Chambers, E., IV; Smith, E.A. Effects of testing experience on performance of trained sensory panelists. J. Sens. Stud. 1993, 8, 155–166. 16. Wolters, C.J.; Allchurch, E.M. Effect of training procedure on the performance of descriptive panels. Food Qual. Prefer. 1994, 5, 203–214. [CrossRef] 17. Bitnes, J.; Ueland, O.; Moller, P.; Martens, M. Reliability of sensory assessors: Issues of retention and learning. J. Sens. Stud. 2008, 23, 852–870. [CrossRef] 18. Chambers, E., IV; Sanchez, K.; Phan, U.T.X.; Miller, R.; Civille, G.V.; Di Donfrancesco, B. Development of a “living” lexicon for descriptive sensory analysis of brewed coffee. J. Sens. Stud. 2016, 31, 465–480. [CrossRef] 19. Jager, G.; Schlich, P.; Tijjsen, I.; Yao, J.; Visalli, M.; de Graaf, C.; Steiger, M. Temporal dominance of emotions: Measuring dynamics of food-related emotions during consumption. Food Qual. Prefer. 2014, 37, 87–99. [CrossRef] 20. Schenker, S.; Heinemann, C.; Huber, M.; Pompizzi, R.; Perren, R.; Escher, F. Impact of roasting conditions on the formation of aroma compounds in coffee beans. J. Food Sci. 2002, 67, 60–66. [CrossRef] 21. Yeretzian, C.; Jordan, A.; Badoud, R.; Lindinger, W. From the green bean to the cup of coffee: Investigating coffee roasting by on-line monitoring of volatiles. Eur. Food Res. Technol. 2002, 214, 92–104. [CrossRef] 22. Baggenstoss, J.; Poisson, L.; Kaegi, R.; Perren, R.; Escher, F. Coffee roasting and aroma formation: Application of different time and temperature conditions. J. Agric. Food Chem. 2008, 56, 5836–5846. [CrossRef] 23. Mayer, F.; Czerny, M.; Grosch, W. Inﬂuence of provenance and roast degree on the composition of potent odorants in Arabica coffees. Eur. Food Res. Technol. 1999, 209, 242–250. [CrossRef] 24. Decazy, F.; Avelino, J.; Guyot, B.; Perriot, J.J.; Pineda, C.; Cilas, C. Quality of different Honduran coffees in relation to several environments. J. Food Sci. 2003, 68, 2356–2361. [CrossRef] 25. Nebesny, E.; Budryn, G. Evaluation of sensory attributes of coffee brews from robusta coffee roasted under different conditions. Eur. Food Res. Technol. 2006, 224, 159–165. [CrossRef] 26. Ross, C.F.; Pecka, K.; Weller, K. Effect of storage conditions on the sensory quality of ground Arabica coffee. J. Food Qual. 2006, 29, 596–606. [CrossRef] 27. Seo, H.S.; Hirano, M.; Shibato, M.; Rakwal, R.; Hwang, I.K.; Masuo, Y. Effect of coffee bean aroma on the rat brain stressed by sleep deprivation: A selected transcript and 2D get-based proteome analysis. J. Agric. Food Chem. 2008, 56, 4665–4673. [CrossRef][PubMed] Beverages 2019, 5, 27 13 of 13

28. Macht, M.; Mueller, J. Immediate effects of chocolate on experimentally induced mood states. Appetite 2007, 49, 667–674. [CrossRef][PubMed] 29. Richins, M.L. Measuring emotions in the consumption experience. J. Cons. Res. 1997, 24, 127–146. [CrossRef]

© 2019 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (http://creativecommons.org/licenses/by/4.0/).