Cup colour influences consumers’ expectations and experience on tasting specialty coffee

Accepted Manuscript Cup colour influences consumers’ expectations and experience on tasting specialty coffee Fabiana M. Carvalho, Charles Spence PII: DOI: Reference:

S0950-3293(18)31001-2 https://doi.org/10.1016/j.foodqual.2019.03.001 FQAP 3673

To appear in:

Food Quality and Preference

Received Date: Revised Date: Accepted Date:

10 December 2018 8 February 2019 4 March 2019

Please cite this article as: Carvalho, F.M., Spence, C., Cup colour influences consumers’ expectations and experience on tasting specialty coffee, Food Quality and Preference (2019), doi: https://doi.org/10.1016/j.foodqual. 2019.03.001

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Cup colour influences consumers’ expectations and experience on tasting specialty coffee

Fabiana M. Carvalho & Charles Spence 1) Fabiana M. Carvalho (corresponding author)

Department of Philosophy, University of Sao Paulo (USP), Sao Paulo, Brazil

Corresponding author at: University of Sao Paulo, FFLCH, Department of Philosophy, Av. Prof. Luciano Gualberto, 315 - Cidade Universit?ria, CEP 05508900 S?o Paulo, Brazil. E-mail address: [email protected]

2) Charles Spence Crossmodal Research Laboratory, Department of Experimental Psychology, University of Oxford, Oxford, UK

Abstract

The present study was designed to investigate the effect of the colour of the cup on sensory and hedonic judgments of specialty coffee by consumers. Altogether, 457 participants took part in one of three experiments. Crossmodal correspondences between the colour of the cup (i.e., an extrinsic cue) and the taste profile of the coffee served (i.e., the contents) were manipulated. Congruent and incongruent colour × taste pairings were created by using four cup colours (white, pink, yellow, and green) and two coffee profiles (sweet Brazilian and acidic Kenyan) to assess whether these manipulations would affect pre-and/or post-tasting ratings. Participants first rated their expectations of sweetness and acidity, and subsequently, their experience of those attributes on tasting the coffees, as well as rating their liking. The results revealed that the colour of the cup exerted a significant influence on both pre- and post-tasting ratings for all attributes measured. Liking ratings significantly decreased in

incongruent pairing conditions – which also increased the unexpected acidity of the Kenyan coffee when tasted from the pink cup. Taken together, these results demonstrate for the first time that the colour of the cup significantly impacts sensory and hedonic judgements of specialty coffee. Our results also show that the contrast between expected and actual experience can result in a negative hedonic response and the enhancement of the unexpected sensory attribute. Implications for the development of coffee cups that can enhance the drinking experience are highlighted. KEYWORDS: SPECIALTY COFFEE; CUP COLOUR; TASTE; EXPECTATIONS; CROSSMODAL.

1. Introduction According to the International Standards Organization, flavour can be defined as a “complex combination of the olfactory, gustatory and trigeminal sensations perceived during tasting” which “may be influenced by tactile, thermal, painful and/or kinaesthetic effects” (ISO 5492, 2008). However, in addition to the so-called consummatory or interoceptive cues which involve the sensory systems mentioned by the ISO definition (i.e., those stimulated once a food or drink product has entered the mouth), anticipatory or exteroceptive cues have also been shown to exert a profound influence over multisensory flavour experiences too (Small Veldhuizen, Felsted, Mak, & McGlone, 2008; Spence, 2015; Stevenson, 2014). Specifically, anticipatory cues are those that help in setting expectations before the food or drink enters the mouth, and include orthonasal smell, vision, audition, and somatosensation. According to the literature, visual information, especially colour cues – provide an especially important source of cues concerning taste/flavour expectations (see Piqueras-Fiszman & Spence, 2015; Spence & Piqueras-Fiszman, 2016, for reviews). Studies conducted over the last three decades in a number of different countries have highlighted a remarkable degree of consistency when people are asked to match colours and basic tastes (see Spence, Wan, Woods, Velasco, Deng, Youssef, & Deroy, 2015, for a review). The pairings between basic tastes and colours have been characterized as an example of a crossmodal correspondence in which apparently unrelated sensory features, or dimensions, are perceived or described as matching (i.e., as a feeling that the stimuli belonging to different dimensions/sensory modalities go together; see Spence, 2011, for a review). Amongst the strong colour-taste correspondences, sour has been associated with yellow and green whereas sweet has been associated with pink and/or red instead. Interestingly, in addition to colour-taste pairings, the saturation of the colour also appears to be related to tastant concentration (Saluja & Stevenson, 2018). These crossmodal correspondences likely arise because of learnt associations between colour and taste. Certain colour-taste associations are probably fairly universal such as, for instance, those related to fruit ripeness in humans’ natural environment (Maga, 1974; Foroni Pergola, & Rumiati, 2016). In this case, prior exposure to differences between red and green nuances would bias natural colour–taste correlation, being the red end of the spectrum associated with sweetness and the green end of the spectrum associated with sourness. In addition to these natural associations, other colour-taste correspondences may be more dependent on different

colours co-occurring with different types of food in different parts of the world. Thus, these particular colour–taste associations are likely to be – at least to a certain extent – culture specific (Shankar, Levitan, & Spence, 2010; Wan, Velasco, Michel, Mu, Woods, & Spence, 2014). Moreover, the saliency of colour cues in taste/flavour priming goes beyond product intrinsic sensory attributes (i.e., those that physically belong to the product itself) and includes extrinsic sources of information – which are not physically a part of the product but are somehow related to it. Product extrinsic colour cues refer to all product-related colour experienced by the consumer – i.e., where the product happens to be sold, served, or consumed, such as tableware, labelling, and packaging (Ares & Deliza, 2010; Harrar & Spence, 2013; Piqueras-Fiszman & Spence, 2012; Spence, 2017). In some situations, even the colour of the environment itself (i.e., contextual cues) has been shown to affect flavour ratings (Oberfeld, Hecht, Allendorf, & Wickelmaier, 2009; Spence, Velasco, & Knoeferle, 2014; Velasco, Jones, King, & Spence, 2013; though see Jiang, Niimi, Ristic, & Bastian, 2017, for an exception). A growing body of empirical research now demonstrates that the physical properties of the serving vessel are strongly tied to the experience of flavour of the beverages being consumed from them (see Spence, 2018; Spence & Wan, 2015, for reviews). The colour of the receptacle has been shown to affect judgments of warmth for coffee (Guéguen & Jacob, 2012), the perceived level of carbonation of water samples (Risso, Maggioni, Olivero, & Gallace, 2015), as well as the perceived intensity of certain flavours and well as basic tastes attributes in several beverages. For instance, early research reported that consumers rated the lemon/lime flavour in 7-Up as more intense when the beverage was tasted from a can that was 15% more yellow in colour than normal (Cheskin, 1957). Presumably the increase in the amount of yellow may have exogenously drawn the consumers’ attention to the lemony flavour already present in the drink. Cheskin dubbed this phenomenon “sensation transference”, when people automatically translate their perception of, or feelings about, the sensory properties of the receptacle into the experience of the contents (see Skaczkowski, Durkin, Kashima, & Wakefield, 2016, for a recent review). Meanwhile, Dichter (1964) carried out a pioneering study into the influence of the colour of the packaging on taste/flavour judgments. Participants were given four cups of coffee to compare and evaluate (see Favre & November, 1979, p. 64, for a summary of the results).

Each cup of coffee was served from a pot having a different colour (brown, red, blue, or yellow). 73% of those tested reported that the coffee served from the brown container was ‘too strong’; Meanwhile, 84% of the female participants suggested the coffee served from the red pot was rich and full-bodied; The aroma of the coffee from the blue jar was rated as having a milder aroma; And the coffee served from the yellow container seemed to have come from a weaker blend. In fact, the coffee in all of the pots was the same. Once again, therefore, these results add to the growing literature illustrating the widespread impact of colour on taste/flavour perception. More recently, the participants in another study were asked to evaluate the receptacle itself (affective and sensory pre-tasting ratings) or the experience of consuming either hot Earl Grey tea or a chilled lemon soft drink from these cups (post-tasting ratings; Schifferstein, 2009). All cups made from different materials and were all pinkish except for one which was transparent glass. One of the assessed attributes was sweetness, and all pink cups, when empty, were rated as significantly sweeter than the transparent cup. Interestingly, however, no differences in sweetness ratings were observed for the actual drinking experiences. It is possible that the participants may have transferred the sensation of the pinkness of the cups (i.e., their beliefs, and expectations, based on previous experience that pink foodstuffs tend to be sweet - Spence et al., 2015) to their impressions of the ‘sweet content’ of the cups themselves. However, the anticipation of sweetness primed by the pink colour was not an indicator of the actual perception of sweet aroma and taste of either of the beverages. One reason for this lack of correlation between expectation and experience could be that the chosen beverages (i.e., Earl Grey tea and lemon soft drink) are not necessarily associated with sweetness in the first place. This study therefore raises a flag concerning the importance of matching the colour of the container (or rather, the taste expectation elicited by that colour) with taste attributes that are actually present in, or have been associated with, the beverage being served in order to search for crossmodal effects. Other recent experiments have assessed the effect of cup colour on flavour attributes and hedonic judgments in hot beverages that include hot chocolate (Piqueras-Fiszman & Spence, 2012) and café latte (Van Doorn, Wuillemin, & Spence, 2014). For instance, in the former study, a hot chocolate drink was served to participants in orange, white, red, or creamcoloured plastic cups. The hot chocolate was rated as having a more intense flavour when served in the orange cup than when served in either white or red cups (with no significant difference from the cream-coloured cup). It was also liked more when tasted from the orange

cup than from the white cup (with no significant difference from the other two coloured cups). The choice of these four colours, according to the authors, was based on the observation that: “those are among the most common colours found among cups available for dispensed (or vended) hot beverages” (Piqueras-Fiszman & Spence, 2012, p. 326). Meanwhile, the results of the latter study revealed that the café latte was rated as more intense in flavour – but also as less sweet – when served in a white mug (or in a transparent mug with a white sleeve) as compared to a blue mug (or in a transparent mug with a blue sleeve) or to a transparent mug. Once again, the colour selection was motivated by the mug types that “are amongst the most commonly used vessels to serve coffee in Australian cafés and restaurants” instead of prior knowledge concerning colour-taste crossmodal effects (Van Doorn et al., 2014, p. 1). Coffee is one of the most commonly consumed beverages worldwide. Traditionally, coffee has been traded as a commodity and consumed as industrial blends (Ponte, 2002). However, coffee trading and consumption has changed profoundly over the last 30 years with new trading schemes appearing as customers become more interested in consuming better quality coffee. Thus, new patterns of coffee drinking have emerged with the growing importance of specialty coffees (Carvalho, Paiva, & Vieira, 2016; Ponte, 2002). Specialty coffee is a term used to refer to those coffees distinguished on the basis of quality and uniqueness of origin, according to the Specialty Coffee Association (SCA) and the international Q Coffee System protocols (Lingle & Menon, 2017). In spite of its formal definition, the concept of specialty coffee has exceeded the narrow meaning based solely on quality. It also refers to the out-ofhome consumption market niche where the content of the consumption experience includes, but is not limited to, high quality coffees. The specialty coffee consumer is likely interested in different coffee flavour profiles, brewing methods, packaging, and ambience (de Luca & Pegan, 2014; Ponte, 2002). Therefore, it is important to understand the relevant factors that might enhance the multisensory experience in the coffee shop. One factor that must be considered when trying to create a truly engaging multisensory atmosphere for specialty coffee consumption is the vessel in which the coffee is served (Bury, 2014; Carvalho & Spence, 2018; de Luca & Pegan, 2014). Many studies have already been published on colour-taste/flavour interactions between the drinking container and its contents. Nevertheless, none have focused on assessing whether the appropriateness of pairing up the colour of the vessel with its corresponding basic taste (e.g., pink or red for sweetness, yellow or green for acidity, black for bitterness – see Spence et al.,

2015) affects pre- and/or post-tasting ratings, as well as hedonic judgments during the drinking experience. Moreover, very little research has been published to date specifically looking at crossmodal influences of the cup on aroma and taste attributes in the case of filter coffee (Carvalho & Spence, 2018; Van Doorn, Woods, Levitan, Wan, Velasco, BernalTorres, & Spence, 2017; see also Guéguen & Jacob, 2012, for crossmodal correspondence between cup colour and the warmth of coffee), whereas the association between the colour of the receptacle and flavour perception in specialty coffee remains essentially unstudied. The study reported here is therefore distinct from previous research in its theoretical approach. We seek not only to examine a possible interaction between the colour of the vessel and the perception of taste/flavour attributes as has been done in the past. In addition, we aim at critically manipulating (i) the colour cue that has been demonstrated to set expectations about taste identity and (ii) the taste profile of the coffee served (i.e., the content). For this, four colours of ceramic cups were selected based on previously documented colour-taste crossmodal correspondences (pink for sweetness; green and yellow for acidity; white as a reference, being the traditional colour for coffee cups in Brazil), and they were paired up with two very distinct coffee flavour profiles. One coffee was high in sweetness (typical of Brazilian specialty coffees) whereas the other was high in acidity (common in Kenyan specialty coffees)1. Thus, the participants were exposed to both congruent and incongruent colour-taste pairings in order to investigate whether the colour of the cup would influence participants’ taste expectations, actual taste perception as well as their hedonic judgements of the coffees served in both conditions (i.e., congruent and incongruent). By following this experimental design based on colour × taste correspondences, hypotheses concerning both sensory and hedonic aspects could be addressed. Drawing upon the aforementioned sensation transference account, and also on observations that sensory expectations alter the actual perception of food products (Piqueras-Fiszman & Spence, 2015), the hypothesis was put forward that cup colour will increase both pre- and post-tasting ratings of its corresponding basic taste in the congruent conditions. For the incongruent conditions,

1

The chemical composition of the raw coffee bean is highly dependent on the post-harvest processing method used, and this composition will determine distinct flavour characteristics. In Brazil, the dry method is often used (i.e., the whole cherry is dried as a raising – so-called ‘natural’ coffees) whereas in Kenya the wet method is more common (i.e., the skin and mucilage are removed from the parchment – so-called ‘fully washed’ coffees). Natural and fully washed coffee beans originate beverages with very distinct flavour profiles but with similar final score. Usually, natural coffee beans have denser body and sweeter taste compared to the fully washed coffees, which tend to have a much cleaner and more acidic taste (Da Rosa et al., 2015; Hameed et al., 2018).

based on both processing fluency and assimilation-contrast theories that accounts for the effect of cognitive effort in processing mismatching multisensory cues (Reber, Winkielman, & Schwartz, 1998; Reber, & Schwarz, 2001) and of disconfirmation of expectations by the actual consumption experience (Cardello & Sawyer, 1992), one of two outcomes might be expected. If the disparity between the expectations and actual perception is within an acceptable range (though note that it is currently unclear how wide the acceptable range here is), the consumer is capable of assimilating the difference, meaning that the stimuli are being processed more fluently or easily. By contrast, in those situations in which the discrepancy (or prediction error – Friston, 2010; Hirsh, Mar, & Peterson, 2012) is outside of this acceptance range, decreased processing fluency may lead to decreased liking ratings. Indeed, the consumer may even overreact by exaggerating the difference between expectation and reality – observed in an evaluation shift in the direction opposite to the original expectation (Piqueras-Fiszman & Spence, 2015; Schifferstein, 2001). In other words, the promise of a sensory experience followed by the delivery of a mismatching one may lead to high ratings of the unexpected taste/flavour as well as greater customer disappointment (Yeomans, Chambers, Blumenthal, & Blake, 2008)2. Given the rationale outlined above, we focus here on testing the hypothesis that the colour of the cup alters both the expected and actual sensory experience as well as hedonic judgments in congruent and incongruent set-ups.

2. Methods 2.1. Participants A total of 457 participants gave their informed consent to take part in one of the three experiments reported in the present study. The study was approved by the Research Ethics Committee of the School of Psychology, University of São Paulo, Brazil. After data quality 2

The brain constantly makes predictions about the likely taste/flavour of that which we are about to consume (i.e., eat or drink). Generally-speaking, we tend to like foods/drinks more if they meet our expectations than if they do not (Woods et al., 2011; Yeomans et al., 2008). However, in addition to the expectation about the food/drink itself, the context in which they are served also plays a not insignificant role in setting the consumer’s expectations towards the eating/drinking experience as a whole. For instance, there is a high demand for novelty in molecular cuisine. In this context, the diners expect the unexpected and so they keep their minds open to surprise (i.e., what might think of as prediction error) (Spence & Piqueras-Fiszman, 2014). On the other hand, surprise is not what a client necessarily wants in a cup of coffee served in a coffee shop. Rather, drinking coffee can be seen as an ordinary experience, i.e., a common, frequent experience within the realm of everyday life (unless, that is, it is part of a molecular gastronomy/modernist meal; though see Spence & Youssef, 2018).

control, the data from 82 participants were included in the final analysis of Experiment 1 (31 female; age: 36.4 ± 9.2, 22 to 61 years-old), 92 in the final analysis of Experiment 2 (40 female; age: 36.8 ± 11.4, 20 to 60 years-old), and 272 in the final analysis of Experiment 3 (142 female; age: 35.0 ± 7.9, 19 to 61 years-old)3. None of the participants in any of the three experiments were informed about the origin or flavour profile of the coffees that they were served.

2.2. Stimulus 2.2.1. Coffee Only single origin Arabica coffees were used in the experiments reported here. All of the coffees were assessed by SCA-certified sensory analysts (i.e., Q grader cuppers). In Experiments 1 and 2, the participants were served only one type of coffee. The coffee used in Experiment 1 came from the Imperio Estate, located in Buritizeiro, Minas Gerais, Brazil (Designation of Origin: Cerrado Mineiro; altitude of 1000m). The cultivar was Acaia Cerrado, and the post-harvest processing method used was dry/natural (Schwan, Silva, & Batista, 2012). The coffee sample received an overall score of 86.5 points on a 0-100 scale (Lingle & Menon, 2017), with high sweetness and low-to-medium acidity4. Regarding overall aroma/flavour characteristics, the coffee was described as having strong notes of red berries and chocolate. The coffee used in Experiment 2 came from Porta do Ceu Estate, located in Santana de Caldas, Minas Gerais, Brazil (Sul de Minas region; altitude of 1250m). The cultivar was Red Catuai and was also processed as natural. The coffee sample was given an overall score of 86 points, with high sweetness and low-to-medium acidity. The coffee aroma/flavour attributes were described as having predominant notes of citrus fruit and molasses. In Experiments 1 and 2, a single shot of espresso coffee (30mL) was served to participants. The coffee was prepared by the baristas working in the two coffee shops in which the 3

Considering medium effect size (d) of 0.75, an intended power ≥ 0.8, and alpha value ≤ 0.05, the minimum sample size is 73 in two-way factorial designs (Gacula & Rutenbeck, 2006). 4

The specialty coffee grading system positively scores the presence of sweetness and ‘vibrant acidity’ – as well as the balance between these two basic tastes, and negatively scores dominant bitterness (Traore et al., 2018). Thus, specialty coffees essentially lack dominant bitterness, but will express varying levels of acidity and sweetness depending on several agricultural and roasting factors (Hameed et al., 2018).

experiments took place, according to their own extraction recipe. The serving temperature was not assessed in either study. In Experiment 3, two different coffees were served to all participants. The first coffee was from Ambiental Fortaleza Estate, located in Mococa, Sao Paulo, Brazil (Mogiana region; altitude of 1300m). The cultivar was Yellow Catuai processed as natural coffee, with an overall score of 88 points, with high sweetness and low acidity. The predominant coffee aroma/flavour notes were butter toffee and prunes. The second coffee came from the Nyeri region, Kenya (altitude of 1600-1700m), and consisted of a blend of four cultivars (SL28, SL34, Ruiru, and Batian). The post-harvest processing method used was the washed process, and the coffee received an overall score of 88 points, with low-tomedium sweetness and high acidity, and with strong floral notes as well as tangerine and blackcurrant. The coffee beverage served to the participants in Experiment 3 was a filter (pour-over) coffee prepared using the Hario V60 Kit (Hario V60; Tokyo, Japan) at a concentration of 77 g·L−1 using mineral water at 92 °C. The mean temperature of the coffee served to the participants was 59.8°C (SD=1.9). 2.2.2. Cups In Experiments 1 and 2, the participants were divided into two testing groups according to the colour of the cup from which the espresso coffee was sampled. All of the cups were white on the inside in order to keep the visual contrast between the coffee and the inside wall constant (Hurlbert, 1996). On the outside, the cups could either be white or light pink, and were handleless (see Fig. 1A). Colour was the only factor varied as all of the cups had the same shape, were made of the same material (ceramic), and had virtually the same weight [Mean(g)±SD for white (105.5±3.1) and pink (103.7±2.9)]. In Experiment 3, all of the participants were exposed to the four coloured cups. All of the cups were white on the inside. On the outside, the cups could be white, light pink, light green, or light yellow, all handled (see Fig. 1B). Once again, colour was the only factor that varied between the cups – shape, material (ceramic), and weight [Mean(g)±SD for white (131.5±2.5), pink (132.7±2.3), green (131.4±2.9), and yellow (133.2±2.1)] were kept constant. It is also worth noting that by keeping the inside of the cup a constant colour, by changing the outer surface of the cup one ends up creating a foreground-background colour combination that may convey a specific taste even more effectively than either of the component colours. So, for example, Woods et al (2016) have demonstrated that people categorize pink on white as sweet more consistently than they do the best of the component colours, white.

2.3. Design and procedure Experiments 1 and 2 were conducted in two different coffee shops. The first shop was located in Sao Paulo – SP (Takko Cafe) and the second in Pocos de Caldas – MG (Ancora Coffee House), Brazil. The testing was carried out over four days during working hours (11am-5pm) and all of the participants were customers/clients of the shops. The potential participant (customer) was invited to take part in the study at the moment he/she ordered the single shot of espresso coffee at the counter, where he/she was informed the study was about tasting and evaluating the espresso coffee that had just been ordered. Upon acceptance, the participant was preferentially led to the same two-seater table where the participant would take one seat and the experimenter took the other. A short questionnaire was used to assess each participant’s familiarity and consumption frequency. Only those amateurs who had been consuming specialty coffee for at least a year were included in the study. None of the participants reported having a cold or any other impairment of their sense of smell or taste at the time of the study. The testing session started by placing a sheet containing the rating scales, a pen, and a glass of water in front of the participant, who then received a two-minute briefing while the coffee was being prepared. The participant received one cup – either white or pink – containing the single shot of espresso (30 mL) that had been ordered from the hands of the barista. At this point, the experimenter left the participant alone in order to evaluate the requested attributes. The participants were instructed to first rate, only by looking at the cup of coffee, its expected sweetness and acidity using the rating scales (with a counterbalanced order across participants). Next, they moved on to sample the coffee and repeat the same ratings after having tasted it, as well as judging how much they liked it. Ratings were performed using a 10-cm visual analog scale anchored at 0 (‘not at all’) and 10 (‘very’). The participants were assigned to each testing condition (i.e., white or pink cup) in order to get a final sampling design that was as balanced as possible. Each testing session lasted for around 6-7 min. Experiment 3 was conducted in Sofa Cafe, a coffee shop and school in Sao Paulo. Data was collected in a quiet, well-lit air-conditioned testing room. The majority of the participants were recruited primarily through social media websites. The online advert made clear the experimental procedure as well as the inclusion criteria (only amateur consumers who had been drinking specialty coffee for at least a year with no sugar added; not having a cold or

any other impairment of their sense of smell or taste at the time of the study could take part). Before the start of the study, all of the participants were informed that they would taste and evaluate four samples of specialty coffee. They filled in a short questionnaire on their familiarity specialty coffee and consumption frequency. The participants were led, in groups of twelve, to the testing room and were then seated around six two-seater tables, with at least one-metre spacing between adjacent tasters. A sheet containing the rating scales, a pen, and a glass of water were placed in front of each participant’s place prior to their arrival at the testing room. At the start of each session, the group of participants received a three-minute briefing in order to ensure that all groups were given the same instructions. To avoid a possible coffee–colour bias in the responses, the tasting procedure followed a within-participant experimental design, and the evaluation of the samples followed a sequential monadic presentation scheme with the order of presentation balanced amongst participants (see Fig. 1). The participants were informed that they were going to evaluate a total of four coffee samples of 50 mL each, one at a time. They were instructed first to look at (but not taste or smell) each one of the coffee cups and evaluate their expected sweetness and acidity using the “expectation” scales. They were then instructed to taste the coffee samples and rate the sweetness, acidity, and how much they liked them using the “perception” scales. The participants were also instructed to rinse their mouths out with water between samples in order to cleanse their palates. The order in which the sweetness and acidity scales were presented was counterbalanced across participants. One supervisor was present during the testing sessions in order to provide guidance if necessary. Upon finishing the study, the participants were thanked for their participation and instructed to leave the room. They also received a small sample of specialty coffee for taking part in the study. Each testing session lasted for around 15-20 min.

2.4. Data analyses Data from Experiments 1 and 2, which followed exactly the same design and procedure, were separately analysed using the same methods. For both expected and post-tasting espresso ratings, a multivariate analysis of variance (MANOVA) was conducted on the dependent variables ‘sweetness’, ‘acidity’, and ‘liking’, with the colour of the cup as a between-

participants factor. A repeated-measures MANOVA was used to compare the expected espresso ratings to the post-tasting ratings, with rating type (before or after tasting) as the within-participants factor, the colour of the cup as the between-participants factor, and sweetness, acidity, and liking as the dependent variables. The data from Experiment 3 were first analysed using repeated-measures MANOVAs on the dependent variables ‘sweetness’ and ‘acidity’ for expected coffee ratings (pre-tasting). A two-way repeated measures ANOVA was performed on the post tasting data in order to assess the main effects of cup colour and coffee type, as well as their interaction effect, on the participants’ perception of sweetness and acidity, and on their ‘liking’ ratings. Finally, a repeated-measures MANOVA was used to compare the expected coffee ratings to the posttasting ratings on the dependent variables ‘sweetness’, ‘acidity’, and ‘liking’, where rating type (pre- or post-tasting) was included as an additional within-participants factor. All posthoc pairwise comparisons were Bonferroni corrected, and differences were considered significant at p ≤ 0.05. In order to further explore the relationship between pre- and post-tasting scores, and hedonic judgments, a Pearson’s correlation analysis was carried out. First, a difference index was generated by subtracting the post-tasting from the pre-tasting rating values of sweetness and acidity for each participant (i.e., ‘expectation’ minus ‘perception’ ratings), for both congruent (i.e., pink cup / Brazilian coffee; green cup / Kenyan coffee) and incongruent (i.e., pink cup / Kenyan coffee; green cup / Brazilian coffee) conditions. Next, the correlations were calculated between the difference indices and the liking scores for each of these four conditions.

3. Results 3.1. Taste expectation (pre-tasting ratings) The MANOVA tests revealed a significant main effect of cup colour on participants’ ratings of the expected taste attributes of the coffee in all three experiments. That is, there was a significant difference in expected sweetness and acidity ratings based on cup colour in

Experiments 1 [F(2,79) = 8.04, p < 0.01, Wilks’ lambda5 = 0.83], 2 [F(2,89) = 4.07, p < 0.05, Wilks’ lambda = 0.92], and 3 [F(3,269) = 268.40, p < 0.0001, Wilks’ lambda = 0.25]. Further pairwise comparisons (Bonferroni corrected) on the data from Experiments 1 and 2 revealed effects between specific groups (i.e., cup colours). In Experiment 1, the participants expected the coffee served in the pink cup to taste sweeter (p < 0.001) and to be less acidic (p < 0.05) than the coffee served from the white cup (see Fig 2a). In Experiment 2, the participants expected the coffee to taste less acidic (p < 0.01) when served from the pink cup than from the white cup, with no effect of cup colour on expected sweetness (see Fig 2b). Post-hoc tests using Bonferroni correction on the data from Experiment 3 revealed that the expected acidity of the coffee differed significantly between the four coloured cups, being green > yellow > white > pink (all p < 0.05). As for sweetness, there were significant differences amongst the four coloured cups being tested, being pink > yellow > white = green (all p < 0.001). In general, the coffee served in the pink cup was expected to taste sweeter and less acidic than any of the other colours whereas the coffee from the green cup was expected to be more acidic than any of the other colours (see Fig 3a).

3.2. Taste perception (post-tasting ratings) 3.2.1. Experiments 1 and 2 The MANOVA tests revealed a significant main effect of cup colour on participants’ ratings of coffee sweetness, acidity, and liking in Experiments 1 [F(3,78) = 29.22, p < 0.001, Wilks’ lambda = 0.47] and 2 [F(3,88) = 19.77, p < 0.0001, Wilks’ lambda = 0.60]. Further pairwise comparisons (Bonferroni corrected) revealed that the coffee, in both experiments, was rated as tasting sweeter (p < 0.001) and less acidic (p < 0.001) when sampled from the pink cup than from the white cup. The coffee was also liked more when tasted from the pink cup as compared to the white cup in both Experiments 1 (p < 0.001) and 2 (p < 0.01) (see Fig. 2a and 2b). 5

Wilks' lambda performs, in the multivariate setting, with a combination of dependent variables, the same role as the F-test performs in one-way analysis of variance. It is a direct measure of the proportion of variance in the combination of dependent variables that is unaccounted for by the independent variable (the grouping variable or factor). If a large proportion of the variance is accounted for by the independent variable then it suggests that there is an effect from the grouping variable and that the have different mean values (Wilks' lambda equal to zero means that there is not any variance not explained by the independent variable – which is ideal) (Todorov & Filzmoser, 2010).

3.2.2. Experiment 3 A two-way repeated measures ANOVA on the data from Experiment 3 revealed significant main effects of the type of coffee (Brazilian or Kenyan) and cup colour (white, yellow, pink, and green), as well as an interaction between these factors, on participants’ ratings of coffee acidity, sweetness, and liking. As for the main effects, coffee type significantly affected participants’ ratings of acidity [F(1,135) = 1486.16, p < 0.0001, Wilks’ lambda = 0.08], sweetness [F(1,135) = 513.29, p < 0.0001, Wilks’ lambda = 0.21], and liking [F(1,135) = 9.71, p < 0.01, Wilks’ lambda = 0.93], with the Brazilian coffee being perceived as tasting sweeter, less acidic, and also being liked more overall than the Kenyan coffee. The colour of the cup also exerted a significant impact on ratings of acidity [F(3,133) = 4.14, p < 0.01, Wilks’ lambda = 0.92], sweetness [F(3,133) = 9.37, p < 0.001, Wilks’ lambda = 0.82], and liking [F(3,133) = 17.81, p < 0.001, Wilks’ lambda = 0.71]. Regarding the acidity of the coffee, the only observed significant difference was between the pink (lower ratings) and green (higher ratings) cups (p < 0.05). As per sweetness ratings, there was no significant difference between pink and yellow, and white and green, but a significant difference between these two pairs was observed (i.e., pink = yellow > white = green) (all p < 0.01). Liking ratings were significantly higher for the yellow cup than for all other three colours (whose ratings did not differ significantly; yellow > white = pink = green; all p < 0.001). A significant interaction effect between coffee type and cup colour was also observed on acidity [F(3,133)=12.91, p < 0.001, Wilks’ lambda=0.77], sweetness [F(3,133)=34.59, p < 0.001, Wilks’ lambda=0.88], and liking [F(3,133)=41.57, p < 0.001, Wilks’ lambda=0.52] ratings. The results of the post hoc tests (Bonferroni corrected) are summarized in Table 1 (see also Fig. 3b). Table 1: Post-hoc comparisons for interaction effects between coffee type and cup colour on all dependent variables assessed in Experiment 3.

Acidity* Sweetness*

Brazilian coffee (high sweetness; low acidity) Yellow, Pink, White > Green White > Pink Pink > Yellow > White > Green

Pink = Yellow > White > Green Liking** * p < 0.05; ** p < 0.001

Kenyan coffee (low-to-medium sweetness; high acidity) Pink, Yellow, Green > White Pink > Green Yellow > White, Pink Green > Pink Green = Yellow > White > Pink

3.3. Taste expectations versus taste perception 3.3.1. Experiments 1 and 2 The repeated measures MANOVA did not reveal a significant main effect of condition (rating type: pre-tasting vs. post-tasting) in either of the experiments, but did reveal a significant interaction effect between rating type and cup colour on participants’ ratings of the sweetness and acidity of the coffee in both Experiments 1 [F(2,79) = 16.85, p < 0.001, Wilks’ lambda = 0.83] and 2 [F(2,89) = 8.26, p < 0.01, Wilks’ lambda = 0.92]. Further pairwise comparisons (Bonferroni corrected) showed specific effects for each cup colour. In Experiment 1, the coffee was rated as tasting sweeter (p < 0.01) and less acidic (p < 0.001) than expected when sampled from the pink cup, and also perceived as more acidic (p < 0.05) than expected when tasted from the white cup. In Experiment 2, the coffee tasted from the pink cup was perceived as sweeter (p < 0.01) than expected whereas the coffee tasted from the white cup was perceived as more acidic (p < 0.05) than expected (see Fig. 2a and 2b). 3.3.2. Experiment 3 As for Experiment 3, the repeated measures MANOVA revealed a significant interaction effect between rating type and cup colour on participants’ ratings of coffee sweetness and acidity [F(3,133) = 28.80, p < 0.0001, Wilks’ lambda = 0.84]. No significant main effect of condition was found. Additional pairwise tests (Bonferroni corrected) as a function of cup colour revealed specific effects for each coffee type. The Brazilian coffee was perceived as sweeter than expected when tasted from the yellow, white, and green cups (all p < 0.001) whereas the Kenyan coffee was perceived as less sweet than expected when tasted from the yellow, white, and pink cups (all p < 0.001). As per the acidity ratings, the Brazilian coffee was perceived as less acidic than expected when tasted from the yellow and green cups (all p < 0.001) whereas the Kenyan coffee was perceived as more acidic than expected when tasted from all four cups (all p < 0.001) (see Fig. 4). A Pearson’s correlation analysis was performed to assess the relationship between the generated difference indices (see Section 2.4) and liking ratings. There was a negative correlation between the acidity difference index and liking ratings in the incongruent condition ‘green cup / Brazilian coffee’ (r = 0.36, n = 133, p < 0.001), and between the sweetness difference index and liking ratings in the incongruent condition pink cup / Kenyan

coffee (r = -0.38, n = 134, p < 0.001) (see Fig 5). No correlation was observed between the tested variables in the congruent conditions. The results of all three experiments are summarized in Tables 2 (sensory ratings) and 3 (hedonic ratings).

4. Discussion Given previous reports documenting the existence of crossmodal correspondences between colours and tastes (see Spence et al., 2015), the three experiments carried out in the present study were designed to investigate whether the colour of the outer surface of coffee cups (white, pink, yellow, and green) would influence expectations and experience of taste attributes (sweetness and acidity). Different groups of amateur consumers of specialty coffee evaluated either espresso (Experiments 1 and 2) or filter coffees (Experiment 3) in pre- and post-tasting conditions. In addition, congruent (pink / sweetness) and incongruent (green and yellow / acidity) colour-taste pairings were constructed to test whether the colour of the cup would influence participants’ expectation and perception of sweetness and acidity, as well as their hedonic judgments of coffees. The results of all three experiments are in agreement with previous studies in showing that that the colour of the vessel affects both pre- and/or posttasting evaluations of several types of beverages (Cheskin, 1957; Schifferstein, 2009; Piqueras-Fiszman & Spence, 2012), including coffee-based beverages (Van Doorn et al., 2014). In Experiments 1 and 2, the participants expected the espresso coffee to taste sweeter and/or less acidic from the pink cup than from the white (reference) cup – expectations which were then confirmed in post-tasting ratings. The espresso was also liked more when tasted from the pink cup. These preliminary results point towards significant modulatory effect of the colour pink on both expectation and perception of sweetness – which was increased – and acidity – which was decreased – in specialty coffee. They are in part in agreement with Schifferstein (2009) who demonstrated that pink cups (against a transparent one) increased participants’ expected sweetness, without observed differences in sweetness ratings during the actual drinking. One reason as to why expectations did not influence post-tasting sweetness in Schifferstein’s study could be that the chosen beverages (i.e., Earl Grey tea and lemon soft drink) are not necessarily associated with sweetness in the first place. Thus, in order to search for crossmodal effects, it is important that the taste primed by the colour of the container is actually present in, or has been associated with, the contents.

The findings reported in Experiment 3 widen our understanding of how the colour of the cup affects pre- and post-tasting ratings of both sweetness and acidity in specialty coffee. The sweet Brazilian coffee was sampled from pink (congruent condition), green or yellow (incongruent conditions) cups. In the same way, an acidic Kenyan coffee was served in pink (incongruent condition), green or yellow (congruent conditions) cups. A white cup was also included for the sake of reference. As anticipated, pre-tasting ratings revealed that the coffee was expected to have the sweetest and least acidic taste when served in the pink cup, and the most acidic and least sweet when served in the green cup. In fact, pink and green are two colours strongly associated with sweet and sour basic tastes, respectively (Spence et al., 2015). One suggested explanation for the effect of colour cues on taste/flavour perception (i.e., crossmodal correspondence) is that many ripening fruits show a transition from colours at the green end of the spectrum to colours at the red end of the spectrum (including pink) (e.g., Maga, 1974; Foroni et al., 2016). Thus, according to this argument, prior exposure to this natural colour–taste correlation (i.e., between redness and sweetness levels in ripening fruits) might help to explain why red-green colouring might impact the perception of sweetness-sourness in a beverage such as coffee. Interestingly, the colour yellow, which has also been reported to be strongly associated with sour tastes/flavours (such as lemony – Zampini Sanabria, Phillips, & Spence, 2007), triggered expectations towards both acidity and sweetness in both coffee samples. One way in which to think about this finding is in terms of the role played by colours in the perception of flavour identity (and its correlation to basic tastes) – flavour identity which sometimes varies as a function of the cultural background of the participant assessed (Shankar et al., 2010; Wan et al., 2014). In Brazil (and many other parts of South America – e.g., including Colombia), green limes are much more abundant, common, and cheaper than yellow lemons (which are there called Sicilian lemons). In addition to that, Brazil is a tropical country with a huge availability of yellow fruits with high levels of sweetness, think here only of mangoes, bananas, and pineapples (but see Velasco, Michel, Youssef, Gamez, Cheok, & Spence, 2016, for the strong red-sweet association also in Brazilians). For instance, a recent study by Torres, Salazar & Salgado (2016) had Colombian participants rate the sweetness and sourness of jellies of three different fruit flavours (mango, lulo, and soursop) presented in three different colours (red, yellow, and green). Intriguingly, sweetness was rated higher when the three fruit jellies were paired up with yellow whereas the colour red only decreased sourness ratings in the lulo jelly. The colour green increased the perception of sourness in both lulo and soursop

jellies. The results of this study can therefore be taken to suggest that participants in tropical countries may have a tendency to associate the colour yellow with sweet taste given contextual and cultural influences. Post-tasting ratings of sweetness and acidity also varied as a function of the colour of the cup. In fact, several studies have demonstrated that people tend to match perceived tastes crossmodally to visual features of serving vessels and packaging (see Spence, 2016, 2018, for recent reviews). The sweet expectations primed by the pink cup actually did carry-over to influence people’s perception when a sweet (Brazilian) coffee was tasted – but not when an acidic (Kenyan) coffee was tasted instead. The Brazilian coffee was perceived as sweeter as well as less acidic, and was also liked more when tasted from the pink cup against the white cup. However, when the acidic Kenyan coffee was tasted from the pink cup (incongruent condition), it was perceived as having the highest acidity and lowest sweetness levels as compared to any other cup colour, and it was also liked the least. Regarding the green cup, the increased expectation of acidity did increase the actual perception of acidity for the Kenyan coffee (when compared to the white cup). The Kenyan coffee was also liked more when sampled from the green cup. However, when the green cup was paired up with the Brazilian coffee (incongruent condition), it was perceived as significantly lower in both sweetness and acidity, and was also liked the least. Presumably due to the occurrence of a strong disconfirmation of expectation the coffees were liked less and, somehow, the most noticeable (but not anticipated) taste was significantly enhanced. Several studies have reported the role of extrinsic cues as sources of product-related expectation, and consequent taste/flavour perception and hedonic judgments (see PiquerasFiszman & Spence, 2015, for a review). Colour-induced expectation effects might influence taste/flavour perception – or more specifically, its identification – by directing the person’s attention towards a particular component of a flavour stimulus (Ashkenazi & Marks, 2004). In so doing, that component may become relatively more salient in the taster’s experience against the background which contains a variety of other competing in-mouth sensations (Stevenson, 2012). In addition to any such saliency effect, the attended flavour component may also be perceived slightly earlier in time than unattended or less well attended stimuli (see Spence & Parise, 2010, for a review). This attentional capture (and possible prior entry triggered) by the colour of the food/drink product extends to the container in which it happens to be presented (e.g., consuming vessel, packaging) as people automatically transfer the sensory properties of the container into the experience of its contents – a phenomenon

called “sensation transference” (Spence & Wan, 2015; Risso et al., 2015; Skaczkowski et al., 2016). Here, it is perhaps worth noting that the consumption behaviour of fruits and food/drink products are quite similar in the sense that they are identified by the colour of their outer shell (i.e., fruit skin or product container) – shell which is normally discarded. A comparison between pre-and post-tasting ratings revealed that some expectations were fulfilled whilst others were disconfirmed. The confirmation of the colour-induced expectations led to higher hedonic judgments in both the Brazilian and Kenyan coffees. In fact, it has been shown that people usually tend to like food and drink more if they meet their expectations than if they do not (Lee, Frederick, & Ariely, 2006), since congruent multisensory cues are likely to be processed more easily (Labroo, Dhar, & Schwarz, 2007), particularly if they are in a context in which coherence is expected (see Footnote 1). In fact, in the incongruent conditions, as the size of the discrepancy between expectations and actual stimuli increased, the linking ratings decreased (see Fig. 5 for the correlation between these two variables). It is interesting to note that the Kenyan coffee was rated as significantly more acidic when tasted from the pink cup than from a cup of any other colour– including green. That is, the pink colour led to an enhancement of the unexpected sensory attribute (i.e. acidity) already present in the coffee. This result suggests that the coffee already high in acidity might have been perceived as unpleasantly too acidic, which was reflected in the lowest liking ratings for the Kenyan coffee. Indeed, this perceived discrepancy could have induced a strong contrast effect which caused the participants to further magnify the already present difference between expectation and real stimulus (Cardello & Sawyer, 1992; Yeomans et al., 2008). It is important to mention that recording expectations prior to tasting make them explicit to the subject (instead of keeping them implicit). Despite several studies have employed this pre- × post-tasting experimental design (Schifferstein, 2009; Van Doorn, Colonna‐Dashwood, Hudd‐Baillie, & Spence, 2015; Wang, Carvalho, Persoone, & Spence, 2017), we are not aware of any study which has specifically assessed the effect of the gap between expectation and experience on flavour and hedonic ratings. Another interesting result to emerge from the present study is that the yellow cup – when compared to the white cup – increased the sweetness as well as the acidity ratings of both Brazilian and Kenyan coffees. In addition, both coffees tasted from the yellow cup were liked as much as when these two coffees were tasted from their congruent coloured cups, i.e., pink and green, respectively. This suggests that the yellow colour was congruently paired up with both sweet and acidic tastes in this case, as also observed in the pre-tasting effects.

Taken together, these results suggest that the colour of the cup strongly impacts consumers’ expectation and experience of tastes in specialty coffee. Nonetheless, it is important to note that the participants were not informed about the country of origin or flavour profile of the coffees they were asked to evaluate from the different coloured cups. In a real coffee shop, the client usually chooses the coffee from a menu list, or is informed by the barista which beans are currently available. A regular specialty coffee consumer would know what to expect regarding the flavour profile when ordering a Kenyan or a Brazilian coffee. In this real-world scenario, the information about the country of origin (i.e., semantic cue) would also likely impact flavour/taste sensory as well as hedonic judgements (Stefani, Romano, & Cavicchi, 2006). However, the effect of labelling information regarding country of origin has not been assessed in the present study. In addition, cups with different shapes were used in Experiments 1 & 2 (handleless, taller, and less round) and Experiment 3 (handled, shorter, and rounder). The potential impact of this shape distinction on both expected and perceived tastes intensities cannot be ruled out (Carvalho & Spence, 2018; van Doorn et al., 2017). However, since context, coffee, and experimental design were almost all different between the three experiments, is it difficult to infer whether and how these differences in shape might have affected the taste ratings. Having said that, various studies have clearly demonstrated the profound impact that the colour of the drinking vessel has on people’s perception of a wide variety of beverages, from wine to hot chocolate through soda and beer, even when the beverage is already familiar to the participant (see Spence & Wan, 2015, for a review). One curious example here comes from the fact that when Coca-Cola released a limited edition white-coloured Christmas can a few years ago to learn that many consumers complained all too vociferously about the change in the taste of their well-known red-canned black Coke drink (Esterl, 2011).6

5. Conclusions Colour plays a central part in consumers’ response towards food/drink, even if the colour is part of the container from which the food/drink is consumed. The results of the present study suggest this is also true for specialty coffee since the colour of the cup affected expected and perceived tastes as well as hedonic judgements in amateur consumers. The colour-induced 6

In fact, in response to this consumer backlash, the cans were soon withdrawn from the shelves, only to be offered to airline passengers who couldn’t so easily complain (see Spence, 2017, on this story).

effects observed in pre- and post-tasting sweetness and acidity ratings corroborate several previous findings on the crossmodal correspondence between colours and basic tastes, and also shed light on cross-cultural effects on this correspondence (i.e., the yellow-coloured cup being appropriately paired up with both sweet and acidic tastes). The present research argues that the drinking vessel is a necessary part of the multisensory coffee drinking experience. Thus, the specialty coffee companies should focus not only on the sensory characteristics of the coffee beverage itself. The sensory properties of the vessel should be much more carefully thought and designed so as to enhance the experience of the consumer – by also taking into account individual preferences of particular coffee taste/flavour.

Acknowledgements The first author is very grateful to all supporters of “The Coffee Sensorium” project. This project owes a great deal to many people from the specialty coffee community. The authors would, in particular, like to thank Tim Wendelboe and Felipe Croce (from Fazenda Ambiental Fortaleza) for donating the coffees used in Experiment 3, Diego Gonzales (Sofa Cafe) for providing the testing location, and Tony Chen for painting the coffee cups. We would also like to thank Takko Cafe and Ancora Coffee House for all their support in Experiments 1 and 2. Finally, thanks to the large crew of baristas who voluntarily helped in the data collection, and to all of the specialty coffee shops in Sao Paulo who helped in recruiting amateur consumers through social media.

References Ares, G., & Deliza, R. (2010). Studying the influence of package shape and colour on consumer expectations of milk desserts using word association and conjoint analysis. Food Quality and Preference, 21, 930-937. Ashkenazi, A., & Marks, L. E. (2004). Effect of endogenous attention on detection of weak gustatory and olfactory flavors. Perception & Psychophysics, 66(4), 596-608. Bury, C. (2014). Does the cup influence the taste of the coffee? http://coffeestrides.blogspot.com.br/2014/03/does-cup-influence-taste-of-coffee.html Accessed 03.10.18. Cardello, A. V., & Sawyer, F. M. (1992). Effects of disconfirmed consumer expectations on food acceptability. Journal of Sensory Studies, 7(4), 253-277.

Carvalho, F. M., & Spence, C. (2018). The shape of the cup influences aroma, taste, and hedonic judgements of specialty coffee. Food Quality and Preference, 68, 315-321. Carvalho, J. M., Paiva, E. L., & Vieira, L. M. (2016). Quality attributes of a high specification product: Evidences from the speciality coffee business. British Food Journal, 118(1), 132-149. Cheskin, L. (1957). How to predict what people will buy. New York, NY: Liveright. Da Rosa, S. D. V., Oliveira, P. D., Giomo, G. S., & Isquierdo, E. P. (2015). Post-harvest effects on beverage quality and physiological performance of coffee beans. African Journal of Agricultural Research, 10(12), 1457-1466. de Luca, P., & Pegan, G. (2014). The coffee shop and customer experience: A study of the US market. In F. Musso & E. Druica (Eds.), Handbook of research on retailer-consumer relationship development (pp. 173-196). Hershey PA: IGI Global. Dichter, E. (1964). Handbook of consumer motivations. New York, NY: McGraw-Hill. Esterl, M. (2011). A frosty reception for Coca-Cola’s white Christmas cans. The Wall Street Journal (Business), December 1. https://www.wsj.com/articles/SB10001424052970204012004577070521211375302 Favre, J.-P., & November, A. (1979). Color and communication. Zurich: ABC-Verlag. Foroni, F., Pergola, G., & Rumiati, R. I. (2016). Food color is in the eye of the beholder: The role of human trichromatic vision in food evaluation. Scientific Reports, 6:37034. Friston, K. (2010). The free-energy principle: A unified brain theory? Nature Reviews Neuroscience, 11(2), 127-138. Gacula Jr, M., & Rutenbeck, S. (2006). Sample size in consumer test and descriptive analysis. Journal of Sensory Studies, 21(2), 129-145. Guéguen, N., & Jacob, C. (2012). Coffee cup color and evaluation of a beverage’s “warmth quality”. Color Research and Application, 39, 79-81. Hameed, A., Hussain, S. A., Ijaz, M. U., Ullah, S., Pasha, I., & Suleria, H. A. R. (2018). Farm to consumer: Factors affecting the organoleptic characteristics of coffee. II: Postharvest processing factors. Comprehensive Reviews in Food Science and Food Safety, 17(5), 11841237. Harrar, V., & Spence, C. (2013). The taste of cutlery: How the taste of food is affected by the weight, size, shape, and colour of the cutlery used to eat it. Flavour, 2(1):21. Hirsh, J. B., Mar, R. A., & Peterson, J. B. (2012). Psychological entropy: A framework for understanding uncertainty-related anxiety. Psychological Review, 119(2), 304-321. Hurlbert, A. (1996). Colour vision: Putting it in context. Current Biology, 6(11), 1381-1384. ISO (2008). Standard 5492: Terms relating to sensory analysis. International Organization for Standardization. Vienna: Austrian Standards Institute. Jiang, W.-W., Niimi, J., Ristic, R., & Bastian S. E. P. (2017). The effects of immersive context and wine flavor on consumer wine flavor perception and emotions elicited. American Journal of Enology & Viticulture, 68, 1-10. Labroo, A. A., Dhar, R., & Schwarz, N. (2007). Of frog wines and frowning watches: Semantic priming, perceptual fluency, and brand evaluation. Journal of Consumer Research, 34(6), 819-831.

Lee, L., Frederick, S., & Ariely, D. (2006). Try it, you'll like it: The influence of expectation, consumption, and revelation on preferences for beer. Psychological Science, 17(12), 10541058 Lingle, T. R., & Menon, S. N. (2017). Cupping and grading—Discovering character and quality. In B. Folmer (Ed.), The craft and science of coffee (pp. 181-203). London, UK: Academic Press. Maga, J. A. (1974). Influence of color on taste thresholds. Chemical Senses, 1(1), 115-119. Oberfeld, D., Hecht, H., Allendorf, U., & Wickelmaier, F. (2009). Ambient lighting modifies the flavor of wine. Journal of Sensory Studies, 24, 797-832. Piqueras-Fiszman, B., & Spence, C. (2012). The influence of the color of the cup on consumers' perception of a hot beverage. Journal of Sensory Studies, 27(5), 324-331. Piqueras-Fiszman, B., & Spence, C. (2015). Sensory expectations based on product-extrinsic food cues: An interdisciplinary review of the empirical evidence and theoretical accounts. Food Quality and Preference, 40, 165-179. Ponte, S. (2002). The latte revolution? Regulation, markets and consumption in the global coffee chain. World Development, 30(7), 1099-1122. Reber, R., & Schwarz, N. (2001). The hot fringes of consciousness: Perceptual fluency and affect. Consciousness & Cognition, 2, 223-231 Reber, R., Winkielman, P., & Schwartz, N. (1998). Effects of perceptual fluency on affective judgments. Psychological Science, 9:1, 45–48. Risso, P., Maggioni, E., Olivero, N., & Gallace, A. (2015). The association between the colour of a container and the liquid inside: An experimental study on consumers’ perception, expectations and choices regarding mineral water. Food Quality and Preference, 44, 17-25. Saluja, S., & Stevenson, R. J. (2018). Cross-modal associations between real tastes and colors. Chemical Senses, 43, 475-480. Schifferstein, H. N. (2001). Effects of product beliefs on product perception and liking. In L. Frewer, E. Risvik, & H. Schifferstein (Eds.), Food, people and society (pp. 73-96). Berlin, Germany: Springer. Schifferstein, H. N. J. (2009). The drinking experience: Cup or content? Food Quality and Preference, 20(3), 268-276. Shankar, M. U., Levitan, C. A., Prescott, J., & Spence, C. (2009). The influence of color and label information on flavor perception. Chemosensory Perception, 2:53 Shankar, M. U., Levitan, C., & Spence, C. (2010). Grape expectations: The role of cognitive influences in color–flavor interactions. Consciousness and Cognition, 19(1), 380-390. Skaczkowski, G., Durkin, S., Kashima, Y., & Wakefield, M. (2016). The effect of packaging, branding and labeling on the experience of unhealthy food and drink: A review. Appetite, 99, 219-234. Small, D. M., Veldhuizen, M. G., Felsted, J., Mak, Y. E., & McGlone, F. (2008). Separable substrates for anticipatory and consummatory food chemosensation. Neuron, 57, 786-797. Spence, C. (2011). Crossmodal correspondences: A tutorial review. Attention, Perception, & Psychophysics, 73(4), 971-995. Spence, C. (2015). Multisensory flavour perception. Cell, 161, 24-35.

Spence, C. (2016). Multisensory packaging design: Colour, shape, texture, sound, and smell. In M. Chen & P. Burgess (Eds.), Integrating the packaging and product experience: A route to consumer satisfaction (pp. 1-22). Oxford, UK: Elsevier. Spence, C. (2017). Gastrophysics: The new science of eating. London, UK: Viking Penguin. Spence, C. (2018). Background colour & its impact on food perception & behaviour. Food Quality & Preference, 68, 156-166. Spence, C., & Parise, C. (2010). Prior-entry: A review. Consciousness and Cognition, 19(1), 364-379. Spence, C., & Piqueras-Fiszman, B. (2014). The perfect meal: The multisensory science of food and dining. Chichester, UK: John Wiley & Sons. Spence, C., & Piqueras-Fiszman, B. (2016). Food color and its impact on taste/flavour perception. In B. Piqueras-Fiszman & C. Spence (Eds.), Multisensory flavor perception: From fundamental neuroscience through to the marketplace (pp. 107-132). Oxford, UK: Elsevier. Spence, C., Velasco, C., & Knoeferle, K. (2014). A large sample study on the influence of the multisensory environment on the wine drinking experience. Flavour, 3:8. Spence, C., & Wan, X. (2015). Beverage perception and consumption: The influence of the container on the perception of the contents. Food Quality and Preference, 39, 131-140. Spence, C., Wan, X., Woods, A., Velasco, C., Deng, J., Youssef, J., & Deroy, O. (2015). On tasty colours and colourful tastes? Assessing, explaining, and utilizing crossmodal correspondences between colours and basic tastes. Flavour, 4(1):23 Spence, C., & Youssef, J. (2018). Assessing the long-term impact of the molecular gastronomy movement on haute cuisine. International Journal of Gastronomy & Food Science, 14, 35-44. Stefani, G., Romano, D., & Cavicchi, A. (2006). Consumer expectations, liking and willingness to pay for specialty foods: Do sensory characteristics tell the whole story? Food Quality and Preference, 17(1-2), 53-62 Stevenson, R. J. (2012). The role of attention in flavour perception. Flavour, 1:2. Stevenson, R. J. (2014). Flavor binding: Its nature and cause. Psychological Bulletin, 140, 487-510. Todorov, V., & Filzmoser, P. (2010). Robust statistic for the One-way MANOVA. Computational Statistics & Data Analysis, 54(1), 37-48. Torres, L. O., Salazar, N. S, & Salgado, C.J. (2016). Effect of colour on exotic fruit flavour perception (Bachelor's thesis, Universidad de La Sabana, Colombia). Traore, T. M., Wilson, N. L., & Fields, D. (2018). What explains specialty coffee quality scores and prizes: A case study from The Cup of Excellence program. Journal of Agricultural and Applied Economics, 50(3), 349-368. Van Doorn, G., Colonna‐Dashwood, M., Hudd‐Baillie, R., & Spence, C. (2015). Latté art influences both the expected and rated value of milk‐based coffee drinks. Journal of Sensory Studies, 30(4), 305-315.

Van Doorn, G., Woods, A., Levitan, C. A., Wan, X., Velasco, C., Bernal-Torres, C., & Spence, C. (2017). Does the shape of a cup influence coffee taste expectations? A crosscultural, online study. Food Quality and Preference, 56, 201-211. Van Doorn, G. H., Wuillemin, D., & Spence, C. (2014). Does the colour of the mug influence the taste of the coffee? Flavour, 3(1):10. Velasco, C., Jones, R., King, S., & Spence, C. (2013). Assessing the influence of the multisensory environment on the whisky drinking experience. Flavour, 2(1):23. Velasco, C., Michel, C., Youssef, J., Gamez, X., Cheok, A. D., & Spence, C. (2016). Colourtaste correspondences: Designing food experiences to meet expectations or to surprise. International Journal of Food Design, 1, 83-102. Wan, X., Velasco, C., Michel, C., Mu, B., Woods, A. T., & Spence, C. (2014). Does the shape of the glass influence the crossmodal association between colour and flavour? A crosscultural comparison. Flavour, 3(1):3. Wang, Q. J., Carvalho, F. R., Persoone, D., & Spence, C. (2017). Assessing the effect of shape on the evaluation of expected and actual chocolate flavour. Flavour, 6(1), 2-7. Woods, A. T., Lloyd, D. M., Kuenzel, J., Poliakoff, E., Dijksterhuis, G. B., & Thomas, A. (2011). Expected taste intensity affects response to sweet drinks in primary taste cortex. Neuroreport, 22, 365-369. Woods, A. T., Marmolejo-Ramos, F., Velasco, C., & Spence, C. (2016). Using single colours and colour pairs to communicate basic tastes II; Foreground-background colour combinations. i-Perception, 7:5. Yeomans, M. R., Chambers, L., Blumenthal, H., & Blake, A. (2008). The role of expectancy in sensory and hedonic evaluation: The case of smoked salmon ice-cream. Food Quality and Preference, 19(6), 565-573. Zampini, M., Sanabria, D., Phillips, N., & Spence, C. (2007). The multisensory perception of flavor: Assessing the influence of color cues on flavor discrimination responses. Food Quality and Preference, 18(7), 975-984.

Figure 1: Cups used in Experiments 1 and 2 (A), and in Experiment 3 (B). Diagram representing the experimental procedure adopted in Experiment 3. Each participant evaluated four samples (cup colour + coffee type) in one of the four possible sequences depicted above (1-4) (C).

Figure 2: Mean ratings (± SE) of sweetness, acidity, and liking for the white (white bars) and pink (grey bars) cups as a function of condition (pre- and post-tasting) in Experiment 1 (A) and Experiment 2 (B). Asterisk indicates statistical significance between white and pink cups on both pre- and post-tasting ratings at p < 0.05 (*) or p < 0.001 (**). Dagger indicates statistical significance for pre- versus post-tasting at p < 0.05 (†) or p < 0.001 (††).

Figure 3: Mean ratings (± SE) of (A) pre-tasting (expectation) and (B) post-tasting (perception) as a function of rating type (sweetness, acidity, and liking) for all four cup colours tested (pink, yellow, white, and green) in Experiment 3. Panel (B) shows separate plots for Brazilian and Kenyan coffees. Asterisks indicates statistical significance at p < 0.05 (*) or p < 0.001 (**).

Figure 4: Mean ratings (± SE) of pre-tasting (white), post-tasting for the Brazilian coffee (grey), and post-tasting for the Kenyan coffee (black) as a function of rating type (sweetness and acidity) and the colour of the cup (pink, yellow, white, and green) in Experiment 3. Asterisk indicates statistically significant difference between pre- versus post-tasting ratings for the Brazilian coffee at p < 0.001 (**). Dagger indicates statistically significance difference between pre- versus post-tasting ratings for the Kenyan coffee at p < 0.001 (††).

Figure 5: Scatter-plots showing the correlation between the difference indices and liking ratings in both incongruent conditions in Experiment 3. The difference indices were created by subtracting the post-tasting from the pre-tasting rating values of sweetness and acidity, respectively. E = Expectation (i.e., pre-tasting scores); P= Perception (i.e., post-tasting scores).

Table 2: Compilation of pre- and post-tasting sweetness and acidity ratings (mean ± SE) across all three experiments.

Experiment 1* Cup Colour

Pre-tasting (mean ± SE)

Experiment 2*

Post-tasting (mean ± SE)

Pre-tasting (mean ± SE)

Experiment 3*

Post-tasting (mean ± SE)

Pre-tasting (mean ± SE)

Post-tasting (mean ± SE) Brazilian Coffee

Post-tasting (mean ± SE) Kenyan Coffee

Sw

Ac

Sw

Ac

Sw

Ac

Sw

Ac

Sw

Ac

Sw

Ac

Sw

Ac

White

5.2±0.1

5.7±0.3

5.1±0.2

6.2±0.1

4.1±0.3

6.0±0.3

4.2±0.3

6.7±0.2

5.3±0.1

5.0±0.1

6.2±0.1

4.4±0.1

4.5±0.1

6.3±0.1

Pink

5.6±0.2

5.1±0.1

6.3±0.1

4.9±0.2

5.5±0.2

5.1±0.3

6.2±0.1

4.6±0.2

6.4±0.1

4.7±0.1

7.1±0.1

3.9±0.1

4.2±0.1

7.2±0.1

Yellow

--

--

--

--

--

--

--

--

5.8±0.1

5.3±0.1

6.5±0.1

4.1±0.1

4.9±0.1

6.9±0.1

Green

--

--

--

--

--

--

--

--

5.2±0.1

5.9±0.1

5.9±0.1

3.6±0.1

4.6±0.1

6.9±0.1

Sw = sweetness; Ac = acidity. For significant differences between mean values, please see Figure 2 (for Experiments 1 and 2), and Figures 3 and 4 (for Experiment 3). *Please note that, since cups, coffee, context, and participants were almost all different between the three experiments, the scores reported here should only be compared within experiments.

Table 3: Compilation of liking ratings (mean ± SE) across all three Experiments Cup colour

Experiment 1*

Experiment 2*

White

7.4±0.2

Pink

Experiment 3* Brazilian coffee

Kenyan coffee

7.8±0.1

6.9±0.1

6.7±0.1

8.0±0.1

8.4±0.1

7.8±0.1

5.6±0.1

Yellow

--

--

7.5±0.1

7.4±0.1

Green

--

--

5.9±0.1

7.5±0.1

For significant differences between mean values, please see Figure 2 (for Experiments 1 and 2), and Figures 3 and 4 (for Experiment 3). *Please note that, since cups, coffee, context, and participants were almost all different between the three experiments, the scores reported here should only be compared within experiments.

Manuscript Highlights Title: Cup colour influences consumers’ expectations and experience on tasting specialty coffee Authors: Fabiana M. Carvalho & Charles Spence

-- Three studies examine the effect of cup colour on specialty coffee taste attributes. -- Pre- and post-tasting ratings of sweetness and acidity are affected by cup colour. -- Cup colour influences hedonic judgements in congruent and incongruent pairings. -- It is important to consider designing receptacles to optimize the drinking experience.