Food Quality and Preference 18 (2007) 975–984 www.elsevier.com/locate/foodqual The multisensory perception of flavor: Assessing the influence of color cues on flavor discrimination responses Massimiliano Zampini a,b,c,*, Daniel Sanabria c,d, Nicola Phillips e, Charles Spence c a Department of Cognitive Sciences and Education, University of Trento, Rovereto, Italy b Center for Mind/Brain Sciences, University of Trento, Rovereto, Italy c Department of Experiment Psychology, University of Oxford, South Parks Road, Oxford, England, UK d Departamento de Psicologı́a Experimental, Universidad de Granada, Spain e Unilever Research & Development Port Sunlight, Bebington, UK Received 14 March 2006; received in revised form 30 March 2007; accepted 2 April 2007 Available online 14 April 2007 Abstract Two experiments are reported that were designed to investigate the influence of visual color cues on people’s flavor discrimination and flavor intensity ratings for a variety of fruit-flavored solutions. In Experiment 1, the participants had to associate specific flavors with solutions of various colors simply by looking at them (i.e., without tasting them). In Experiment 2, the participants tasted the solutions and had to discriminate the flavor of solutions that had been colored either ‘appropriately’ or ‘inappropriately’, or else presented as colorless solutions. The participants were explicitly informed that the colors of the solutions provided no useful information regarding the actual flavor identity of the solutions. The participants also rated the flavor intensity of the solutions. The accuracy of participants’ flavor discrimination performance was significantly lower when the solutions were colored inappropriately than when they were colored appropriately (or else were presented as colorless solutions). These results show that the modulatory effect of visual cues on flavor perception can override participants’ awareness that the solutions would frequently be colored inappropriately. Ó 2007 Elsevier Ltd. All rights reserved. Keywords: Flavor perception; Color; Visual dominance; Fruit-flavored beverages; Labelled magnitude scale (LMS) 1. Introduction The perception and evaluation of food and drink is an inherently multisensory experience. Gustatory, olfactory, visual, oral-somatosensory, auditory, and even nociceptive cues can all play a role in determining our perception of what we eat and drink (see Delwiche, 2004; Spence, 2002; Stillman, 2002). For instance, our perception of the pleasantness of a food is influenced not only by its look, smell, and taste, but also by its oral texture and by the sound that * Corresponding author. Address: Department of Cognitive Sciences and Education, University of Trento, Rovereto, Italy. Tel.: +39 0464 483604; fax: +39 0464 483554. E-mail addresses: [email protected] (M. Zampini), [email protected] (C. Spence). 0950-3293/$ - see front matter Ó 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.foodqual.2007.04.001 it makes in the mouth when we eat it (e.g., Amerine, Pangborn, & Roessler, 1965; Vickers, 1983; Zampini & Spence, 2004, 2005). Flavor perception is also influenced by interactions between oral texture and both olfactory and gustatory cues. For example, Christensen (1980a, 1980b) has demonstrated that changes in the viscosity of a solution can modify its perceived taste and flavor and vice versa (see also Bult, de Wijk, & Hummel, 2007). The visual appearance of food and drink can also provide another important determinant of flavor perception. Over the years, many different researchers have attempted to evaluate the influence of color cues on people’s flavor perception (see Clydesdale, 1993; Delwiche, 2004; for reviews). Color cues have been shown to dramatically affect people’s perception of a variety of different foods and drinks (e.g., DuBose, Cardello, & Maller, 1980; Duncker, 976 M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 1939; Garber, Hyatt, & Starr, 2000; Johnson & Clydesdale, 1982; Morrot, Brochet, & Dubourdieu, 2001; Philipsen, Clydesdale, Griffin, & Stern, 1995; Roth, Radle, Gifford, & Clydesdale, 1988; Stillman, 1993; Wheatley, 1973; Zellner & Durlach, 2003). For example, in one oft-cited study, DuBose et al. reported that participants were less accurate in identifying the flavor of a fruit-flavored beverage when they were unable to perceive its color (e.g., 20% correct flavor detection responses in the masked color condition vs. 100% correct responses under normal viewing conditions for an orange-flavored beverage). More importantly, 40% of the participants reported that a cherry-flavored beverage actually tasted of orange when the beverage was inappropriately colored orange (compared to 0% orange-flavor responses when the drink was appropriately colored red). Other researchers have also reported a similar visual modulation of odor discrimination responses (e.g., Blackwell, 1995; Davis, 1981; Zellner, Bartoli, & Eckard, 1991; Zellner & Kautz, 1990; Zellner & Whitten, 1999). While the potential influence of color cues on flavor identification responses is by now well-documented, the evidence regarding the impact of changes in color intensity on perceived flavor intensity is less clear. For instance, the addition of a red coloring to cherry- and strawberry-flavored sucrose solutions has been found to increase the perceived sweetness of these solutions (Johnson & Clydesdale, 1982; Johnson, Dzendolet, & Clydesdale, 1983; Johnson, Dzendolet, Damon, Sawyer, & Clydesdale, 1982). Furthermore, a number of years ago, Maga (1974) made the intriguing suggestion that those colors that are typically associated with the natural ripening of fruits (e.g., red, yellow, etc.) may be particularly effective in modulating sweetness perception (see also Strugnell, 1997). However, Frank, Ducheny, and Mize (1989) found that strawberry odor, but not red color, affected sweetness perception in their study. In one of the earliest studies in this area, Pangborn (1960) reported that people perceived a green-colored pear nectar as being less sweet than a colorless pear nectar. However, Pangborn and Hansen (1963) failed to replicate these results, instead reporting that green coloring had no effect on the perceived sweetness of pear nectar, despite an overall increase in sensitivity to sweetness when color was added to the solutions. Ambiguous results have also been reported in studies where participants have been asked to rate the flavor intensity of solutions varying in the intensity of the color added. DuBose et al. (1980) also found that overall flavor intensity was affected by color intensity, with higher levels of color intensity resulting in stronger flavor evaluation responses by participants for the orange-flavored but not for the cherry-flavored beverages tested in their study (see Clydesdale, 1993; for a comprehensive review). At present, no clear explanation has been provided for the lack of convergent results between studies that have investigated the influence of color intensity on flavor perception. It is possible that the lack of convergence might reflect the use of inadequate scaling procedures for rating taste or flavor intensity in many of the previous studies in this area. One plausible suggestion here is that a more consistent pattern of responding might have been observed if a more sophisticated means of rating perception were to be used, such as that embodied by the Labelled Magnitude Scale (LMS; Green, Shaffer, & Gilmore, 1993). The LMS is a semantically-labelled scale with a quasi-logarithmic spacing between the verbal labels that has been specifically developed for the rating of gustatory and oral-somesthetic sensations. The LMS has the characteristics of a category scale (i.e., a presentation that is easily accepted by naı̈ve participants) yielding data with ratio-level properties. The LMS is usually presented as a vertical line with its upper limit labelled ‘‘strongest imaginable” and the lower limit ‘‘no sensation” (the intermediate labels are: ‘‘very strong”, ‘‘strong”, ‘‘moderate”, ‘‘weak”, ‘‘barely detectable”). In a typical study using the LMS, participants have to rate the intensity of a stimulus by comparing it with all other stimuli found in daily life along this vertical line. Unlike other magnitude scales, the LMS reduces the possibility of ceiling effects by using the ‘‘strongest imaginable” adjective as the upper limit, thus not limiting one to a fixed number range (e.g., Bartoshuk, 2000). Furthermore, Green et al. (1993) have argued that the use of more traditional numerical scaling methods (e.g., magnitude estimation), can emphasize inter-individual differences, due to people’s idiosyncratic usage of numbers. This problem is reduced when participants use the LMS scale. Moreover, Kurtz and White (1998) have shown that the evaluation of complex stimuli such as flavors is particularly stable between raters using the LMS. The primary aim of our main experiment (Experiment 2) was to investigate the influence of color intensity on the perception of flavor intensity using the LMS procedure. The LMS procedure has never been used for this kind of task before, and given that it seems more suitable than other scaling procedures, we thought that its use might help to reduce some of the discrepancies highlighted amongst previous research in this area. However, before investigating the impact of color on flavor perception directly, an initial experiment was conducted in order to analyze people’s expectations concerning flavor based solely on visible color cues. In Experiment 1, the participants were asked to simply identify the flavor that was most closely associated for them with the color of the solutions presented by inspecting them visually (but without tasting them). The aim of this preliminary study was to determine whether there are certain colors that are more reliably associated with one flavor than with another. The hypothesis was that these colors might have a greater impact on flavor perception in our main experiment (Experiment 2) where participants evaluated the solutions by both viewing and tasting them. It seemed plausible that this factor – the different color–flavor associations held by different individuals – might also help to explain the inconsistent effect of color intensity on flavor intensity ratings that have been documented in previous studies. M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 2. Experiment 1 2.1. Methods 2.1.1. Participants Eleven female participants aged between 25 and 42 years (mean age of 34 years) took part in both of the experiments reported here. All of the participants were naı̈ve as to the purpose of the study, and varied in their previous experience of psychophysical testing procedures. The participants were selected from a panel of untrained individuals recruited from the staff at Unilever Research & Development Port Sunlight. All of the participants reported having normal gustatory and olfactory abilities and normal, or corrected-to-normal, vision. All of the participants had normal color-vision as assessed by means of the Ishihara test for color blindness (Ishihara, 1943). None of the participants reported having a cold or other respiratory tract infection either on the days of the testing sessions or in the week prior to testing. The participants gave their informed consent prior to the start of the experiment. Both of the experiments reported in this study had ethical approval from Unilever Research & Development Port Sunlight Laboratory. The participants received £20 gift voucher each in return for taking part in the study. 2.1.2. Apparatus and stimuli The participants were seated comfortably in front of a table in an individual testing booth at the Port Sunlight testing facilities. The participants were given a transparent plastic cup of a flavorless solution (40 ml) on each trial. The solutions were colored red, green, orange, yellow, blue, grey, or else were colorless. The solutions were prepared from 1 l of distilled water by adding concentrated food colorings (Supercook, Leeds, UK): 16 ml of yellow food coloring for the yellow solutions; 16 ml of blue food coloring for the blue solutions; and 16 ml of red food coloring for the red solutions. The green solutions were obtained by mixing 4 ml of blue food coloring with 8 ml of yellow food coloring; the orange samples by adding 8 ml of red and 16 ml of yellow; and the grey samples by adding 8 ml of yellow, 8 ml of red, and 20 ml of the blue food coloring. All of the food colorings used in the present study were flavorless. The colorless samples consisted of distilled water and nothing else. 977 2.1.3. Design The one within-participant factor was the color of the solutions (red, green, orange, yellow, blue, grey, or colorless). Each of the seven colored solutions was presented 5 times within a single block of 35 randomly-ordered trials. The experiment lasted for approximately 30 min. 2.1.4. Procedure The participants were given written instructions describing the experimental procedure to read in their own time. The experimenter then explained the procedure to the participants and demonstrated the equipment. The participants were instructed to look at each solution from above, and to identify the flavors that they perceived as corresponding most closely to the color of the solution. They had to indicate their response by making a tick on a list using a paper-based response sheet. The following options were available on the list: Strawberry, pear, lemon, melon, lime, orange, aniseed, spearmint, liquorice, cherry, lettuce, vanilla, toffee, cream soda, apple, peach, raspberry, pineapple, yoghurt, blackcurrant, flavorless, or other. If participants indicated this last option they were prompted to suggest the specific flavor that they had in mind. The participants were also informed that they could choose the same option on more than one occasion. No time limits were imposed for the completion of a participant’s responses. 2.2. Results and discussion The analyses reported in this study were conducted using SPSS (SPSS Inc., Chicago, USA). Participants’ responses to each of the colored solutions were analyzed using v2 analysis. The results (summarized in Table 1) showed that participants more often associated the green colored solutions with lime flavor (M = 69%), the orange colored solutions with an orange flavor (M = 91%), the yellow solutions with the lemon flavor (M = 89%), the blue solutions with spearmint flavor (M = 86%), the grey solution with the flavor of blackcurrant (M = 53%) or liquorice (M = 40%), and the colorless solution with the flavorless response (M = 51%). Finally, it is important to note that the red coloring was not especially associated with any particular flavor (i.e., strawberry, M = 46%, raspberry, and cherry flavor both M = 27%) over the others (although Table 1 Results of v2 analysis for each of the colors used in Experiment 1 Color v2 p Associated flavors Green Orange Yellow Blue Grey Red Colorless 32.33 82.01 120.64 67.34 41.94 3.64 69.85 <.001a <.001a <.001a <.001a <.001a .16 <.001a Lime (69%) > apple (20%), melon (11%) Orange (91%) > aniseed (5%), toffee (4%) Lemon (89%) > pear (5%), apple (4%), melon (2%) Spearmint (86%) > raspberry (9%), cream soda (5%) Blackcurrant (53%), liquorice (40%) > cherry (4%), aniseed (4%) Strawberry (46%), rasperry (27%), cherry (27%) Flavorless (51%) > cream soda (16%), vanilla (15%), aniseed (15%), spearmint (2%), melon (2%), pear (2%) a Significant post-hoc comparisons (Bonferroni corrected with p = .05 before comparison) to test for particular color–flavor associations are shown in the last column together with the percentage of choices for each flavor. 978 M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 note that this finding does not necessarily rule out the possibility that the red color may actually have had an equally strong association with several of the flavor terms on the list provided). The results of Experiment 1 demonstrate that solutions having certain colors are strongly associated with particular flavors. The orange, yellow, and blue solutions were frequently associated with the flavors of orange, lemon, and spearmint, respectively. There was more variability in the flavor associations generated for the red solutions: Participants typically associated the red solutions with the flavors of strawberry, raspberry, and cherry. It would seem reasonable to suggest that colored solutions whose color is more robustly associated with one particular flavor should have a greater influence on flavor identification responses when participants actually taste the solutions than for colored solutions where the color–flavor link is less apparent (or is less commonly shared). This hypothesis was investigated in the second experiment by presenting fruit-flavored solutions (orange, lime, and strawberry) each colored green, orange, or red. Preliminary testing had revealed that these three flavors were roughly equally identifiable.1 The question of whether or not the intensity of the color would influence the perception of flavor intensity was also investigated using a LMS procedure that had been specifically designed to evaluate oral sensation (e.g., Green et al., 1993). In Experiment 2, all of the participants were explicitly informed that the flavored solutions would frequently be colored ‘inappropriately’. This experimental manipulation allowed us to investigate whether or not the visual cues would still influence flavor perception when the participants were aware of the lack of any meaningful correspondence between color and flavor. To our knowledge, this issue has never been addressed previously in research on the influence of color cues on flavor perception. 3. Experiment 2 3.1. Methods 3.1.1. Apparatus and stimuli The participants were seated in front of a disinfected sink (the disinfectant used was odorless) and received a 1 Participants were presented with colorless solutions having the following flavors: strawberry, lime, orange, blackcurrant, yoghurt, apple, raspberry, and flavorless. The participants were instructed to taste the solutions by moving the liquid around in their mouths, but not to swallow at any stage. The participants were asked to spit the solutions into the provided sink. Next, participants were asked to identify the flavor of the solution, by ticking a list (the options were the same as those used in Experiment 1). Finally, they were asked to rinse their mouths thoroughly with artificial saliva, without swallowing, and spit this into the sink provided. The accuracy of participants’ responses was as follows: flavorless (89%), apple (65%), strawberry (40%), lime and orange (33%), blackcurrant (20%), raspberry (15%) and finally yoghurt (12%). plastic cup of colored and flavored solution on each trial. The plastic cups were transparent and specifically designed for food use. The solutions were presented at room temperature (20 ± 2 °C). The solutions in this experiment were colored red, green, or orange. Strawberry (House of Flavors, reference: 01/10114), lime (01/10071), and orange (01/10085) fruit flavors were used. The solutions were prepared by adding 1.5 ml of concentrated fruit flavor to each litre of distilled water used. In each litre of solution, 62.5 g of sucrose (GMP sugar, Aarhus United, Hull, UK) was dissolved. Each of the flavorings was equiprobably associated with each of the different colors (red, green, orange, and colorless). This meant that, for example, the strawberry flavored solutions were just as likely to be colored red, green, orange, as to be presented as a colorless solution. Consequently, the color of the solutions did not have any predictive validity regarding the likely flavor of the solutions being tasted. The participants were told that they would often be tricked by the color of the solutions (i.e., that the color of the solutions would often not correspond to the flavor typically associated with that color). All of the participants were explicitly informed that the colors of the solutions that they were to taste provided absolutely no useful information with regard to the likely flavor of the solutions (i.e., that the colors and flavors were fully crossed in the experimental design). Flavorless samples were also presented and they could be colored or colorless just as for the flavored solutions. The volume of coloring added to the solutions varied between either a standard concentration (i.e., 8 ml of red coloring for the red solutions; 2 ml of blue coloring and 4 ml of yellow coloring for the green solutions; and 4 ml of red coloring and 8 ml of yellow coloring for the orange solutions) and a double concentration (i.e., by doubling the volumes of each of the colorings added). The bottled liquids were prepared on the morning of the study, used during the day and discarded at the end of the day. The experimenter dispensed 40 ml of the solutions from the airtight glass bottles into the clear plastic cups immediately prior to testing using a sterile syringe. The cups were sealed during the period between the dispensing of the liquids and the beginning of the experimental testing session. The experimenter gave participants 40 ml of flavorless palate cleansing solution (‘artificial saliva’) between each trial in order to wash their mouths out before tasting the next beverage. This palate cleansing solution was made from deionised water, 1.865 g/l of potassium chloride (KCl), and 0.210 g/l of sodium bicarbonate (NaHCO3; see O’Doherty, Rolls, Francis, Bowtell, & McGlone, 2001), and was stored in separate glass bottles from the flavored solutions. 3.1.2. Design There were three within-participants factors: The flavor of the solutions (strawberry, lime, orange, or flavorless), M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 the color of the solutions (red, green, orange, or colorless), and the amount of food coloring added (standard or double). These factors were fully crossed (with the exception that the volume of coloring had no meaning for the colorless solutions) resulting in 28 conditions, which were each presented 4 times. The colored samples were added to the colorless–flavorless samples (which were presented 12 times) resulting in a block of 124 randomly-ordered trials in total. The experimental session lasted for approximately 90 min. 3.1.3. Procedure The participants were given both written and oral instructions to view the solutions from directly above the plastic cup and then to taste them by moving the liquid around in their mouth, but not to swallow the liquid at any stage.2 After tasting each of the solutions, the participants were asked to spit the solutions into the sink provided. The participants were instructed to try and determine the identity and intensity of the flavor, bearing in mind that these might have been manipulated so that the color should not be considered to provide a reliable cue to flavor identity. After having tasted each solution, the participants were instructed to indicate the flavor that they perceived by ticking an answer sheet (the options were the same as in Experiment 1) and to evaluate their perception of flavor intensity for each solution using the LMS (e.g., Green et al., 1993). The participants were fully informed and trained (but not experienced) in the use of the scale. In particular, they were instructed on how to interpret the verbal descriptors (e.g., strongest imaginable sensation). They were then asked to rinse their mouths thoroughly with the palate cleansing solution, again without swallowing, and to spit this into the sink provided before testing the next solution. 3.2. Results The data regarding flavor identification responses from Experiment 2 are highlighted in Fig. 1. The accuracy data regarding participants’ flavor detection responses were submitted to a two-way within-participants repeated-measures ANOVA with the factors of Flavor (flavorless, lime, orange, or strawberry) and Color (colorless, green-standard, green-doubled, orange-standard, orange-doubled, 2 The principal reason for instructing the participants not to swallow was that they might have found it rather unpleasant to potentially drink such a large quantity of liquid. The participants actually tasted 124 cups 40 ml = about 5 l of liquid in total. However, it is important to note on this point that previous studies have shown that people can vary quite considerably in terms of the retro-nasal delivery of odors/flavors when swallowing is prevented (see Buettner, Beer, Hannig, & Settles, 2001; Hodgson, Linforth, & Taylor, 2003). The fact that the participants did not swallow the drinks in the present study represents one limitation in terms of the ecological validity of our experimental design given that people normally swallow the beverages/foodstuffs that they put in their mouths. 979 red-standard, or red-doubled). For all of the analyses reported here, post-hoc comparisons used Bonferroni-corrected t-tests (where p < .05 prior to correction). The analysis of the accuracy data revealed a significant main effect of Flavor (F(3, 30) = 5.49, p < .005), reflecting the fact that participants detected the flavorless solutions (mean of 93% correct ‘flavorless’ responses) more accurately than when they were lime- (M = 55%), orange- (M = 62%) or strawberry-flavored (M = 51%; none of the post-hoc comparisons between the flavored solutions reached significance). The significant interaction between Flavor and Color (F(18, 180) = 5.01, p < .001), revealed that the lime-flavored solutions were identified more accurately when they were colored green-standard (M = 73%), green-doubled (M = 70%) or were colorless (M = 73%) than when they were colored orange-standard (M = 54%), orange-doubled (M = 31%), red-standard (M = 45%), or red-doubled (M = 36%). Moreover, the orange flavor was detected more accurately in solutions that were colored orange-standard (M = 82%), orange-doubled (M = 79%), or colorless (M = 69%) than in either the green-standard (M = 45%), green-doubled (M = 50%), red-standard (M = 60%), or red-doubled (M = 48%) colored solutions (the post-hoc comparisons between the orange and colorless solutions and between the green and red solutions were not significant). The accuracy with which the participants were able to detect the strawberry-flavored solutions was unaffected by their color (M = 56% for red-standard; M = 59% for red-doubled; M = 47% for green-standard; M = 49% for green-doubled; M = 46% for orange-standard; M = 53% for orange-doubled; and M = 48% for the colorless solutions). There was a borderline-significant main effect of Color (F(6, 60) = 2.17, p = .058), with participants detecting the flavors of the solutions more accurately when they were colorless (M = 72%), orange-standard (M = 68%), orangedoubled (M = 66%), than when they were colored red-doubled (M = 58%; however, note that none of the post-hoc comparisons between these conditions were significant). A similar ANOVA was performed on the flavor intensity LMS ratings (see Fig. 2). The participants’ individual mean responses were log 10-transformed before conducting this analysis because the distribution of LMS ratings is typically log-normal (Green & George, 2004). The analysis of the flavor intensity log 10-converted data revealed a significant main effect of Flavor (F(3, 30) = 77.22, p < .001), attributable to the fact that participants rated the intensity of the flavorless solutions (mean rating of – 1.150) as being significantly weaker than the lime (M = 1.295), orange (M = 1.118), or strawberry (M = 1.355) solutions, as might have been expected. There was also a significant difference between the strawberry- and orange-flavored solutions with the strawberry solutions being rated as having a more intense flavor than the orange solutions. There was no significant difference between the intensity ratings for the orange- and lime-flavored solutions, nor for the comparison between the lime- and strawberry-flavored solutions. 980 LIME 100 c 100 75 75 Correct responses (%) Correct responses (%) a M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 50 25 0 STRAWBERRY 50 25 0 Greenstandard Greendoubled Orangestandard Orangedoubled Redstandard Reddoubled Greenstandard Colorless Greendoubled Color of the solutions Orangestandard Redstandard Reddoubled Colorless Reddoubled Colorless Color of the solutions ORANGE FLAVORLESS d 100 75 75 Correct responses (%) b 100 Correct responses (%) Orangedoubled 50 25 50 25 0 0 Greenstandard Greendoubled Orangestandard Orangedoubled Redstandard Reddoubled Colorless Color of the solutions Greenstandard Greendoubled Orangestandard Orangedoubled Redstandard Color of the solutions Fig. 1. Mean percentage of correct flavor detection responses for the lime (a), orange (b), strawberry (c), and flavorless (d) solutions presented in Experiment 2. The error bars represent the between-participants standard errors of the means. There was no main effect of Color [F(6, 60) = 1.18, p = .33], although the interaction between Flavor and Color was significant [F(18, 180) = 2.84, p < .001]. This interaction reflects the fact that participants rated the lime-flavored solution as having a less intense flavor when the color was double orange (M = 1.095) than when the solutions were uncolored (M = 1.394) or the color was standard orange (M = 1.371). No effect of color intensity was found for either the green or red colored lime-flavored solutions (standard green coloring, M = 1.318; double green coloring, M = 1.327; standard red coloring, M = 1.298; and double red coloring, M = 1.260). 3.3. Discussion The most important finding to emerge from Experiment 2 was that color cues can modulate flavor identification responses even when participants are informed that these colors will often be misleading. In particular, flavors associated with inappropriate colors (i.e., lime-flavored drinks that were colored either red or orange; orange-flavored drinks that were colored either green or red) were perceived less accurately than when they were presented with the appropriate color (i.e., lime flavor–green color; orange fla- vor–orange color). These results appear to show that inappropriate coloring tended to lead to an impairment of flavor discrimination responses, while appropriate coloring did not improve flavor discrimination accuracy (at least when compared to the flavor discrimination accuracy for the colorless solutions). Interestingly, no significant effect of color was shown for the strawberry-flavored solutions. The identification of strawberry flavor did not decrease when the solutions were colored incongruently (i.e., when strawberry-flavored solutions were colored green or orange). Finally, it is worth noting that adding color to flavorless solutions never induced significant ‘illusory’ flavor responses (cf. Zellner & Kautz, 1990). Another finding to emerge from Experiment 2 was that increasing the intensity of the color of the solutions did not result in an increase of participants’ flavor intensity evaluations. That is, the participant did not rate the double colored solutions as having a flavor that was significantly more intense than the standard colored solution. The concentration of coloring in the solutions did not influence participants’ ratings either when the solutions were appropriately colored or when they were inappropriately colored (cf. Johnson et al., 1982, 1983). It is worth noting that this null effect of color intensity on flavor intensity emerged despite the fact that an LMS scale, a rating procedure that M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 c LIME 1. 2 Perceived flavor intensity (log10) Perceived flavor intensity (log10) a 1. 6 0. 8 0. 4 0 -0 .4 -0 .8 -1 .2 981 STRAWBERRY 1. 6 1. 2 0. 8 0. 4 0 -0 .4 -0 .8 -1 .2 -1 .6 -1 .6 -2 -2 Greenstandard Greendoubled Orangestandard Orangedoubled Redstandard Reddoubled Greenstandard Colorless Greendoubled Color of the solutions b Orangestandard Orangedoubled Redstandard Reddoubled Colorless Reddoubled Colorless Color of the solutions d 1. 6 ORANGE FLAVORLESS Perceived flavor intensity (log10) Perceived flavor intensity (log10) 1. 6 1. 2 0. 8 0. 4 0 -0 .4 -0 .8 -1 .2 1. 2 0. 8 0. 4 0 -0 .4 -0 .8 -1 .2 -1 .6 -1 .6 -2 -2 Greenstandard Greendoubled Orangestandard Orangedoubled Redstandard Reddoubled Colorless Color of the solutions Greenstandard Greendoubled Orangestandard Orangedoubled Redstandard Color of the solutions Fig. 2. Mean log 10 flavor intensity ratings for the lime (a), orange (b), strawberry (c), and flavorless (d) solutions presented in Experiment 2. The error bars represent the between-participants standard errors of the means. has been argued to be less influenced by inter-individual differences between participants (see Green et al., 1993; Kurtz & White, 1998; this point will be discussed further below), was used. 4. General discussion The results of Experiment 1 demonstrated that solutions of certain colors are more often associated with particular flavors (than with the other flavor options presented on the flavor list) when people had to link the color of the solution with a flavor by inspecting it visually (but without tasting it). In particular, green-colored solutions were associated with lime flavor, the orange-colored solution with an orange flavor, the blue-colored solutions were frequently associated with a spearmint flavor, and the yellow-colored solutions with a lemon flavor (see Table 1). By contrast, the red- and grey-colored solutions were not so strongly associated with any particular flavor (rather, the red color appeared to have equally strong associations with a number of the flavor options that were provided on the response list). The participants associated red-colored solutions with the flavor of strawberry, raspberry, and cherry, while the grey-colored solutions were associated with the flavors of blackcurrant and liquorice more often than with any of the other flavors.3 The results of Experiment 2 showed that it is possible to impair flavor discrimination performance by coloring fruitflavored solutions inappropriately (cf. DuBose et al., 1980). In particular, the lime- and orange-flavored solutions were discriminated less accurately when the solutions were colored inappropriately (i.e., lime-flavored solutions colored orange or red, and orange-flavored solutions colored green or red). By contrast, the discrimination of strawberry flavor was not affected by the color of the solutions. That is, the inappropriate coloring of the strawberry-flavored solutions did not reduce the participants’ ability to recognize the actual flavor. One possible explanation for this result is that those flavors that are more strongly associated with 3 It is worth highlighting the possibility that the list of flavors provided to participants in the present study might have influenced the profile of their responses. In particular, it is possible that the specific color/flavor associations reported in Experiment 1 might to some extent have been prompted by the options available on the response list given to participants. For instance, there were many realistic, commonly available flavor options on the list that could be associated with the red color (e.g., apple, cherry, raspberry, strawberry, apple). By contrast, there were fewer possible flavor options that could sensibly be associated with the orange color. 982 M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 a color are more difficult to identify when presented in inappropriately colored solutions. The results of Experiment 1 demonstrated that the relationship between color and a specific flavor was stronger for the orange- and green-colored solutions than for the red-colored solutions. That is, the participants more often matched the orange color with the flavor of orange and the green color with the flavor of lime. By contrast, the red color was associated with strawberry, raspberry and cherry flavors. The results of Experiment 2, demonstrating the effect of inappropriate food coloring on orange- and lime-flavored solutions but not on strawberry-flavored solutions, would therefore seem to confirm the view that inappropriate colors decrease the detection of flavor only when there is a strong flavor–color association. It is important to note that the present results show that people can still be misled by the inappropriate coloring of a solution even if they know that the color does not provide a reliable guide to the flavor of the solution. By contrast, the participants in the majority of previous research in this area (e.g., DuBose et al., 1980; Johnson & Clydesdale, 1982; Morrot et al., 2001; Oram et al., 1995; Philipsen et al., 1995; Roth et al., 1988; Stillman, 1993; Zellner & Durlach, 2003) were not explicitly informed that the flavors of the solutions might not be paired with the appropriately colored solutions. It can therefore be argued that the apparent visual influence over flavor perception reported in many of these earlier studies might simply have reflected a decisional bias introduced by the obvious variation in the color cues (cf. the literature on the effectiveness of the color of medications on the placebo effect; e.g., de Craen, Roos, de Vries, & Kleijnen, 1996), rather than a genuine perceptual effect (i.e., of the color cues actually modulating flavor perception itself; though see also Garber, Hyatt, & Starr, 2001; for an alternative perspective from the field of marketing). In other words, if participants found it difficult to discriminate the identity of the flavor on the basis of gustatory and olfactory cues then they may simply have decided to respond on the basis of the more easily discriminable color cues instead. Therefore, it might be hypothesized that participants in these previous studies may have been influenced by decisional processes, whereas the design of the present experiment was more focused on uncovering any genuinely perceptual interactions by explicitly informing participants that the color–flavor link would often be misleading (i.e., that the solutions would frequently be presented in an inappropriate color; cf. Bertelson & Aschersleben, 1998). The results of Experiment 2 show that the modulatory role of visual information on multisensory flavor perception was so robust as to override any awareness that participants might have had (e.g., as informed by the experimenter) about the possibility that solutions were colored inappropriately. The second important result to emerge from the analysis of Experiment 2 was that changing the intensity of the colors added to the solutions did not change participants’ perception of flavor intensity even when the participants made their intensity rating responses using the LMS. The participants were not influenced by the different concentration of coloring in the solutions, either when the color was appropriate to the flavor or when it was not. As noted in Section 1, contradictory results have also been reported by researchers in previous studies regarding the influence of variations in color intensity on the perception of taste and flavor intensity, with some researchers reporting that perceived flavor intensity could be enhanced by increasing the level of color intensity (e.g., DuBose et al., 1980; Johnson & Clydesdale, 1982; Johnson et al., 1982, 1983; Pangborn, 1960); whereas others failed to confirm these results (e.g., Frank et al., 1989; Pangborn & Hansen, 1963; see Clydesdale, 1993; for a review). In Section 1, it was argued that one possible reason for the discrepancy reported in previous studies was that a numerical scale was used for rating taste or flavor intensity in all of the studies. As has been noted elsewhere, the use of such scales appears to be somewhat limited. In particular, greater inter-individual differences have been reported when numerical scaling methods are used, presumably because of the idiosyncratic number usage encouraged by the use of such scales (e.g., Green et al., 1993). Moreover, ceiling effects are often reported in studies using numerical scales and this can make the interpretation of results more difficult (Bartoshuk, 2000). In the present study, these methodological limitations were overcome by using an LMS to assess flavor intensity ratings (Bartoshuk, 2000; Green et al., 1993). Nevertheless, no effect of variations in color intensity on flavor intensity ratings by the participants were found (see the results of Experiment 2). It should also be noted that the results show quite a low variance for each flavor/color combination. It could be argued that this result might reflect a problem associated with the convergence of responses that has sometimes been reported to occur when using LMS scales. In particular, it has been pointed out that the use of the ‘Strongest Imaginable’ label has a tendency to suppress intensity ratings constraining them to a relatively small portion of the response scale (e.g., see Bartoshuk, Fast, & Snyder, 2005). Therefore, the possibility cannot be unequivocally ruled out that the use of the LMS scale by participants in the present study might have been in part responsible for the lack of effect of color intensity on flavor intensity ratings. Future studies should seek to address this issue. It is, however, worth noting that more intense coloring does not necessarily always equate with increased flavor intensity ratings in beverages. For example, Chan and Kane-Martinelli (1997) reported that the perceived flavor intensity for certain foods (such as chicken bouillon) was higher with the commercially available color sample than when the samples were given in a higher-intensity color (see also Clydesdale, 1993; on this point). Note also that if the discrepancy between the intensity of the color and the intensity of the flavor is too great participants may experience a disconfirmation of expectation (or some form of dissonance between the visually- and gustatorily-determined M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 flavor intensities) and the color and taste cues will no longer be linked (e.g., Clydesdale, 1993; cf. Ernst & Banks, 2002). Finally, it is also worth noting that the colors that are present in food and drinks (such as in the fruit-flavored solutions presented in the current study) might affect people’s flavor identification at both a perceptual and at a more semantic level. With regard to the nature of any semantic interactions between visual cues and flavor perception, the notion of redness could presumably be communicated to participants either by coloring the drink red, or else by packaging the drink in a red container (e.g., see Garber et al., 2000; Hine, 1995; Spence, 2007), or even by viewing the word red (cf. Demattè, Sanabria, & Spence, 2006; on this point). In future research, it will be interesting to discriminate between genuinely perceptual versus semantic crossmodal interactions in flavor perception. Acknowledgements This investigation was supported by a research grant from Unilever Research & Development. M.Z. was also supported by a returning grant ‘‘Rientro dei cervelli” from the MIUR (Italy). References Amerine, M. A., Pangborn, R. M., & Roessler, E. B. (1965). Principles of sensory evaluation of food. New York: Academic Press. Bartoshuk, L. M. (2000). Psychophysical advances aid the study of genetic variation in taste. Appetite, 34, 105. Bartoshuk, L. M., Fast, K., & Snyder, D. J. (2005). Difference in our sensory world: Invalid comparisons with labelled scales. Current Directions in Psychological Science, 14, 122–125. Bertelson, P., & Aschersleben, G. (1998). Automatic visual bias of perceived auditory location. Psychonomic Bulletin and Review, 5, 482–489. Blackwell, L. (1995). Visual cues and their effects on odour assessment. Nutrition and Food Science, 5, 24–28. Buettner, A., Beer, A., Hannig, C., & Settles, M. (2001). Observation of the swallowing process by application of videofluoroscopy and realtime magnetic resonance imaging-consequences for retronasal aroma stimulation. Chemical Senses, 26, 1211–1219. Bult, J. H. F., de Wijk, R. A., & Hummel, T. (2007). Investigations on multimodal sensory integration: Texture, taste, and ortho- and retronasal olfactory stimuli in concert. Neuroscience Letters, 411, 6–10. Chan, M. M., & Kane-Martinelli, C. (1997). The effect of color on perceived flavor intensity and acceptance of foods by young adults and elderly adults. Journal of the American Dietetic Association, 97, 657–959. Christensen, C. M. (1980a). Effects of taste quality and intensity on oral perception of viscosity. Perception and Psychophysics, 28, 315–320. Christensen, C. M. (1980b). Effects of solution viscosity on perceived saltiness and sweetness. Perception and Psychophysics, 28, 347–353. Clydesdale, F. M. (1993). Color as a factor in food choice. Critical Reviews in Food Science and Nutrition, 33, 83–101. Davis, R. (1981). The role of nonolfactory context cues in odor identification. Perception and Psychophysics, 30, 83–89. de Craen, A. J. M., Roos, P. J., de Vries, A. L., & Kleijnen, J. (1996). Effect of colour of drugs: Systematic review of perceived effect of drugs and their effectiveness. British Medical Journal, 313, 1624–1626. 983 Delwiche, J. (2004). The impact of perceptual interactions on perceived flavor. Food Quality and Preference, 15, 137–146. Demattè, M. L., Sanabria, D., & Spence, C. (2006). Crossmodal associations and interactions between olfaction and vision. Chemical Senses, 31, E50–E51. DuBose, C. N., Cardello, A. V., & Maller, O. (1980). Effects of colorants and flavorants on identification, perceived flavor intensity, and hedonic quality of fruit-flavored beverages and cake. Journal of Food Science, 45, 1393–1399, 1415. Duncker, K. (1939). The influence of past experience upon perceptual properties. American Journal of Psychology, 52, 255–265. Ernst, M. O., & Banks, M. S. (2002). Humans integrate visual and haptic information in a statistically optimal fashion. Nature, 415, 429–433. Frank, R. A., Ducheny, K., & Mize, S. J. S. (1989). Strawberry odor, but not red color, enhances the sweetness of sucrose solutions. Chemical Senses, 14, 371–377. Garber, L. L., Jr., Hyatt, E. M., & Starr, R. G. Jr., (2000). The effects of food color on perceived flavor. Journal of Marketing Theory and Practice, 8(4), 59–72. Garber, L. L., Jr., Hyatt, E. M., & Starr, R. G. Jr., (2001). Placing food color experimentation into a valid consumer context. Journal of Food Products Marketing, 7(3), 3–24. Green, B. G., & George, P. (2004). ‘Thermal taste’ predicts higher responsiveness to chemical taste and flavor. Chemical Senses, 29, 617–628. Green, B. G., Shaffer, G. S., & Gilmore, M. M. (1993). Derivation and evaluation of a semantic scale of oral sensation with apparent ratio properties. Chemical Senses, 18, 683–702. Hine, T. (1995). The total package: The secret history and hidden meanings of boxes, bottles, cans, and other persuasive containers. New York: Little Brown. Hodgson, M., Linforth, R. S. T., & Taylor, A. J. (2003). Simultaneous real-time measurements of mastication, swallowing, nasal airflow, and aroma release. Journal of Agricultural and Food Chemistry, 51, 5052–5057. Ishihara, S. (1943). The series of plates designed as tests for color blindness (9th ed. (Complete Edition)). London: H.K. Lewis & Co Ltd. Johnson, J. L., & Clydesdale, F. M. (1982). Perceived sweetness and redness in colored sucrose solutions. Journal of Food Science, 47, 747–752. Johnson, J. L., Dzendolet, E., & Clydesdale, F. M. (1983). Psychophysical relationships between sweetness and redness in strawberry-drinks. Journal of Food Protection, 46, 21–25. Johnson, J. L., Dzendolet, E., Damon, R., Sawyer, M., & Clydesdale, F. M. (1982). Psychophysical relationships between perceived sweetness and color in cherry-flavored beverages. Journal of Food Science, 45, 601–606. Kurtz, D. B., & White, T. L. (1998). Metrics of odorant dissimilarity: Labeled magnitude scale vs. magnitude estimation. Annals of the New York Academy of Sciences, 855, 638–640. Maga, J. A. (1974). Influence of color on taste thresholds. Chemical Senses and Flavor, 1, 115–119. Morrot, G., Brochet, F., & Dubourdieu, D. (2001). The color of odors. Brain and Language, 79, 309–320. O’Doherty, J., Rolls, E. T., Francis, S., Bowtell, R., & McGlone, F. (2001). Representation of pleasant and aversive taste in the human brain. Journal of Neurophysiology, 85, 1315–1321. Oram, N., Laing, D. G., Hutchinson, I., Owen, J., Rose, G., Freeman, M., et al. (1995). The influence of flavor and color on drink identification by children and adults. Developmental Psychobiology, 28, 239–246. Pangborn, R. M. (1960). Influence of color on the discrimination of sweetness. American Journal of Psychology, 73, 229–238. Pangborn, R. M., & Hansen, B. (1963). The influence of color on discrimination of sweetness and sourness in pear-nectar. American Journal of Psychology, 76, 315–317. Philipsen, D. H., Clydesdale, F. M., Griffin, R. W., & Stern, P. (1995). Consumer age affects response sensory characteristics of a cherry flavored beverage. Journal of Food Science, 60, 364–368. 984 M. Zampini et al. / Food Quality and Preference 18 (2007) 975–984 Roth, H. A., Radle, L. J., Gifford, S. R., & Clydesdale, F. M. (1988). Psychophysical relationships between perceived sweetness and color in lemon- and lime-flavored drinks. Journal of Food Science, 53, 1116–1119, 1162. Spence, C. (2002). The ICI report on the secret of the senses. London: The Communication Group. Spence, C. (2007). Creating innovative packaging that appeals to all the senses. Paper presented at the PACE: Packaging and converting executive forum meeting. Paris, 8–11 February. Stillman, J. (1993). Color influences flavor identification in fruit-flavored beverages. Journal of Food Science, 58, 810–812. Stillman, J. A. (2002). Gustation: Intersensory experience par excellence. Perception, 31, 1491–1500. Strugnell, C. (1997). Colour and its role in sweetness perception. Appetite, 28, 85. Vickers, Z. M. (1983). Pleasantness of food sounds. Journal of Food Science, 48, 783–786. Wheatley, J. (1973). Putting color into marketing. Marketing, 67, 24–29. Zampini, M., & Spence, C. (2004). The role of auditory cues in modulating the perceived crispness and staleness of potato chips. Journal of Sensory Studies, 19, 347–363. Zampini, M., & Spence, C. (2005). Modifying the multisensory perception of a carbonated beverage using auditory cues. Food Quality and Preference, 16, 632–641. Zellner, D. A., Bartoli, A. M., & Eckard, R. (1991). Influence of color on odor identification and liking ratings. American Journal of Psychology, 104, 547–561. Zellner, D. A., & Durlach, P. (2003). Effect of color on expected and experienced refreshment, intensity, and liking of beverages. American Journal of Psychology, 116, 633–647. Zellner, D. A., & Kautz, M. A. (1990). Color affects perceived odor intensity. Journal of Experimental Psychology: Human Perception and Performance, 16, 391–397. Zellner, D. A., & Whitten, L. A. (1999). The effect of color intensity and appropriateness on color-induced odor enhancement. American Journal of Psychology, 112, 585–604.
© Copyright 2025 Paperzz