Background

Shape can play an important role in our perception of food. Many studies have shown that people can reliably match basic tastes to shapes with varying degrees of angularity (see [1], for a review). Most prominent amongst these correspondences is the association between round shapes and sweetness [2–4]. Round shapes have even been shown to increase participants’ sensitivity for sweetness at threshold levels, compared to angular shapes [5, 6]. In the realm of chocolate, people from Western cultures tend to associate chocolate with a higher cacao content (e.g. dark chocolate, which tend to be more bitter) with more angular shapes and chocolate with lower cacao content (e.g. milk chocolate, which tend to be sweeter) with rounder shapes [7–9]; interestingly, the remote Himba tribe in Northern Namibia makes the opposite mapping, with milk chocolate mapped to more angular shapes [10]. Food perception can even be influenced by the shape of the plate it is served on. For instance, different foods have been shown to taste sweeter when sampled from a round plate than from a square one ([11, 12]; see [13], for a review; though see [14] for a null effect of plate shape on taste ratings).

Furthermore, such shape-taste associations can influence consumer expectations of food products. For instance, round packaging has been shown to give rise to higher expected sweetness than packaging that is more angular (e.g. [15]; see [16], for a recent review). Speaking more generally, round packaging tends to be preferred over angular packaging [17]. People also expect a sweeter-tasting product on viewing a rounder design and more sour tastes when exposed to a more angular design ([18]; see also [16, 19] for reviews). The ability for shape to influence expectations may be explained via affective priming. Round shapes are perceived as more pleasant than angular shapes [20, 21]. Furthermore, there is electrophysiological evidence that geometric shapes may automatically activate an emotional response [22].

Finally, such sensory expectations have been shown to alter the sensory perception and hedonic evaluation of the actual food product (see [23, 24] for reviews). Many theories have been put forward to account for the effect of disparity between expectations and the subsequent eating experience. In particular, the theory of assimilation contrast has been used by food science researchers [25]. According to the theory, if the difference between expectation and reality is within the consumer’s limit of acceptance, the consumer would, consciously or otherwise, try to change their perception of the product to bring it in line with expectations. On the other hand, if the difference is sufficiently great, the consumer exaggerates the difference between expectations and reality, and their product evaluation shifts in the opposite direction than originally expected [26, 27]. For instance, a slight understatement of quality in advertising might lead to higher consumer satisfaction with the product, and an overstated promise of quality for a poor product might lead to greater customer disappointment.

With all this in mind, it certainly seems plausible that a change in shape might influence the way in which people perceive and experience food. We chose to focus on chocolate since it is a popular food item that is sold in a variety of shapes. In the present study, therefore, we set out to test the hypothesis that shape might alter both the expected and actual sensory and hedonic experience of eating chocolate.

Results

Expectations

The RM-MANOVA test revealed a significant main effect of shape (see Fig. 1) on participants’ ratings of the chocolates’ expected flavour attributes and preference (F(4,97) = 14.64, p < 0.001, Wilks’ lambda = 0.62), but no effect of cacao content (F(4,97) = 1.52, p = 0.20). Further ANOVAs revealed significant effects of chocolate shape on measures of sweetness (F(1,100) = 33.62, p < 0.001, η partial 2 = 0.25), bitterness (F(1,100) = 23.41, p ≤ 0.001, η partial 2 = 0.19) and creaminess (F(1,100) = 50.56, p < 0.001, η partial 2 = 0.34), but not for chocolate liking (F(1,100) = 0.03, p = 0.87). In general, round shapes were rated as sweeter, less bitter and creamier than the angular-shaped chocolates.

Fig. 1
figure 1

Two shapes of chocolates used during the study. The round shape was hemispherical while the angular shape was a seven-sided prism. Note that the 71 and 80% cacao chocolates appear to have the same colour

Full size image

Post-tasting ratings

The RM-MANOVA test revealed a significant main effect of cacao content (71 or 80%) on participants’ chocolate ratings (F(4,97) = 3.51, p = 0.01, Wilks’ lambda = 0.87), but no effect of shape was found (F(4,97) = 1.57, p = 0.19). Further ANOVAs revealed significant effects of cacao content on bitterness ratings (F(1,100) = 7.50, p = 0.007, η partial 2 = 0.07), where the 71% chocolate was rated as more bitter than the 80% chocolate. There were no significant effects of cacao content on sweetness (F(1,100) = 2.58, p = 0.11), creaminess (F(1,100) = 0.65, p = 0.42) or chocolate liking (F(1,100) = 0.10, p = 0.76).

Expected vs. post-tasting ratings

The RM-MANOVA test revealed significant main effects of rating type (F(4,98) = 6.61, p < 0.001, Wilks’ lambda = 0.79). Further ANOVAs revealed significant effects of rating type (expected or actual) on measures of sweetness (F(1,101) = 21.27, p < 0.001, η partial 2 = 0.17), bitterness (F(1,101) = 6.37, p = 0.01, η partial 2 = 0.06) and liking (F(1,101) = 9.39, p = 0.003, η partial 2 = 0.09), but not for creaminess (F(1,101) = 1.50, p = 0.22).

More specifically, the RM-MANOVA test also revealed a significant interaction effect between rating type and shape (F(4,98) = 7.73, p < 0.001, Wilks’ lambda = 0.76) (see Fig. 2), with regard to sweetness (F(1,101) = 19.71, p < 0.001, η partial 2 = 0.16), bitterness (F(1,101) = 12.24, p = 0.001, η partial 2 = 0.11) and creaminess (F(1,101) = 22.27, p < 0.001, η partial 2 = 0.18). Further post hoc testing showed that the round chocolates were less sweet, more bitter, less creamy and liked less than expected (p < 0.01 for all comparisons); on the other hand, angular chocolates were more creamy than expected (p = 0.05). These ratings are shown in Fig. 2.

Fig. 2
figure 2

Mean values of a expected and b actual ratings of sweetness, bitterness, creaminess and liking, for both angular and round chocolate shapes. Error bars indicate the standard error. Asterisk ‘*’ indicates statistical significance at p < 0.05

Full size image

Discussion

The results of the present study demonstrate that, compared to angular-shaped chocolate, participants expected the rounder chocolate to be significantly sweeter, less bitter and creamier. However, when it comes to post-taste chocolate ratings, there were no significant differences between shapes in the present study. Furthermore, the round chocolates were significantly less sweet, more bitter, less creamy and liked less than expected, while the post-taste ratings for angular chocolates were more creamy than expected, but otherwise not significantly different from expectations. This discrepancy between expectations and reality could have induced a contrast effect, whereby the participants would have further magnified the difference between expectations and reality (see [26, 28]). The fact that the round chocolate was liked less than expected is also in line with the contrast hypothesis, according to which the disconfirmation of expectations would lead to a decreased hedonic evaluation [29]. Overall, these results are consistent with the literature showing that round shapes can prime specific taste/flavour expectations, which could then go on to influence the subsequent tasting experience [15].

While the rounder-shaped chocolates did indeed prime expectations that chocolate will taste sweeter and creamier, those expectations did not carry over to influence the actual tasting experience. One possible explanation is that the chocolate samples used in the present study were simply too dark and differed too much from participants’ expectations. As a case in point, the mean expected sweetness of the round chocolates was 4.64/7, whereas the actual sweet rating was only 3.48/7. As we have seen in previous studies, expectations have a greater influence on sensory perception when the discrepancy between expectations and reality is not too large [30, 31]. In future research, it would be worth repeating the present study with exactly the same shapes, but using a sweeter milk chocolate, in order to test for an assimilation effect of the enhanced sweetness and creaminess ratings.

The fact that the 71% chocolate was rated to be more bitter than the 80% chocolate might initially seem surprising, but the effect is small [32], and we did not control for supertaster status between the two groups of participants who tasted 71 vs. 80% see (Fast K: Developing a scale to measure just about anything: comparisons across groups and individuals, unpublished M.D. thesis, for evidence of differing chocolate bitterness ratings as a function of taster status, and [33], on the general problem of between-group comparisons of perceived taste intensity given differences in taster status). Another plausible explanation is that participants might have found the 71% chocolate to taste more acidic and confounded acidity in the chocolates with bitterness (e.g., [34]). Furthermore, it is plausible that the crystalline structureFootnote 1, and therefore the mouthfeel, of the chocolates could have varied across samples. Although, as the chocolates were produced in the workshop of a master chocolatier, it is likely that chocolate texture was consistent across all the samples.

Another potentially confounding factor in the present study is whether the expectation values were driven by angularity or asymmetry, since the angular chocolate used here also happened to be asymmetrical. It has been shown that, along with angular shapes, asymmetrical shapes also tend to be associated with unpleasantness and sourness (as opposed to pleasantness and sweetness) by participants from both the UK and Colombia [35]. Therefore, it would be interesting in future studies to control for symmetry when selecting chocolate shapes.

Conclusions

In summary, the present study demonstrates the importance of product shape on setting consumer expectations. Chocolates with round shapes were expected to be sweeter, less bitter and more creamy than angular-shaped chocolates. In the age of constant product upgrades, it is increasingly important for food manufacturers and designers to take such considerations into account. For instance, Cadbury has recently updated two of its products to have smoother contours, with the claim that the new smoother contours will be a better fit for the mouth [36]. In addition, Cadbury claims the new shapes allow the chocolates to melt in the mouth slightly before biting, for maximum flavour and prolonged enjoyment (see [37] for evidence of chocolate shape's influence on texture and flavour perceptions). It remains to be seen how this new product updating will affect the overall consumer’s experience (see [38]).

Methods

Participants

One hundred two participants (58 women, 44 men) aged between 16 and 74 years (M = 35.64, SD = 16.81) took part in the study. All of the participants gave their informed consent prior to taking part. The participants did not have a cold nor any other known impairment of their sense of smell, taste or hearing at the time of the study, by self-report. The participants were informed that they would be tasting chocolate before the start of the study.

Food stimuli

Two bitter chocolate formulas were prepared for use in this study (basic ingredients: cocoa mass, sugar, cocoa butter and natural vanilla flavour). One formula had 71% cocoa and the other 80%. Furthermore, each formula was presented in two different shapes, one rounded and the other angular (see Fig. 3). The chocolates were developed at The Chocolate Line factory in Bruges, under the supervision of the award-wining Belgian chocolatier Dominique Persoone (www.thechocolateline.be). Note that all of the chocolate samples had the same dark brown colour (although due to its geometry, the round shape looks more shiny) and similar volume (approximately 2.5 cm3).

Fig. 3
figure 3

Mean values of pre-taste (expected) and post-taste ratings of sweetness, bitterness, creaminess and liking, for a angular and b round chocolate shapes. Error bars indicate the standard error. Single asterisk ‘*’ indicates statistical significance at p ≤ .05 and double asterisks ‘**’indicate statistical significance at p < .005

Full size image

Design

The study was approved by the Social Ethics Committee at KU Leuven-SMEC (Protocol G-2016 03 519). Participants were given two pieces of chocolate with different shapes (but with identical formulas) and rated, first, the expected sweetness, bitterness, creaminess and liking for both pieces of chocolate. After tasting, they repeated the same ratings for both chocolates. Half of the participants tasted the 71% cacao formula and the rest tasted the 80% cacao formula.

Procedure

The ninth floor of Music Instrument Museum in Brussels (MIM) was chosen as the site for the study. Due to its independent location inside the museum, being located between the museum’s restaurant on the top floor and the rest of the exhibitions below, it was possible to maintain a reasonably well-controlled environment during the study.

Each participant was seated in front of a computer screen and given two chocolates on labelled plates (1 and 2), tap water, a computer mouse and a keyboard to interact with the survey. The participants were randomly assigned to taste either the 71% or the 80% chocolates. First, the participants were instructed to look at (but not taste) the first piece of chocolate and evaluate its expected sweetness, bitterness, creaminess and liking, on 7-point Likert scales. They then moved on to the second piece of chocolate and repeated the procedure. Next, the participants were instructed to go back to the first piece of chocolate, taste it and then rate its sweetness, bitterness, creaminess and how much they liked it. After rinsing their mouths out with water to cleanse their palates, the participants then tasted and rated the second piece of chocolate. The order in which the chocolates were tasted was randomised amongst participants.

Together with the written guidelines concerning the study, at least one supervisor was present during the study in order to provide guidance and support. Upon finishing the study, the participants were then instructed to leave the room without discussing any details with the next group of participants. The study lasted for around 10 min.

Data analysis

For both expected and post-tasting chocolate ratings, a repeated-measures multivariate analysis of variance (RM-MANOVA) was conducted on the measures of sweetness, bitterness, creaminess and liking. Shape (round or angular) was the within-participant factor, and cacao content (71 or 80%) was the between-participant factor. Expected chocolate ratings were also compared to the post-tasting ratings via a RM-MANOVA, with rating type (before or after tasting) and chocolate shape as within-participant factors and sweetness, bitterness, creaminess and liking as dependent variables. All post hoc pairwise comparisons are Bonferroni corrected.