The Influence of Color on the Perception of Taste of People with Autism Spectrum Disorders

Griber Yuliya Alexandrovna Doctor of Cultural Studies Professor, Director of Color Laboratory at Smolensk State University 214000, Russia, Smolensk region, Smolensk, Przhevalskogo str., 4 y.griber@gmail.com Other publications by this author

Elkind Grigorii Vitalevich Lecturer, Smolensk Teacher Training College 214018, Russia, Smolensk region, Smolensk, Rayevsky str., 2 spedkoll@mail.ru

IntroductionWhen we see food, the color of the products or packaging in most cases is for us the only source of information about its qualities.

Focusing solely on color, we often draw conclusions about the composition, nutritional value and usefulness (see, for example: ^[1]).

The strongest connections are noted between color and taste (see the review of studies ^[2]). Even with a cursory glance at an edible product, involuntary associations and expressed expectations about the connection between color and taste are involved. For example, we expect that red drinks are likely to taste like strawberries or cherries, and green ones – lime, mint or apple ^[3].

Such associations are cross–modal correspondences - the transfer of the sensation of one modality (visually perceived color) to another (taste). In most cases, they are built on the principle of "color replaces the object" and are formed as a result of observation of statistical patterns of the surrounding world. As a rule, these are the most common pairs between colors and flavors (4; 5). The meaning of such connections is acquired as a result of associative learning (see for example: ^[6]): we see again and again that cucumber is green, and gradually the taste of cucumber begins to be associated with this color.

Over time, on the basis of previous experience, a special color identity is formed for each taste, some tastes are more intense (for example, the taste of strawberries), others are less pronounced (for example, the taste of chicken) (see podr.: ^[4]).

Experimental studies of recent decades have shown that intentional or unintentional violation of the expected connections between color and taste significantly affects cognitive functions. If you change the color of a well-known product, it becomes difficult for a person to find out what kind of taste he is actually experiencing (see for example: ^[7]). The changed color noticeably affects how we describe the taste of the product. For example, if white wine is colored red, even professional sommeliers begin to use "red" reference objects to characterize the taste - they compare it with cherries, plums, black pepper ^{[8; 9]}.

Violation of expected connections also affects visual search. In this case, an effect similar to the Stroop effect is observed: a person searches for a product with the right taste longer if the color of its packaging does not match the taste (for example, blue for strawberries), and finds it much faster when the color of the packaging meets his expectations (for example, for strawberries – red or pink) ^[10].

Of particular interest for the study of cross-modal correspondences between color and taste is the study of similar associations in people with autism spectrum disorders.

Due to color vision disorders, the specifics of sensory processing, characteristic hypersensitivity to visual information in general and the lack of multisensory integration, people with autism perceive colors differently from neurotypical people (see e.g.: ^{[11; 12]}). They distinguish shades worse, but they remember them better ^[13]. They form a different structure of color preferences and affective reactions to color ^[14]. Extreme affective reactions are often observed – obsession with a certain color (in many cases, green) (see, for example: ^[15]) or, conversely, aversion to certain shades (primarily yellow) ^[16] and even phobias ^[12]. Researchers associate the characteristic extreme reactions to color with increased sensitivity to sensory stimulation in autism in general, which is enhanced under the influence of certain chromatic characteristics (for example, due to the high brightness of yellow) ^[16]. Color obsessions and phobias in people with autism spectrum disorders are also explained by their unusually strong ability to associate colors with objects, as well as uneven perception of shades (hypo- and hypersensitivity) in various areas of the color space ^[12].

Despite the large number of studies devoted to various aspects of processing color information by people with autism spectrum disorders, the study of color–related taste expectations in autism is a relatively new topic for cognitive research. The only empirical study known to us is an online experiment with Japanese respondents in which people with autism spectrum disorders compared 5 basic tastes (sour, sweet, salty, bitter and umami) with 11 color stimuli corresponding to the main color categories (black, blue, brown, green, gray, orange, pink, purple, red, white, yellow) ^[17]. The study showed that in people with autism spectrum disorders, associations between color and the five main tastes are less conventional, which the experimenters explain by the reduced effect of prior knowledge during associative learning. The most noticeable connections were found in pink – with sweet, yellow – with sour, green – with bitter, blue – with salty and bitter, red – with umami.

Continuing the study of cross-modal correspondences in people with autism spectrum disorders, we decided to change the format of the study: on the one hand, to make it free and, due to this, significantly expand the range of associations included in the analysis (see subsp.: ^{[18; 19]}), on the other, to use non–abstract color patterns as stimuli samples, and packaging layouts, thus increasing the accuracy of perceptual images.

The object of the study is people with autism spectrum disorders, the subject is the expectations of taste that are formed in them when they perceive a certain color.

The hypothesis of the study is that due to the specifics of sensory processing and the lack of multisensory integration, crossmodal correspondences between color and taste in people with autism will differ from associations of neurotypical people not only in the composition and prevalence of individual connections, but also in quantity, cognitive significance and semantics of choice.

MethodParticipants.

The experiment involved 20 respondents with autism spectrum disorders (7 men and 13 women) aged 18 to 20 years. Their responses were compared with the results of the control group (N=20), with the same socio-demographic characteristics.

Incentives. To conduct the study, we used computer stimuli created according to a template for the packaging design of a well-known brand of a bar, from which all information about the product and manufacturer was removed. The experimental stimuli had five different colors – green, yellow, red, pink and blue (Table 1) – and were shown to participants on a Philips 271V8L/01 computer monitor with a screen diagonal of 27 inches.

Table 1. Chromatic characteristics of experimental stimuligreen

Experiment procedure. In each of the groups, participants were asked to choose among the bars on the screen one that they would eat first. Then they were asked to explain the reason for their choice and name the taste with which they associate the color of the selected package. After that, the selected bar disappeared from the screen, the participants were asked to make a choice again, again answer the question about the reason for their preference and determine the taste. The procedure was repeated until a single stimulus remained on the screen. At the end of the experiment, all five stimuli reappeared on the screen and the subject was asked to indicate the bar that, in his opinion, looks more useful than all the others, is associated with healthy food.

The experiment was conducted with each participant separately. The order of presentation of stimuli of different colors in all three groups was random. The participants' answers were recorded on a dictaphone.

Data analysis. To assess the degree of diversity of the obtained color associations, the Margalef index was used in the study (see appendix: ^[19]), which was calculated by the formula:

d=(s–1)/lnN,

where s is the number of types of associations, N is the number of responses.

The cognitive significance of associations was assessed using the index (CSI) proposed by W. Sutrop for the analysis of data from experiments with the compilation of lists ^[20] and well-proven in experimental studies of color names (see, for example: ^[21]). This indicator was calculated using the formula:

CSI(i) = ni  / N /mri,

where ni is the number of people who have a certain associative relationship (for example, yellow – banana), N is the number of participants, mri is the average rank of a specific association between color and taste.

The cognitive significance index shows how important this or that associative connection is in a certain culture. In relation to our data, it is convenient because it allows us to take into account two important indicators at once: (1) the frequency of occurrence of a certain association of color with taste and (2) the average rank of this association in the list. The more common the association is and the higher its rank, the greater the cognitive significance.

The cognitive significance index can take values in the range from 0 to 1. At the same time, the maximum value (CSI = 1) means that all participants used a certain association (ni = N) and each named it first (ri = 1).

Visualization of the frequency of the most popular tastes (Fig. 2) was carried out using an online service WordClouds.com , designed to create a "word cloud" – a weighted list in which the font size corresponds to the frequency of occurrence of a certain association in the data array.

ResultsSince each participant of the experiment associated color with taste 5 times and explained his choice the same number of times, in each of the groups we received 100 associations between color and taste and the same number of ratings.

(1) The length of the lists and the variety of associations. At the first stage of the analysis, for each of the groups, we compiled a list of all the tastes with which the participants associated colors.  In the group of participants with autism spectrum disorders, the list was significantly longer than in the control group. In the experimental group, we recorded 51 different associations, 18 of which (35%) were nonuniform. The list of participants from the control group included 41 associations, of which 9 (22%) were nonuniform. The Margalef index, which we used to compare the diversity of the obtained associations, in the experimental group was dE = 10.86 compared with dQ = 8.69 in the control group (Table 2).

Table 2. Margalef Diversity IndexMargalef index, d

We also compared the variety and length of lists compiled separately for each of the five colors. The participants named the most flavors (12) in both groups for red (dE = dK = 3.67). The shortest lists of associations with green (8 and 5 flavors, respectively) (dE = 3.00; dK = 2.00). At the same time, the lists differed fundamentally in the number of repeated (non–single) connections: in the control group, the green color had only one non-single connection, the other colors had two; in the experimental group, the number of repeated connections for each color was at least twice as large (Fig. 1).

Figure 1. Number of associations of color with taste in the experimental (left) and control (right) groups

(2) The contents of the lists.

The lists compiled for the two groups differed markedly in content (Fig. 2). In the control group, all colors except blue and pink were associated exclusively with natural vegetable flavors. Participants from the control group almost always used fruits and vegetables, nuts and berries as reference objects. In people with autism, colors in many cases were associated with products that could not be hidden behind the shown packaging (ketchup, borscht, meat, semolina porridge, whipped cream). Their list included more artificial food (jelly, jelly, candy, cake). They called abstract tastes (bitter), drinks (coffee, water) and even concepts that denote something completely inedible (blood, rotten vegetables).

Most often, to characterize the taste, participants from both groups named apple, strawberry, raspberry, blueberry, banana. For the same taste associations, the most intense connection with a specific color was noted (more than 30% of respondents).

Figure 2. Associations between color and taste in the experimental (left) and control (right) groups; word size shows its frequency(3) Frequency of occurrence of associations.

At the next stage of the study, we compared the compiled lists of associative pairs by the frequency of associations represented in them and determined the dominant ones, highlighting in the ratings those connections that in each of the groups accounted for half of the responses. In people with autism, the group of dominant associations was 2 times larger than in the control group: 10 and 5 associations, respectively. In the control group, the list included one association for each color: green was associated with an apple in 16 people, blue was associated with the taste of blueberries in 10 people, yellow was associated with a banana, and pink with raspberries, for 8 participants red was the taste of strawberries. The same couples were also in the lead in the ranking of participants with autism spectrum disorders. However, in addition to them, the list includes associations of yellow with lemon, blue with plum and coffee, red with blood, pink with jelly. At the same time, only the association of green with apple was repeated in at least half of the participants (12).

(4) Cognitive significance of associations. To compare the structure of associative fields and identify the main and less significant connections in it, we calculated cognitive significance indices (CSI) for each of the pairs of associations in both groups (Fig. 4).

In the experimental group, seven flavors turned out to be two- and even three-color. These are apple (green, red and yellow), blueberry (blue and pink), strawberry (red and pink), candy (pink and yellow), watermelon (red and green), grapes (blue and green), chocolate (blue and red). At the same time, in two cases (candy and chocolate), the participants of the experiment most likely associated the taste with the color of the packaging. The color image of the watermelon turned out to be interesting, the taste of which was associated not only with edible red flesh, but also with inedible green peel among participants with autism.  In the control group, we found only 4 associations of the same taste with different colors: apple (green and red), strawberry (red and pink), banana (yellow and blue), orange (red and yellow). In all cases, the association of taste with at least one of the two colors was single and, accordingly, had much less cognitive significance.

Figure 3. Cognitive significance Index (CSI) in the experimental (top, square) and control (bottom, circle) groups; the color of the marker corresponds to the color of the stimulus(5) Semantics of choice.

The distribution of color preferences among participants from both groups for the first choice (Fig. 4 on the left) completely coincided: the most preferred were green and pink, which most respondents associated with apple and raspberry, respectively. At the same time, participants from different groups explained their choice in fundamentally different ways. In the control group, the main reasons for the choice were sympathy for the color (favorite color; color that you like; pleasant color) and a physiological or affective response to its chromatic characteristics (bright, saturated, interesting color). The experimental group was dominated by atmospheric associations (sun color, summer color, night color, wall color, shirt color) and associations with tastes (sweet, delicious color).

Figure 4. Color preferences of the participants of the experimental (dense fill) and control (hatching) groups; the color of the segments corresponds to the color of the stimuliThe colors that aroused the least interest and which the participants of the experiment chose last were noticeably different in the two groups (Fig. 4 in the center):

in the experimental group, it was pink and blue; in the control group, it was pink and yellow. However, the reasons for choosing this time coincided: in both groups, the participants did not like this color for some reason, it seemed ugly and dark.

As in previous studies (see, for example, ^[22-25]), in both groups, the overwhelming majority of respondents (16 out of 20) associated green with healthy food (Fig. 4 on the right). Most of the participants explained their choice by saying that green is the color of plants, foliage, grass and greenery, vegetables and fruits. In the experimental group, associations with summer and nature, tranquility, kindness and harmony, inspired by memories, were again added to the reasons for choosing.

Discussion and conclusions

The experiment showed that when perceiving color, people with autism spectrum disorders have fundamentally different expectations of taste compared to neurotypical people.

The main difference is that colors in most cases cause them not conventional logical associations with taste, which dominate in the control group, but creative – expressive and hidden. This is indicated by a large number of original (non-repeating) the connections between taste and color, and the "multicoloredness" of some associations, and the presence of "inedible" concepts in the list of associations, which at first glance are not related to taste in any way.

Along with statistically determined crossmodal correspondences dominating in the control group, associations in people with autism spectrum disorders in many cases have an emotional nature. In contrast to the control group, abstract color preferences have a noticeable effect on the choice of color associated with products in autistic people. In particular, as in abstract preferences, the least chosen color is yellow, which, according to the results of previous experiments, quite often causes extreme negative reactions in people with autism spectrum disorders (cf.: ^[16]).

Finally, the expectations of taste from color in the participants of the experimental group are much less influenced by the context, the importance of which in the cognitive mechanism of associative connections, researchers prove from the point of view of the theory of semantic discriminativeness ^{[26; 27]}. Under the influence of context, the correlation of color with a certain concept is often determined on the basis of other stimuli in the comparison group, rather than based directly on the strength of the underlying association. This is clearly seen in the control group: given that the packaging of the bar is on the screen, neurotypical participants did not associate color with all possible tastes, but tried to choose only those that are acceptable for the bar. In people with autism, the rule of semantic discriminativeness does not work so strictly, and their associations of color with taste are much more free.

The data obtained on the number and composition of cross-modal correspondences between taste and color in people with autism spectrum disorders, the prevalence of individual associations, their cognitive significance and semantics of choice are important for understanding the cognitive mechanisms of color exposure and establishing causal relationships between a given color stimulation and its effect on an individual. The results can be used in clinical practice, will be in demand in the education system and the activities of social institutions to create a barrier-free color environment.