Taste experience can be introduced to different types of virtual environments, ranging from curved computer-based screens to AR/VR HMDs (Wang et al., 2021). There is an increasing trend of using virtual environments in human-food interaction research because they allow researchers to easily add environmental features or properties and minimize certain negative effects of real settings (Bangcuyo et al., 2015; Delarue et al., 2019). For instance, compared to traditional booths, the hedonic data collected in a virtual coffee house can be more identifiable and reliable as a predictor for future coffee evaluation (Bangcuyo et al., 2015). Delarue et al. (2019) have shown that a multi-sensory immersive room could minimize the differences in product testing results produced at different periods in a lab setting. Moreover, introducing the taste experience to VR also benefits the immersive experience. Eating and drinking unify both virtual and physical sensory cues (Spence, 2016). Thus, they can increase users’ presence in the virtual environments, i.e., their sense of being there (Slater and Wilbur, 1997). Wang et al. (2021) also suggested that introducing physical interaction mechanisms that the users are used to in the physical world into the immersive virtual environments can enhance the psychological feeling of presence. Furthermore, Gallace et al. (2012) stated that increasing the number of senses stimulated in VR can increase users’ enjoyment, memorability, and presence. Since taste is one of the major sensory modalities familiar to people in real-life settings, it is beneficial to have it in VR. Eating and food-orientating activities may also become integral to the virtual experience in social contexts as VR becomes more available and allow people to meet on platforms like VRChat ² for socialization and Horizon Workrooms ³ for business meetings.

On the other hand, taste-related results in virtual environments may be different from those obtained in physical environments. For example, the virtual environment may significantly influence users’ hedonic responses to foodstuffs. Beef can be rated higher in terms of liking when consumed in a VR restaurant compared to a traditional sensory laboratory condition (Crofton et al., 2021). Stelick et al. (2018) also found that VR had a significant influence on the taste perception of blue cheese, especially the level of pungency. Ledoux et al. (2013) showed that food craving produced by VR was marginally greater than a neutral cue in a physical setting.

Cross-modal correspondence (CMC) is about the interactions among different senses. People combine information captured from various senses to form a more comprehensive view of the external world or activity they are doing (e.g., when people see and smell food in front of them). Interaction between these systems is an essential part of this process (Stein and Meredith, 1993). In particular, based on CMC, people can map features or sensations in one sensory modality onto features of other modalities in a consistent manner (Spence, 2011). For example, the taste of sweetness can correspond to a round shape (visual) or high pitch (audio). Surprisingly, CMC is not confined to one group (cultural or otherwise) but is often universally shared (Knöferle and Spence, 2012). Taste is one of the most common sensations that people experience daily. Basic taste perceptions include sweetness, sourness, bitterness, saltiness, and savor (umami) (Keast and Breslin, 2003), which can often be formed with other senses, such as smell or sound (Knöferle and Spence, 2012; Kerruish, 2019; Halabi and Saleh, 2021).

Color is an essential aspect of CMC between vision and taste. Color-taste correspondence is of great interest to the food industry, especially in helping design food product packaging because of its direct influence on consumers’ expectations (Cardello, 1994; Koch and Koch, 2003). To a large extent, it serves as a sensory property of the product before its consumption (Spence and Velasco, 2018). Color is a focused factor that significantly affects taste perception (Piqueras-Fiszman and Spence, 2014). For example, from the studies of color itself, Koch and Koch (2003) have stated that red is connected to sweetness while yellow is connected to sour, citrus, and fruity tastes. In addition, Wan et al. (2014) have shown that some strong CMCs are similar in different cultures, such as mapping black with bitterness and pink with sweetness.

Factors influencing multi-sensory flavor perception can be divided into either intrinsic to the food (e.g., texture, aroma) or extrinsic to it (e.g., related to the packaging, receptacle, or external environment) (Wang et al., 2019), both of which have been studied by previous research on color-taste CMC. When color represents the original feature of the foodstuff, it is considered an intrinsic visual stimulus feature (Spence, 2019b). For example, Johnson and Clydesdale (1982) found that applying red color to water could increase the sweet taste. Similarly, Maga (1974) showed that green statistically increased sweet taste threshold sensitivity while yellow decreased taste sensitivity. Different from intrinsic features, extrinsic factors are outside of the food itself. For example, Piqueras-Fiszman and Spence (2012) have shown that orange and dark-cream cups could enhance the chocolate flavor of a drink. Harrar et al. (2011) investigated popcorn in four different colored bowls and found that the sweet popcorn, in addition to being sweet, was perceived as saltier when consumed in a colored bowl than in a white bowl.

Specifically, we noticed that the color of ambient light is a type of extrinsic factor that can influence taste. Spence and Carvalho (2020) stressed the importance of the environment where the beverages are consumed. The ambient color of an environment could significantly affect the beverage’s flavor and taste (Oberfeld et al., 2009). Oberfeld et al. (2009) showed that wine was perceived to be spicier in blue or green ambient lighting than in red or white. Spence et al. (2014) explored people’s taste perception of red wine in a room that changes ambient color and music and found a significant difference in taste and beverage liking. Their results provide empirical support for the claim about the influence of ambient color on flavor from Oberfeld et al. (2009).

In general, there are three possible mechanisms underpinning the CMCs (Spence, 2011; Parise and Spence, 2013): 1) structural correspondence, described from the neural science side, is the byproduct of the innate cognitive system; 2) statistical correspondence, describes a learned process associating the established experience and the environment; and 3) semantic correspondence, stresses that the same terms can be used to describe different stimuli linguistically. For color-taste CMC, there are mainly two types of specific explanations for their correspondence. The first type is the association with the source object(s), meaning that the color-taste pairs result from associating specific colors with certain gustatory cues (Wang et al., 2020). Moreover, Spence and Levitan (2021) found that some color-taste associations are not limited to a particular source object. The second explanation is users’ emotional mediation. For example, the pink-sweet association arises because both stimuli are linked with happiness independently and are connected by people (Spence and Levitan, 2021). It is important to note that some colors with positive emotional valance in abstract scenarios may lead to negative emotions when matching with food. For instance, blue is a favorite color to many people, but blue-colored food can trigger negative feelings (Spence, 2019a).

The color-related research in the physical world has inspired some researchers to apply them in VR. Wang et al. (2020) designed an experiment manipulating the color of coffee in VR so that their participants saw either a dark brown or light brown liquid as they consumed the real coffee. Similarly, Huang et al. (2019) designed three experiments to study the influence of actual tea color simulation in VR on users’ taste ratings of Chinese red and green tea. Moreover, Chen et al. (2020)’s study went beyond colors. They created a sweet, bitter, and neutral scene integrating shape, color, and visual textures, and assessed their influence on perceived sweetness and environment liking. Results showed that the sweet-congruent environment significantly increased the perceived sweetness of the beverage they tested. However, these studies either focus on manipulating the intrinsic color of the foodstuff in VR or dealing with complex VR scenes.

As discussed in the previous section, the color of the ambient light is an effective stimulus studied in the physical environment. This inspired us to consider if similar effects can be reproduced and leveraged in VR environments. The virtual environments may cost-effectively control or simulate the ambient color rooms while providing fully colored immersive environments. Firstly, the color of the ambient light is a simple parameter that can be controlled in VR and be compared with previous works done in the physical setting. Although not related to taste, researchers have already suggested using VR for ambient color studies. For example, Oberfeld and Hecht (2011) studied the influence of environmental color on the perceived height and width of interior rooms virtually. von Castell et al. (2018) also studied the effects of the hue, saturation, and luminance of color on the perceived height of interior spaces via a VR HMD. Lipson-Smith et al. (2020) did a study focusing on the mood and preference towards the virtual environments varying in colors. In the meantime, applying ambient light as an environmental parameter to VR leverages the immersion provided by this medium well. In VR development tools, ambient light hue is an important parameter in creating the visual appearance of the scene and its objects. Changing ambient light hue can effectively alter the color scheme of all the 3D models (objects) in the VR scene (Unity, 2019). Based on these, we chose ambient light varying in colors as our stimuli to study its influence on taste perception in VR. While ambient light is essential to and a foundation building block to all VR environments, its effect on taste perception has not been explored in detail. Our research aims to fill this important but underexplored area.

In a VR environment, color may induce Visually Induced Motion Sickness (VIMS) and make the application unusable (Gusev et al., 2016). Bonato et al. (2004) simulated the optokinetic drum (spinning sphere) with specific color patterns in a virtual environment. They suggested that the chromaticity might increase the inconsistency between visual and vestibular inputs. However, So and Yuen (2007) found the change in the colors would not influence VIMS significantly in VR. Besides, colored environments may lead to a higher VIMS compared to monochrome environments (Gusev et al., 2018). Based on the literature, it is not clear whether the ambient color in VR would cause VIMS and, if so, to what extent the VIMS would be. To avoid discomfort and bias in our color-taste experiment, we conducted the first experiment to adjust the ambient colors in the virtual environments to ensure they were not harmful to the experience. Color recognition can be another potential issue for VR color-taste studies. It is important to consider whether people could depict the ambient color hue in a VR environment as they perceive it in the real world (Lipson-Smith et al., 2020). We also want to filter out this issue before running the color-taste experiment.

Selecting a suitable tastant is also important for a color-taste study. Some of the existing studies in CMC between taste and other stimuli used word-based descriptions or images of the food. Crisinel and Spence (2009) used Implicit Association Test and applied food images as visual stimuli to investigate the association between basic taste and pitch (auditory parameter). Similarly, Wan et al. (2014) used graphics with words referring to taste as the visual stimuli in their online study. The participants dragged the image to the word they thought it was associated with. On the contrary, Crisinel and Spence (2010) did a repeated experiment using real tastants. However, they got similar taste perception results as those evoked by the food names. As reported by Crisinel and Spence (2009) in their work, although it was convenient to use the names of food or drinks in the study, it entailed several disadvantages. First, between-participant differences might exist, such as the same word “beer” may associate with various distinctive tastes. Second, excluding real tastants prevented them from studying the direct interaction between the sensory modalities. As such, in our study, we did not use words and instead used a real tastant because this is more aligned with the use of VR, as a platform that brings users to experience virtual environments but allows them to still be closely associated with the physical world.

Previous studies in taste and virtual environments have utilized a variety of tastant choices. We collected some representative ones and listed also in Table 1. Among these, we found that a beverage is the most frequently used tastant. In addition, drinking is relatively more convenient and safer than eating. Thus, we used beverages as our taste stimuli. We anticipated that a beverage with a very strong taste would lead towards one specific taste and bias the results. Due to this, a neutral tastant that would not affect participants’ taste perception is required. However, limited research has discussed the rationale behind choosing the tastants and whether their tastes are neutral. We decided to run a pilot study to choose a neutral tastant before our formal color-taste experiment.

Table 1 summarizes the related works mentioned above. This study has three main differences compared to prior studies:

1) Unlike prior work about taste perception in immersive virtual scenes (Section 2.1), our focus was CMC-related stimuli instead of realistic scene simulation.

We utilized the colors described in Wan et al. (2014)’s taste-color synesthesia, including black, green, grey, orange, pink, purple, red, yellow, white, blue, and brown. These colors were used as the hue of the ambient light in the VR environments. The RGB values of these colors are also listed in Figure 1. Specifically, we created 11 hexagonal virtual rooms to demonstrate these ambient colors (see Figure 1). In each room, we placed a spotlight at the top middle of the room to make sure it illuminated the room (i.e., the ambient color covered the entire interior of the room), and assigned the RGB value of the color. The spot angle of the light was 129.8°. We set the range for controlling how far the light is emitted from the center of the object from 0 to 20. The default light intensity for each room is 10. We left the rooms empty to prevent any confounding factors.

Twelve healthy participants (6 females, 6 males; mean age = 21.2 ± 1.25 years) from a Sino-Chinese university located in a mid-size city in China were recruited for the first experiment. The participants were not paid and joined the experiment voluntarily. They did not have color blindness and did not report any issues regarding color recognition. Seven participants had prior experiences with VR HMD.

We used an Oculus Rift S VR HMD to display the virtual environments. The HMD has a resolution of 1,280 × 1,440 px per eye, a refreshing rate of 80 Hz, and a 115° field of view. Participants could move their heads in 6DOF. It was connected to an Intel Core i7-6700HQ processor laptop with an NVIDIA GeForce GTX 1070 GPU card. Participants could interact with the virtual environment via an Oculus hand-held controller. The virtual environments were created using the Unity3D platform (version 2019.1.14f1) and the related scripts were programmed in C#. The experiment was conducted in an empty and quiet room where the participants were equipped with the VR HMD and controllers. Figure 2 shows the setup.

A within-subjects design was used in this experiment. Participants would enter all the virtual rooms to complete the given tasks. When first entering a room, the initial intensity of the tested ambient light was set at 10. Participants were required to adjust the light intensity up and down starting from this value until they found the lighting condition that was the most comfortable. For experimental purposes, we set the adjustment range of the light intensity from 0 (dimmest light) to 20 (brightest light for our 3D assets). Participants could press Button A or B to make small adjustments to the light intensity (increased or decreased by 0.5 per press) and push the thumbstick to make big adjustments (see Figure 2). Once they settled down the light intensity for the current room, they stopped interacting with the controller for about 10 s. The finalized light intensity was then recorded by the program. After that, they needed to answer the question “what ambient color do you think is the current environment?” and complete a Virtual Reality Sickness Questionnaire (VRSQ) (Kim et al., 2018) in oral form. The experimenter asked these questions from VRSQ one by one and manually recorded the answers. Then the participants moved to the next room. The order of the room for each participant was counterbalanced using a Latin-Square design. Note that in our experimental design, participants’ answers toward color recognition might not have a relationship with their adjustment of light intensity. Although each participant had encountered several levels of intensity, they were exposed to the same intensity condition at first and were informed that they would make the adjustment for one specific color each time.

The whole experiment lasted approximately 30 min for each participant. Before the formal experiment began, participants were asked to fill in a questionnaire for demographic background. We asked about their prior VR experience and if they had difficulty recognizing colors. Then participants were introduced to the research aims, tasks, controls of the experiment, and the questions that would be asked. Then they wore the VR HMD and started the experiment. After they adjusted the light intensity for all the ambient colors, they would take off the HMD and have a break. After the break, we asked participants to adjust the light intensity for the ambient colors in reverse order. This was intended to eliminate the residue effects from the previous colored room.

The average light intensities for each ambient color are 14.63 for black, 5.72 for green, 11.97 for grey, 6.82 for orange, 7.98 for pink, 6.63 for purple, 3.68 for red, 4.34 for yellow, 4.69 for white, 4.09 for blue, and 12.75 for brown. As mentioned before, the values are between 0 (the dimmest condition) to 20 (the brightest condition).

In total, we collected 24 color recognition results (12 participants × 2 rounds). Figure 3 summarizes the recognition accuracy for each ambient color. The recognition accuracy for the ambient color of green, blue, red, purple, pink, and orange was high. However, the ambient color of brown and grey was low (29% and 41.5%, respectively). Though not low in recognition accuracy, white was frequently mixed with light pink (4 times). For grey and brown, there was a grey-green, grey-blue confusion, and a brown-pink confusion. Figure 4 is the summary of the average VRSQ score for each ambient color following the calculating method provided by Kim et al. (2018). A higher score indicates that the simulator sickness caused by the ambient color environment is more severe. Participants regarded brown, grey, and white as more comfortable than other colors in the VR environment after self-adjusting the light intensity. However, the difference was not significant.

Colors with low recognition accuracy were excluded from the follow-up experiment. We excluded white as it was mixed with pink while keeping pink since pink could be recognized successfully in the virtual environment. Grey and brown were excluded for the same reason. Although the color recognition accuracy of black was not high, it is an important color showing a high correlation with the bitter taste in the physical environment. So, it was useful to keep it and see its effectiveness in the virtual environment.

The results of recognition accuracy show that gaps may exist between the ambient colors that a developer wants the audience to perceive and the actual perceived color by this audience, especially when creating virtual environments. Developers are suggested to pay attention to color recognition accuracy, especially when color is an important factor in the virtual environment. Another contributing result was the comfortable intensity for the 11 ambient colors in the virtual environment with their VRSQ value, working as a guideline for developers who want to utilize these colors in their designs.

We invited the same 12 participants in Experiment 1 to complete this pilot study. The participants filled in a demographic questionnaire before the formal experiment. In the questionnaire, we inquired about their attention to taste in the form of scaling questions from 0-low to 6-high, “to how much extent do you pay attention to taste during your daily dieting?”. Their average ratings on taste attention were: 5.08 for sweetness, 4.69 for sourness, 3.38 for bitterness, and 5.06 for saltiness. We prepared four common beverages, including apple juice (brand: Huiyuan Juice), black tea (brand: Nongfu spring), coconut water (brand: Goodfarmer), and yogurt soda (brand: Energetic Forest), all of which could be bought at a local grocery shop. Our main consideration was to provide several possible choices of drinks that do not have a strong taste and are also familiar to users. After some trials before the pilot study with a number of possibilities, the four chosen ones represent a wide range. The pilot study then helped us narrow done to one.

We opted to use mini paper cups (100 ml) to ensure that it was convenient to drink the beverage while wearing an HMD in the main experiment later. The cups were covered with lids to prevent participants from recognizing the beverage by seeing the content or by smelling the odor (see Figure 5). Participants were asked to drink each of the four beverages, holding it in their mouth for at least 3 s before swallowing it. Then they needed to rate the taste of the beverage in terms of sweetness, sourness, bitterness, saltiness, and neutrality using a 7-scale questionnaire. Water was given to participants between two beverages to remove any residual, left-over taste. The sequence of the delivered beverages was counterbalanced. Note that, water was not used as the tastant because it was used as the “palate cleanser” to clear participants’ taste from the previous drink, and as such using it as the tastant is not appropriate (Huang et al., 2019; Chen et al., 2020; Wang et al., 2020).

Black tea was rated the most neutral among the four beverages. The remaining tastants have strong sweet tastes based on the ratings. Therefore, we used black tea as the taste stimulus in our experiment.

We removed the colors showing low recognition accuracy from our Experiment 1 (as described in Section 3.5). Thus, the color stimuli in the main study included black, green, orange, pink, purple, red, yellow, and blue. Although white was excluded, we set it as a starting, ending, and transition color for the experiment to avoid the residue effects from the previous colored room. A similar approach can be found in Spence et al. (2014). Figure 1 shows a preview of each virtual environment with the RGB parameter and intensity level of each ambient color. Notice that we did not include the 3D model of the beverage in the virtual environment because we wanted to focus on the influence of the immersive environment only. We used black tea as the tastant based on the results of the pilot study. Like the pilot study, the tastant was served in 100 ml paper cups covered with lids.

We recruited another 16 participants (7 females, 9 males; 22.2 ± 1.66 years, ranging from 21 to 27 years). Thirteen participants had used VR HMD before the experiment and all of them had normal or corrected-to-normal vision and had no history of color blindness. Regarding their drinking behaviors, thirteen participants reported drinking beverages several times a week and twelve participants pay much attention to the taste of the beverage during their daily drinking activities. We used the same VR apparatus as in Experiment 1.

Figure 6 provides an overview of Experiment 2. We applied a within-subjects design in this experiment with ambient color as the independent variable. As mentioned, we used a white room as a starting, ending, and transition room between the two rooms. The order of ambient color conditions was counterbalanced with a Latin-Square design. In each condition, the participants would be exposed to the virtual room for 30 s, starting from when they entered the scene and ending when they heard a notification to drink the tastant. They were asked to have a sip from the 100 ml tastant in their mouths for 3 s before swallowing it. After they drank the tastant, they were then asked to report their feelings according to the following questions:

• Q2-1. Can you rate the taste of the drink in the just-finished scene in terms of sweet, sour, bitter, and salty? Ratings are from 1 (very weak) to 7 (very strong).

Before going to the next ambient color room, they were exposed to a white room and delivered a mini cup of pure water for palate cleansing. After the participants experienced all eight conditions, they were asked to take the VR HMD off, and answer a post-experiment questionnaire to collect their demographic information.

The participants were first briefed about the task and procedure of the experiment. They were encouraged to raise any questions if something was unclear. They were also informed that they could stop at any point during the experiment if they felt uncomfortable. The participants then put the VR HMD on and were asked not to take it off before the end of the experiment. Thus, they could not see and did not receive any information about the drinks during the whole experiment. They were first exposed to a white room and were given a cup of pure water to accommodate the ‘blinded’ drinking action and cleansing palate. After that, they were given the mini cup containing the tastant and kept it in their hand. Next, the experimenter switched the scene to the first room to start the formal experiment. After the participants were immersed in the ambient color virtual environment for 30 s, the background voice began to remind the participant of starting drinking. After the drinking is finished, participants were asked for answering the above questions orally. Before going to the next ambient color room, they were exposed to a white room and the cups in their hands were replaced with new ones containing pure water. They then drank the pure water for palate cleansing. After that, they entered the next ambient color room with a new mini cup of tastant delivered, replacing the cup containing pure water. Noticing that, when immersed in the virtual environment without seeing the content in the physical world, most of the participants were unaware that they were drinking the same tastants throughout the study.

After the participants experienced all eight conditions, they were asked to take the VR HMD off, drink the taste stimuli once more, and then report their feelings. Finally, we gave participants a post-experiment questionnaire to collect their demographic information, past VR experience, and beverage consumption habits.

SPSS (version 26) was used for data analysis. The results were transformed with Aligned Rank Transform (Wobbrock et al., 2011). We performed one-way Repeated Measure-(RM-) ANOVA with the transformed data for Q1-1 and Q1-4. If the transformed results violated the assumption of sphericity, we report the results with Greenhouse-Geisser or Huynh-Feldt corrections according to the case (ϵ < 0.75 or > 0.75 , respectively). We used Fisher’s exact test to examine whether there is a significant color-taste association (i.e., Q1-2). Post-hoc pairwise comparisons were run with Bonferroni corrections if the above tests revealed significant differences or associations.

In this work, we focused on exploring how environmental colors would affect people’s taste perception in VR than verifying a psychological phenomenon in VR. We used VR as a platform for providing new perceptions and a higher level of immersion to study the cross-modal correspondence (CMC) experience.

As listed in the Related Work (Section 2), our research can be positioned differently from prior work in physical and VR settings. Our results in taste association present an interesting comparison with the results given by studies conducted using a desktop computer screen as a medium for showing colors (Wan et al., 2014). Pink was significantly associated with sweetness in both VR and the real world. However, different from Wan et al. (2014)’s results from experiments in the real world where green was found to be more associated with sourness and black with a bitter taste, our results in VR showed that green can also be associated with a bitter taste. This indicates that the same color studied via different mediums may bring the same or different results in taste perception.

Unlike prior work regarding taste-related CMC in virtual environments, our selection of ambient colors as visual stimuli is novel and has not been explored systematically before. Choosing a parameter that is not related to the environment (e.g., color change of the foodstuff, change of the container shape) may lose the benefit of the immersion provided by VR headsets. In addition, our focus is on the basic aspect of VR than the complex ones. The color of ambient light suits these requirements. Our work is different from Chen et al. (2020) which used VR environments involving a diversity of visual elements, surface textures, and types of rooms. Moreover, while the experiment by Wang et al. (2020) and Huang et al. (2019) focused on the changing the color of the drink in VR (i.e., an intrinsic property of the drinks), our study dealt with the ambient color of the VR environment (i.e., an extrinsic property of the drinks). In our experiment, the sweetness of the beverage seems to be increased by the pink environment, and the orange VR environment was preferable. These findings can be compared with Wang et al. (2020)’s results showing that the intrinsic beverage color did not influence the perceived sweetness or liking of the beverage.

Participants’ answer to the associated scene in the VR environment shows that most participants’ color-taste associations were based on specific scenarios. This mainly includes foodstuff and non-food-related scenarios. According to the explanation by Spence (2019b) (see Section 2.2), people can often point to specific source object(s) that embody both color and taste, which is also in line with the associated scenarios regarding foodstuff in our experiment. The ambient color may trigger participants’ recall of real-life experiences with food, and then the association with taste would come up based on these experiences. The variety of answers indicates the association may be personal and subjective, which may also be a topic for further investigation.

Some answers about scenarios other than foodstuff could be explained by emotional mediation (Spence, 2011) (see Section 2.2) of the relation between taste and the words regarding emotion in a particular culture. Categorized by taste association, some answers such as “chemical poison”, “nursing house”, “horror movie” appeared under bitter association while “vineyard”, “club”, “children’s room” appeared under sweet association. When categorized by color, there were some negative words like “prison” for the black virtual environment, “horror” for the red virtual environment, and “depressed” for the green virtual environment. This kind of negative association observed with certain ambient colors may be in line with some recent studies that have explored negative emotions in VR (Lavoie et al., 2021; Magdin et al., 2021). These studies showed that virtual environments involving a higher level of absorption may increase negative emotional responses. Also, Elliot et al. (2007) suggested that attention must be paid to how color can act as a subtle environmental cue that has important influences on human behavior. Regarding color study in virtual environments, research work has shown that the context in a virtual environment where the ambient color was assigned can also influence emotion and valance (Lipson-Smith et al., 2020). This leads to a future question to be discussed since, in this experiment, we did not include 3D models in the scenes. Based on this, we suggest that future researchers or designers applying ambient color in VR systems can pay special attention to the emotional association the color can trigger.

Our experiment results could contribute to the design of human-food applications, presenting the possibility of how the dieting experience can be altered by using specific environmental colors. Researchers have looked into the use of diet activities as positive psychological interventions to increase everyday happiness and well-being (Fischler, 2011; Cook et al., 2020). Moreover, existing works have explored what and how advanced technology can support healthy dieting using wearable devices and mobile apps (Dong and Biswas, 2013; Pan et al., 2019). Our work shows the possibility of using VR to change people’s taste perception, especially the sensation of sweet taste. Given the health issues caused by high levels of sugary drink intake (Essman et al., 2021; Leung et al., 2021), our findings could have applications to support healthier drinking habits. We found that modifying the ambient color in a VR environment can enhance users’ sensation of the sweet taste of the beverage. This approach provides a novel beverage-drinking experience and a cost-effective approach to changing users’ diet behavior (by providing users’ perception of sweetness with drinks that are lower in actual sugar content). Future research could look into this potential application via longitudinal experiments.