Virtual Environment, Real Impacts: A Self-determination Perspective on the use of Virtual Reality for Pro-environmental Behavior Interventions

ABSTRACT

Plastic waste is a serious environmental problem worldwide. Effective environmental communication is key to mobilizing the public to adopt pro-environmental behaviors for reducing plastic waste. While virtual reality (VR) is proposed as a viable tool that could overcome several challenges facing environmental communication, certain limitations exist in the extant literature, making it unclear whether it is the modality of VR itself, rather than some extraneous factor, that accounts for pro-environmental outcomes. Hence, adopting a between-subjects experimental design, this study improves on past research by comparing the efficacy of a VR game with an equivalent computerized quiz in increasing participants’ pro-environmental behavioral intention. Our results indicate that VR (vs. computerized quiz) has a comparative advantage for increasing participants’ behavioral intention to learn about plastic waste (BIL). Furthermore, guided by the self-determination theory, we find that perceived autonomy and autonomous motivation serve as serial mediators in the relationship between modality and BIL.

Key Policy Highlights

• Policymakers could adopt VR technologies to increase public members’ interest in learning about environmental issues.

• In designing pro-environmental behavioral interventions, policymakers should focus on facilitating individuals’ autonomous motivation by giving them a sense of control.

KEYWORDS:

• virtual reality
• self-determination theory
• plastic waste
• pro-environmental behavior
• autonomous motivation

The world is currently facing many environmental challenges. However, despite an apparent consensus around the reality and impacts of issues like climate change (Carlton et al., 2015), there is a great disparity between people’s concerns and their behavior regarding issues like plastic consumption (Kautish et al., 2021). Due to growing demands in the markets, plastic consumption has risen by over 300% in the last three decades, while only 9% of plastic is recycled successfully (OECD, 2022). Moreover, the centrality of plastic in human activities and our daily living has contributed to a failure to enact meaningful regulatory changes in plastic production, use, and disposal (Smith & Brisman, 2021). Scientists and scholars routinely highlight the damaging impacts that plastic consumption has on the environment. For example, microplastic pollution and its subsequent environmental impacts have received increasing attention as these pervasive fragments from plastic development and breakdown have been found in locations ranging from the highest terrestrial peaks to the deepest trenches in the ocean – and have even been detected inside human bodies (Garcia-Vazquez & Garcia-Ael, 2021). This dire situation has prompted the United Nations to acknowledge plastic waste as a global crisis (UN Environment Programme, 2022).

Despite efforts to increase its overall waste recycling rate, Singapore has encountered setbacks due to the lack of general recycling knowledge and consumer awareness among the public (Singapore Environment Council, 2018). On average, Singapore uses 1.76 billion plastic items each year, with much of the plastic consumption originating from single-use bags and takeaway containers (Singapore Environment Council, 2018). Juxtaposing the ballooning of plastic consumption, Singapore’s plastic recycling rate remains at a consistently low level of 6% (National Environmental Agency, 2022). Prompted by concerns over sustainable production and consumption, there is an imminent need to examine new strategies for raising public awareness and directing behavior change to reduce plastic waste.

Virtual reality (VR) presents itself as a viable tool for improving awareness about environmental issues and pro-environmental behaviors. For example, some research suggests that interactive VR experiences like games can encourage connectedness to nature (Klein & Hilbig, 2018), offer direct experiences of how one’s actions influence the environment (Fox et al., 2020), and improve mastery and self-efficacy over pro-environmental behaviors (Stenberdt & Makransky, 2023). Moreover, by requiring players to actively engage with the medium, VR experiences can facilitate greater involvement with and learning from the environmental message compared to traditional communication media.

Yet, given the diversity of pro-environmental behaviors, it is difficult to assume similar positive impacts across different contexts. For example, Kleinlogel et al. (2023) observed that VR exposure only improved pro-environmental energy-saving behaviors that were directly addressed in the stimulus. Consequently, the study found that the specific pro-environmental knowledge acquired from the VR experience did not transfer to other types of energy-saving behaviors not mentioned in the stimulus. As little research has been conducted on VR experiences that focus on plastic waste reduction, it is valuable to understand how effective the medium might be in encouraging relevant pro-environmental behaviors.

In addition, due to inconsistencies across extant research, it is uncertain if the positive effects of VR exposure are a result of VR’s modal affordances or other features not unique to the medium. Indeed, several studies provide comparisons that introduce confounding factors outside of modality (e.g. Ahn et al., 2014) or identify no significant differences between VR exposure and exposure to an equivalent comparison (e.g. Soliman et al., 2017). Given the high costs of developing a VR experience, it is important to ascertain if the potential positive effects of VR exposure are a result of modal differences, or due to extraneous factors that may be replicated in another communication medium. To that end, the objectives of the study are to (a) compare the efficacy of our VR intervention with an equivalent computerized quiz and (b) investigate the psychological mechanisms underlying the effects of VR for behavioral intervention.

Virtual reality for environmental communication

Fauville et al. (2020) identified several challenges facing environmental communication, most of which are rooted in the fact that the impacts of one’s actions on the environment are hardly visible to and immediately experienced by oneself. This creates a disconnect between people and the environment, making them less receptive to environmental communication efforts. Fauville et al. (2020) argue that these challenges can be overcome using VR due to its affordances of presence and immersion. Presence refers to a feeling of actually being at the place presented to one in the virtual world, while immersion refers to a perceived state where one is absorbed in an environment and actively interacts with the stimuli in the environment (Witmer & Singer, 1998). By creating immersive virtual environments that afford its users a sense of presence, individuals can experience and learn how their actions can affect the environment in a psychologically realistic manner. For example, studies have suggested that VR experiences can reduce the perceived time–space distance between individuals and environmental threats like climate change (Fox et al., 2020) and encourage their connectedness to and appreciation of the natural environment (Soliman et al., 2017).

Scholars have developed VR experiences that showed promising results in driving sustainable changes. For instance, Markowitz et al. (2018) found that a VR experience featuring an underwater world was effective in increasing participants’ knowledge about ocean acidification and interest in learning about the topic. Ahn et al. (2016) showed that embodying animals in VR (vs. watching the experience on video) had a larger positive effect on perceived connection with nature, which tends to be positively associated with pro-environmental behavior (Soliman et al., 2017). Hsu et al. (2018) have demonstrated that their VR game could significantly affect individuals’ cognition and behavior intention toward water conservation. Similarly, Ahn et al. (2014) have found that participants who experienced cutting down a tree in VR reduced their consumption of paper more than those who read a written description or watched a video about tree cutting. In the same study, those in the VR condition also reported more pro-environmental behaviors one week after the VR experience compared to those in the other conditions (Ahn et al., 2014). In the context of Singapore specifically, a community project provided preliminary evidence suggesting that VR can be an effective communication medium in raising awareness about microplastics and promoting behaviors toward the reduction of plastic waste (Vasey et al., 2019).

However, we identified some limitations in the current literature on VR for environmental communication. First, the VR environments developed in most studies were detached from the daily actions of a regular person and may not easily translate into actual behavior (Meijers et al., 2022). For example, Soliman et al. (2017) found that participants who viewed nature-related virtual environments – such as forests, mountains, rivers, and wildlife – experienced increased connectedness to nature but did not exhibit subsequent pro-environmental behaviors not highlighted in the stimulus. The measured behaviors included signing up for a monthly newsletter on nature and sustainability as well as receiving a download link for a relevant sustainability strategic plan. As such, while some VR experiences may be effective in raising awareness due to their novelty, their lack of connection with one’s real-life actions may limit their impact on one’s daily behavior. It is thus imperative to design and explore the impact of VR experiences featuring scenarios that individuals are likely to encounter in their everyday lives so that participants can learn the specific actions they could take to mitigate the problem of plastic waste.

Furthermore, in studies that compared the effects of VR with other modalities, the VR condition often required active interaction (e.g. cutting down a virtual tree; Ahn et al., 2014) while participants in the comparison groups did not engage in a similarly interactive experience (e.g. reading a description or watching a video; Ahn et al., 2014). This makes it unclear whether the effects demonstrated in the VR condition were due to the modality of the content presented or extraneous qualities not exclusive to VR technologies, such as interactivity. In their study comparing the effects of watching a nature video using either VR or a regular computer screen, Soliman et al. (2017) found that the modality used for video watching did not influence the effects of the video on nature connectedness. Their results demonstrate that VR does not deliver additional benefits compared to regular computers when the amount of interaction required of participants is minimal, given that both conditions in their study involved passive video viewing. Thus, it remains unclear if VR would be a more effective modality for environmental communication, compared to a less immersive one, when active interaction is required for both conditions.

The current study aims to bridge this gap by experimentally comparing a serious game hosted in VR (i.e. the “VR condition”) with an equivalent computerized quiz (i.e. the “non-VR condition”), where participants were required to actively answer a series of questions in both conditions. Given the considerable cost involved in developing a VR-based intervention, it is important to ascertain that VR is a significantly more effective modality than a less costly alternative. An important practical contribution of the current study is thus that it may provide an empirical basis for justifying the deployment of VR for environmental communication or behavior change at large. Based on existing results from VR research, we posit:

H1: Participants in the VR condition will show higher levels of pro-environmental behavioral intention than those in the non-VR condition.

We investigate two types of pro-environmental behavioral intention: behavioral intention to learn about plastic waste reduction (BIL) and behavioral intention toward plastic consumption and disposal (BICD). BIL indicates participants’ intention to equip themselves with knowledge about plastic waste reduction. It is a key dependent variable in this paper as the Singapore Environment Council (2018) has highlighted the lack of general recycling knowledge and consumer awareness as the main challenges in tackling plastic waste. BICD indicates participants’ intention to engage in actual behaviors serving to mitigate the issue of plastic waste.

Another limitation of extant research lies in the lack of explicit reference to theories explaining the effects of VR on cognition or behavior (Fox et al., 2020). Past studies tended to predict the effectiveness of VR based on the unique affordances of this technology, such as immersion (e.g. Hsu et al., 2018) and body transfer (e.g. Ahn et al., 2016), without a theoretical justification for the psychological mechanisms underlying the effects. The absence of theory renders extant research largely exploratory and may lower the generalizability of past studies. As such, another aim of the current study is to investigate the psychological mechanisms underlying the effects of VR experience on pro-environmental behavioral intention. Guided by the self-determination theory (SDT), we examine whether the effects of VR on pro-environmental behavioral intention can be explained by its ability to promote autonomous motivation. Additionally, we explore agency-related concepts as alternative explanations to the effects of VR. In doing so, this study contributes to the literature by providing a theoretical framework guiding the development and evaluation of VR experience for environmental communication in future research.

Self-determination theory

Within the context of pro-environmental behavior, it is essential that individuals are motivated by their intrinsic goals (van der Linden, 2015). Where motivation is concerned, SDT has been established as a prominent way of differentiating between the typologies of motivation. SDT posits that when individuals find tasks enjoyable or challenging, these tasks promote growth and individuals become intrinsically motivated (Deci & Ryan, 2000; Osbaldiston & Sheldon, 2003). Conversely, when individuals participate in tasks governed by external sources of control such as the environment (i.e. reward or punishment), individuals become extrinsically motivated (Pelletier, Tuson, Green-Demers, Noels, & Beaton, 1998). In other words, SDT contends that human motivation is influenced by contextual factors (Deci & Ryan, 2000). SDT further differentiates extrinsic motivation into four types depending on the extent to which they are internally versus externally regulated (Ryan & Deci, 2000). As such, this theory views different types of motivation as ranging along a continuum based on their locus of regulation, with intrinsic motivation as the most internally regulated, or autonomous, form of motivation (Ryan & Deci, 2000). Accordingly, some scholars prefer to compute a single relative autonomy index by weighing these motivation types by their relative positions on the continuum (e.g. Radel et al., 2013; Sheldon et al., 2017), such that the higher the score, the more autonomous a person’s motivation is.

Facilitating autonomous motivation is crucial for successful behavior intervention (Deci et al., 1991; Ryan & Deci, 2000). Past studies using behavioral learning strategies such as reinforcement to induce pro-environmental behavior have seen limited success because any changes such strategies produced, if any, were only short-term (Geller, 1989; Green-Demers et al., 1997). Using the lens of SDT, reinforcement provides external incentives for individuals to engage in pro-environmental behavior, which cultivates extrinsic rather than autonomous motivation. Deci and Ryan (1985) argue that higher autonomous motivation predicts better outcomes for behavior intervention. Consistent with this, Green-Demers et al. (1997) found autonomous motivation to be positively associated with a range of pro-environmental behaviors, with the association stronger for more difficult behaviors. Similarly, de Groot and Steg (2010) found that the more individuals were autonomously motivated to do things for the environment, the more likely they would choose a car based on its environmental impacts and prefer the more environmentally friendly options. In the same study, autonomous motivation was also positively associated with one’s intention to donate to an environmental rather than a humanitarian cause. In contrast, external regulation, the most externally controlled form of extrinsic motivation, was negatively associated with one’s intention to donate to an environmental cause.

SDT contends that to cultivate autonomous motivation, satisfaction of one’s psychological needs for autonomy, competence, and relatedness are key (Ryan & Deci, 2000). Autonomy refers to perceived freedom and control over one’s actions (Decharms & Carpenter, 1968). Competence is defined as the propensity to interact effectively with the environment and confidence in achieving desired outcomes (Harter, 1978). Lastly, relatedness refers to the propensity to interact with and be connected to others in an experience (Baumeister & Leary, 1995). Since the VR game in this study does not involve social interaction, the aspect of relatedness is outside the scope of our investigation.

Notably, past research has identified presence and immersion – key affordances of VR – to be positively associated with perceived autonomy and competence (Ijaz et al., 2020; Ryan et al., 2006). As perceived autonomy relates to one’s ability to claim ownership over their actions, presence and immersion may support individuals’ perceived autonomy by enhancing their sense of control over their virtual behaviors and decision-making. Presence and immersion can further bolster individuals’ experience of naturalistic feedback to their actions, thus demonstrating how their actions in VR can influence the environment (Jung, 2011). Likewise, individuals’ perceived competence may be supported when they can effectively and convincingly interact with, explore, and influence the virtual environment (Ioannou et al., 2019). Given the ability of VR technology to induce a heightened sense of presence and immersion, we expect VR to be more effective in fulfilling individuals’ needs for autonomy and competence than traditional communication media. In support of this, Kern et al. (2019) found that a VR program for gait rehabilitation (vs. a traditional rehabilitation program using treadmills) was more efficacious in improving participants’ perceived decision freedom and competence. Hence, we posit:

H2: Participants in the VR condition will show higher levels of perceived autonomy than those in the non-VR condition.

H3: Participants in the VR condition will show higher levels of perceived competence than those in the non-VR condition.

Based on the connection between psychological needs and autonomous motivation, and between autonomous motivation and behavior intention, we posit:

H4: Perceived autonomy and autonomous motivation will serially mediate the relationship between modality and pro-environmental behavioral intention.

H5: Perceived competence and autonomous motivation will serially mediate the relationship between modality and pro-environmental behavioral intention.

Alternative pathways linking VR to pro-environmental behaviors

While the focus of this paper is on SDT, the authors hoped to explore other potential explanations for the hypothesized effects of VR as a better medium for environmental communication. The pathways grounded in SDT focus on the importance of intrinsic motivation to facilitating pro-environmental behaviors; this is particularly pertinent to the context of serious games where gamification is used to make an otherwise boring task interesting. For this reason, the concepts related to the SDT (i.e. perceived autonomy, perceived competence, and motivation) were measured with reference to the VR game or the quiz, depending on which condition participants were assigned to. For example, one item used to measure perceived competence was “I think I am pretty good at this game / quiz”, and a sample item for measuring intrinsic motivation was “(I played the game / did the quiz) Because I enjoy the game / quiz”. Hence, by proposing the SDT pathways, we aim to test the hypothesis that the motivational appeal of a VR game is a mechanism through which participants’ pro-environmental behavioral intention can be enhanced.

However, past research has shown that the use of VR could directly influence participants’ environmental beliefs. These environmental beliefs then serve as antecedents to pro-environmental behaviors. This means that change in environmental beliefs, instead of enhanced autonomous motivation, could act as an alternative explanation for the hypothesized greater effect of VR (vs. non-VR equivalent) on participants’ pro-environmental behaviors. To test for this alternative explanation, we hence investigated two environmental belief variables, self-efficacy and locus of control, as potential mediators linking VR intervention to pro-environmental behaviors. These two variables were chosen because they have been found to be strong predictors of pro-environmental behaviors in extant research (e.g. Bamberg & Möser, 2007; Fox et al., 2020) and may be effectively facilitated by VR intervention due to the affordances of this technology (Makransky et al., 2019). However, we want to stress that they are in no way the only alternative pathways.

Self-efficacy, which refers to confidence in one’s ability to perform a certain task (Bandura, 1977), is a strong determinant of pro-environmental behavior (Fox et al., 2020; van Valkengoed & Steg, 2019). In this paper, we are specifically interested in environmental self-efficacy, which is participants’ confidence in their ability to perform pro-environmental behaviors. Bandura’s self-efficacy theory highlights the crucial role of self-efficacy in stimulating behavior change (Bandura, 1977). Accordingly, this factor is prominently featured in many socio-cognitive theories and models applied for behavioral interventions, such as the theory of planned behavior (Ajzen, 1991) and the model of private proactive adaptation to climate change (Grothmann & Patt, 2005).

The unique affordances of VR may enable it to enhance one’s environmental self-efficacy, hence promoting pro-environmental behavior. By allowing users to experience firsthand the real-time impacts of their decisions and actions in a virtual world, VR could promote performance accomplishments (Plechatá et al., 2022), which is posited as one of the most important determinants of self-efficacy (Bandura, 1977). In their cognitive affective model of immersive learning, Makransky and Petersen (2021) argue that the technical features of interactivity and immersion allow VR to induce a sense of presence and agency, or perceived control over one’s actions, which leads to increased self-efficacy that facilitates learning and behavior change. In support of this, many empirical studies have shown that VR increases self-efficacy significantly more than other modalities (e.g. Makransky et al., 2019; Shu et al., 2019; Zeng et al., 2017). Furthermore, Fox et al. (2020) found that a serious game hosted in VR that afforded more (vs. less) interactivity increased participants’ pro-environmental behavior, and the relationship was mediated by self-efficacy. Similarly, Plechatá et al. (2022) compared two versions of VR interventions, which differed by their interactivity, and found that the more interactive alternative had a greater influence on pro-environmental behavior intention, and the effect was fully mediated by self-efficacy. While the authors of both papers compared two forms of VR interventions, we have not found studies examining the mediating role of self-efficacy in the relationship between VR intervention (vs. non-VR equivalent) and pro-environmental behavior. To bridge the gap, we posit:

H6: Participants in the VR condition will show higher levels of environmental self-efficacy than those in the non-VR condition.

H7: Environmental self-efficacy will mediate the relationship between modality and pro-environmental behavioral intention.

In addition, we explore the role of environmental locus of control as a mediator. It is defined as a perceived source of control (internal vs. external) over the desirable outcomes of pro-environmental behavior (Kim et al., 2022). While some scholars argue that self-efficacy and internal locus of control are conceptually similar (Fauville et al., 2020; Plechatá et al., 2022), the distinction between the two is that self-efficacy focuses on one’s ability to perform a task (e.g. engaging in pro-environmental behaviors), which does not necessarily lead to the desirable outcome, while internal locus of control speaks directly to one’s perceived control over the desirable outcome (e.g. improving the quality of the environment). Furthermore, in the current study, these two concepts differ from perceived competence introduced in the earlier section in that both self-efficacy and internal locus of control are beliefs about pro-environmental behaviors, whereas perceived competence is a belief about one’s proficiency in the intervention task (i.e. either the VR game or the computerized quiz; see supplementary material for more details).

Studies have shown internal environmental locus of control to be among the strongest predictors of pro-environmental behaviors (Bamberg & Möser, 2007). Moreover, a study showed that a VR intervention (vs. equivalents in print or video) led to more pro-environmental behaviors via increased internal environmental locus of control (Ahn et al., 2014). Hence, we posit:

H8: Participants in the VR condition will show higher levels of internal environmental locus of control than those in the non-VR condition.

H9: Internal environmental locus of control will mediate the relationship between modality and pro-environmental behavioral intention.

Our hypotheses are summarized (Table 1) below.

Table 1. Summary of Proposed Hypotheses.

No.	Hypothesis
H1	Modality → Pro-environmental behavioral intention
H2	Modality → Perceived autonomy
H3	Modality → Perceived competence
H4	Modality → Perceived autonomy → Autonomous motivation → Pro-environmental behavioral intention
H5	Modality → Perceived competence → Autonomous motivation → Pro-environmental behavioral intention
H6	Modality → Environmental self-efficacy
H7	Modality → Environmental self-efficacy → Pro-environmental behavioral intention
H8	Modality → Internal environmental locus of control
H9	Modality → Internal environmental locus of control → Pro-environmental behavioral intention

Method

Participants

We recruited 63 undergraduates from a university in Singapore for the study. Participants’ ages range from 21 to 34 (M = 22.98, SD = 2.48) and 58.7% (n = 37) of them were female. The majority of the participants were Singapore citizens (82.5%, n = 52) and ethnic Chinese (88.9%, n = 56). Participants were compensated with either $10 in cash or 5 course credits by choice. This study was approved by the University’s Institutional Review Board.

VR stimulus

The VR game was developed in collaboration with professional VR developers and an interdisciplinary team of researchers specializing in communication, education, and environmental engineering. The VR game was created using Unity 2021.2.7f1 and comprised an introductory scene, four interactive scenarios, and a closing scene.

Each interactive scenario presented an everyday task (e.g. dining, grocery shopping, and recycling; Figure 1) in an immersive environment. Participants were asked to make a series of choices within each scenario. Each choice carried a score that reflected its environmental impact, such that choices that contributed more toward mitigating plastic waste were given higher scores. After making a choice, participants were shown the score and some explanation about the environmental impact of the choice made.

Figure 1. Interactive Scenarios Presented in the VR Game.

Note: From left to right, the screenshots correspond to the dining scene, grocery shopping scene, and recycling scene.

The score assigned to each choice was established through a simplified life cycle assessment (LCA; Muralikrishna & Manickam, 2017). LCA evaluates the ecological consequences of a specific product from its production to its final disposal, including the extraction of raw materials, the manufacturing process, and the product’s usage and eventual disposal. To conduct the LCA, we used the GaBi 10.6 software, integrated with the ecoinvent 3.5 database, using standards set by the International Organization for Standardization (ISO) in the 14040 series (ISO, 2006). The actual environmental impact of each product was evaluated against 11 environmental impact categories based on the CML method. After ranking each in-game choice based on its actual environmental impact, we assigned standardized scores that ranged from 1 (for the least environmentally friendly choice in that scenario) to 10 (for the most environmentally friendly choice in that scenario).

After playing through the four scenarios, participants viewed an underwater environment that reflected the overall environmental impacts of their choices (Figure 2). High scorers viewed a clean underwater environment filled with flora and fauna while low scorers viewed a murky underwater environment devoid of lives.

Figure 2. Underwater Environments for Participants with Low Scores (Left), Medium Scores (Middle), and High Scores (Right).

Participants viewed the VR game through a Meta Quest 2 standalone headset with built-in headphones and used a set of controllers to interact with the VR game. Participants were able to move around the virtual environment in a controlled 2-meter by 2-meter space.

Experimental procedure

Participants were randomly assigned to the VR condition (n = 30) and the non-VR condition (n = 33). Those in the VR condition played the VR game described above. Participants in the non-VR condition completed an equivalent computerized quiz, administered via Qualtrics. This quiz presented participants with the same scenarios and decision choices as in the VR condition. For example, in the grocery shopping scene (second screenshot in Figure 1), participants in both conditions had to choose among ‘bananas pre-packaged in plastic”, “bananas with paper banding”, and “bananas without packaging”. The difference lay in modality: while those in the VR condition made the choice in a VR environment, those in the non-VR condition made the choice in response to a multiple-choice question presented in a computerized quiz. After a selection of choices was made, similar to those in the VR condition, participants in the non-VR condition were shown the score associated with each choice and an explanation for the score given.

In both conditions, participants were first asked to read a study information sheet and fill out a consent form. Those in the VR condition were then taught how to put on the VR headset and use the controllers. Those in the non-VR condition were led to a desktop computer where they would complete the quiz.

After the respective experimental treatments, participants completed an online questionnaire in the laboratory that captured our dependent variables. All participants were sent a debriefing email after concluding the experiment.

Measures

All measures used a five-point Likert scale (1 = strongly disagree; 5 = strongly agree). We provide the exact item wordings and descriptive statistics of all variables in the supplementary material.

Perceived autonomy (α = .78) was measured using seven items (e.g. “I felt like it was my own choice to play this game / do this quiz”¹) from the perceived choice subscale in the Intrinsic Motivation Inventory (IMI; see Ryan et al., 1983).

Perceived competence (α = .86) was measured using six items (e.g. “I think I am pretty good at this game / quiz.”) from the perceived competence subscale in the Intrinsic Motivation Inventory (IMI; see Ryan et al., 1983).

To measure autonomous motivation, we employed the comprehensive relative autonomy index (C-RAI; Sheldon et al., 2017). The C-RAI incorporates measures for six types of motivation, namely intrinsic motivation (α = .89), identified regulation (α = .88), positive introjected regulation (α = .92), negative introjected regulation (α = .91), external regulation (α = .86), and amotivation (α = .92). Participants were requested to indicate why they played the VR game / did the quiz by expressing their agreement with each of the 26 responses in C-RAI (e.g. “Because the game / quiz is interesting” for intrinsic motivation and “I don't know why I would play the game / do the quiz” for amotivation). We first created a composite index for each motivation type. With the six composite indices, we computed a relative autonomy index (RAI) using the following formula: RAI = 3 * intrinsic motivation + 2 * identified regulation + positive introjected regulation – negative introjected regulation – 2 * external regulation – 3 * amotivation (Sheldon et al., 2017). The weights assigned to each motivation type were determined by their theoretical position on the self-determination continuum as proposed by Ryan and Deci (2000).

Environmental self-efficacy (r = .70) was measured using two items (e.g. “I think that I am capable of protecting the environment by means of my personal plastic waste reduction”) adapted from Reese and Junge (2017).

Internal environmental locus of control (α = .82) was measured using nine items (e.g. “My individual actions would improve the quality of the environment if I were to buy and use products that do not have plastic packaging”) adapted from Ahn et al. (2014).

BIL (α = .89) was measured using four items (e.g. “I will continue to learn about plastic waste reduction in the future”) adapted from Chai et al. (2020). BICD was measured using 15 items adapted from Barr et al. (2001). These items consisted of three clusters of behaviors, with 5 items for each cluster, which targeted participants’ intention to reduce (α = .70; e.g. “reduce purchases of produce which has a lot of packaging”), reuse (α = .70; e.g. “reuse old containers, like ice cream tubs or margarine boxes”), and recycle (α = .80; e.g. “sort out recyclable and non-recyclable trash and throw them into corresponding bins”). After creating the respective composite scores for the three clusters of behavior, we created the variable for overall behavioral intention toward plastic consumption and disposal using the mean of the three composite scores (α = .88).

To test our hypotheses, we employed paired sample t-tests and the “PROCESS” macro (Hayes, 2017). All analyses were completed using SPSS.

Results

To assess the success of our random assignment, we analyzed participants’ ages using a t-test and their genders using a chi-square test across different conditions. The results showed no significant differences in either age (t(61) = −1.07, p = .29) or gender (χ²(1, N = 63) = .04, p = .85) composition between the two conditions, indicating a successful random assignment. The descriptive statistics and zero order correlations of our variables are presented in Table 2.

Table 2. Zero Order Correlations.

Variable	M	SD	2	3	4	5	6	7
1. Autonomy	4.22	.65	.15	.53***	.28**	.36***	.29**	.18*
2. Competence	3.71	.69	–	.23**	.22*	.11	.17	.12
3. Autonomous Motivation	11.80	6.71	–	–	.35***	.55***	.45***	.50***
4. Self-efficacy	4.35	.67	–	–	–	.55***	.56***	.43***
5. Locus of control	4.37	.51	–	–	–	–	.56***	.51***
6. BICD	4.26	.53	–	–	–	–	–	.58***
7. BIL	4.24	.64	–	–	–	–	–	–

Note: *p < .05, **p < .01, ***p < .001

One of the primary objectives of this study was to investigate if modality affected participants’ pro-environmental behavioral intention. Results from our t-test showed that participants in the VR condition (M = 4.49, SE = .11), compared to those in the non-VR condition (M = 4.05, SE = .10), had higher BIL (t(61) = −2.96, p < .01).² Due to concerns regarding the relatively small sample sizes in this study, we conducted a post hoc power analysis using G*Power to determine if the study had sufficient statistical power to detect a significant difference in participants’ BIL. With sample sizes of 30 and 33 for the two groups, an alpha level (α) of .05, and an observed effect size (Cohen's d) of .75, the analysis indicated that an independent samples t-test possessed a satisfactory power (.83) to detect the observed effect size. Participants in the VR condition (M = 4.40, SE = .10) did not differ from those in the non-VR condition (M = 4.27, SE = .08) on BICD (t(54) = −1.00, p = .32).³ Hence, H1 was partially supported. Since modality only had an effect on BIL, subsequent mediation analyses focused on this behavioral intention variable (i.e. it was used as the dependent variable).

Participants in the VR condition (M = 4.33, SE = .11), compared to those in the non-VR condition (M = 4.00, SE = .12), had higher perceived autonomy (t(61) = −2.11, p < .05). Participants in the VR condition (M = 3.76, SE = .12), compared to those in the non-VR condition (M = 3.48, SE = .12), had marginally higher perceived competence (t(61) = −1.57, one-sided p = .06). Hence, H2 was supported while H3 was marginally supported.

To investigate if psychological needs, followed by autonomous motivation, mediated the effect of modality on BIL, we ran two serial mediation models (Process model 6, 5000 bootstraps; Hayes, 2017). The first model (Figure 3) showed that perceived autonomy and autonomous motivation serially mediated the relationship between modality (VR condition = 1; non-VR condition = 0) and BIL (B = .12, SE = .06, 95% CI = .02 to .23, $R^2$ = .46). Within the serial mediation model, the mediation paths by perceived autonomy (B = -.04, SE = .04, 95% CI = -.13 to .04) or autonomous motivation (B = .13, SE = .10, 95% CI = -.04 to .36) alone were not significant. Hence, H4 was supported. The second model showed that perceived competence and autonomous motivation did not serially mediate the relationship between modality and BIL (B = .02, SE = .02, 95% CI = -.02 to .06, $R^2$ = .46). Within the serial mediation model, the mediation path by perceived competence alone (B = .01, SE = .03, 95% CI = -.06 to .07) was not significant while that by autonomous motivation alone was significant (B = .20, SE = .11, 95% CI = .02 to .46). This suggests that autonomous motivation mediated the relationship between modality and BIL, but contrary to our hypotheses, perceived competence was not a significant serial mediator. Hence, H5 was not supported.

Figure 3. Serial Mediation Model with Perceived Autonomy and Autonomous Motivation as Mediators.

Note: *p < .05, **p < .01, ***p < .001.

We explored two other mediators, self-efficacy and internal environmental locus of control, as alternative mechanisms explaining the effect of modality on BIL. First, t-tests showed that participants in the VR condition (M = 4.52, SE = .10 for self-efficacy; M = 4.46, SE = .09 for internal environmental locus of control), compared to those in the non-VR condition (M = 4.29, SE = .11 for self-efficacy; M = 4.28, SE = .08 for internal environmental locus of control), had marginally higher levels of self-efficacy (t(61) = −1.52, one-sided p = .07) and internal environmental locus of control (t(61) = −1.51, one-sided p = .07). Hence, H6 and H8 were marginally supported. Next, we ran two mediation models with self-efficacy and internal environmental locus of control as the mediator respectively (Process model 4, 5000 bootstraps; Hayes, 2017).⁴ The results showed that neither self-efficacy (B = .05, SE = .04, 95% CI = -.03 to .14, $R^2$ = .16) nor internal environmental locus of control (B = .12, SE = .09, 95% CI = -.04 to .30, $R^2$ = .39) mediated the effect of modality on BIL. Hence, H7 and H9 were not supported.

The mediation results are summarized in Table 3 below.

Table 3. Mediation Results for the Relationship between Modality and BIL.

Mediator(s)	Indirect Effects	SE	95%CI
Perceived Autonomy → Autonomous Motivation	.12	.06	.02 –.23
Perceived Competence → Autonomous Motivation	.02	.02	−.02 –.06
Self-efficacy	.05	.04	−.03 –.14
Internal Environmental Locus of Control	.12	.09	−.04 –.30

Subsequently, we conducted similar mediation analyses with BICD as the dependent variable for exploration purposes. The results showed that the effect of modality on BICD was not mediated by any of the tested variables (Table 4).

Table 4. Mediation Results for the Relationship between Modality and BICD.

Mediator(s)	Indirect Effects	SE	95%CI
Perceived Autonomy → Autonomous Motivation	.05	.03	−.00 –.12
Perceived Competence → Autonomous Motivation	.01	.01	−.01 –.04
Self-efficacy	.09	.06	−.03 –.22
Internal Environmental Locus of Control	.12	.08	−.04 –.29

Discussion

This study primarily aimed to investigate the efficacy of a VR game (vs. an equivalent non-VR quiz) for promoting pro-environmental behavioral intentions (BIL and BICD) related to plastic consumption and waste. We demonstrate that modality affects participants’ BIL, with VR being the preferred modality, and that this relationship can be explained through a serial pathway involving participants’ perceived autonomy and subsequently, autonomous motivation. These findings indicate a comparative advantage of using VR for encouraging pro-environmental behavioral intention as well as the use of SDT for explaining the theoretical mechanisms underlying such behavioral intention.

An unexpected finding was that participants did not significantly differ in their BICD between the VR and non-VR conditions. In our study, we provided scenarios that participants are likely to have encountered and navigated in their daily lives and offered corresponding explanations about the environmental impact of each choice they made. This was in line with previous research that suggested that the simulation of real-life scenarios was an important component of digital (vs. non-digital) environments for improving pro-environmental behavior (e.g. Fox et al., 2020; Janakiraman et al., 2021; Sajjadi et al., 2022). However, BICD may be driven by psychological mechanisms not addressed in our VR game.

Scurati and colleagues (2021) suggest that there are three directions in which VR experiences can be designed to influence pro-environmental behavior: emotional, rational, and practical. Emotionally guided VR experiences would aim to shift feelings like affection, fear, and connection with the environment, rational ones would be associated with shaping knowledge and understanding, while the practical direction pertains to evaluating and comparing solutions for environmental issues. As our VR game largely targets more practical goals of teaching appropriate actions for managing plastic consumption and disposal, it could be that BICD is better addressed through a different direction of game development. While previous research has also demonstrated that emotional appeals may be insufficient for encouraging pro-environmental behavior in the form of choosing to save paper and receive more information about nature and sustainability (Soliman et al., 2017), scholars also suggest that different pro-environmental behaviors are motivated by distinct mechanisms (e.g. Larson et al., 2015; Lee et al., 2014). As such, it could be that the significantly higher levels of perceived autonomy and the marginally higher levels of competence, self-efficacy, and environmental locus of control engendered by our practically led VR game design do not motivate BICD.

Comparatively, studies investigating underlying mechanisms for BIL have instead considered the interest, enjoyment, and autonomous motivation afforded by VR as contributors (Cheng & Tsai, 2020). In line with this, we found autonomous motivation as a mediator explaining the positive effect of VR on BIL.

Furthermore, our results revealed that perceived autonomy is a key variable contributing to BIL. These findings contrast with suggestions in prior studies that the interactivity of gamified environments and the self-control they necessitate (such as when making choices or actions) may have negative effects on various pro-environmental behavioral intentions, such as one’s willingness to learn about deforestation (Maltseva et al., 2019). When compared to a non-gamified survey, these authors argue that the increased interactivity – and thus autonomy – afforded by a gamified survey may exhaust one’s cognitive resources for and interest in pro-environmental behaviors.

However, the present study showed that environmental communication through a VR game (vs. a non-VR quiz) produced better results in increasing participants’ BIL, and this effect could be explained by higher levels of perceived autonomy and subsequently, autonomous motivation facilitated by VR. This suggests that the increased interactivity afforded by VR may be beneficial to behavioral intervention in the environmental domain. Another distinct advantage of VR is that it can effectively visualize and translate distant and complex topics, such as the impact of plastic consumption and waste. Prior research has identified that VR environments can reduce both intrinsic and extraneous cognitive load – referring to the perceived difficulty of the topic presented and the additional mental processing required by the format through which the topic is presented respectively (Dan & Reiner, 2017). In other words, by enabling direct, contextualized, and realistic experiences, VR can promote greater interest in learning more about environmental issues (Liu et al., 2022).

As Maltseva and colleagues (2019) did not manipulate the modality, but rather the interactivity, of their survey task, their chosen investigative topic of deforestation may have been challenging or cognitively taxing to visualize through a computerized survey. Thus, we suggest that pro-environmental behavior interventions are effective when they promote one’s perceived autonomy without necessitating extraneous cognitive load on tasks such as visualization. Here, by simulating realistic and immersive environments, VR can serve as an effective tool for encouraging pro-environmental behavioral intentions.

Moreover, although the VR condition (vs. the non-VR condition) had a marginal effect on environmental self-efficacy and internal locus of control, neither variable mediated the relationship between modality and BIL. Again, this may be attributed to the distinct psychological mechanisms underlying BIL compared to other pro-environmental behavior intentions – such as one’s intention to recycle or donate to an environmental organization. Moreover, both environmental self-efficacy and internal locus of control in this study measured participants’ confidence in or perceived control over various actions relevant to reducing their plastic waste or improving the quality of the environment. Hence, the scope of the measures used might have been conceptually distant from one’s intention to learn more about plastic waste.

Nonetheless, our study verifies the efficacy of VR experiences, like games, for pro-environmental behavior interventions when it comes to encouraging BIL. Learning can play an important role in encouraging actual pro-environmental behavior by fostering long-term, active, and intrinsic interest in the environmental impacts of plastic consumption and plastic waste (Meyer, 2015; Wi & Chang, 2019). Theoretically, the study offers an empirically based explanation for VR experiences’ comparative advantage over non-VR formats, such as a computerized quiz. Beyond the technical affordances of presence and immersion, this study contributes to existing VR literature by suggesting that psychological factors, namely perceived autonomy and autonomous motivation, act as more proximal factors in explaining the effects of VR intervention on pro-environmental behavioral intentions.

Moreover, the findings of this study are in line with past SDT research that supporting perceived autonomy benefits autonomous motivation and subsequent BIL – verifying the importance of SDT-informed approaches to pro-environmental behavioral interventions. Rather than diminishing one’s cognitive resources, we identified that the VR modality supported autonomy through its unique capacity to provide interactive and realistic experiences.

Another interesting contribution to research at the intersection of SDT and VR is the relationship between self-determination and different pro-environmental behaviors. We identified that while autonomy and autonomous motivation serially mediated the relationship between modality and BIL, psychological needs and autonomous motivation did not mediate the relationship between modality and BICD. This corresponds with past research suggesting that some pro-environmental behaviors might be more self-determined than others (Cooke et al., 2016).

As such, the findings also evince the importance of distinguishing between different pro-environmental behaviors in measurement, stimulus development, and analysis. Although BIL and BICD are both types of pro-environmental behavioral intentions, there may be distinct psychological processes underlying each behavioral intention. For example, Ho et al. (2015) found that the antecedents of pro-environmental behavioral intentions differed between two different sub-types of pro-environmental behavior (i.e. intention to engage in green-buying behavior and intention to engage in environmental civic action).

Practically, this study supports the use of VR for encouraging some pro-environmental behavior intentions when compared to interactive non-VR formats. Specifically, we recommend that when practitioners identify a lack of autonomy to be a barrier to certain behaviors, like learning about plastic waste, VR can be a suitable medium due to its higher interactivity compared to other modalities. To that end, encouraging the public to engage in learning and information-seeking is one route for fostering behavior change that is sustainable and intrinsically motivated.

Moreover, our results indicate that younger demographics such as university students may benefit from VR-based interventions for improving pro-environmental behavioral intentions. This is salient as younger people are often the focus of environmental campaigns – as they will eventually inherit and live with the potential degradation of ecological environment as well as become key actors in determining environmental policy (Grønhøj & Hubert, 2021; Kadic-Maglajlic et al., 2019). Yet, current research suggests an incongruence between younger people’s concerns about sustainability and their engagement in pro-environmental behaviors (Ágoston et al., 2024) and that they may be less committed to carrying out common, pro-environmental household behaviors compared to older people (Grønhøj & Hubert, 2021). As such, VR can be a promising avenue that practitioners can leverage for encouraging younger people to learn more about and engage with pro-environmental behaviors.

Practitioners and scholars may also examine if the beneficial effects of VR on pro-environmental behaviors translate to other demographics, like older adults or young children. Past research has demonstrated that older adults, when compared to younger adults, may similarly benefit from VR experiences and training (Dilanchian et al., 2021; Dobrowolski et al., 2021). The use of VR with young children, however, should be approached carefully due to age-specific concerns of using immersive technology during stages of developmental changes. For example, VR might engender shifts in children’s social and cognitive responses (Bailey et al., 2019) or foster the formation of false memories (Segovia & Bailenson, 2009).

However, practitioners should be cautious that the use of VR is suitable for their needs and would not induce extraneous cognitive load that could detract from the users’ experience. While our study recommends the use of direct, contextualized, and realistic VR experiences, there may be other qualities of VR that increase or decrease extraneous cognitive load. For example, Ahn and colleagues (2022) found that high spatial presence – a user’s perceived experience of being located in an interactive, mediated (VR) environment – may detract from the available resources a user has to process information related to learning goals. The novelty and relative complexity of VR, compared to more familiar media, may also induce technostress, whereby the new interactions and interfaces can contribute to greater mental load and processing (Souchet et al., 2023). Practitioners interested in developing VR experiences should also be aware of the antecedents they are targeting and look to existing research to clarify the variables that motivate specific pro-environmental behaviors.

We suggest that while VR experiences can be beneficial for specific pro-environmental behaviors, they should not be viewed as a universal solution for addressing environmental issues; there may be cases where non-VR modalities might be similarly or more effective for promoting pro-environmental behaviors. Hence, in situations where the affordances of VR, such as mediated interactivity, do not offer additional benefits to more feasible, accessible, and convenient modalities, VR should not be viewed as a cure-all. While not highlighted in this study, it should also be recognized that there are potential disadvantages of using VR, such as how the technology’s acceptability and accessibility might differ across different populations (e.g. Mott et al., 2020), the possibility of triggering cybersickness (i.e. VR-induced motion sickness; Hamad & Jia, 2022), and the need for a sophisticated technology infrastructure to support VR use (Cevikbas et al., 2023). As such, VR should be considered as an addition to a wide range of potential media for hosting pro-environmental behavioral interventions.

Limitations and future research

This study is not without its limitations. First, in explaining the effect of VR on pro-environmental behaviors, since the focus of this study is on self-determination theory, the alternative mediation pathways (H6 – H9) did not receive as much attention in the conceptualization stage and replication studies to test them are warranted. As highlighted in the discussion, we only measured participants’ self-efficacy and locus of control related to plastic waste reduction. This could have limited our scope of analysis when investigating antecedents to and explanatory pathways predicting BIL. Future studies should thus include self-efficacy measures specific to learning and information-seeking when investigating BIL. Moreover, studies can also seek to investigate more nuanced and diverse variations of pro-environmental behaviors and their antecedents relevant to the environmental issue of interest.

Furthermore, the generalizability of the findings could have been limited by the chosen sample of undergraduates as participants. As prior studies have identified differences in individuals’ experience of presence, cybersickness (Dilanchian et al., 2021), and emotional responses (Liu et al., 2020) in VR experiences across demographic variables like age, it is important to ascertain if the positive effects of VR experience on BIL are applicable to samples more representative of the general population.

Finally, due to the nature of the VR game used in this study, we did not explore the role of relatedness in the mediation pathway linking modality and pro-environmental behavior. As an important psychological need contributing to one’s experience of autonomous motivation, we suggest that future studies employ stimuli that enable interaction and connectedness with other social agents or actors in VR.

CRediT author statement

Sherry R. Xiong: Methodology, Investigation, Formal Analysis, Writing – Original Draft Shirley S. Ho: Conceptualization, Methodology, Writing – Review & Editing, Supervision, Funding acquisition Wenqi Tan: Conceptualization, Methodology, Investigation, Writing – Original Draft Benjamin J. Li: Conceptualization, Methodology, Writing – Review & Editing Grzegorz Lisak: Conceptualization, Methodology.

Acknowledgments

The authors would like to thank Dr. Shanti Divaharan for her contributions to conceptualizing the study and experimental materials.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

Data will be made available on request.

Additional information

Funding

This study was approved by the Nanyang Technological University’s Institutional Review Board. The IRB reference number for this manuscript is: IRB-2022-885. This research was supported by Nanyang Technological University’s Accelerating Creativity and Excellence (ACE) Award [NTU-ACE2021-05] and the Ministry of Education, Singapore, under its MOE AcRF Tier 3 Award [MOE-MOET32022-0006]. The funders did not play a role in the study conceptualization, in the collection or analyses of data, in the writing of the manuscript, or in the decision to publish the results.

Notes

1 The phrasing was customized to each condition.

2 Unless otherwise specified, all the p-values reported are two-sided p-values.

3 Equal variances not assumed, hence a different degree of freedom here.

4 These were single-mediator models with either self-efficacy or internal environmental locus of control in them as the mediator.