Scoping Review of Research on Augmented Reality in Environmental Education

During the past decade, there has been an exponential growth of AR in education (Al-Ansi et al., 2023; Garzón, 2021; Garzón & Acevedo, 2019). AR offers new learning opportunities and provides students with visual and auditory stimulations that diverge from traditional forms of teaching. With AR acting as a bridge between virtual and real worlds (Klopfer & Sheldon, 2010), learners can visualize abstract concepts and complex spatial relationships (Hung et al., 2017) and experience phenomena that are not easily accessible in the real world (Klopfer & Squire, 2008). There is emerging evidence for the positive effects of AR on the motivation and engagement of middle-school students (Di Serio et al., 2013), on the motivation, engagement, and STEM interest of senior high-school students (Hsu et al., 2017), and on the motivation, attention and concept skills of preschool children (Aydoğdu, 2022). However, students who used AR for language learning and social studies indicated higher learning motivation than students who utilized AR for science learning. This is according to results of one meta-analysis of quasi-experimental AR studies from 2012–2021 (Chang et al., 2022). In this meta-analysis, effect sizes were interpreted using Cohen’s guidelines where a 0.2 effect size is considered small, 0.5 is medium, 0.8 is large, 1.2 is very large, and 2.0 is huge (1992).

Meta-analytic results report AR’s significant effects on multiple learning outcomes (i.e., knowledge, skill, and performance) as well, with AR exhibiting the largest mean effect size on performance according to the meta-analysis described above (Chang et al., 2022). Another meta-analysis of AR studies from 2010–2018 found AR to have a medium effect (also based on Cohen’s d) on students’ learning gains, with undergraduate students and the engineering field displaying the most benefits (Garzón & Acevedo, 2019). Secondary school students in one study also benefited from reading AR storybooks by contributing to the improvement of their reading comprehension scores (Şimşek and Direkçi, 2023). A meta-analysis of the effects of AR technology as used in science education showed its moderate-to-large positive effect (0.737) on students’ academic achievement (Xu, 2022). Yet, another meta-analysis found insufficient support for AR’s impact on learning outcomes and pointed out the need for more well-designed, randomized-controlled trials on AR for more conclusive evidence (Bölek et al., 2021). These two meta-analyses both used Cohen’s criterion to quantify the effect size in their sample (Cohen, 1988). Notwithstanding these mixed results, AR has established itself as a key emerging educational technology for both educators and researchers (Wei et al., 2021).

Emerging evidence for the benefits of AR in education has highlighted it as a prospective tool for environmental education (Ducasse, 2020). The interactive and engaging nature of AR enhances students’ interest and motivation in environmental topics (Ladykova et al., 2024). As an example, in a marine learning program in Taiwan that integrated AR technology, lower grade primary school students shared their perception of the interactive multimedia to be “fun, engaging, interactive, and participatory” (Lu & Liu, 2015, p. 537), results that were also reflected in their confidence and satisfaction scores. The program also effectively improved the performance of low achievers in the study and impacted many learning outcomes. This is consistent with studies that showed AR’s positive impact on students’ learning of environmental concepts, through providing visuals of complex processes such as climate change, pollution, and ecosystem dynamics (Dunleavy & Dede, 2014; Ladykova et al., 2024). Affective outcomes, such as students’ attitudes and emotional connection to environmental issues, are also enhanced (Ladykova et al., 2024). In another study, an AR digital picture book was developed and incorporated into environmental education learning activities of sixth grade students from national elementary schools in Taiwan and Japan (Chen, 2022). The researchers attributed the non-significant effect of science learning self-efficacy on environmental behaviors to the AR digital picture book having been based on Taiwanese Dao indigenous culture, making it harder for Japanese students to comprehend. On the other hand, results of an EcoMOBILE project that tested a combined AR application (i.e., FreshAIR™) and an environmental probeware demonstrated improvements in the self-efficacy of sixth graders on a school field trip and their understanding of what scientists do (Kamarainen et al., 2013). Teachers in the study also observed high levels of engagement among students.

In another evaluation of a mobile (i.e., tablet), Greek-based, culture-specific, interactive AR game for fourth grade Greek students, the students had a generally positive attitude towards the AR game called "Save Elli, Save the Environment" (Koutromanos et al., 2018). This AR program was specifically designed as a game (a game-based learning tool) used in selected locations that were within walking distance to the school where the AR game was tested. The students found the game useful and enjoyable due to various interactions that arose among team members. The interactions during the game promoted collaboration and increased interest in learning. However, technical problems (e.g., net connectivity, GPS) and environmental conditions in the field (e.g., strong winds and bright sunshine) posed some challenges in terms of reading and listening to the content of the AR game from the students’ tablets. These findings were echoed by a book chapter that discussed AR for outdoor environmental education (Ducasse, 2020). The author pointed out that a major issue in designing technology-enhanced learning activities for environmental education is that they are usually implemented in informal settings and outdoors. Despite this, the author acknowledged the vast potential of mobile-based AR in promoting learning in context, by elaborating on various applications of learning theories (i.e., situated learning, place-based learning, experiential learning and inquiry-based learning) and how mobile-based AR can enable a more efficient transfer of theoretical knowledge from school to real-life situations. One good example of the usefulness of AR in learning about the environment is its potential to be used by students in simulating environmental scenarios to investigate the influence of different actions on ecosystems (Dunleavy & Dede, 2014).

The studies reviewed thus far indicate the varied demographics of students who are catered to by AR, the different characteristics of the AR tools (e.g., learning content, learning context/modes of delivery, gamification elements or game-based learning, devices used), and the diverse learning outcomes assessed in AR for environmental education research. These studies demonstrate how AR could expand the pedagogical activities that students engaging in environmental education might learn from.

A recent review highlighted the potential of mobile AR for early childhood and primary education (Criollo-C et al., 2024). While this review used the term sustainable education, the article was about the use of mobile AR in these educational levels between 2012–2022, and did not zoom in on a particular area of study. In the present review, we focus on AR in environmental education as defined earlier. Our work also differs from this recent review since we did not place any restrictions in terms of publication year and educational levels. Clarifying the scope of studies on the use of various types of AR tools in environmental education across all educational levels is essential to identify important gaps and trends in the literature (Linnenluecke et al., 2020). First, we determined what are the demographic characteristics of participants in the AR studies included in this review. Second, we identified the research methods and approaches employed in studies about AR in environmental education. Third, we identified the characteristics of the AR tools examined in the studies. We determined their characteristics based on learning content (culture-specific or global), presence of gamification/game-based learning, subjects/courses the AR tools were used in, learning context/mode of delivery, duration/length of intervention/program and devices used. For clarity, we define culture-specificity as the possession of unique characteristics that distinguish one culture from another including traditions, social practices, language, beliefs, and values (Hall, 1989). We distinguished between culture-specific and global AR tools given evidence for the possible effect of misaligned and non-culturally relevant AR content on student outcomes (Chen, 2022), and the existence of AR technology with content tied to specific contexts (Chen, 2022; Koutromanos et al., 2018). We also distinguished between gamification and game-based learning and recognize the importance of documenting how many of the AR tools evaluated in the included studies have these elements. Gamification optimizes the benefits of external rewards by adding game elements (e.g., points, badges, leaderboards) to non-game contexts to enhance engagement and motivation (Lampropoulos et al., 2022), while game-based learning involves the use of actual games that were specifically developed as learning activities (Yu et al., 2022). We also deemed it significant to gather information about the subjects the AR tools were used in given the apparent need for multidisciplinary solutions to environmental problems (Edwards, 2019). In the case of learning context or mode of delivery, we define experiential learning as the process of obtaining knowledge and skills by hands-on, immersive and interactive experiences (Arduini & Chiusaroli, 2019), that could be done in an unrestricted/unsupervised setting, with teacher supervision outdoors or in an informal setting, or within class in a formal setting. We also thought it important to note the length of the evaluated programs and the various hardware devices used to apply AR technology, acknowledging that duration and required equipment for implementation could vary considerably. Fourth, we aimed to review findings on the learning and behavioral outcomes measured in quantitative studies about using AR in environmental education, to identify the ones most studied. Lastly, given the call to base educational technology research on sound theoretical frameworks (Hew et al., 2019), the current review aims to gather information about theories employed in studies on the use of AR in environmental education specifically. Identifying frameworks used will aid in examining whether the current approaches used in evaluating AR in environmental education are measuring the most appropriate learning and behavioral outcomes in students. Doing so will help in determining outcomes that need to be targeted in students when proposing, implementing, and evaluating programs and interventions. The aims of this review are aligned with United Nations Sustainable Development Goals (SDGs), by directing attention to how education and innovative approaches help address an issue of global importance impacting generations to come.

This scoping review was conducted to answer the following research questions:

Research Question 1: What were the demographic characteristics of the participants in the AR studies reviewed [i.e., educational levels, ethnicity, study setting (country)]?

Research Question 2: What are the research methods and approaches employed in studies about AR in environmental education?

Research Question 3: What were the characteristics of the AR tools used in environmental education [i.e., name of AR application (if available), learning content (culture-specific or global), presence of gamification/game-based learning, subjects/courses the AR tools were used in, learning context/mode of delivery (in-class or formal setting, independent homework or unrestricted setting, or experiential/mixed or informal setting), duration/length of intervention/program and devices used (mobile/smartphone, head-mounted display, tablet, computer, or projector)]?

Research Question 4: What were the learning and behavioral outcomes measured in studies that evaluated the use of AR in environmental education?

Research Question 5: What were the theoretical frameworks employed in the studies included in the review?

Scopus, Web of Science, and ScienceDirect were used as the electronic databases for this review. These databases were chosen due to their multidisciplinary nature and their credibility as sources of comprehensive collections of publications (Marja et al., n.d.; Pranckutė, 2021). Specifically, Scopus and Web of Science are both known to have very good coverage for natural science, mathematics, and engineering, and good coverage for articles in quantitative social science (UC Davis DataLab, 2023). The search was conducted with keywords (“augmented reality”) AND (“environmental education” OR “climate education” OR “climate change education” OR “environmental conservation” OR “sustainable education” OR “environmental literacy”). No restrictions on publication year and educational levels were placed. The search strategy was adapted to the syntax and specific characteristics of each database for advanced searches (Siddaway et al., 2019). The initial search of the databases yielded 305 articles. After applying filters to refine the search, 223 articles were left to review. After removing duplicates, 185 articles were retained. The titles and abstracts of the studies were screened using the following criteria: 1) empirical articles from peer-reviewed journals, book chapters, theses and conference proceedings that were available in full-text, 2) focused on the use of AR in environmental education, 3) written in English.

Based on Arksey and O' Malley's framework, the third stage is identifying studies to be included in the scoping review (Arksey and O'Malley, 2005). Full-text articles that were retained after title and abstract screening (N = 49) were reviewed, and 21 articles were dropped in this process. Reasons for exclusion are the following:

To gather more relevant articles, Google Scholar and the references of book chapters and some of the retained articles were searched. Fourteen additional articles were found from this process, bringing the total number of articles to be included in the analysis to 42.

Figure 1 depicts the systematic search procedure and selection process followed in this scoping review.

Full size image

A data extraction sheet was employed to collect information needed to address the research questions. The content analysis technique (Hsu et al., 2013) was applied to extract the data from the included studies. The following information were charted: title, authors, year of publication, sample size, educational level, ethnicity, study setting, sample size, research method, study design, name of AR application, learning content, scope of learning content, presence of gamification and/or game-based learning, subject/course the AR tool is used in, learning environment or mode of delivery, duration of intervention, devices used, learning or behavioral outcomes assessed, and theoretical framework employed in the studies.

A top-down approach was used in the analysis of the data. The categories formed were based on the information required to address the research questions. Data were charted from the articles using a data extraction sheet, and frequencies of responses were counted when possible. Consultations among members of the research team were held to resolve disagreements. Details of the studies included are provided in Table 1.

The first published study that evaluated the use of AR in environmental education was in 2013, and the highest number of articles published on the topic was in 2023 (see Fig. 2). This indicates an emerging interest in researching the use of AR in environmental education. The finding that most AR technology in the reviewed studies had been implemented and evaluated among younger students (in primary and secondary education) is not surprising, considering previous findings that participants in such studies were mostly primary school students (Ladykova et al., 2024). Years spent in primary and secondary school are children’s formative years (Pollard & Bourne, 2002), and an opportune time for them to explore and satisfy their curiosities about the world they live in. Add to this young children’s preference for novel and innovative tools that tap into their interests (Radich, 2013), the use of context-sensitive AR in educating them about the environment thus holds some promise. There is some initial support for the potential of mobile AR for early childhood and primary education (Criollo-C et al., 2024). However, successful and effective use of AR, particularly in environmental education, relies on proposed programs and interventions being supported with evidence from continuous and well-designed evaluation studies.

We also noted how most of the included studies only had students as their participants, with very few articles having teacher samples. While this was expected since students are the main users and beneficiaries of AR-supported pedagogies, it is also important to obtain the insights and perspectives of teachers as the decision-makers and implementers of teaching innovations (Mouza, 2018; Simon & Zeng, 2024). Thus, we highly recommend the inclusion of teachers as samples in future studies examining the use of AR in environmental education, in recognition of the inseparable nature of teaching and learning (Simon, Jiang, et al., 2024). A comprehensive evaluation model of educational technology that includes teachers as key participants has also been proposed for a more holistic understanding of a particular tool’s impact on education (Lai et al., 2022). A multi-faceted approach that combines both students’ and teachers’ insights and experiences in educational technology assessments, with the use of empirical surveys, qualitative methods (e.g., interviews, focused group discussions, classroom observations), and systematic reviews would be a welcome addition to what is currently written about the use of AR in environmental education.

We also recommend more inclusion and diversity in future studies on the use of AR in environmental education are called for, given that the largest number of studies were conducted in Asia, with underrepresentation in continents other than Europe. It is also worth noting how all the studies have been conducted in developed or upper middle-income nations that have more access to resources and funding for developing and testing technological innovations such as AR, hinting at possible inequities. One point to ponder on is whether AR technology would be an appropriate solution for economically disadvantaged contexts such as in Global South countries (Nalipay et al., 2023). The cost-effectiveness of an AR tool can therefore be one consideration if we are aiming to increase the reach and accessibility of such innovations and to prevent further digital divide between wealthy and economically disadvantaged nations. This is most likely the reason why most of the AR interventions in the reviewed studies made use of smartphones as they are relatively cost-effective and accessible device options.

As expected, only a few studies were purely qualitative in nature. While experiments and quasi-experiments were common methods used in the included studies, they are best triangulated with qualitative methods to obtain richer insights into how AR technologies are applied in environmental education. A relevant article noted how challenges of using AR in environmental education are underreported (Ladykova et al., 2024), which can be primarily attributed to the dominance of experimental methods in this research area as found in this study and in the current review. Method triangulation was done in forty percent of the included studies, using synchronous assessment in the tool, student and teacher interviews, focused group discussions, and observations, adding qualitative components to their evaluations. Quantification of user retention or the number of players who completed each quest (Wang et al., 2021) and micro-ethnographic line-by-line examination of video data (Zimmerman et al., 2015) were some of the methods that supplemented self-report questionnaires. These can be good alternative options to examine real-time and actual behavioral patterns of students’ responses within the AR program as they can address the biases and limitations inherent in self-report data.

Regarding the scope of learning contents, it is unsurprising that most of the AR technology in the studies utilized a global perspective (with content relevant anywhere in the world) given the global nature of environmental problems. We acknowledge the value of incorporating culture-specific elements into learning contents to make them more engaging and relevant to students (Xu et al., 2023). But the downside of presenting content that is too culture-specific was made evident in the study by Chen (2022) where Japanese sixth grade students demonstrated less comprehension than their Taiwanese counterparts, due to the content of the AR app consisting of references to Taiwanese Dao indigenous culture unfamiliar to the Japanese students. Thus, we would like to highlight the importance of evaluating the appropriateness of culture-specific tools and to ensure that they get applied to populations that would benefit most from what they offer.

The wide range of subjects and courses involved in environmental education confirms that solutions to environmental problems are multidisciplinary in nature (Edwards, 2019). While the included studies integrated elements into the AR tools that were meant to increase students’ learning and engagement (e.g., experiential learning, game-based learning, gamification), the evaluation approaches and frameworks used were often not aligned with these objectives. We discuss the implications of such misalignment in the sub-section below.

As expected, the theories that guided the studies also dictated the outcomes measured. Since the use of AR in learning required students to be explorers and active participants in their learning process, it made sense to employ Situated Learning Theory (Lave & Wenger, 1991) as framework in three of the AR studies. For example, in one study that made use of Situated Learning Theory, students’ knowledge and understanding of what happens and how things happen in nature and the environment, science, and knowledge specific to the content of the lessons were assessed (Kamarainen et al., 2013). Another study went a step beyond and measured the impact of the educational resource on students’ motivation (attention, relevance, confidence, satisfaction) (Garzón et al., 2020). Similarly, the Theory of Planned Behaviour (Ajzen, 1991) was utilized in three of the included studies. For instance, one study measured attitude, subjective norms, perceived behavioural control and intention to adopt sustainable behaviors, as well as knowledge outcomes that pertain to recycling, afforestation, water contamination, and energy conservation (Wang et al., 2021). One trend we have observed is the practice of applying more than one theory or framework, such as in the case of studies that combined Theory of Planned Behaviour with Technology Acceptance Model (Koutromanos et al., 2018; Mei & Yang, 2019). However, it is quite concerning that the most employed framework among all the included studies was the Technology Acceptance Model. It has been found that too much focus on usage rather than engagement in technology-enhanced learning research tends to be misleading (Dunn & Kennedy, 2019). Although widely used in EdTech research, the Technology Acceptance Model does not directly address learning through the technology. Thus, the model is not well aligned with what should be the goal of education, which is to stimulate students’ interest and to improve their learning. We therefore recommend the use of well-established learning frameworks previously used in evaluating the effectiveness of AR in teaching and learning such as social constructivist learning theory (Dunleavy & Dede, 2014), or cognitive load theory that both pay attention to constructs relevant to learning through technology. Social constructivist learning theory focuses on social interactions, collaborative learning, guided instruction, and cultural context and how these aid in the development of cognitive abilities. Cognitive load theory recognizes the limited capacity of working memory and recommends managing these limitations through simplifying complex information, using visuals, minimizing distractions, and scaffolding learning to reduce learners’ cognitive load. These factors can be applied in designing and considered in evaluating AR technology to optimize learning in the tool.

According to the outcome-based approach to teaching and learning (Biggs, 2012; Biggs & Tang, 2011), the teaching and learning activities (which AR is part of), as well as the assessment strategies that evaluate students’ learning as a consequence of engaging with the activities, should be constructively aligned with the intended learning outcomes. The current review gathered information on studies that evaluated the use of AR in environmental education. Constructive alignment did not appear to be a priority in some of the studies: i.e., not all studies incorporated learning or knowledge outcome measures that are supposed to assess students’ learning in the program. Measuring constructs such as usability, perceived usefulness, and perceived benefits in AR evaluation studies has its place. However, placing premium on measuring such constructs over measuring students’ learning is an issue that needs to be addressed by this emerging field. We therefore recommend a more education-driven approach to the evaluation of digital technologies (Laurillard, 2008). This can be done by including assessment of outcomes most relevant to student learning through the technology in future evaluation studies.

Overall, the findings of this review point to various opportunities for future research and practice, some of which will be discussed in the subsequent section.

This scoping review suggests vast opportunities for future research and practice. As findings point to important gaps, researchers and educators must work together to design an AR program that captures the learning and behavioral outcomes that are most important for students. Enlisting the help of seasoned instructional designers, educational technologists, and other content experts can ensure that the appropriate knowledge and expertise are tapped in the development of a program that maximizes the affordances of AR technology for environmental education.

For more focused efforts, we recommend inclusion of outcomes that are relevant to students’ environmental knowledge acquisition using AR. We also recommend that evaluations always include an examination of how the programs/interventions stimulated their interest, motivation, and engage students in learning about the environment. Similarly, self-efficacy, an important construct with strong associations with learning outcomes (Richardson et al., 2012; Schneider & Preckel, 2017), has been neglected in the reviewed studies and thus would be good to include in future research.

In terms of measuring the quantity of students' engagement with AR, supplements to traditional self-report data from surveys and interviews are highly recommended. Contextually sensitive and more ecologically valid approaches such as experience sampling methods (ESM; Xie et al., 2020) can be viable alternatives. While still self-report in nature, this kind of data can be more nuanced and accurate as cognitions, emotions, and behaviors are captured as they occur in real-time. Students’ actual engagement with AR tools and in situ behavior can also be derived from log file data and other learning analytics embedded in the software. These can offer better ways of inferring student engagement processes with digital technologies (Tempelaar et al., 2020).

Given the nature of AR, it made sense to apply the technology within experiential learning, which is what most researchers in the reviewed studies did. We recommend the development of apps that can be flexibly used in the school setting, in the field, and as part of students’ homework. Furthermore, as teachers are already laden with various responsibilities, care must be taken to not add to their burden in introducing pedagogical innovations. Thus, institutions are encouraged to invest in proper training and professional development opportunities for instructors to enable them to maximize the functions of AR to drive their students’ learning. Also, to successfully integrate AR in lessons, it is important to obtain not just students’ but also 2teachers’ perspectives in studies that examine the use of AR in environmental education.

The use of AR in environmental education is nascent. We would like to foreground the importance of conducting more research in the area, as there is already emerging evidence of AR’s potential as an effective pedagogical tool (AlNajdi, 2022). We recommend expanding research on the use of AR in environmental education to underrepresented regions globally. However, such research should be conducted with careful consideration of the appropriateness of the content of the tools in particular contexts, the costs attached to, and the resources required for integrating AR tools in school curricula.

In the introduction, we have briefly described the advantages of AR over other digital tools in environmental education. In acknowledgement of the importance of comparing different types of educational media, we strongly recommend backing up the assumptions presented with more empirical evidence. Comparative studies would allow for contrasting the benefits and pitfalls of different educational technologies with properly controlled key variables. This enables identification of which specific affordances of the various media are more effective in enhancing learning and engagement (Lawson et al., 2024).

Lastly, interest is hypothesized as developing through stages, over time, as learners maintain their engagement in the activities relevant to their interest (Hidi & Renninger, 2006; Renninger & Hidi, 2022). Longitudinal studies and evaluation of whole programs and courses that integrate AR in students' lessons are therefore necessary for gathering more robust evidence of AR’s effectiveness -in environmental education.