Implementing Augmented Reality Models in the Classroom Environment Using Merge Cubes: A Quantitative Study of the Effects on Students’ Cognitive Load and Motivation

Abstract

The present study investigates the extent to which the use of Merge Cubes as haptic AR tools in the classroom—realized in construction technology lessons at a vocational college as an exemplary case—influences the cognitive load and motivation of learners. A quasi-experimental field study was conducted using a questionnaire in a pre-post design including a control group at a vocational college in Germany (North Rhine-Westphalia). During the intervention phase, the students in the experimental group worked with materials such as textbooks and worksheets that were specifically expanded to include the Merge Cube AR learning tool, while the students in the control group only used conventional learning materials. In both the pre- and post-test, the cognitive load and motivation of the learners were recorded using questionnaires. The results indicate that the use of Merge Cubes can reduce cognitive load: the extraneous cognitive load of the experimental group decreased over the course of the intervention, whereas that for the control group increased significantly in comparison. In addition, the germane cognitive load increased slightly in the experimental group, whereas that for the control group decreased. With regard to the intrinsic motivation of the learners, both groups recorded an increase, although the difference between the two groups was not significant. Based on these results, further factors influencing the effect on learning and implications for the practical use of the Merge Cube in the classroom are discussed, the concrete validation of which requires further research.

1. Introduction

Augmented reality (AR) is likely a technology that many children, teenagers, and young adults use on their smartphones and tablets without realizing it. AR technology makes it possible to perceive the real environment and superimpose or combine it with virtual objects in real time (Azuma, 1997). In the game Pokémon GO, for example, virtual Pokémon are embedded in the player’s real environment. AR is also used in the social media sector, mostly via camera filters that project glasses, flowers, symbols, or other objects onto the user’s face.

AR is currently employed in a wide range of environments and domains (Lampropoulos, 2025). In addition to its entertainment value, future implementation of AR in various professions is expected to enable more productive and effective work. Settings for the specific use of AR can arise in production and logistics, where AR can enable more precise assembly and more efficient warehouse management by displaying relevant data directly in a worker’s real field of vision. In the area of maintenance and repair, AR applications can support technicians with superimposed real-time instructions, which can lead to faster problem-solving and reduced downtime. Specific technical, physical, and psychological effects associated with the use of new and innovative technologies must be identified. In this context, Barta et al. (2024) have identified the effects of AR as one of the most relevant fields of research.

As almost all modern smartphones and tablets support AR, the technology can be integrated easily in various contexts (Bitkom/Bundesverband Informationswirtschaft, Telekommunikation und neue Medien e. V., 2023). The low threshold to access also favors its implementation in the education sector, where AR is increasingly gaining ground (Lampropoulos, 2025). The education system is confronted with the task of redesigning teaching and learning settings in order to meet the challenges posed by digitalization (Thissen, 2017). In particular, the aim is to impart digital skills to pupils of all ages (Fehrmann, 2022, 2024). To achieve this, it is necessary—among other things—to “use the possibilities of new technologies and the internet sensibly and appropriately, i.e., to network the physical learning spaces, which need to be rethought, with the virtual spaces” (Thissen, 2017, p. 3, translated). Such networking can be supported by AR technology.

The Merge Cube is a concrete tool for implementing AR in the classroom. When the physical cube is held in front of the camera of a tablet, it allows users to hold and manipulate digital 3D objects and interactive content in their hands, creating a multi-sensory learning experience in various areas of education. While there are already numerous international studies and research projects on the implementation of AR in the education sector that have identified the potential of this technology, only a select few studies have analyzed the effects of specific AR tools in the classroom on learning processes and aspects of learning psychology to date. This study addresses this research gap in the form of a media comparison, using the Merge Cube as an example of AR technology.

2. Theoretical Background

2.1. Augmented Reality—A Definition

Augmented reality (AR) describes an increasingly transformative technology that extends the real, physical world with virtual information and objects (Nijholt, 2023). In contrast to virtual reality (VR), in which users are completely immersed in the virtual world, the physical environment remains visible in AR, while virtual elements such as text, graphics, or 3D animations are superimposed or linked to the environment in real time. The aim of AR is to combine digital and real components in such a way that they appear as natural and interactive as possible for users. As a variant of VR (Azuma, 1997), AR technology supplements reality, rather than replacing it completely.

This relationship has also been described as the reality–virtuality continuum, according to Milgram et al. (1994): the continuum serves as a theoretical framework to understand and categorize different degrees of blending of real and virtual elements in immersive technologies. Between the real environment, which exclusively comprises real, physical objects, and the virtual environment, which exclusively uses digital objects, augmented reality (AR) is defined as an extension of the real world with virtual content and augmented virtuality (AV) as a virtual environment with real elements.

The decisive factor for assigning an environment to one of the defined gradations is the extent to which elements of the real environment are combined with those of the virtual environment (Figure 1). In augmented reality, the proportion of real elements outweighs the proportion of virtual elements (Dörner et al., 2019), meaning that digital elements are added to the real world. AR can thus be described as positioned closer to the real world (Lampropoulos, 2025).

In addition to the combination of virtual and real elements, Azuma (1997) defined AR as the real-time interaction and three-dimensional registration of content. The superimposition of virtual content on the real environment enables the user to perceive reality in direct combination with virtual objects and to experience it as a coherent whole, whereby experiences and interactions are possible in real time (Dörner et al., 2019). With the help of special markers, the virtual content can be registered in a three-dimensional space (Sakti & Sejati, 2024); as such, the virtual objects appear to be firmly connected to reality and behave according to the user’s movements and changes in perspective (Dörner et al., 2019):

“Augmented reality (AR) is a (direct and interactive) perception of the real environment enriched with virtual content (for any senses) in real time, which is oriented as far as possible towards reality in its characteristics and appearance, so that in extreme cases (if this is desired) a distinction between real and virtual (sensory) impressions is no longer possible”.

(Dörner et al., 2019, p. 21, translated)

Recent developments in the field of AR have enabled more sophisticated interactions and the integration of autonomous (partly AI-supported) virtual content that improves the user experience beyond mere overlays (Nijholt, 2023). Considering these contemporary advancements in the domain of augmented reality (AR) technology, Nijholt (2023) underscored that, in contradistinction to technological facets, the definition and utilization of AR should prioritize human interactions with virtual objects. Consequently, when engaging with AR, the primary focus should be on sensory stimulation and the associated human experiences. This encompasses, inter alia, the integration of multiple senses beyond vision. Furthermore, Nijholt underscored that the control of AR content does not necessarily have to reside with the user. Instead, content can be introduced and controlled dynamically by the system (Nijholt, 2023). In the context of an AR-based simulation, for instance, virtual events such as weather changes or traffic simulations can be configured to operate autonomously, leveraging AI to enhance the overall AR experience. Meanwhile, AR is presently used in many areas of everyday life; for example, AR applications are being used in the field of e-commerce in the form of virtual try-ons (VTOs), enabling users to virtually put on glasses frames or project furniture into their own living room before making a purchase (Barta et al., 2024). Furthermore, face filters or AR games such as Pokémon GO are frequently used in the social media sector, and the first textbooks and learning applications that integrate AR are also being developed in the educational context.

2.2. AR in the Classroom

2.2.1. Potential and Challenges

The use of AR in an educational context is of great value, as it can be used to illustrate processes and structures that would otherwise not be accessible to learners or could not be experienced interactively (Knutzen & Howe, 2021; Sakti & Sejati, 2024). For example, AR can be used to visualize chemical reactions, make the internal organs of living organisms explorable without dissection, or allow historical developments to be experienced interactively. Through enabling learning through experience, AR promotes constructivist learning (Dunleavy & Dede, 2014; Hellriegel & Čubela, 2018). In AR learning environments, students are particularly encouraged to actively and physically interact with the learning objects, in order to interact with them intensively and perceive them in a multi-sensory manner (Schäfer et al., 2023). At the level of learning, further potential can also be derived from AR, such as a better understanding of complex concepts (Schweiger et al., 2022), the promotion of digital literacy, and increased motivation among pupils (Radu, 2014). However, it should be noted at this point that the increased motivation observed when using AR may be due to novelty-related effects, and might not last (M. Akçayır & G. Akçayır, 2017). Particularly regarding learning at vocational colleges, AR makes it possible to offer students a wide range of experiences from the world of work that were previously only associated with considerable material expenditure (Buehler & Kohne, 2020) or physical risks (Dietze, 2020). For example, work processes on machines can be simulated without time pressure, hazards, or burdening ongoing production processes with training (Fehling, 2017). At the same time, potentially dangerous situations or disasters can be simulated in a risk-free manner, allowing students to learn how to behave in an emergency.

The main challenge relating to the use of AR in a school context is that only a few service providers currently support hardware and software for learning with AR. In addition, schools must have the appropriate hardware, such as end devices and a fast internet connection (Fehling, 2017). Not only is the relevant technology under ongoing development and implementation, but the construction of suitable learning environments and learning content also ties up time and organizational resources (Thissen, 2017). Notably, AR simulations can often only be developed, to a certain extent, by teachers themselves (Buehler & Kohne, 2020) and must also be integrated into existing curricular learning content. Regarding financial aspects, it should be noted that, despite the long-term savings that can be associated with the use of AR, the technical implementation of end devices and learning environments can entail high costs (Fehling, 2017). Health risks are also a challenge hindering the use of such technology in the classroom. For example, the use of AR elements can cause simulator sickness (Christou, 2010) or lead to eye strain (Zender et al., 2022). There may also be risks of injury due to the nature of the real spaces in which the AR technology is to be used (Knoll & Stieglitz, 2022).

Based on current and technological developments, Lampropoulos et al. (2024) specifically highlighted the technical need to analyze processes of affective data processing in the context of AR, as the affective states of students in these environments may differ from those in traditional learning environments. The authors recommended, for example, considering the emotions, personality, characteristics, knowledge, and preferences of learners to enable adaptive and personalized learning and customized learning environments. The use of artificial intelligence can be particularly helpful in this regard; in particular, the combination of artificial intelligence (AI) and augmented reality (AR) makes it possible to optimize content and activity recommendations, generating individualized experiences in which interactive digital information is embedded in the user’s individual physical environment (Lampropoulos, 2025). This synergy between AI algorithms and AR technology facilitates the creation of customized interactions that are tailored to the user’s specific preferences and needs (Lampropoulos, 2025).

2.2.2. AR and Its Influence on the Cognitive Load in Learning Processes

In order to successfully complete learning processes, the cognitive characteristics of learners for processing and storing information, as well as motivational characteristics such as their own confidence in success or expectation of failure, are particularly relevant (Hasselhorn & Gold, 2013).

The Cognitive Load Theory (CLT), according to Chandler and Sweller (1991), is based on the assumption that the human working memory has limited capacity and that learners require capacity in their working memory to retrieve knowledge from the long-term memory, transfer it to the working memory and link it to new information there (Niegemann et al., 2008). Maresch also pointed out that the working memory has a limited working capacity (Maresch, 2006). Learning environments should, therefore, be designed in such a way that learning is facilitated by imposing a low cognitive load, thus enabling effective learning. A distinction is made between three types of cognitive load: intrinsic cognitive load (ICL, load resulting from the complexity of the content of the learning material), extraneous cognitive load (ECL, load resulting from the design of the learning material, the learning environment and the framework conditions for learning), and germane cognitive load (GCL, load resulting from the effort required by the learner to understand the learning material and construct new schemata). These three types of cognitive load add up to the total load on the working memory. An increase in one type of load leads to a reduction in the other two, given that the capacity of the working memory is limited.

AR- and VR-based learning environments can be equipped with a wide range of information at different levels of representation, allowing different sensory channels to be addressed for learning. However, the possible variety in the combination of information can also lead to an overload of the working memory and reduced learning success (Hamann et al., 2020; split-attention effect, redundancy effect).

In terms of cognitive load theory, it is relevant for the integration of AR in the classroom, in order to design learning environments in such a way that they optimally balance the cognitive load of students. Mayer (2005) recommended combining texts and images (multimedia effect); placing related elements as close together as possible (contiguity effect); avoiding unnecessary sounds, images, and words (coherence effect); presenting graphics and animations with auditory explanations (modality effect); avoiding duplication of content (redundancy effect); and highlighting important information (signaling effect). Through applying the principles of CLT, the effectiveness of AR technologies in education can be increased, which contributes to sustainable knowledge transfer.

2.2.3. AR and Its Influence on Motivation in Learning Processes

The motivation of learners is also relevant for successful learning processes, as it is a psychological process that serves to initiate, control, maintain, and evaluate goal-oriented action (Dresel & Lämmle, 2017; Edelmann & Wittmann, 2012). Motivation thus regulates the willingness of learners to engage with a learning subject with intensity, perseverance, and commitment (Hasselhorn & Gold, 2013). A distinction is made between two forms of motivation (Ryan & Deci, 2000): extrinsic motivation is caused by external stimuli and influences and is usually aimed at a specific outcome such as praise, recognition by third parties, or the maintenance of social status. Intrinsic motivation, on the other hand, is based on personal interest, results from one’s own drive, and leads to feelings of satisfaction and self-realization, which is why it is particularly important for effective learning processes.

Initial studies have shown that the use of AR can have a positive influence on the intrinsic motivation of learners. For example, through a systematic review of 68 studies, M. Akçayır and G. Akçayır (2017) illustrated that the use of AR in educational contexts achieves an increase in learning motivation and satisfaction as well as an improvement in learners’ attitudes towards learning. Radu (2014) and Probst et al. (2021, 2022) also found that the motivation of learners is increased by the immersive design of AR applications and that students enjoy working with AR applications more than conventional learning materials, even though they are sometimes more complicated to use.

2.2.4. The Merge Cube as a Learning Medium

The Merge Cube, developed by Merge Labs, is a concrete learning medium that makes AR experiences possible in the classroom. The Merge Cube is a small (7 × 7 × 7 cm) cube that can be purchased (consisting of foam), printed using a 3D printer, or assembled from paper (Figure 2).

When learners look at the cube through the camera of a tablet or smartphone, they can interactively hold digital 3D objects that are projected over the cube in their hands (Figure 3). For technical realization, a silver-colored pattern of markers reminiscent of hieroglyphics is printed on the surface of the Merge Cube (Matteson, 2018). Recognizing these symbols as fixed coordinate points in AR apps and overlaying them with the model coordinate points, the cube can be visually transformed into 3D objects such as skeletons, cells of the body, or historical buildings (Sperlich, 2022). When learners rotate the cube, the model that is visible on the screen also rotates. The models are supplemented with information in the form of text, sound, and images. This makes it possible, for example, to add real-time weather data to a model of the globe or to interactively display population distribution data. The Merge Cube can be used with three of the manufacturer’s apps—Merge Object Viewer, Merge Explorer, and Merge Holo Globe—which make it possible to project various models onto the cube, including self-designed ones. This learning approach, which combines haptic, visual, and auditory approaches, appeals to different learning styles and can render complex phenomena and models visible (Voštinár & Ferianc, 2023).

The fact that the Merge Cube makes it possible to integrate very large objects, such as a globe, up to microscopically small structures, such as those found in a plant cell, appears to be helpful for learning processes. In contrast to two-dimensional graphics, the models can be experienced interactively, opening up new learning experiences. In this way, the learning content can be anchored more sustainably through multi-sensory experiences (Knoll & Stieglitz, 2022). In addition, many models can be made available at the same time by using several Merge Cubes in the classroom. The option to build the Merge Cubes means they can be made available at low cost. In terms of infrastructure, a key challenge is that enough tablets with Wi-Fi must be available. In addition, individual models may be behind payment barriers, incurring further costs. In terms of lesson integration, the didactic and methodological integration involves effort, as new materials may have to be developed and teaching methods adapted.

2.2.5. Effects of Using the Merge Cube on Learning

To date, only a few empirical findings on the didactically effective embedding of the Merge Cube and concrete learning effects resulting from the use of the Merge Cube have been reported internationally. With regard to didactic embedding, Taufiq et al. (2021) used a self-developed module on the solar system, which was evaluated by various expert groups according to the ADDIE model, in order to show that the Merge Cube can be expected to promote problem-solving skills. Probst et al. (2021, 2022) have shown that the Merge Cube has significantly positive effects on learner motivation, school self-efficacy expectations, and learning success in chemistry lessons at secondary school, indicating its benefits for teaching practice with regard to the concrete potential of its use for learning. With regard to an extended understanding of complex concepts in the field of biomedicine, Lin (2021) was able to determine that more than 90% of the students surveyed rated the use of the Merge Cube as helpful. The aim of this study is to investigate the learning psychological effects associated with using the Merge Cube in the classroom, as indicated by both theoretical implications and the results of an initial study, with a focus on the effects on learner motivation and changes in cognitive load.

3. Empirical Study: Psychological Effects on Learning Based on the Use of the Merge Cube in Construction Technology Lessons

3.1. Research Question and Aim of This Study

As there is little existing research on the specific learning effects of AR tools, this study aims to answer the following question, using the Merge Cube as an example in vocational colleges: To what extent does the use of the Merge Cube in the classroom influence the cognitive load and motivation of learners?

The aim of this quantitative longitudinal study—in which students from experimental and control groups were surveyed at two measurement points (pre and post) using an identical questionnaire (Döring & Bortz, 2016)—was to determine whether the use of Merge Cubes leads to a significant reduction in extraneous cognitive load (which hinders learning), a significant increase in germane cognitive load (which promotes learning), and a significant increase in intrinsic motivation among students when compared to traditional learning methods.

The relevance of this research at the technological level is based on the fact that the multi-sensory interaction made possible by the Merge Cube can lead to a reduction in cognitive load, which is conducive to learning. This relief, which may be due to the technological features, needs to be verified. Despite initial indications of increased motivation with the use of AR elements, there are few controlled studies that have been conducted in an educational context. Therefore, it is important to identify causal effects resulting from the use of different AR technologies and to determine the best possible conditions for using AR in education, including, for example, the choice of specific learning tools and an optimal intervention duration. The objective of the present study is to generate preliminary findings on the learning effects associated with the use of the Merge Cube, as described in the following section. In view of these findings, specific recommendations for the processing of further research objectives are to be established, which include the design of interventions and the definition of AR design principles (Barta et al., 2024).

3.2. Design and Methodology

A quasi-experimental field study was conducted using a pre-post design involving an experimental group and a control group. The learners were surveyed in two class groups in the first year of training, which ensured internal validity, but did not involve randomization. Between the points of measurement, an intervention was carried out in which the experimental group used the Merge Cubes as AR tools in addition to the conventional learning materials in class.

The use of a pre-post design is justified by the fact that the primary aim was to measure the fundamental effects of the use of Merge Cubes in terms of direct cognitive and motivational responses. Cognitive load theory explicitly points out that extraneous cognitive load and germane cognitive load occur during the learning process (Chandler & Sweller, 1991) and can be measured in a valid manner immediately after the intervention. For this reason, a follow-up measure was not used in this study. Other reasons for not using a third post-intervention measurement are the relatively short duration of the intervention (as described in Section 3.3), the measurement instruments used (which record concrete states rather than longer-term changes), and the practical conditions of the classroom.

The use of the Merge Cube as a learning medium is the independent variable. In the experimental group, the use of the Merge Cube as a learning medium was explicitly requested via the work materials and tasks while, in the control group, no AR tools were used. The work materials differed only in the use of the Merge Cube, in order to minimize the need for co-funding. Intrinsic motivation and cognitive load (germane cognitive load, extraneous cognitive load) served as dependent variables.

As the cognitive load in learning processes and the students’ learning motivation, in particular, were to be recorded at the first and second points of measurement, a fully structured written digital survey was used on the basis of standardized questionnaires, in order to obtain self-reports (Döring & Bortz, 2016). In particular, the Cognitive Load Questionnaire (CLQ; Klepsch et al., 2017) was used to measure cognitive load. Its eight-item structure (e.g., “During the tasks in class, you had to process many things in your head at the same time”) with a seven-point Likert response format results in three scales to determine intrinsic cognitive load (ICL), extraneous cognitive load (ECL), and germane cognitive load (GCL). To determine the level of motivation, the short scale of intrinsic motivation (KIM, an abbreviation for the German “Kurzskala intrinsischer Motivation”; Wilde et al., 2009), which comprises 12 items (e.g., “I was able to control the activity in class myself.”), was used, to which a seven-point Likert scale was added for synchronization with the first questionnaire module, thus relieving the participants’ cognitive load (Döring & Bortz, 2016).

3.3. Survey Implementation and Sample

During the survey and intervention conducted in June 2024 at Wilhelm-Emmanuel-von-Ketteler-Berufskolleg Münster (vocational college), N = 37 students from carpentry classes who were first-year apprentices were surveyed. One class acted as the experimental group (n = 20; 17 male, 2 female, 1 non-binary) and one class as the control group (n = 17; 14 male, 3 female). Between pre-test and post-test, both classes took part in a 20 h intervention phase, which lasted from 3 to 13 June 2024.

3.4. Intervention and Learning Materials

The first module of the two-part teaching unit focused on promoting skills for the graphic representation of solids, in the context of which the pupils worked on 11 three-panel projection and parallel projection exercises. The students were familiar with the structure of both representation methods, such that their spatial thinking and technical drawing were to be trained and deepened. The experimental group was given the opportunity to use two Merge Cube models for each exercise task in addition to the 2D model printout: one model comprised the pure building structure, while the second model was the same but digitally supplemented with dimensions, labels, and other aids. These and all other materials differed between the experimental group and the control group only in that students in the experimental group had access to the three-dimensional Merge Cube model in addition to the two-dimensional image, in order to minimize any confounding effects.

In the second module, the students were asked to plan a garden tool shed. A prefabricated model, which was provided to the control group in isometric form (i.e., as a three-dimensional printout; Figure 4, left), was available to the experimental group as a Merge Cube model (Figure 4, right). The use of the Merge Cube by the students in the experimental group was actively encouraged by the teacher.

This teaching unit contributed in particular to the promotion of professional competence, in that the pupils had to solve the drawing and planning tasks on the basis of their professional knowledge and skills in a goal-oriented, independent, and appropriate manner. Self-competence was promoted by the fact that the students worked independently, questioned their decisions, and had to take responsibility for their results. Collaboration during the project work also trained social and communication skills.

3.5. Data Entry and Evaluation

Two scales for measuring germane and extraneous cognitive load, as well as a scale for measuring intrinsic motivation, were created at the two points of measurement in accordance with the layout of the original instruments. In addition to descriptive evaluations, t-tests were carried out to identify potential differences between the experimental and control groups.

4. Presentation of Selected Results

4.1. Results for the First Point of Measurement

The mean scores in the three scales recorded at the first measurement point are shown below, broken down by group (

Table 1. Comparison of pre-assessments and test scores between the experimental and control groups based on mean scores in pre-test.

	EC		CG
	M	SD	M	SD	t	FG	p	d
GCL_Pre	5.28	0.60	5.33	1.17	−0.16	22.85	0.875	0.91
ECL_Pre	3.72	0.93	3.59	1.15	0.38	35	0.710	1.04
IM_Pre	4.22	0.62	4.48	1.02	−0.95	35	0.349	0.83

Note: EG: n = 20, CG: n = 17, p-value = two-sided, * p < 0.05.

). The difference in means between the groups was classified using a t-test.

With regard to the comparison of the self-assessments of germane cognitive load and extraneous cognitive load between the students in the experimental and control groups, it can be seen that the students in the experimental group rated their learning-related cognitive load (GCL_Pre, M = 5.28, SD = 0.60) similar to that of the students in the control group (GCL_Pre, M = 5.33, SD = 1.17) [t(22.85) = −0.16, p = 0.88, d = 0.91]. There was no significant difference between the groups.

The same applied to the additional, non-learning-related cognitive load (ECL); in particular, the values of the experimental group (ECL_Pre, M = 3.72, SD = 0.93) did not differ significantly from those of the control group (ECL_Pre, M = 3.59, SD = 1.15) [t(35) = 0.38, p = 0.71, d = 1.04].

With regard to the assessment of intrinsic motivation, it is also clear from the data that the two groups assessed themselves similarly. The assessment of the students in the experimental group with regard to their intrinsic motivation (IM_Pre, M = 4.22, SD = 0.62) was comparable to that of the students in the control group (IM_Pre, M = 4.48, SD = 1.02) [t(35) = −0.95, p = 0.35, d = 0.83]. According to Cohen (1988), there is a strong effect.

The evaluation of the t-test indicates that the experimental and control groups were comparable, in terms of the three assessed areas of learning psychology, before the intervention phase.

4.2. Results for the Second Point of Measurement

The mean scores in the three scales recorded at the second measurement point are shown below, broken down by group (

Table 2. Comparison of post-assessments and test scores between the experimental and control group based on mean scores in post-test.

	EC		CG
	M	SD	M	SD	t	FG	p	d
GCL_Post	5.37	1.01	4.92	0.84	1.43	34	0.163	0.94
ECL_Post	3.10	0.69	3.98	1.09	−2.94	34	0.006 *	0.89
IM_Post	4.68	1.02	4.53	0.74	0.51	34	0.614	0.91

Note: EG: n = 20, CG: n = 17, p-value = two-sided, * p < 0.05.

). The difference in the means between the groups was classified using a t-test.

With regard to the comparison of the self-assessed GCL between the students in the experimental and control groups at the second point of measurement (i.e., after the intervention), it was found that the students in the experimental group rated their learning-related cognitive load higher (GCL post, M = 5.37, SD = 1.01) than those in the control group (GCL post, M = 4.92, SD = 0.84) [t(34) = 1.43, p = 0.16, d = 0.94]. According to Cohen (1988), this difference is large. Despite the high effect size, the difference was not statistically significant (p < 0.05).

There was a significant difference with regard to the additional non-learning-related cognitive load (ECL). The students in the experimental group rated their extraneous cognitive load significantly lower in the post-test (ECL_Post, M = 3.10, SD = 0.69) than the students in the control group (ECL_Post, M = 3.98, SD = 1.09) [t(34) = 2.94, p = 0.006, d = 0.89]. According to Cohen (1988), the effect size can be classified as high.

Assessment of the motivation at the second point of measurement also demonstrated that the students in the experimental group rated their intrinsic motivation higher (IM_Post, M = 4.68, SD = 1.02) than those in the control group (IM_Post, M = 4.53, SD = 0.74) [t(34) = −0.51, p = 0.61, d = 0.91]. According to Cohen (1988), there is also a strong effect here; however, the difference between the groups was not significant.

4.3. Results of the Changes Between the First and Second Points of Measurement

The changes in the assessments of cognitive load and motivation between the first and second point of measurement were analyzed. The differences in the mean values between the pre- and post-test measurements were used for this purpose, as shown below (

Table 3. Comparison of changes in assessments and test scores between the experimental and control group using independent t-tests, based on the difference in pre-post mean scores.

	EC		CG
	M	SD	M	SD	t	FG	p	d
GCL_Diff	0.08	1.13	−0.52	0.76	1.83	34	0.076	0.98
ECL_Diff	−0.62	1.13	0.52	0.73	−3.48	34	0.001 *	0.97
IM_Diff	0.46	1.20	0.10	0.83	1.04	34	0.308	1.05

Note: EG: n = 20, CG: n = 17, p-value = two-sided, * p < 0.05.

).

The results indicate that the experimental group tended to show greater changes in the measured scales when compared to the control group.

The experimental group showed a slight increase in germane cognitive load (GCL_Diff, M = 0.08, SD = 1.13), while the control group showed a decrease (GCL_Diff, M = −0.52, SD = 0.76) [t(34) = 1.83, p = 0.076, d = 0.98] (Figure 5). The difference in development was not significant in the extent of its difference. Analyses of variance revealed that group affiliation (experimental group, control group) had no significant influence on the development of the GCL [F(1,34) = 3.36, p = 0.076, η_p² = 0.090, n = 37].

Significant differences were found in particular regarding the extraneous cognitive load and, thus, the additional non-learning-related cognitive load. The extraneous cognitive load significantly differed for the students in the experimental group (ECL_Diff, M = −0.62, SD = 1.13), when compared to the control group (ECL_Diff, M = 0.52, SD = 0.73) with a strong effect (Cohen, 1988) [t(34) = −3.48, p < 0.01, d = 0.97] (Figure 6). Overall, the ECL value in the experimental group showed a decrease after the intervention (M < 0), whereas that of the control group increased (M > 0). Analyses of variance revealed that group affiliation (experimental group, control group) had a significant influence on the development of ECL [F(1,34) = 12.12, p < 0.001, η_p² = 0.263, n = 37].

With regard to the changes in the learners’ motivation, it should be emphasized that the students in both groups experienced an increase in their intrinsic motivation (M > 0). Notably, the increase in the intrinsic motivation of the students in the experimental group (IM_Diff, M = 0.46, SD = 1.20) was not significantly minimally higher than that of the students in the control group (IM_Diff, M = 0.10, SD = 0.83) [t(34) = 1.04, p = 0.308, d = 1.05] (Figure 7). Analyses of variance revealed that group affiliation (experimental group, control group) had no significant influence on the development of intrinsic motivation [F(1,34) = 1.07, p = 0.308, η_p² = 0.308, n = 37].

5. Discussion and Outlook

5.1. Classification of Central Results in the Discourse

Hoffmann (2023) has pointed out that AR technologies have the potential to increase the Germane Cognitive Load (GCL) if they are effectively integrated into the classroom. According to Maresch (2006), learning materials should be designed in such a way that the GCL is maximized in the best possible way. The results of the present study show that the GCL, which is conducive to learning, was above the theoretical mean in both groups before the intervention. After the intervention, the GCL in the experimental group increased slightly, while that in the control group decreased. Although there was only a slightly positive, non-significant increase compared to the control group when using the Merge Cube, this small increase may indicate that the use of the Merge Cube had a positive effect on the learning process. The reduction in the GCL in the control group could indicate that traditional learning methods have a less activating and learning-promoting effect, and that the students were not adequately challenged, such that their working memory retained free capacity. The non-significant differences in the comparison between the groups could indicate that the implementation of Merge Cubes in construction technology lessons still needs to be further optimized, despite the basically methodically varied learning material. Possible reasons for the non-significance of the GCL development may lie in the short duration of the intervention, which suggests the need for longer-term research designs analyzing the concrete effects of using the Merge Cube.

Furthermore, it is important to reduce the Extraneous Cognitive Load (ECL), which hinders learning, in order to ensure a successful learning process that is as individualized as possible, as ECL takes up unnecessary cognitive resources and distracts learners from the actual learning process (Maresch, 2006). The presented results illustrate that the ECL before the intervention was below the theoretical mean value in both groups. After the intervention, the development of ECL differed significantly between the groups, with the ECL in the experimental group decreasing and that in the control group increasing. The finding that the ECL was reduced for students in the intervention group using the Merge Cube is consistent with the results of M. Akçayır and G. Akçayır (2017) and Santos et al. (2014), who argued that AR applications can reduce cognitive load by making information more accessible and understandable. The results obtained are also in line with those of Hoffmann (2023), who proved that AR applications can help reduce ECL. Although Hamann et al. (2020) pointed out, with regard to the design of VR and AR learning environments, that split-attention effects and redundancy effects can lead to learners having difficulties concentrating on the actual learning objective due to the large amount of information and stimuli offered, it can be deduced from the results presented here that the use of Merge Cubes may counteract these potentially negative effects. The targeted presentation of information on the cube reduces unnecessary non-learning-related cognitive stress.

Chiang et al. (2014) have shown that the use of AR technology has the potential to increase student motivation. The trends recorded in the present study, which derive from the use of the Merge Cube, tend to be congruent with the positive effects of AR on student motivation in the educational context reported in the literature, although the change in both groups was not significant in the comparison. Intrinsic motivation in both groups was above the theoretical mean value before the intervention. During the intervention phase, intrinsic motivation increased in the experimental group, whereas it decreased slightly in the control group.

Probst et al. (2021, 2022) have empirically demonstrated that the use of the Merge Cube resulted in a significant increase in motivation among secondary school students. The duration of the intervention, in particular, can be cited as a possible influence for the fact that the motivation of the vocational college students who took part in this study did not increase significantly when compared to the control group, while the total duration of the investigation in the study by Probst et al. (2022) was only 80–85 min and the investigation in the study by Probst et al. (2021) lasted over 6 lessons, the students in the experimental group in the present empirical study engaged with the Merge Cube as an AR learning tool over a period of 20 lessons within 11 days. M. Akçayır and G. Akçayır (2017) have drawn attention to the possible existence of habituation effects; for example, a significant increase in intrinsic motivation could be observed in shorter intervention phases, whereas no significant increase in motivation could be achieved in a somewhat longer intervention phase, as was the case in this study. Further studies should analyze whether a change in the form of motivation may have occurred, as described by Ryan and Deci (2000), in which the initial intrinsic enthusiasm changed to a motivation determined by external circumstances.

Other side effects that could explain the lack of a significant change in motivation may be due to the design of the work material (Meyer, 2022). Although attention was paid to the development of models with the highest possible level of multimedia, adaptivity and interactivity when designing the work materials and two variants of the same three-dimensional model were typically offered with different levels of support, in order to take into account the heterogeneity of the students, the design of the work materials and the didactic implementation may not have been sufficiently optimized to determine the full effect of the AR technology. As such, an even more differentiated adaptation of the learning materials to the individual needs of the learners may be necessary in order to significantly increase intrinsic motivation in the long term.

5.2. Limitations of the Study

The study design, in the form of a quasi-experimental field study with a pre–post design, was considered to be fundamentally effective in answering the questions. In follow-up studies, the implementation of a third follow-up point of measurement would be desirable, allowing medium- and long-term effects to be assessed (Döring & Bortz, 2016). In order to use alternative treatments as comprehensively as possible to obtain adequate control groups (Spörer & Glaser, 2010), it would be interesting to form a further study group in addition to the experimental and control group, which works with worksheets and tablets, in order to record further side effects.

The fact that this study was conducted as a comparative media study should be viewed critically. The challenges posed by comparative media studies result from the fact that they primarily attempt to record the effect of a digital medium, but often pay too little attention to the diverse, layered, and deep structures comprising the didactic and methodological design of a learning situation. A future research design to assess the impact of analysis of the Merge Cube should therefore primarily focus on how, when, and under what conditions learning with AR produces positive effects (Buchner & Kerres, 2022).

The sample used in this study was limited to students from two non-randomized learning groups from the construction technology sector at a vocational college, which restricts the generalizability of the results to other types of schools and courses of education. Nevertheless, the selected sample can be considered suitable for answering the research question and the study groups can be classified as comparable, as the results obtained at the first point of measurement indicated that the experimental and control groups were sufficiently comparable before the intervention phase. For future surveys, it would be desirable to recruit a larger sample that represents a more balanced proportion of students from a wider range of training occupations and vocational college courses.

The research instrument used is generally considered to be effective, although it should be noted regarding data collection that the students were aware during the reactive survey that they were taking part in a scientific study. As such, resulting distortions cannot be ruled out (Döring & Bortz, 2016). Regarding the intervention design and implementation, it should be noted that the survey at the first point of measurement took place directly in the morning at the start of lessons, while the survey at the second point of measurement took place during the last lesson of the day. It is possible that the learners were no longer as concentrated as they were at the beginning of the day, given that the school day had already progressed. Further limitations may result from the duration and length of the intervention. Regarding the inferential statistical analysis of the collected data, no further limitations were identified in addition to the limitations of the study design, sample generation, and questionnaire construction resulting from the methodological decisions already presented.

5.3. Research Desiderata and Practical Teaching Limitations

The present empirical study confirmed that the use of Merge Cubes as AR learning tools in vocational school lessons leads to a significant reduction in extraneous cognitive load when compared to traditional learning methods. This result supports the assessment frequently found in the literature that AR technology can reduce cognitive load by making information more accessible and easier to understand. Despite a positive trend, no statistically significant increase was found regarding the germane cognitive load, which promotes learning. These results indicate that the implementation of Merge Cubes in the classroom and the design of learning materials must be further optimized to maximize the cognitive resources that are conducive to learning in students.

To fully exploit further potential with regard to effective learning, increased learner motivation, and to integrate the Merge Cube as a learning tool into lessons in the long term, it is necessary to develop or adapt suitable teaching materials at a meta-level and to empirically evaluate the interplay between didactic design and the effects of the AR tool.

On a meta-level, the results also suggest—in comparison with the international discourse—that it is worthwhile for educational institutions to invest in the technology and train their teachers in the use and application of Merge Cubes. An important prerequisite for the implementation of the AR tool in the German education system is the development and provision of specific teaching materials tailored to the Merge Cube, in order to fully exploit the potential of this innovative technology.

There are research desiderata regarding the use of AR in the classroom, particularly regarding the long-term intervention effects and possible differences based on the type of school and the age of the pupils. In this regard, it would be desirable to develop empirical test procedures that could be used at different class levels and in subjects with a similar structural design.