Virtual reality (VR) has gained substantial attention as a transformative technology for STEM education, with meta-analyses reporting positive effects on learning outcomes across diverse contexts (Cromley et al., 2023; Radianti et al., 2020). The immersive, three-dimensional nature of VR offers unique affordances for visualizing complex spatial relationships, which are particularly valuable in disciplines such as anatomy, chemistry, and engineering, where traditional two-dimensional representations may limit student understanding (Makransky and Petersen, 2021).

However, a critical examination of the VR education research literature reveals significant methodological limitations that constrain the applicability of findings to authentic classroom contexts. First, student exposure to VR is frequently restricted to brief sessions–often 15–60 min–limiting students’ ability to overcome initial learning curves with VR interfaces and capitalize on the technology’s educational affordances (Lawson et al., 2024). Second, the predominance of experimental and quasi-experimental designs with pre- and post-comparisons treats VR as a fixed intervention rather than a complex, evolving instructional medium (Radianti et al., 2020). Third, instructors are typically excluded from research designs despite their pivotal role in facilitating technology integration and shaping classroom learning experiences.

Beyond methodological concerns, practical barriers persist in impeding the widespread adoption of VR in higher education. These include significant hardware and software costs, physical discomfort and motion sickness experienced by some users, variation in display quality across price points, and substantial learning curves associated with VR interfaces (Jensen and Konradsen, 2018; Saredakis et al., 2020). Students’ prior VR experience–whether through gaming or other contexts–may not transfer to proficiency with educational VR platforms, as recreational and educational VR often require different skills and orientations (Makransky et al., 2019).

These limitations highlight the need for qualitative research that examines extended VR use in authentic classroom settings, considers both instructor and student perspectives, and captures the evolving dynamics of technology integration over time.

Three specific gaps in the research literature motivate this study. First and foremost, there is a temporal gap. The predominance of short-term interventions hinders understanding of how VR integration dynamics evolve over extended instructional periods, such as academic semesters. Secondly, there is a perspective gap. The systematic exclusion of instructors from VR research designs obscures the ways in which their knowledge, beliefs, and pedagogical decisions shape student experiences. Lastly, the theoretical gap, characterized by the limited application of established frameworks–particularly those connecting instructor knowledge to student motivation–constrains the field’s theoretical development.

To address these gaps, this study employs a focused ethnographic approach (Andreassen et al., 2020; Rashid et al., 2019) to examine the semester-long integration of VR in an undergraduate neuroanatomy laboratory course. We applied two complementary theoretical frameworks: the Technological Pedagogical Content Knowledge (TPACK) framework (Mishra and Koehler, 2006), to analyze instructor knowledge profiles, and the Expectancy-Value Theory (EVT) (Eccles and Wigfield, 2002), to examine student motivational dynamics. By integrating these frameworks, we aim to illuminate how instructor knowledge shapes student motivation and experience in VR-enhanced learning environments.

Three research questions (RQs) guided this investigation.

RQ1: How does the instructor’s TPACK profile shape the design and implementation of semester-long VR-integrated instruction?

The TPACK framework, introduced by Mishra and Koehler (2006), provides a comprehensive lens for understanding the complex knowledge required for effective technology integration in teaching. Building on Shulman’s (1986) construct of Pedagogical Content Knowledge (PCK), TPACK suggests that teaching with technology requires the interplay of three primary knowledge domains: Technological Knowledge (TK), Pedagogical Knowledge (PK), and Content Knowledge (CK).

The TPACK framework identifies seven interrelated constructs (Table 1). Of particular relevance to VR integration is Technological Pedagogical Knowledge (TPK), which encompasses knowledge of how teaching and learning are affected when specific technologies are used. Prior research suggests that TPK is often underdeveloped among faculty who adopt immersive technologies, as they may understand VR capabilities and possess strong pedagogical skills separately but struggle with effectively scaffolding technology-mediated learning (Bower et al., 2020; Shin and Kim, 2024).

For VR specifically, TPK encompasses critical pedagogical competencies that extend beyond basic technology operation (TK) or understanding VR’s representational affordances (TCK). These include structuring progressive guidance sequences for VR skill development, managing troubleshooting and technical support in immersive learning environments, scaffolding the integration of multimodal content (visual, spatial, haptic), and adapting assessment approaches for immersive contexts. While TK and TCK are necessary foundations, they are insufficient without TPK knowledge of how to orchestrate the technology-mediated learning process effectively.

Notably, no formalized TPACK-VR framework currently exists in peer-reviewed literature. Researchers have adapted the original framework to immersive contexts rather than applying a standardized VR-specific version (Lee and Wu, 2024; Thohir et al., 2023). This study contributes to this emerging area by operationalizing TPACK for semester-long VR integration and identifying VR-specific manifestations of each construct.

To understand student motivational dynamics in VR-integrated learning environments, we draw on Expectancy-Value Theory (Eccles and Wigfield, 2002), which has been recently updated as Situated Expectancy-Value Theory (SEVT) to emphasize the context-bound nature of motivation (Eccles and Wigfield, 2020). EVT believes that achievement-related choices and performance are predicted by two key factors: expectancy beliefs (individuals’ beliefs about how well they will perform) and subjective task values (the reasons for engaging in the task).

EVT identifies four value components: intrinsic value (enjoyment derived from the task itself), utility value (usefulness for future goals), attainment value (personal importance tied to identity), and cost (negative aspects of engagement). Following Barron and Hulleman (2015), we treat cost as a distinct component, not merely the inverse of value, encompassing effort cost, opportunity cost, psychological cost (often includes ego cost), and emotional cost.

Direct EVT applications to VR education remain limited but instructive. Ferdinand et al. (2023) conducted an experimental intervention manipulating perceived usefulness (utility value) before a VR biology lesson with 196 10th-grade students, finding significantly greater learning achievement when utility value was primed. While this secondary school context differs from our undergraduate neuroanatomy setting, the underlying motivational dynamics, particularly how value perceptions shape engagement, provide theoretical grounding for examining VR-integrated learning. Broader EVT applications in STEM education offer additional support for our framework. Perez et al. (2019) identified distinct expectancy-value-cost profiles predicting STEM persistence, with students showing moderate expectancy/values combined with high costs demonstrating the worst outcomes. This conceptualization proves particularly relevant for technology-mediated learning contexts as students navigate content learning and technology mastery simultaneously (Perez et al., 2019). The dual learning demand can create cognitive and emotional burdens that compete with perceived value, potentially undermining motivation even when students recognize the technology’s potential benefits. While the specific technologies differ (VR headsets vs. laboratory equipment or software), the underlying EVT mechanism, i.e., how cost perceptions can overwhelm value perceptions, should operate similarly across technology-intensive learning environments. Our study examines whether these theoretically grounded EVT dynamics manifest in extended VR integration and how they interact with instructor knowledge profiles. Specifically, this study identifies five constructs that depict students’ EVT profiles in a VR-integrated learning environment (Table 2).

A notable gap in the research literature is the lack of frameworks that connect instructor knowledge to student motivation in technology-mediated contexts. TPACK and EVT are coupled in this study to capture both instructor and student dimensions of VR integration. Specifically, we examine how the instructor’s TPACK profile, particularly their TPK regarding knowledge of how to scaffold technology-mediated learning, relates to the EVT dynamics experienced by students. We contend that effective VR integration requires alignment between instructor TPK and student EVT profiles, including their expectancy beliefs, value perceptions, and cost experiences. Importantly, we do not assert that TPK directly causes specific EVT outcomes. Rather, we propose a mediating pathway: instructor TPK shapes pedagogical decisions (scaffolding approaches, success criteria communication, support structure design), which in turn influence the quality of students’ technology-mediated learning experiences, thereby affecting their EVT dynamics. Misalignment in this relationship may explain why technically successful implementations, where VR functions properly and content is well-designed, can still produce negative student experiences.

We hypothesize specific TPK-EVT alignment mechanisms for three EVT constructs that are most responsive to in-class pedagogical decisions: 1) TPK and Expectancy Beliefs: When instructors lack TPK knowledge of how to scaffold VR learning progressively, they may fail to establish clear success criteria or provide appropriate performance feedback. Students receiving ambiguous guidance about what constitutes successful VR task completion will struggle to develop accurate expectancy beliefs about their ability to succeed, regardless of their actual technical competence. 2) TPK and Intrinsic Value: When instructors possess TPK knowledge of how to leverage VR’s unique affordances pedagogically, not just understanding what VR can represent (TCK) but how to structure engaging exploration sequences, students are more likely to experience intrinsic value through meaningful interaction with immersive content. Conversely, poor TPK may lead to VR activities that feel like technology obstacles rather than learning opportunities. 3) TPK and Cost: When instructors lack TPK of VR classroom management, including how to structure technical support, troubleshoot efficiently, and minimize downtime, students experience elevated costs from time spent on technical issues rather than learning. These costs include effort cost (cognitive load from simultaneous technology and content learning), time cost (class periods consumed by troubleshooting), and emotional cost (frustration from technical barriers).

Two EVT constructs, i.e., utility value and attainment value, were not included in our initial hypotheses because they are shaped primarily by factors outside instructor control during a single semester. Utility value depends on students' pre-existing career goals and perceptions of how course content connects to those goals, while attainment value reflects identity development and personal importance typically established before course enrollment. While instructors can attempt to communicate utility (e.g., explaining career relevance), students' perceptions of utility are often determined by broader curricular context and life goals. We remain open to discover how TPK might interact with these constructs but focus our framework on the three EVT components most malleable through instructional practice: expectancy, intrinsic value, and cost. We envision this TPACK-EVT integration provides a theoretical foundation for examining how instructor knowledge shapes student motivation in VR-integrated instruction in this current study.

This study employs a focused ethnographic case study approach. While traditional ethnography involves extended cultural immersion over months to years, focused ethnography adapts ethnographic methods for bounded contemporary settings where time constraints and specific research questions necessitate more concentrated data collection (Knoblauch, 2005; Pink and Morgan, 2013). In educational research, focused ethnography is suitable for examining interaction patterns within specific institutional contexts, such as classrooms, over defined periods, such as academic semesters (Wall, 2015).

Focused ethnography is characterized by: (a) intensive, time-limited observation; (b) data-intensive methods (video, audio, detailed notes) that compensate for temporal brevity; (c) focus on specific interactions rather than entire cultural systems; and (d) researcher “alterity” and shared background knowledge enabling efficient and intersubjective interpretation (Knoblauch, 2005). These characteristics align well with this study, which examines the integration of VR in higher education. The semester provides natural boundaries, VR interactions generate rich observational data, and researchers typically share an educational background with participants.

The study was conducted at a public comprehensive university in Central California during the Spring 2024 semester. The focal course was a neuroanatomy laboratory, paired with a lecture course taught by a different instructor. The laboratory instructor was a neuroscientist by training who had participated in a year-long extended reality (XR) faculty learning community (XR-FLC) the previous year, where she was introduced to VR and piloted initial integration (Wu et al., 2025).

Despite formal faculty development preparation through the XR-FLC, including technical training and pedagogical support, the instructor still encountered substantial TPK challenges during semester-long implementation. This suggests that the 1-year professional development, while valuable for TK and TCK development, may still be insufficient for developing robust TPK for extended VR integration. The persistent TPK gaps observed in our study, despite intentional preparation, underscore the need for ongoing, iterative faculty support rather than one-time training interventions.

The course integrated VR throughout the semester using syGlass, a specialized third-party VR platform for volumetric data visualization (Figure 1). Students worked with actual MRI brain scans rather than simplified anatomical models, a deliberate design choice by the instructor to provide authentic clinical training experiences. The semester included two major VR-integrated components: (1) individual work identifying gross anatomical structures and neural pathways in VR, and (2) collaborative team projects creating narrated VR tutorials about brain function disorders using real patient data provided by syGlass.

The laboratory instructor was a female faculty member with a doctoral degree in behavioral neuroscience and extensive experience integrating technology into research and teaching. She had a strong background in physics and biophysics that informed her technological orientation.

Thirty-eight students were enrolled across two laboratory sections (n = 19 each). Students were primarily biology, health sciences, and chemistry majors taking the neuroanatomy course as an elective. Six students (three from each section) participated in the focus group. Participants were purposively sampled to represent diversity across multiple dimensions: gender (4 female, 2 male), academic major (biology, health sciences, and chemistry), and observed engagement levels during classroom observations (ranging from highly engaged to visibly struggling). This sampling strategy aimed to capture a range of EVT experiences rather than assuming uniform student responses to VR integration. Focus group recruitment occurred toward the end of multiple classroom observations, allowing the research team to identify students representing different engagement patterns observed across the semester.

Institutional Review Board approval was obtained. The instructor provided written informed consent for observation, interview, and use of course materials. Focus group participants provided separate informed consent. All participants were assigned pseudonyms.

Multiple data streams were collected to enable triangulation (Table 3), which entails: 1) Observation Protocol: The principal investigator and two graduate research assistants conducted classroom observations using a structured field note template that included timestamps, descriptive notes, analytical memos, and theory-linked observations (TPACK and EVT indicators). Observations captured the physical environment, instructional activities, participant behaviors, verbal interactions, and technical events. 2) Interview Protocol: The instructor interview followed a semi-structured protocol addressing educational and professional background, technology integration philosophy, TPACK-related knowledge domains, course design decisions, reflections on implementation challenges, and plans for future iterations. 3) Focus Group Protocol: The student focus group addressed: course expectations, VR technology experience, EVT components (expectancy, values, costs), challenges encountered, and recommendations for improvement. 4) Course Documents: The instructor provided the research team with the course’s Canvas shell. Access to syllabus, assignments, grading rubrics, and announcements captured classroom management dynamics and instructional design intentions.

Data analysis followed an iterative process integrating deductive and inductive approaches (Miles et al., 2014). The first cycle employed a priori codes derived from TPACK (seven constructs) and EVT (five constructs). Codebooks were developed collaboratively by the research team, specifying: (a) code definitions from theoretical literature (e.g., TPK defined per Mishra and Koehler, 2006), (b) operational definitions for the VR context (see Tables 1, 2), (c) inclusion and exclusion criteria for each code, and (d) illustrative examples from pilot coding.

To illustrate our coding approach: When the instructor stated, ‘I wanted them to be exposed into the real MRI scans… put that knowledge into the real live data,’ this was coded as TCK because it demonstrated understanding of how VR uniquely represents neuroanatomical content in clinically authentic ways. When a student stated ‘I had no idea what my expectations were of me. And then it just felt like I was falling short of them constantly,’ this was coded as low expectancy belief (EVT) because it reflected uncertainty about success criteria and diminished confidence in ability to meet unstated performance standards. Field observation noting ‘She [instructor] is running around putting out fires all class. She has not been able to sit down once’ was coded as TPK gap (specifically, inadequate support structure planning) because it illustrated consequences of insufficient pedagogical knowledge for managing VR classroom logistics. The second cycle involved axial coding to identify relationships among codes and between data sources. For example, we examined how instructor TPK patterns (from interview and observations) related to student EVT dynamics (from focus group and observations). The third cycle employed selective coding to develop assertions addressing each research question, explicitly requiring triangulation across at least two data sources for each major claim.

To ensure inter-coder reliability, the principal investigator and the graduate research assistants independently coded a subset of data (one observation, one interview excerpt, one focus group excerpt) using the theoretical coding scheme. Disagreements were resolved through discussion until a consensus was reached, with the facilitation of the external research partner (remote). The reconciled coding scheme was applied to complete coding of the remaining data with bi-weekly peer debriefing.

With multiple streams of data, triangulation was performed by verifying convergent evidence from at least two data sources for each major finding. Findings were documented with supporting quotes from multiple sources, and disconfirming evidence was actively sought. Throughout the analysis, researchers maintained analytical memos that documented emerging interpretations, connections to theory, and methodological decisions, thereby providing an audit trail for the research process.

In a prior funded project, the principal investigator led the XR-FLC in which the instructor participated, creating a collegial relationship that facilitated access but required attention to potential bias. In the current study, the PI maintained a non-participant observer role during classroom observations and employed peer debriefing with graduate research assistants unfamiliar with the instructor to challenge interpretations. The research team brought backgrounds in educational technology, STEM education, and qualitative methods, informing and potentially shaping analysis.

Corresponding to RQ1, the instructor’s TPACK profile was characterized by strong Technological Content Knowledge (TCK), i.e., a clear understanding of how VR could uniquely represent neuroanatomical content, but developing Technological Pedagogical Knowledge (TPK), creating a gap between her vision for VR integration and her capacity to scaffold that vision in practice.

Corresponding to RQ2, student experiences were systematically shaped by EVT dynamics, particularly the interaction between low expectancy beliefs, misaligned value perceptions, and disproportionate costs that exceeded perceived benefits.

Corresponding to RQ3, the relationship between instructor TPACK and student EVT was characterized by misalignments, specifically, the instructor’s TPK gaps directly shaped the negative EVT dynamics students experienced. Both parties recognized this misalignment but were unable to fully resolve it during the semester.

As a focused ethnography study, the research team also conducted a temporal analysis (Figure 2) supported by the comprehensive collection of multiple streams of data and cross-data triangulation. The temporal analysis contributes to a better understanding of how TPACK and EVT patterns evolved across the semester, categorized into four different phases, including Phase 1: Early/VR Introduction (Weeks 1–4), Phase 2: Skill Development (Weeks 5–8), Phase 3: Intensive/Team Project Work (Weeks 9–12), and Phase 4: Culmination (Weeks 13–16).

This research study is ultimately directed at understanding how STEM instructors can integrate VR in classroom settings more effectively. Based on our TPACK-EVT framework and empirical findings, we offer actionable recommendations for three stakeholder groups.

Several limitations should be acknowledged and shape the interpretation of our findings. First, this single-case study conducted at one institution with one instructor limits statistical generalizability. Our findings should be understood as an in-depth examination of one VR integration implementation rather than patterns guaranteed to replicate across all contexts. The theoretical framework we propose (TPACK-EVT integration) requires testing across multiple implementations to establish broader validity. The instructor’s specific background (neuroscience PhD, physics training, prior XR-FLC participation) may not represent typical faculty attempting VR integration. Second, the student focus group (n = 6), while purposively sampled for diversity, may overrepresent students with strong opinions, either particularly positive or negative, who were motivated to participate. Students with moderate or ambivalent experiences may be underrepresented. While we triangulated focus group data with classroom observations to mitigate this limitation, claims about cohort-wide patterns should be interpreted cautiously. The focus group provides depth of understanding about EVT dynamics but cannot establish prevalence of specific experiences across all 38 enrolled students.

Third, the syGlass platform’s specific characteristics may not generalize to other VR applications. The instructor deliberately chose to use authentic MRI data and real patient cases rather than simplified anatomical models, a decision reflecting her TCK vision for clinical authenticity. This choice likely elevated cost (working with messy real data) while enhancing perceived authenticity. VR platforms using simplified, game-like models might produce different TPK challenges and EVT dynamics. Similarly, VR applications in other disciplines (chemistry, engineering, geology) may involve different content-technology relationships that alter the TPACK-EVT dynamics we observed. Fourth, the researcher’s prior relationship with the instructor through the XR-FLC may have influenced interpretations. To mitigate this, we employed peer debriefing with graduate research assistants who had no prior relationship with the instructor, actively sought disconfirming evidence during analysis, and maintained transparent audit trails of analytical decisions. However, peer researchers should be aware of this positionality. Finally, the focused ethnographic approach, while appropriate for semester-long investigation, captures only one implementation cycle. The instructor explicitly noted she would approach a second iteration differently (“Now I did it, I know how it is going to be”). Longitudinal research following instructors across multiple VR integration cycles would illuminate TPK development trajectories we could not capture.

Despite these limitations, we believe this study makes valuable contributions. Our detailed examination of TPACK-EVT dynamics over an extended period provides insights unavailable from experimental studies with brief VR exposure. The proposed TPACK-EVT integration framework, while requiring further validation, offers a conceptual tool for analyzing VR implementations that previous research lacked. Future research should address these limitations through comparative case studies and quantitative hypothesis testing.