From Stable to Screen: How Setting Shapes Student Stress Responses to Campus-Based Virtual Equine-Assisted Intervention

Abstract

As student mental health concerns intensify, campuses are exploring wellness initiatives including equine-assisted intervention (EAI). Because maintaining equines on campus is not always feasible, virtual reality (VR) has emerged as an alternative method for delivering equine experiences. Although equine environments alone may reduce stress, limited research has examined how environmental setting influences outcomes during VR EAI. Therefore, the objective of this exploratory study was to compare stress levels, as observed through heart rate and cortisol concentrations, for students participating in VR EAI within equine and classroom settings. Students participated in one of the following on-campus VR EAI programs: (1) classroom (n = 10) or (2) equine (n = 14) environment. Although both environmental types demonstrated a significant difference between minimum-maximum heart rate during the session (p < 0.05), only the classroom led to a significant reduction in heart rate post-session (p = 0.02). Neither environmental type observed a significant difference between pre-post cortisol concentrations (p > 0.05). Environmental type was an influencing factor for cortisol concentrations (p = 0.04). These findings suggest that environmental context may influence physiological responses during VR EAI; however, further research is warranted that includes larger samples, more controlled designs, and integrated subjective and physiological measures.

1. Introduction

College students are experiencing increased levels of stress, anxiety, and overall mental health challenges, leading universities to explore supportive programs beyond traditional counseling services (Anderson et al., 2010; Sulaiman et al., 2022). One approach that has gained attention is equine-assisted intervention (EAI), which has been associated with improvements in emotional regulation and reductions in stress. Studies have shown that participation in live EAI can influence physiological stress indicators such as cortisol concentrations and heart rate, suggesting measurable effects on the body’s stress response (Malinowski et al., 2018; Naber et al., 2025; Pendry et al., 2014). These effects are thought to arise from multiple therapeutic components, including direct interaction with the horse, engagement in structured activities, and the social and sensory experiences associated with the intervention. In addition, indirect exposure to horses, such as observing or being present in an equine environment, has been associated with increased emotional safety and positive learning outcomes among college students, even without direct therapeutic interaction (Holtcamp et al., 2023).

Although promising, traditional EAI programs require horses, specialized facilities, and trained personnel, limiting accessibility on college campuses. However, virtual reality (VR) offers a practical alternative by providing immersive equine experiences without these logistical barriers. Unlike live EAI, VR-based interventions do not involve direct interaction with a living animal but instead rely on immersive audiovisual simulation to evoke psychological and physiological responses. Across a variety of therapeutic applications, VR interventions have demonstrated the ability to reduce stress and anxiety while influencing physiological markers such as heart rate (El-Qirem et al., 2022; Kim et al., 2021). Emerging research also suggests that virtual human–animal interaction may provide some psychological benefits comparable to those observed during live animal interaction (Na & Dong, 2023; Schwab et al., 2025; Willms et al., 2026).

In addition to the intervention itself, the physical environment in which VR is experienced may independently influence stress responses. Environmental psychology and stress reduction theories propose that natural settings can promote parasympathetic nervous system activity, reduce sympathetic arousal, and facilitate physiological recovery from stress through multisensory environmental cues, even when those cues are not the primary focus of attention (Ulrich et al., 1991). Consistent with this perspective, exposure to natural or equine environments has been associated with reduced perceived stress and altered physiological arousal independent of direct horse interaction (Deconinck, 2023; Friend et al., 2023). During a VR-based equine experience, the surrounding physical environment may therefore provide additional sensory information, including visual, auditory, olfactory, and tactile cues, that complements or competes with the virtual experience. This, in return, can potentially influence immersion and physiological stress responses. However, research is lacking examining whether the physical setting in which VR EAI is delivered modifies physiological outcomes, highlighting the need to better understand how environmental context contributes to the effectiveness of VR-based equine interventions.

Therefore, the aim of this exploratory study was to investigate environmental influence on physiological stress responses, measured through heart rate and salivary cortisol, in college students participating in VR EAI conducted in two settings, the equine environment and the classroom environment. Based on previous findings that equine environments (Friend et al., 2023) and indirect equine interaction (Holtcamp et al., 2023) independently reduce stress, we hypothesize that participants completing VR EAI in an equine environment would demonstrate greater reductions in heart rate and salivary cortisol compared to those in a classroom setting.

2. Materials and Methods

2.1. Study Design and Participants

College students enrolled at Mississippi State University were recruited during the Fall 2024 semester through departmental emails, classroom announcements, flyers, and social media posts. Participation was voluntary, and no exclusion criteria were imposed. Participants self-selected into one of two scheduled on-campus VR EAI sessions based on availability, with sessions conducted in either a classroom or equine environment; therefore, participants were not randomly assigned to treatment conditions. A total of 27 students participated in the VR EAI program, of whom 24 consented to physiological biomarker collection (classroom environment, n = 10; equine environment, n = 14). To reduce procedural variability, all participants completed the same standardized VR EAI protocol using the identical VR video, headset and equipment, session procedures, intervention personnel, and procedures related to data collection. Consequently, the physical setting in which the VR session was delivered (classroom or equine environment) represented the primary experimental difference between groups. Heart rate and salivary cortisol were collected immediately before and after the VR session to assess physiological responses. Because participants self-selected into intervention sessions, no formal assessment of baseline equivalence between groups was conducted beyond the pre-session physiological measurements collected immediately before the VR session.

The sessions for both environmental types were offered during the afternoon for one day. For both settings, two headsets were made available for participants. Students checked in with a member of the research team and completed a demographic questionnaire while waiting for a headset to be available. Variables collected within the questionnaire included age, sex, environment associated with childhood upbringing (urban, suburban, or rural), and prior animal agriculture and VR experience. These demographic variables were incorporated into statistical analysis to evaluate associations between participant characteristics and physiological stress biomarkers.

The VR EAI within the equine environment consisted of students taking part in the VR experience at the on-campus equine facilities. The headsets were set up on a table at the covered, open-sided riding arena where equine activities were taking place. The equines within the environment were out of reach from the students but equines could be seen before and after the VR session. The VR EAI held within the classroom setting was done on campus. The room was a small seminar classroom that had the door closed during VR sessions.

2.2. Virtual Reality Intervention

The VR experience consisted of three-dimensional 360° video footage recorded at the Mississippi State University equine facility using a GoPro 360° camera (GoPro, Inc.; San Mateo, CA, USA). The footage included mounted trail riding and ground activities working with a horse with the footage edited for duration and clarity before being uploaded to a Meta Quest 1 VR All-in-One- headset (Meta Platforms, Inc.; Menlo Park, CA, USA). Original audio was retained.

The final video lasted approximately 10 min and was set on a loop. Video included walking, trotting, and cantering sequences in an open trail ride setting and ground handling activities that included grooming and leading a horse. Participants remained seated during headset use in both environmental conditions to minimize physical exertion as a confounding factor. Students could select how long they participated in the VR experience with maximum time allowed being one hour. Students from both groups participated in the VR session approximately 15–30 min.

2.3. Physiological Measures

Two biomarkers associated with stress response were evaluated:

2.3.1. Heart Rate

Heart rate was continuously recorded during VR exposure using a Fitbit wearable device. Pre- and post-session heart rate values along with minimum and maximum heart rate values during the session were extracted for statistical comparison.

2.3.2. Salivary Cortisol

Salivary cortisol was collected using Salimetrics adult swabs (Salimetrics, State College, PA, USA) following procedures described by Friend et al. (2023). Participants in both treatment conditions provided samples within 5 min before and after the VR session. Samples were immediately placed on dry ice and stored at −80 °C until analysis. Cortisol concentrations were determined using a Salimetrics enzyme-linked immunosorbent assay according to manufacturer guidelines. Samples were thawed, vortexed, and centrifuged at 1500 rpm for 15 min, and 25 µL aliquots were assayed in duplicate. Intra- and inter-assay coefficients of variation were 9% and 12%, respectively.

2.4. Statistical Analysis

Descriptive statistics (means ± standard deviations) were calculated using SAS software (Version 9.4; SAS Institute Inc., Cary, NC, USA). Data were assessed for normality prior to inferential analyses. Within-group pre- and post-session changes in heart rate and salivary cortisol concentrations were evaluated using paired two-tailed t-tests when normality assumptions were met or Wilcoxon signed-rank tests for non-normally distributed data. Because participants self-selected a single intervention setting, between-group comparisons of physiological responses were conducted using the Kruskal–Wallis test.

Effect sizes (Cohen’s d) were calculated for both within-group and between-group comparisons to quantify the magnitude and direction of observed effects. For within-group analyses, Cohen’s d was calculated using the mean pre-post change divided by the standard deviation of the paired differences. Within-group effect sizes were calculated as post-session values minus pre-session values, such that positive effect sizes indicated an increase following the VR interaction and negative effect sizes indicated a decrease. Between-group effect sizes were calculated using the difference between group means divided by the pooled standard deviation. For environmental comparisons, effect sizes were calculated as equine environment values minus classroom environment values, such that positive effect sizes indicated higher physiological responses in the equine environment and negative effect sizes indicated higher physiological responses in the classroom environment. Cohen’s d values of approximately 0.20, 0.50, and 0.80 were used as general benchmarks for small, moderate, and large effects, respectively. Ninety-five percent confidence intervals (95% CIs) were calculated and reported for estimated mean differences and effect sizes, where appropriate, to quantify the precision of observed effects. Given the exploratory nature of the study and limited sample size, effect sizes and confidence intervals were interpreted in conjunction with statistical significance testing rather than as independent indicators of clinical or biological relevance.

A single significance threshold of p ≤ 0.05 was applied to all statistical analyses. The p-values reported for each test represent the probability associated with the specific comparison being evaluated and therefore vary according to the observed data, sample variability, and statistical test used. Values between p > 0.05 and p ≤ 0.10 were reported as statistical tendencies to indicate results that may be biologically meaningful but did not meet the predefined criterion for statistical significance.

3. Results

3.1. Demographics

A total of 24 participants were included in the analysis, with 10 in the classroom environment and 14 in the equine environment (

Table 1. Demographic information associated with participants from the virtual reality equine-assisted intervention sessions offered in the classroom and equine environments.

Question Category	Answer Options	Classroom Environment Responses (n = 10)	Equine Environment Responses (n = 14)
Age	18–25	9	11
	26–30	1	3
Sex	Male	3	5
	Female	7	9
Setting 1	Urban	0	2
	Suburban	4	7
	Rural	6	5
Virtual Reality Experience 2	Yes	0	1
	No	10	13
Animal Agriculture Experience 3	Yes	6	8
	No	4	6

¹ Setting: Reported location where participants spent the majority of their lives. ² Virtual Reality Experience: Participants’ response given as to prior experience utilizing the virtual reality headset. ³ Animal Agriculture Experience: Participants’ response given as to prior experience working around livestock animals including horses.

). The majority of participants were between the ages of 18–25 (87% classroom; 79% equine) and female (70% classroom; 64% equine). Most participants reported a suburban or rural upbringing, with rural backgrounds being more common in the classroom group (60%) and suburban backgrounds more common in the equine group (50%). The majority of participants had no prior experience with virtual reality (100% classroom; 93% equine), while over half reported prior experience with animal agriculture (60% classroom; 57% equine).

3.2. Heart Rate

A trend toward an increase in heart rate was observed following the VR EAI session for participants in the equine environment (p = 0.06), corresponding to a small negative within-group effect (d = −0.32, 95% CI: −0.87 to 0.08). In contrast, participants in the classroom VR EAI group demonstrated a significant reduction in heart rate following the interaction (p = 0.02, Figure 1), corresponding to a moderate within-group effect (d = 0.46, 95% CI: 0.19–1.01).

Minimum heart rate for both environmental conditions fell within the range of normal resting heart rates for adults (Nealen, 2016). Maximum heart rate during the VR EAI session for both environmental conditions fell below the target heart rate and aerobic threshold for adults as reported by Nealen (2016). Both maximum and minimum values for both environmental conditions occurred during the session and are presented in Figure 2. A significant difference between minimum and maximum heart rate values was observed within both environmental conditions (equine: p = 0.002; classroom: p = 0.0001), demonstrating within-session variability in physiological responses during VR EAI exposure. The comparison of minimum-to-maximum heart rate responses between environments revealed a small effect of environmental context (d = 0.32, 95% CI: 0.02–0.63), indicating a modest difference in heart rate variability between the equine and classroom VR settings. Despite this difference in response range, overall heart rate measurements were not significantly influenced by environmental type or participant demographic characteristics (p > 0.10).

3.3. Cortisol Concentrations

Salivary cortisol concentrations increased following the VR interaction in both environmental conditions (Figure 3); however, within-group changes did not reach statistical significance (p > 0.05). A tendency toward increased cortisol concentrations was observed in the classroom VR environment (p = 0.07), corresponding to a negligible within-group effect (d = −0.04, 95% CI: −0.70 to 0.62), suggesting minimal evidence of a consistent change in cortisol following VR exposure. Participant demographic characteristics were not associated with cortisol concentrations (p > 0.10). Environmental condition was associated with differences in cortisol concentrations, with higher cortisol levels observed in the equine environment compared with the classroom environment (p = 0.04). This difference represented a small effect of environmental context (d = 0.32, 95% CI: 0.02–0.63).

4. Discussion

As student mental health concerns continue to rise, campuses are increasingly recognizing the need for effective, accessible on-campus interventions. Although EAI has demonstrated benefits for mental health of young adults, many institutions lack access to live horse interactions. Consequently, VR-based equine experiences have emerged as a potential alternative; however, the novelty of this approach warrants further investigation to determine optimal implementation strategies. The objective of this exploratory study was to examine whether environmental context influences physiological stress responses during VR-based EAI in college students. Results from this study indicate that both VR EAI environmental settings elicited measurable physiological responses, although the direction and magnitude of responses varied by environmental context. These preliminary findings suggest that the setting in which VR EAI is delivered may contribute to differences in physiological activation; however, the variability observed across outcomes highlights the need for further investigation using larger samples, controlled experimental designs, and integrated subjective and physiological measures (Deconinck, 2023; Kim et al., 2021).

4.1. Environmental Context and Physiological Responses

Contrary to the original hypothesis, heart rate responses differed according to environmental context. Participants in the classroom VR EAI environment demonstrated a significant reduction in post-session heart rate, whereas participants in the equine VR EAI environment exhibited a trend toward increased heart rate that did not reach statistical significance. Analysis of minimum and maximum heart rate values further demonstrated significant within-session cardiovascular variability in both environmental conditions, suggesting that physiological engagement occurred throughout the VR EAI experience regardless of setting. Together, these findings indicate that VR EAI may elicit dynamic autonomic responses, but the direction and magnitude of these responses may depend on environmental characteristics, participant engagement, and the broader context in which the virtual experience is delivered.

The influence of environmental context on VR responses may be partly explained by differences in immersion and presence. Virtual reality experiences are not determined solely by the digital content presented to users; rather, psychological presence and engagement emerge from the interaction among technological characteristics, sensory information, environmental cues, and individual perception (Slater, 2009; Slater & Wilbur, 1997). Greater immersion, defined as the objective capability of a virtual system to deliver sensory-rich experiences, may contribute to stronger feelings of presence, or the subjective experience of “being there” within the virtual environment (Cummings & Bailenson, 2016; Witmer & Singer, 1998). In the current study, differences between the classroom and equine environments may have altered the overall perceptual context of the VR experience, potentially influencing attentional engagement and physiological activation. Research examining environmental context in VR suggests that surrounding physical conditions, including sensory cues and congruence between the physical and virtual environments, can influence presence, emotional responses, and physiological outcomes (Diemer et al., 2015; Makransky & Petersen, 2021).

Previous research has demonstrated that VR-based interventions can influence autonomic responses including heart rate through changes in emotional engagement, stress perception, and environmental simulation (El-Qirem et al., 2022; Kim et al., 2021). Similarly, EAI has been associated with alterations in physiological arousal and stress-related outcomes (Holtcamp et al., 2024; Malinowski et al., 2018; Pendry et al., 2014). However, the current findings indicate that an equine context does not necessarily produce uniform reductions in physiological activation during VR-based interaction. One possible explanation is that the equine environment introduced additional sensory and contextual stimuli, including animal movement, sounds, visual complexity, and increased ecological cues. This may have enhanced engagement or attentional demands, contributing to increased autonomic activation associated with interest, emotional involvement, or environmental salience (Shaffer & Ginsberg, 2017; Thayer et al., 2012).

Interpretation of short-duration heart rate responses, however, requires consideration of the temporal characteristics of autonomic measures. Heart rate provides a relatively rapid indicator of cardiovascular activation and may respond within seconds to changes in attention, emotion, movement, or environmental stimulation (Berntson et al., 1997). Therefore, the observed heart rate differences may reflect immediate responses to the VR experience and its surrounding context, rather than sustained changes in stress regulation. Conversely, salivary cortisol reflects hypothalamic–pituitary–adrenal (HPA) axis activity and generally demonstrates slower temporal dynamics. Responses are influenced by sampling timing, circadian variation, individual differences, and the magnitude and duration of the stimulus (Adam & Kumari, 2009; Friend et al., 2023; Kirschbaum & Hellhammer, 1994). These physiological systems therefore provide complementary, but distinct information. As such, differences between heart rate and cortisol findings should be expected, rather than interpreted as contradictory.

Salivary cortisol concentrations within the current study did not demonstrate significant pre- to post-session changes within either environmental condition. The absence of significant within-group changes indicates that a single, relatively short-duration VR EAI exposure may not consistently alter acute cortisol concentrations. Nonetheless, a tendency toward increased cortisol concentrations was observed in the classroom environment. Further, cortisol concentrations differed between environmental conditions, as higher values were observed in the equine environment. These findings suggest that environmental context may contribute to differences in endocrine responses. It is important to note, however, that methodological considerations are particularly important when interpreting cortisol findings in short-duration intervention studies such as the case within the current study. Cortisol responses depend strongly on the timing of sample collection relative to the intervention, the duration of exposure, and whether appropriate control conditions are included (Dickerson & Kemeny, 2004; Hellhammer et al., 2009). As such, future VR EAI studies should consider multiple post-exposure sampling points to better characterize cortisol response trajectories, rather than relying on a single post-intervention measurement.

The higher cortisol concentrations observed in the equine environment may reflect increased physiological activation associated with environmental complexity, novelty, or engagement with equine-related cues, rather than a negative stress response. Previous studies examining natural environments and intervention strategies utilizing animals have demonstrated that physiological responses can vary according to environmental features, individual perception, and intervention characteristics (Malinowski et al., 2018; Ulrich et al., 1991). Because cortisol is a nonspecific indicator of HPA-axis activation, increased concentrations cannot be interpreted independently as evidence of psychological distress. Instead, cortisol findings should be considered alongside autonomic measures, subjective experiences, and behavioral outcomes to better understand the multidimensional nature of participant responses. When considered together, the heart rate and cortisol findings demonstrate that physiological responses to VR EAI are complex and may depend on both the characteristics of the virtual intervention and the physical context in which it is experienced. Heart rate appeared more sensitive to immediate environmental differences and short-term autonomic activation, whereas cortisol responses appeared more variable and influenced by the slower dynamics of endocrine regulation. These findings support the importance of incorporating multiple physiological measures when evaluating immersive interventions and highlight the need to consider both the virtual environment and the physical environment surrounding the user. Neither heart rate nor cortisol responses were significantly associated with participant demographic characteristics, suggesting limited evidence that demographic factors influenced physiological outcomes within this sample. However, conclusions regarding generalizability should be limited by the small sample size, exploratory design, and need for more rigorous experimental controls.

4.2. Study Limitations

As previously stated, several limitations should be considered when interpreting the findings of this exploratory study. The relatively small sample size and unequal group sizes limited statistical power and reduced the precision of the estimated effects, as reflected by the width of several 95% confidence intervals. Consequently, although effect size estimates provided useful information regarding the magnitude and direction of physiological responses, these estimates should be interpreted cautiously because they are subject to considerable uncertainty. Larger, adequately powered studies are needed to provide more precise estimates of the effects of VR EAI across different environmental contexts. Accordingly, the emphasis of the present study is on estimation, rather than definitive hypothesis testing, with effect sizes and confidence intervals providing preliminary evidence to inform future research and sample size calculations.

Furthermore, participation was voluntary and no exclusion criteria were applied, increasing the potential for selection bias. Due to the small sample size, similarities across participants were observed across sex, age range, and prior VR and animal agriculture experiences. A larger sample size would allow for a more diversified representation of the student population. It is important to note, however, that only 10 students participated in the classroom setting, which may indicate a limitation in recruitment for this type of setting compared to the equine setting. Mental health programming is often associated with a stigma, and the use of the classroom environment simulates the setting of conventional therapeutic environments (Holtcamp et al., 2024). This may reflect a value in the use of the equine environment in which other alternative outdoor settings may further offer the ability to recruit students less willing to engage in traditional settings for mental health programming, and as such, should be of interest in future VR research targeting student mental health.

Virtual reality exposure duration varied across participants, which may have influenced physiological responses (Kim et al., 2021), and the single-session VR exposure may have limited acclimation, particularly for participants unfamiliar with VR technology. Although demographic variables, including prior VR experience, did not influence outcomes, the specific types of prior VR exposure were not documented, and unfamiliarity with the headset used may have required additional adjustment time. Because participants self-selected into intervention sessions and were not randomly assigned, the potential influence of unmeasured participant characteristics cannot be ruled out. Future studies should employ randomized assignment or matching procedures to better isolate the effects of environmental setting. Additionally, reliance on only two physiological biomarkers may not fully capture the complexity of the stress response, which is influenced by interacting physiological and psychological factors (El-Qirem et al., 2022; Kim et al., 2021). Along with exploring additional biomarkers, future research should include subjective measures of stress, anxiety, and emotional state alongside physiological biomarkers as this would provide a more comprehensive understanding of participant response (El-Qirem et al., 2022; Kim et al., 2021; Holtcamp et al., 2024). Future research should incorporate additional biomarkers that have been used in stress and VR research, such as electrodermal activity (EDA) and salivary alpha-amylase, which provide further insight into autonomic nervous system activity and acute stress responses (Critchley, 2002; Nater & Rohleder, 2009; Shaffer & Ginsberg, 2017). Assessment of additional biomarkers should occur in conjunction with subjective measures of stress, anxiety, and emotional state, as this would provide a more comprehensive understanding of participant response (El-Qirem et al., 2022; Kim et al., 2021; Holtcamp et al., 2024).

5. Conclusions

This exploratory study provides preliminary evidence that VR EAI can elicit measurable physiological responses in college students; however, variability in heart rate and salivary cortisol responses suggests that physiological effects may differ according to environmental context and individual characteristics. These findings should not be interpreted as evidence of definitive therapeutic effectiveness, but rather as an initial evaluation of physiological responses associated with VR EAI exposure. Given the limited sample size and the need for more rigorous experimental approaches, including clearly defined randomization procedures and larger, adequately powered cohorts, further research is needed to confirm these findings. Future studies should incorporate controlled randomized designs, repeated VR exposures, and integrated subjective and physiological measures to better characterize the potential role of VR EAI in supporting student well-being and to identify environmental factors that may enhance or attenuate responses. Overall, VR EAI represents a promising and accessible approach for continued investigation, particularly in settings where live animal-assisted interventions are not readily available.