Abstract: Future military operations training increasingly relies on virtual and mixed reality. Yet stress—a central determinant of performance and resilience—is often treated as an assumed by-product of scenario intensity rather than as an explicit design dimension. Drawing on the TrainOps and EviS projects, this short essay introduces the concept of stress-literate training and outlines a three-layer pipeline that connects (1) operationally grounded stress cues in XR scenarios, (2) robust biosignal acquisition (e.g., heart rate, EDA, eye tracking), and (3) interpretable stress metrics and instructor dashboards.

Problem statement: How can armed forces systematically design and measure stress in XR training?

So what?: Military training centres and XR technology providers should co-develop stressor taxonomies, sensing strategies, and interpretable stress metrics, integrating them into training platforms to enable stress-literate assessment, debriefing, and adaptive training design.

Why Stress Must Be Designed, Not Assumed

Virtual and mixed reality technologies have become an established component of contemporary military training. XR-based simulations are increasingly used to train tactical decision-making, coordination, and situational awareness in environments that would be costly, dangerous, or logistically infeasible to reproduce in the real world. Especially for high-risk professions, XR offers repeatability, controllability, and safe exposure to complex operational situations.[1] As a result, immersion and scenario realism have become dominant design goals in virtual training environments.

For officer education, this shift matters because emerging training technologies not only change what is trained, but also how leadership and decision-making are cultivated under uncertainty, time pressure, and responsibility. Stress is therefore not a peripheral side effect of XR realism but a core variable shaping the training of tomorrow’s officers–particularly in contexts where command decisions must be made under acute operational pressure.[2]

Despite this development, stress remains insufficiently addressed in XR training. Scenario intensity is often equated with stress, although empirical evidence shows that subjective workload and physiological stress responses often diverge, particularly in simulated environments.[3] Similarly, studies on VR-based training highlight substantial variability in stress responses across individuals exposed to comparable scenarios.[4]

For military training, this mismatch is problematic. Stress is not merely an accompanying effect of demanding situations but a central variable influencing perception, decision-making, and learning under pressure. Training environments that fail to account for how stress is elicited and experienced risk either underchallenging trainees or overwhelming them in pedagogically counterproductive ways. Nevertheless, stress in XR training is typically treated as an emergent side effect of scenario design rather than as a phenomenon that can be deliberately shaped and assessed.

Conceptual Grounding

Stress is a central construct in military training, yet it is often treated implicitly rather than conceptually. To design and evaluate stress in XR-based training environments, it is necessary to clarify what is meant by stress in operational contexts and how it manifests in acute situations. Equally important is the understanding of why XR is simultaneously a promising and fragile medium for stress-oriented training.

Stress as a Transactional and Context-Dependent Process

Within psychological literature, stress is not defined solely by the presence of demanding stimuli but by the relationship between environmental demands and an individual’s perceived resources. The transactional stress model proposed by Lazarus and Folkman conceptualises stress as a dynamic process that emerges when individuals appraise situational demands as exceeding their coping capacities.[5] Central to this model is the notion of cognitive appraisal of the situation’s demands relative to available coping resources: the same external stressor may elicit markedly different stress responses depending on how it is interpreted, whether it is perceived as controllable, and which coping resources are available.

This perspective is particularly relevant for XR-based training. Virtual scenarios can replicate operational demands with high visual or auditory fidelity, yet the resulting stress response is not determined solely by stimulus intensity. Research on stress emphasises that predictability, controllability, and meaning attribution strongly shape physiological and psychological responses.[6] Consequently, a scenario that appears “intense” from a design perspective may fail to elicit stress if the trainee perceives it as artificial, predictable, or inconsequential.

From a training standpoint, this implies that stress cannot be reduced to a static property of a scenario. Instead, stress must be understood as a dynamic and relational phenomenon that emerges from the interaction between designed demands and the trainee’s appraisal of the situation. This conceptualisation forms a key requirement for XR training systems that aim to intentionally elicit, modulate, or measure stress.

Acute Stress in Military Operations: CSR and COSR

In military contexts, the focus is typically not on chronic stress but on acute operational stress reactions. These are often described as Combat Stress Reaction (CSR) or Combat and Operational Stress Reaction (COSR), referring to immediate psychophysiological responses to perceived life threat, uncertainty, or extreme task demands during operations.[7] Such reactions are characterised by rapid activation of the autonomic nervous system and can manifest as heightened arousal, narrowed attention, impaired decision-making, or, in extreme cases, functional shutdown.

Importantly, acute stress reactions are not inherently negative. Moderate levels of stress can enhance alertness and task engagement, whereas excessive stress can degrade performance and learning.[8] The relationship between stress and performance is therefore non-linear and context-dependent. For training, this implies that the objective is not to maximise stress, but to calibrate it in a way that supports learning and operational readiness.

Military psychology literature further emphasises that acute stress reactions are shaped by multiple categories of stressors, including task-related demands, uncertainty, social and interpersonal factors, and environmental conditions. These stressors vary across phases of an operation and interact with individual experience and preparedness. As a result, training environments that focus exclusively on task difficulty or time pressure risk neglecting other stress-relevant dimensions critical to real operations.

Why XR Is Both Promising and Fragile for Stress-Oriented Training

XR technologies offer unique advantages for stress-oriented training. Virtual environments enable the controlled presentation of complex, hazardous, or ethically sensitive situations without exposing trainees to real-world danger. They allow stressors to be introduced, removed, or combined in a repeatable manner, making XR particularly attractive for stress inoculation and decision-making training.[9] In principle, XR also allows for fine-tuned manipulation of contextual factors such as uncertainty, time pressure, or social dynamics, which are difficult to standardise in live training.

At the same time, XR is a fragile medium for stress induction. The perceived artificiality of virtual environments can attenuate stress responses, especially when trainees do not fully suspend disbelief or when scenarios lack personal relevance. Moreover, XR systems often emphasise audiovisual realism while neglecting other dimensions that contribute to stress appraisal, such as social evaluation, responsibility, or ambiguity.[10] Prior work on VR-based stress training highlights that generic stressor categories, such as “information overload,” are insufficient unless they are translated into concrete, context-sensitive design elements.[11] XR can produce visual intensity without inducing operationally meaningful stress, complicating both training validity and physiological interpretation.

From Operational Stressors to Stress-Cues (EviS)

If stress is understood as a dynamic and context-dependent process rather than a fixed property of scenarios, XR-based training must move beyond generic notions of “high intensity.” Instead, it requires a systematic approach to designing stress, grounded in operational reality and stress theory. The EviS project addresses this challenge by focusing on identifying, classifying, and translating military-relevant stressors into concrete, implementable design elements for XR training. In concrete terms, EviS (Evidence-based Stressors in VR) is a three-year Bundeswehr research project carried out by the Wehrtechnische Dienststelle 91 (WTD 91) in cooperation with Hologate GmbH and the AIT Austrian Institute of Technology. It combines a literature synthesis with workshops and interviews among soldiers and trainers to derive a prioritised list of operational stressors, which are translated into specification sheets that define each stressor’s audiovisual, temporal, and spatial properties for XR implementation. The resulting stress cues are validated in two successive studies—first in isolated VR environments, then in full training scenarios—so that each can be assigned an empirically grounded intensity rating.

From Operational Stressors to a Design Problem

Military psychology has long recognised that operational stress emerges from multiple interacting sources rather than a single trigger. Campbell and Nobel identify categories including task-related demands, uncertainty, social and interpersonal pressures, and environmental stressors.[12] Such frameworks provide an important conceptual foundation for understanding stress in military operations.

For XR training, however, these taxonomies create a practical challenge. Categories such as uncertainty, information overload, and social evaluation describe sources of stress but do not specify how to implement them within a virtual environment. Prior reviews of VR-based stress training have highlighted this gap, noting that stressors are frequently discussed at an abstract level without providing guidance for scenario design.[13]

To achieve this, EviS combines existing military stress research with workshops and interviews involving military personnel and instructors. Building on established stressor taxonomies, participants assess the operational relevance, training value, and practical feasibility of individual stressors. The result is a prioritised set of stressors that are both operationally meaningful and suitable for implementation in XR-based training.

This process forms the foundation for the development of stress cues: concrete, observable, and controllable scenario elements that operationalise otherwise abstract stressor categories. Rather than treating stress as an emergent property of immersive environments, EviS seeks to transform it into a deliberate and measurable component of training design.

A Taxonomic Approach to Stressor Identification

As a first step, EviS builds on existing military stress research to structure the identification of stressors relevant for XR training. Drawing on Campbell and Nobel’s framework, stressors are grouped into higher-level categories (e.g., task, social/interpersonal, psychological environment, physical environment), which serve as an initial scaffold for further refinement.[14] Importantly, this taxonomic approach is not treated as a fixed classification but as a starting point for contextualisation. Through workshops and interviews with military personnel and instructors, stressors are further specified in terms of their operational relevance, perceived severity, and training value. This qualitative grounding reflects the transactional view of stress, acknowledging that stressors gain meaning through appraisal and experience rather than through objective properties alone.[15]

The outcome of this process is a prioritised set of stressors that are both operationally relevant and amenable to XR implementation. Stressors that are theoretically relevant but technically infeasible or pedagogically undesirable are deliberately deprioritised. This prioritisation step is essential to avoid overloading training environments with stressors that cannot be meaningfully controlled or evaluated.

Translating Stressors into XR-Implementable Design Elements

A central contribution of the EviS project is the concept of stress cues. Stress cues translate abstract stressors into concrete, observable, and implementable elements within an XR training environment.  For example, a stress cue may take the form of a time-pressured radio call, an ambiguous civilian contact, or a sudden loss of communications—controllable scenario elements that operationalise a stressor category such as uncertainty or social evaluation. The concept builds on prior work that introduced stress cues as a means of operationalising real-world stressors in VR-based police training.[16]

In EviS, stress cues are specified along multiple design dimensions. These include the sensory modality through which the stressor is expressed (e.g., visual, auditory, or multisensory), its temporal characteristics (e.g., sudden versus gradual onset), its spatial properties (e.g., fixed or moving within the environment), and its social embedding (e.g., presence of bystanders or evaluative observers). This multi-dimensional specification ensures that stress cues are not treated as isolated stimuli but as situated elements embedded in the broader scenario context.

Combination, Habituation, and Training Validity

Early implementations within EviS highlight several challenges that underscore the necessity of a design-driven approach to stress. First, stressors rarely occur in isolation during real operations. XR training, therefore, requires careful consideration of combined effects, as multiple stress cues may interact in non-additive ways. Without systematic design and validation, combining stress cues risks either overwhelming trainees or masking the effects of individual stressors.

Second, habituation poses a significant challenge for repeated XR training. Stress cues that are effective during initial exposure may lose their impact over time, particularly if they are predictable or lack contextual variation. This aligns with broader stress research emphasising the role of novelty and uncertainty in stress appraisal.[17] Designing stress cues that remain effective across repeated training sessions, therefore, requires deliberate variation and escalation strategies.

Finally, there is a tension between realism and control. While high ecological validity is desirable, overly complex or uncontrolled stress induction undermines the pedagogical goal of calibrated training. The EviS project addresses this tension by treating stress cues as designed training instruments rather than as attempts to fully replicate real-world stress in all its complexity. That is, cues are deliberately selected and dosed for their didactic effect on a specific learning objective rather than optimised for maximum realism: a cue that reliably evokes the intended appraisal (e.g., uncertainty) is preferred over one that is more realistic but pedagogically diffuse.

From Biosignals to Actionable Metrics (TrainOps)

While the EviS project addresses how stress can be intentionally designed in XR-based military training, the TrainOps project focuses on the complementary challenge of measuring stress in a way that is robust, interpretable, and useful for training practice. Without reliable measurement, stress-cue design remains largely speculative, and instructors lack evidence-based insight into how trainees actually respond to stress during XR training.

Measuring Acute Stress in Dynamic Training Contexts

Stress measurement in military training must account for the characteristics of acute operational stress. Unlike chronic stress, acute stress reactions unfold rapidly, fluctuate over short time scales, and are tightly coupled to task demands and situational appraisal.[18]  In other words, acute stress in this context is situation-specific: it arises from the interplay of a concrete scenario and the individual’s subjective appraisal of it, rather than from a general trait or baseline level. For XR-based training, this implies that stress measurement must be continuous, temporally sensitive, and capable of operating under movement and cognitive load.

Prior research on stress detection identifies a broad range of physiological indicators associated with acute stress, including cardiac activity, electrodermal activity, pupil dynamics, and neural measures.[19] However, not all these measures are equally suitable for applied training contexts. Highly invasive or movement-sensitive methods, such as laboratory-grade electroencephalography (EEG), pose practical challenges in XR training environments, where freedom of movement and usability are critical.

TrainOps, therefore, adopts a pragmatic measurement strategy that prioritises non-invasive, wearable biosensors capable of delivering sufficiently robust data under realistic training conditions. This approach reflects a key insight from applied XR training research: measurement quality must be balanced against trainee burden and ecological validity to avoid undermining the training itself.[20]

Robustness over Completeness

At the core of the TrainOps measurement setup is cardiac activity, operationalised through heart rate and heart rate variability. Cardiac measures are well-established indicators of autonomic stress responses and have been shown to provide relatively stable signals even during dynamic tasks when appropriate preprocessing is applied.[21] In contrast, measures such as electrodermal activity or pupil size can offer valuable complementary information about arousal and attentional load but are more susceptible to noise, motion artefacts, and contextual confounds.[22] TrainOps prioritises feasible and robust signals over technical completeness.

From Biosignals to Stress Metrics

Raw physiological signals are rarely meaningful for instructors or trainees. TrainOps, therefore, focuses on transforming biosignals into stress metrics that reflect relevant aspects of acute stress while remaining interpretable in a training context. Building on prior work in XR-based police and military training, the project explores multimodal modelling approaches that combine cardiac measures with additional physiological and contextual signals.[23]

Machine learning techniques, such as ensemble-based models, are employed as tools for integrating heterogeneous data sources rather than as black-box predictors.[24] The objective is not to infer fine-grained psychological states but to classify stress levels into coarse, actionable categories that align with training needs. This aligns with findings from applied stress measurement research, which emphasise that overly complex or opaque models hinder trust and usability in operational settings.[25]

A central design constraint in TrainOps is interpretability. Stress metrics are designed to support instructors’ situational awareness rather than replace their judgment. This reflects prior findings showing that trainers benefit most from aggregated indicators that highlight trends, thresholds, or deviations, rather than from detailed physiological time series.[26]

Visualising Stress for Training Practice

To bridge the gap between physiological measurement and training action, TrainOps includes the design of an instructor-facing stress dashboard. The dashboard visualises trainees’ stress metrics in real time and during after-action reviews, using deliberately simplified representations such as colour-coded stress levels. This approach builds on earlier work demonstrating that reduced, traffic-light-style visualisations can effectively communicate stress information without overwhelming users.[27]

Importantly, the dashboard is not intended to provide an objective “truth” about trainee stress. Instead, it functions as a decision-support tool, enabling instructors to identify moments of elevated stress, compare responses across trainees, and reflect on how stress evolved during specific phases of a scenario. This framing acknowledges the inherent uncertainty of physiological stress measurement and aligns with the broader goal of stress-literate training. Operationally, TrainOps addresses this uncertainty through a multi-modal measurement strategy: physiological signals (e.g., heart rate, heart-rate variability, electrodermal activity) are combined with accelerometry-derived movement data to compute motion-corrected indicators, and a machine-learning model trained on data from dedicated pilot and data-collection studies fuses these signals into a coarse real-time classification (e.g., a green/amber/red traffic-light indicator), so that instructors retain interpretative authority rather than treating any single value as ground truth.

From Stressful Scenarios to Stress-Literate Training

The preceding sections highlighted two complementary challenges in XR-based military training: the intentional design of stress (EviS) and the robust measurement of stress (TrainOps). Taken in isolation, each approach addresses only part of the problem. Designed stressors without measurement remain speculative, while physiological measurement without a design context remains difficult to interpret. This section integrates both perspectives and introduces the notion of stress-literate training as a guiding framework for future XR-based military training systems.

A Three-Layer Pipeline for Stress-Oriented XR Training

Based on the combined insights of EviS and TrainOps, stress-oriented XR training can be conceptualised as a three-layer pipeline:

• designed stress cues,
• measured physiological signals; and
• interpretable stress metrics.

The first layer consists of intentionally designed stress cues grounded in operational stressor taxonomies and translated into implementable XR elements. As discussed in Section 3, stress cues provide a structured way to operationalise abstract stressors such as uncertainty, time pressure, or social evaluation within virtual scenarios.[28] Crucially, this layer establishes design intent: it specifies which forms of stress are expected, when they occur, and how they are scaled.

The second layer captures trainees’ physiological responses to these stress cues using wearable biosensors. Cardiac activity, electrodermal signals, and related measures provide continuous, time-sensitive indicators of autonomic stress activation during training.[29] As emphasised in TrainOps, this layer must be robust to movement and task load to remain viable in operationally realistic XR settings.[30]

The third layer transforms physiological signals (e.g., heart rate, heart rate variability, electrodermal activity, respiration rate, pupil diameter) into actionable stress metrics. Rather than exposing instructors to raw biosignals, aggregated indicators summarise stress levels in ways that support situational awareness and reflection. Prior work demonstrates that simplified, interpretable visualisations – such as coarse stress categories (i.e., broad ordinal levels such as low, moderate, or high arousal) – are more useful for training practice than detailed physiological traces.

Together, these three layers create a closed loop between design intent and observed response. Stress is no longer an assumed by-product of immersive scenarios but a phenomenon that can be deliberately shaped, observed, and discussed.

Central Trade-offs in Stress-Oriented XR Training

Integrating stress design and stress measurement surfaces several fundamental trade-offs that must be addressed explicitly rather than implicitly.

One key tension lies between realism and measurability. Increasing ecological validity by adding complexity, sensory richness, or physical activity can improve perceived realism but often degrades signal quality and interpretability. Physiological measures such as heart rate are particularly susceptible to confounding effects from movement and exertion.[31] Stress-literate training, therefore, requires deliberate compromises, balancing realism with the need for interpretable data rather than maximising either dimension in isolation.

A second trade-off concerns automation versus instructor control. Automated stress metrics and dashboards can support instructors by highlighting trends or critical moments, but they risk shifting authority away from human judgment if treated as objective ground truth. Prior research on XR training support systems emphasises that trainers value physiological data primarily as a complementary source of information, not as a replacement for their expertise.[32] Stress-literate systems must therefore be designed as supplemental decision-support tools that preserve instructors’ interpretive role.

Both trade-offs underscore a central insight: stress-oriented XR training is not primarily a technical optimisation problem but a design problem involving pedagogical goals, human judgment, and system constraints. Beyond technical trade-offs, stress-literate XR training also mitigates leadership-relevant training pitfalls. Without calibration, trainees may experience “stress theatre” – visually intense scenarios that do fail to trigger meaningful operational stress – or, conversely, uncontrolled stress-overload that primarily trains avoidance rather than decision-making under responsibility.[33] Likewise, uncontextualised physiological metrics risk automation bias if interpreted as ground truth — that is, the tendency of human operators to over-trust automated or quantitative outputs and to under-weight contradictory contextual evidence. Stress-literate designs reduce these risks by coupling stress cues with interpretable metrics and by explicitly positioning physiological feedback as instructor support rather than evaluation, thereby preserving human judgment in training decisions.[34]

Defining Stress-Literate Training

Against this background, stress-literate training can be defined as an approach in which stress is treated as an explicit, observable, and adjustable dimension of training rather than as an implicit side effect of scenario intensity. Stress-literate training environments are characterised by three properties.

• First, stress is intentionally designed. Stressors are selected based on operational relevance and translated into controllable stress cues rather than emerging accidentally from scenario complexity;[35]
• Second, stress is measured and contextualised. Physiological responses are interpreted in relation to specific stress cues and scenario phases, acknowledging that stress responses are dynamic and context-dependent;[36] and
• Third, stress is made actionable. Stress information is presented in forms that support instructional decisions, reflection, and after-action review, rather than remaining hidden in raw data streams.[37]

This perspective reframes stress from something that merely happens to trainees into a resource that can be deliberately used to shape learning trajectories and operational readiness.

Implications for Military Training Practice

For training centres, stress-literate XR training requires shared frameworks and standards. Stressor taxonomies, sensor setups, and stress metrics should not be developed ad hoc for individual scenarios but aligned across training programs to support comparability and longitudinal development.

For instructors, stress-literate systems offer new forms of situational awareness. Rather than relying solely on behavioural observation or subjective impressions, trainers can access complementary physiological indicators to inform decisions about escalation, recovery, or debriefing. Importantly, this does not reduce the instructor’s role but expands it, enabling more nuanced reflection on when stress supports learning and when it undermines it.

For XR system designers, integrating stress design and measurement demands closer collaboration between content design, sensing infrastructure, and interface design. Stress-oriented XR systems must support parameterizable stress cues, reliable data acquisition, and interpretable feedback mechanisms as first-class features rather than as afterthoughts.

Limitations and Outlook

While stress-literate XR training offers clear advantages over implicitly stressful scenarios, several limitations and open challenges remain. A primary limitation concerns the interpretation of physiological stress metrics. Biosignals such as heart rate variability or electrodermal activity are influenced not only by psychological stress but also by physical exertion, fatigue, and individual autonomic differences.[38] Although incorporating movement data can reduce some confounding effects, stress metrics remain probabilistic indicators rather than definitive measures of stress.[39] Consequently, physiological data must always be interpreted in relation to scenario context and training intent. For instance, IMU- or video-derived activity estimates can be used to regress out the exertion component of heart-rate changes, and signals can be normalised against an individual’s pre-scenario baseline. Residual variance is then interpreted alongside behavioural markers (e.g., hesitation or communication breakdown) and scenario timestamps, so that a spike is only flagged as stress-relevant when it coincides with a meaningful scenario event and cannot be explained by physical load alone.

A second challenge lies in inter-individual variability. Stress responses depend strongly on experience, appraisal, and coping resources, leading to substantial differences between trainees exposed to identical stress cues.[40] While stress-literate training explicitly acknowledges this variability, it complicates the definition of universal thresholds (i.e., fixed physiological cutoffs, such as a specific heart rate or HRV value assumed to indicate stress across individuals and contexts) or standardised stress levels across trainees.

Finally, there is a risk of over-automation and misinterpretation. Physiological stress indicators may be perceived as objective ground truth, despite their contextual and indirect nature. Prior work emphasises that stress dashboards should support instructor judgment rather than replace it. Clear communication of uncertainty and careful integration into training practice are therefore essential.

These limitations underline that stress-literate XR training is not a finished solution but an evolving design space that requires iterative validation and close collaboration between researchers, developers, instructors, and trainees.

Conclusion

The growing adoption of XR technologies in military training has created new opportunities to prepare personnel for complex and demanding operational environments. Yet realism and immersion alone do not guarantee meaningful stress exposure. As this paper has argued, stress is neither a reliable by-product of scenario intensity nor a phenomenon that can be inferred from visual realism alone. Rather, stress emerges through the interaction between operational demands, individual appraisal, and contextual factors.

Drawing on the EviS and TrainOps projects, this article proposed a stress-literate approach to XR training that integrates three essential elements: the deliberate design of operationally relevant stress cues, the robust measurement of physiological responses, and the transformation of biosignals into interpretable metrics that support instructors and trainees alike. Together, these elements create a framework in which stress can be intentionally calibrated, observed, and discussed rather than merely assumed.

The broader implication is that future XR training systems should not be evaluated solely by their fidelity or technological sophistication. Their value will increasingly depend on their ability to generate appropriate levels of stress, provide meaningful feedback, and support learning under pressure. In this sense, stress is not simply a training outcome but a design variable.

Ultimately, the objective is not to maximise stress but to calibrate it. Armed forces that can systematically design, measure, and interpret stress within XR environments will be better positioned to prepare personnel for the cognitive and emotional demands of contemporary operations. Stress-literate training offers a practical pathway toward more evidence-informed, adaptive, and operationally relevant military training. The future of XR training will not be defined by how realistic virtual environments become, but by how effectively they help instructors understand, calibrate, and develop performance under stress.

[1] Lasse Jensen and Flemming Konradsen, “A Review of the Use of Virtual Reality Head-Mounted Displays in Education and Training,” Education and Information Technologies 23, no. 4 (2018): 1515–29, https://doi.org/10.1007/s10639-017-9676-0; Guido Makransky and Gustav Bøg Petersen, “Investigating the Process of Learning with Desktop Virtual Reality: A Structural Equation Modeling Approach,” Computers & Education 134 (June 2019): 15–30, https://doi.org/10.1016/j.compedu.2019.02.002; Federica Pallavicini et al., “Virtual Reality Applications for Stress Management Training in the Military,” Aerospace Medicine and Human Performance 87, no. 12 (2016): 1021–30.

[2] Sean T. Hannah et al., “A Framework for Examining Leadership in Extreme Contexts,” The Leadership Quarterly 20, no. 6 (2009): 897–919, https://doi.org/10.1016/j.leaqua.2009.09.006; James E. Driskell, Eduardo Salas, and Joan H. Johnston, “Does Stress Lead to a Loss of Team Perspective?” Group Dynamics: Theory, Research, and Practice 3, no. 4 (1999): 291–302, https://doi.org/10.1037/1089-2699.3.4.291.

[3] J. A. Healey and Rosalind W. Picard, “Detecting Stress During Real-World Driving Tasks Using Physiological Sensors,” IEEE Transactions on Intelligent Transportation Systems 6, no. 2 (2005): 156–66, https://doi.org/10.1109/TITS.2005.848368; Giorgos Giannakakis et al., “Review on Psychological Stress Detection Using Biosignals,” IEEE Transactions on Affective Computing 13, no. 1 (2022): 440–60, https://doi.org/10.1109/TAFFC.2019.2927337.

[4] Simone Grassini and Karin Laumann, “Questionnaire Measures and Physiological Correlates of Presence: A Systematic Review,” Frontiers in Psychology 11 (2020): 349, https://doi.org/10.3389/fpsyg.2020.00349.

[5] Richard S. Lazarus and Susan Folkman, Stress, Appraisal, and Coping (New York: Springer Publishing Company, 1984); Richard S. Lazarus and Susan Folkman, “Transactional Theory and Research on Emotions and Coping,” European Journal of Personality 1, no. 3 (1987): 141–69.

[6] Jaap M. Koolhaas et al., “Stress Revisited: A Critical Evaluation of the Stress Concept,” Neuroscience & Biobehavioral Reviews 35, no. 5 (2011): 1291–1301.

[7] Donald J. Campbell and Orly Ben-Yoav Nobel, “Occupational Stressors in Military Service: A Review and Framework,” Military Psychology 21, supp. 2 (2009): S47–67, https://doi.org/10.1080/08995600903249149.

[8] Campbell and Nobel, “Occupational Stressors in Military Service,” S47–67; Koolhaas et al., “Stress Revisited,” 1291–1301.

[9] Pallavicini et al., “Virtual Reality Applications for Stress Management Training in the Military,” 1021–30.

[10] Albert Rizzo et al., Human Emotional State and Its Relevance for Military VR Training (2005).

[11] Pallavicini et al., “Virtual Reality Applications for Stress Management Training in the Military,” 1021–30.

[12] Campbell and Nobel, “Occupational Stressors in Military Service,” S47–67.

[13] Pallavicini et al., “Virtual Reality Applications for Stress Management Training in the Military,” 1021–30.

[14] Campbell and Nobel, “Occupational Stressors in Military Service,” S47–67.

[15] Lazarus and Folkman, “Transactional Theory and Research on Emotions and Coping,” 141–69.

[16] Quynh Nguyen et al., “Stress Out: Translating Real-World Stressors into Audio-Visual Stress Cues in VR for Police Training,” in Human-Computer Interaction – INTERACT 2021, vol. 12933, ed. Carmelo Ardito et al., Lecture Notes in Computer Science (Cham: Springer International Publishing, 2021), 551–61, https://doi.org/10.1007/978-3-030-85616-8_32.

[17] Koolhaas et al., “Stress Revisited,” 1291–1301.

[18] Koolhaas et al., “Stress Revisited,” 1291–1301.

[19] Giannakakis et al., “Review on Psychological Stress Detection Using Biosignals,” 440–60.

[20] Jakob C. Uhl et al., “Threat Assessment in Police VR Training: Multi-Sensory Cues for Situation Awareness,” in 2022 IEEE International Conference on Metrology for Extended Reality, Artificial Intelligence and Neural Engineering (MetroXRAINE) (2022), 432–37.

[21] Fred Shaffer and J. P. Ginsberg, “An Overview of Heart Rate Variability Metrics and Norms,” Frontiers in Public Health 5 (2017): 258, https://doi.org/10.3389/fpubh.2017.00258; Lotte Linssen et al., “Using Accelerometry and Heart Rate Data for Real-Time Monitoring of Soldiers’ Stress in a Dynamic Military Virtual Reality Scenario,” Multimedia Tools and Applications 81, no. 17 (2022): 24739–56.

[22] Healey and Picard, “Detecting Stress During Real-World Driving Tasks Using Physiological Sensors,” 156–66; Giannakakis et al., “Review on Psychological Stress Detection Using Biosignals,” 440–60.

[23] Uhl et al., “Threat Assessment in Police VR Training,” 432–37; Olivia Zechner et al., “Mind the Heart: Designing a Stress Dashboard Based on Physiological Data for Training Highly Stressful Situations in Virtual Reality,” in Human-Computer Interaction – INTERACT 2023, ed. José Abdelnour Nocera et al. (Cham: Springer Nature Switzerland, 2023), 209–30.

[24] Leo Breiman, “Random Forests,” Machine Learning 45, no. 1 (2001): 5–32; Tianqi Chen and Carlos Guestrin, “XGBoost: A Scalable Tree Boosting System,” in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining (2016), 785–94, https://doi.org/10.1145/2939672.2939785.

[25] Giannakakis et al., “Review on Psychological Stress Detection Using Biosignals,” 440–60.

[26] Markus Murtinger et al., “Assist the VR Trainer – Real-Time Dashboard and After-Action Review for Police VR Training,” in 2022 IEEE International Conference on Metrology for Extended Reality, Artificial Intelligence and Neural Engineering (MetroXRAINE), October 26, 2022, 69–74, https://doi.org/10.1109/MetroXRAINE54828.2022.9967532.

[27] Zechner et al., “Mind the Heart,” 209–30.

[28] Campbell and Nobel, “Occupational Stressors in Military Service,” S47–67; Nguyen et al., “Stress Out,” 551–61.

[29] Shaffer and Ginsberg, “An Overview of Heart Rate Variability Metrics and Norms,” 258; Giannakakis et al., “Review on Psychological Stress Detection Using Biosignals,” 440–60.

[30] Uhl et al., “Threat Assessment in Police VR Training,” 432–37; Linssen et al., “Using Accelerometry and Heart Rate Data for Real-Time Monitoring of Soldiers’ Stress,” 24739–56.

[31] Linssen et al., “Using Accelerometry and Heart Rate Data for Real-Time Monitoring of Soldiers’ Stress,” 24739–56.

[32] Murtinger et al., “Assist the VR Trainer,” 69–74.

[33] Campbell and Nobel, “Occupational Stressors in Military Service,” S47–67; Koolhaas et al., “Stress Revisited,” 1291–1301.

[34] Murtinger et al., “Assist the VR Trainer,” 69–74.

[35] Campbell and Nobel, “Occupational Stressors in Military Service,” S47–67; Nguyen et al., “Stress Out,” 551–61.

[36] Lazarus and Folkman, Stress, Appraisal, and Coping; Koolhaas et al., “Stress Revisited,” 1291–1301.

[37] Zechner et al., “Mind the Heart,” 209–30.

[38] Shaffer and Ginsberg, “An Overview of Heart Rate Variability Metrics and Norms,” 258; Giannakakis et al., “Review on Psychological Stress Detection Using Biosignals,” 440–60.

[39] Linssen et al., “Using Accelerometry and Heart Rate Data for Real-Time Monitoring of Soldiers’ Stress,” 24739–56.

[40] Lazarus and Folkman, Stress, Appraisal, and Coping; Koolhaas et al., “Stress Revisited,” 1291–1301.