Towards Intelligent VR Training: A Physiological Adaptation Framework for Cognitive Load and Stress Detection DOI: XXXXXXX.XXXXXXX Conference: 33rd ACM International Conference on User Modeling, Adaptation and Personalization; June 16–19, 2025; NYC, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing

Mahsa Nasri · 2025

Read Towards Intelligent VR Training: A Physiological Adaptation Framework for Cognitive Load and Stress Detection DOI: XXXXXXX.XXXXXXX Conference: 33rd ACM International Conference on User Modeling, Adaptation and Personalization; June 16–19, 2025; NYC, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing by Mahsa Nasri (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2504-06461.txt.

Introduction

Mahsa Nasri designs virtual reality training that reads the trainee's body and adjusts itself. Eye tracking and heart rhythm can betray rising mental load and stress, but training systems rarely use them live. Her framework runs in three phases: first, volunteers perform a classic attention task in VR while their signals are recorded and labeled, teaching models to spot overload.

Next, the models are tuned to new users and wired into a training world that eases or raises difficulty as load changes. A third phase builds a privacy-aware version of the detector for comparison. Nasri aims the work at education, skills training, and health care, with privacy treated as a design duty.

Catalog record

CREATORMahsa Nasri
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE21 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-06461

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Towards Intelligent VR Training: A Physiological Adaptation Framework for Cognitive Load and Stress Detection DOI: XXXXXXX.XXXXXXX Conference: 33rd ACM International Conference on User Modeling, Adaptation and Personalization; June 16–19, 2025; NYC, NY ISBN: 978-1-4503-XXXX-X/2018/06 CCS: Computing methodologies Classification and regression trees CCS: Human-centered computing Virtual reality Mahsa Nasri Abstract . Adaptive Virtual Reality (VR) systems have the potential to enhance training and learning experiences by dynamically responding to users’ cognitive states. This research investigates how eye tracking and heart rate variability (HRV) can be used to detect cognitive load and stress in VR environments, enabling real-time adaptation. The study follows a three-phase approach: (1) conducting a user study with the Stroop task to label cognitive load data and train machine learning models to detect high cognitive load, (2) fine-tuning these models with new users and integrating them into an adaptive VR system that dynamically adjusts training difficulty based on physiological signals, and (3) developing a privacy-aware approach to detect high cognitive load and compare this with the adaptive VR in Phase two. This research contributes to affective computing and adaptive VR using physiological sensing, with applications in education, training, and healthcare. Future work will explore scalability, real-time inference optimization, and ethical considerations in physiological adaptive VR. 1. Introduction Virtual Reality (VR) is more than a technological advancement—it is a transformative medium that redefines how we perceive, interact with, and adapt to digital environments. VR provides an immersive, safe, and credible simulation of real-life training environments ( Choi et al. 2015 ; Bailenson 2018 ; Bigonah et al. 2025 ) . However, most current VR training applications provide unvaried training to all users and do not consider individual differences, such as cognitive abilities or emotional responses. This lack of adaptation can lead to cognitive overload, increased stress, and frustration, ultimately reducing learning efficiency and user engagement ( Sweller et al. 2019 ; Sandi 2013 ) . In my interdisciplinary PhD research, I explore the intersection of affective computing, VR design, and physiological sensing. Affective computing seeks to develop systems capable of recognizing, interpreting, and adapting to human emotions. At the same time, VR design provides the tools to craft experiences that can meaningfully engage and influence users. Physiological sensing bridges these fields, offering real-time insights into users’ cognitive and emotional states through physiological signals such as heart rate variability (HRV) and eye-tracking ( Souchet et al. 2022 ; Thayer 2009 ; Moon et al. 2023 ) . High cognitive load has been shown to induce stress and frustration, which negatively impact user performance and motivation. When mental demands exceed cognitive resources, users experience a loss of control, leading to emotional distress ( Paas et al. 2003 ) . Additionally, excessive cognitive demands activate the autonomic nervous system, increasing physiological arousal markers such as pupil dilation and reduced HRV ( Laborde et al. 2017 ) . This can create a feedback loop where stress further impairs cognitive function, reducing engagement and learning outcomes ( Sweller et al. 2019 ) . To address this, adaptive VR environments dynamically adjust task difficulty and feedback mechanisms based on users’ real-time cognitive and emotional states. By integrating Machine Learning (ML) models trained on physiological data, these systems can prevent cognitive overload, regulate stress, and enhance user experience ( Shin 2009 ; Kritikos et al. 2021 ) . Additionally, adaptive VR offers solutions for diverse applications, from education, which can adjust difficulty levels to optimize learning, to healthcare, and serve as a non-invasive, cost-effective alternative for pain management or stress reduction ( Sweller et al. 2019 ) . This research aims to develop a physiological adaptation framework based on cognitive load detection that adjusts VR training difficulty in real time. Specifically, this study: • Develops a machine learning-based model using HRV and eye-tracking data to detect cognitive load and emotional stress. • Conducts a three-phase user study to train and fine-tune ML models using the Stroop task as ground truth. • Evaluates the efficiency of adaptive VR on user learning, engagement, and stress mitigation compared to privacy-aware adaptive and regular VR. By addressing the interplay between cognitive load, emotions, and physiological responses, this research contributes to affective computing, adaptive VR, and human-centered design. 2. Related Work A systematic literature review was conducted using the Web of Science database, covering ACM, IEEE, and Springer publications from 2019 to 2024 to identify research gaps in affective computing, physiological sensing, and VR design. The search combined terms related to adaptation (e.g., "personalized," "adaptive") with biofeedback-related terms (e.g., "EEG," "GSR," "eye tracking") and "virtual reality." A total of 164 papers were identified, with 117 deemed relevant after the selection of the title and abstract. After filtering, 38 papers met the inclusion criteria, focusing on real-time physiological data use rather than post-analysis. 2.1. Cognitive Load, Emotion, and Physiological Responses in VR Cognitive load significantly influences users’ emotions, particularly stress and frustration. When cognitive demands exceed a user’s working memory capacity, stress levels increase, reducing performance and engagement ( Sweller et al. 2019 ; Paas et al. 2003 ) . Physiological responses such as pupil dilation and reduced HRV correlate with high cognitive load, making them valuable signals for real-time adaptation ( Laborde et al. 2017 ) . Additionally, cognitive overload can trigger emotional distress, leading to a feedback loop where increased stress further impairs cognitive function ( Sweller et al. 2019 ) . Understanding these physiological responses allows computational models to detect cognitive load and dynamically adjust VR experiences to mitigate stress and enhance engagement. 2.2. VR Adaptation Mechanisms VR adaptation mechanisms can be categorized into two primary approaches: rule-based adaptation and data-driven adaptation . Rule-Based Adaptation: These approaches use predefined heuristics to modify VR parameters such as task difficulty, environmental features, and interaction mechanics. Some studies adjust difficulty dynamically to maintain engagement by modifying complexity ( Blum et al. 2019 ; Dey et al. 2019 ; Izountar et al. 2021 ) . Others adapt experience mechanics by altering users’ virtual positions or modifying stimulus characteristics ( Uyan and Celikcan 2024 ; Apicella et al. 2024 ; Baldini et al. 2024 ) . Environmental adaptations, such as adjusting colors or introducing elements like fog, also play a role in immersive adaptation ( Gomes et al. 2021 ; Gupta et al. 2024 ; Prabhu et al. 2020a ) . Data-Driven Adaptation: More recent approaches integrate physiological and behavioral data to adapt experiences. These models leverage physiological signals such as heart rate, EEG, and eye tracking to infer user state and modify the environment accordingly. For instance, Baldini et al. ( Baldini et al. 2024 ) developed a VR system that mitigates cybersickness by adapting navigation speed and scene complexity based on EEG signals. Some studies use NPC behavior adaptation, such as streamlining NPC interactions based on user responses ( Chiossi et al. 2024 ; Chiossi et al. 2022 ) . While these adaptation mechanisms enhance immersion, most rely on predefined heuristics. To enhance adaptability, computational models that leverage real-time physiological sensing have been explored. 2.3. Computational Models for Adaptive VR Computational models for adaptation in VR fall into two primary categories: direct physiological adaptation and ML-based adaptation . Direct Physiological Adaptation: Some studies use real-time physiological signals to trigger immediate environmental changes. For

Read the complete text