Virtual reality (VR) provides immersive learning experiences, yet accurately assessing self-regulated learning (SRL) within these environments remains a significant challenge. Traditional subjective and post hoc assessment techniques frequently fail to capture the dynamic, real-time nature of SRL processes. Unlike prior research that typically integrates multimodal traces, such as logs, physiological data, or eye-tracking, in controlled laboratory settings, we propose a novel approach that seamlessly embeds eye-tracking data as an intrinsic, real-time input to the game’s mechanics, creating a unified, three-component (Cognition, Metacognition, Motivation) assessment framework. This design represents a significant departure from controlled lab studies, allowing immediate, unobstructive, and holistic assessment of SRL metrics. Drawing on Zimmerman’s cyclical model of SRL, the game operationalizes cognition, metacognition, and motivation through interactive mechanics, including information collection, strategic planning, and reflective self-reporting. This study presents a system that collects real-time data through eye-tracking within a virtual environment to measure SRL. Currently, the system captures and analyzes learner behaviors and cognitive engagement in real-time, while adaptive feedback mechanisms will be developed in future iterations to provide personalized guidance based on the collected data. The multidimensional structure of SRL is validated through Structural Equation Modeling (SEM) along with Convergent and Discriminant Validity analyses, i.e., Confirmatory Factor Analysis (CFA) and Variance Inflation Factor (VIF), confirming that Cognition, Metacognition, and Motivation components are distinct yet interrelated components at the data level. An experimental study involving 103 students shows promising correlations between in-game metrics and established external instruments: cognition correlates with General Academic Prerequisites (OSP) test (r = 0.504, p <Emphasis Type="BoldItalic"><</Emphasis> 0.001), metacognition by human raters and AI-based assessments (r = 0.673, p <Emphasis Type="BoldItalic"><</Emphasis> 0.001), and a strong correlation between in-game parameters and expert-determined metacognition (r = 0.783, p <Emphasis Type="BoldItalic"><</Emphasis> 0.001). Motivational indicators correlate with Intrinsic Motivation Inventory (IMI) scores (r = 0.430, p <Emphasis Type="BoldItalic"><</Emphasis> 0.001). These findings provide preliminary evidence supporting the potential validity of the proposed tool, requiring further validation via future studies with more diverse populations. This study presents a foundational framework for objective, real-time, and gamified assessment of SRL, thereby providing preliminary evidence supporting the potential of the proposed tool as a base for developing personalized learning strategies. The findings need cautious interpretation and further validation in diverse populations.

Integrating virtual reality and eye-tracking as a gamified assessment tool for self-regulated learning

期刊论文
出版模式：

开放获取
发表日期：

2026年3月26日

查看全文

David Vanecek,
Inam Ur Rehman,
Martin Dobias

摘要

Virtual reality (VR) provides immersive learning experiences, yet accurately assessing self-regulated learning (SRL) within these environments remains a significant challenge. Traditional subjective and post hoc assessment techniques frequently fail to capture the dynamic, real-time nature of SRL processes. Unlike prior research that typically integrates multimodal traces, such as logs, physiological data, or eye-tracking, in controlled laboratory settings, we propose a novel approach that seamlessly embeds eye-tracking data as an intrinsic, real-time input to the game’s mechanics, creating a unified, three-component (Cognition, Metacognition, Motivation) assessment framework. This design represents a significant departure from controlled lab studies, allowing immediate, unobstructive, and holistic assessment of SRL metrics. Drawing on Zimmerman’s cyclical model of SRL, the game operationalizes cognition, metacognition, and motivation through interactive mechanics, including information collection, strategic planning, and reflective self-reporting. This study presents a system that collects real-time data through eye-tracking within a virtual environment to measure SRL. Currently, the system captures and analyzes learner behaviors and cognitive engagement in real-time, while adaptive feedback mechanisms will be developed in future iterations to provide personalized guidance based on the collected data. The multidimensional structure of SRL is validated through Structural Equation Modeling (SEM) along with Convergent and Discriminant Validity analyses, i.e., Confirmatory Factor Analysis (CFA) and Variance Inflation Factor (VIF), confirming that Cognition, Metacognition, and Motivation components are distinct yet interrelated components at the data level. An experimental study involving 103 students shows promising correlations between in-game metrics and established external instruments: cognition correlates with General Academic Prerequisites (OSP) test (r = 0.504, p < 0.001), metacognition by human raters and AI-based assessments (r = 0.673, p < 0.001), and a strong correlation between in-game parameters and expert-determined metacognition (r = 0.783, p < 0.001). Motivational indicators correlate with Intrinsic Motivation Inventory (IMI) scores (r = 0.430, p < 0.001). These findings provide preliminary evidence supporting the potential validity of the proposed tool, requiring further validation via future studies with more diverse populations. This study presents a foundational framework for objective, real-time, and gamified assessment of SRL, thereby providing preliminary evidence supporting the potential of the proposed tool as a base for developing personalized learning strategies. The findings need cautious interpretation and further validation in diverse populations.