Immersion Assessment of Digital Caves Combining Eye Tracking and First-Person Perspective

Abstract

The evaluation of user experience within immersive virtual environments has traditionally relied on subjective questionnaires, which suffer from post-hoc rationalization and interruption of the presence state. This paper proposes a novel multimodal assessment framework for Digital Cave Automatic Virtual Environments (CAVE) that integrates mobile eye tracking with first-person perspective video analysis. By synchronizing physiological gaze data with the user's visual field, we aim to establish objective correlates of immersion. We investigate the relationship between visual attention patterns—specifically fixation duration, saccadic amplitude, and pupillary response—and selfreported presence scores in a projection-based virtual environment. Our methodology circumvents the physical intrusiveness of headmounted displays, preserving the naturalistic viewing conditions inherent to CAVE systems. The study employs a dual-layer analysiswhere quantitative eye-movement metrics are mapped onto qualitative scene content to determine how specific design elements trigger immersive engagement. Results indicate a significant positive correlation between focused visual exploration of depth cues and heightened states of presence. The findings suggest that non-intrusive physiological monitoring can provide a continuous, real-time index of immersion, offering a robust alternative to static psychometric scales in the design and evaluation of virtual reality systems.