Sense of Presence in Virtual Reality: Exploration and Validation of Neurophysiological Markers

Background Introduction

In recent years, Virtual Reality (VR) technology has been widely applied in fields such as medicine, training, and rehabilitation. The core of VR lies in the user’s “sense of presence,” which refers to the immersive experience of “being there” within a virtual environment. However, current assessments of presence primarily rely on subjective questionnaires, such as the ITC-SOPI (ITC-Sense of Presence Inventory) and SUS (Slater-Usoh-Steed) questionnaires. These methods are prone to subjective bias and struggle to capture unconscious responses. Therefore, developing an objective assessment method based on neurophysiological signals has become an urgent research need.

The aim of this study was to identify neurophysiological markers associated with the sense of presence in VR environments using multimodal physiological signals (e.g., EEG, ECG, and EDA) and to develop a synthetic index based on Mutual Information (MI), called the Sense of Presence Mutual Information Index (SOPMI), to achieve an objective quantification of presence.

Paper Source

The paper was co-authored by Vincenzo Ronca, Fabio Babiloni, and Pietro Aricò from the Department of Computer, Control, and Management Engineering and the Department of Physiology and Pharmacology at Sapienza University of Rome, Italy. It was published in 2025 in the journal IEEE Transactions on Biomedical Engineering and received support from the PRIN 2022 PNRR “Fit2Work” project.

Research Workflow

1. Participants and Experimental Equipment

The study recruited 18 healthy male volunteers (average age 26.5 ± 3.2 years) and was conducted in the laboratory of Sapienza University of Rome. The experimental equipment included an Oculus Rift DK2 head-mounted display (HMD), a Logitech G27 racing wheel, a Nexus-10 MKII system (for recording skin conductance activity, EDA), and a Galileo BePlus electroencephalography (EEG) device. The experimental scenario was a racetrack where participants drove a virtual car from a first-person perspective and collected target objects under different task conditions.

2. Experimental Design

The experiment designed two conditions—high perceptual load and low perceptual load—adjusting task difficulty through weather conditions and the number of targets. Participants completed four multisensory stimulation conditions: visual only (V), visual + auditory (VA), visual + vibrotactile (VV), and visual + auditory + vibrotactile (VAV). In each condition, participants performed a 3-minute driving task and then completed the NASA Task Load Index (NASA-TLX) and a presence questionnaire after the task.

3. Signal Recording and Processing

EEG signals were recorded using 38 electrodes at a sampling frequency of 256 Hz. Signal preprocessing included a 50 Hz notch filter, a 2–40 Hz band-pass filter, and independent component analysis (ICA) to remove ocular and muscle artifacts. Skin conductance activity (EDA) was recorded using the Nexus-10 MKII system and processed with the LEDALAB toolkit for continuous decomposition analysis to extract the skin conductance level (SCL). Electrocardiogram (ECG) signals were recorded using three electrodes, and heartbeats were detected using the Pan-Tompkins algorithm to calculate heart rate variability (HRV).

4. Data Analysis

The study first validated the experimental design using repeated-measures ANOVA, analyzing subjective parameters related to task load and sense of presence. Subsequently, preliminary analyses were conducted on EEG and autonomic signals (e.g., SOP1, SOP2, SCL, HR, and HRV) to identify neurophysiological features associated with the sense of presence. Finally, a multivariate analysis method based on mutual information (MI) was used to develop the synthetic presence index (SOPMI), and its sensitivity and specificity across different experimental conditions were evaluated.

Main Results

1. Subjective Assessment

NASA-TLX analysis revealed that task difficulty significantly affected mental workload (p < 0.001), with low-load conditions perceived as less demanding. Presence questionnaire analysis showed that immersion levels significantly influenced the sense of presence (p < 0.001), with presence scores in the VAV condition significantly higher than in other conditions.

2. Neurophysiological Assessment

EEG features (SOP1 and SOP2) showed significant differences in both task difficulty and immersion levels (p < 0.001). Skin conductance level (SCL) was significantly higher in high-load and high-immersion conditions (p = 0.02), while heart rate (HR) and heart rate variability (HRV) did not show significant differences.

3. Mutual Information-Based Presence Index (SOPMI)

The SOPMI index was significantly higher in the VAV condition compared to other conditions (p < 0.01) and was unaffected by task difficulty. This index showed a significant positive correlation with subjective presence ratings (r = 0.559, p < 0.007), indicating its ability to objectively reflect the user’s sense of presence experience.

Conclusion

This study successfully identified neurophysiological markers associated with the sense of presence in VR environments and preliminarily validated the MI-based presence index (SOPMI) as an objective assessment tool. This method effectively captures the immersive experience of users in VR without being influenced by task difficulty or other cognitive modulations, providing new tools for the design and optimization of VR applications.

Research Highlights

1. Multimodal Physiological Signal Analysis: For the first time, this study combined EEG, ECG, and EDA signals and developed a synthetic presence index using a multivariate analysis approach, overcoming the limitations of traditional subjective assessment methods.
2. Application of Mutual Information Technology: The MI-based method effectively extracts common patterns among different physiological signals, enhancing the specificity of presence assessment.
3. Practical Application Value: This research provides a theoretical foundation for VR applications in fields such as healthcare, training, and Industry 5.0, helping to develop more immersive and user-friendly VR experiences.

Other Valuable Information

The study also highlighted limitations in the current experimental design, such as the singularity of the experimental scenarios and the lack of scalability of high-density EEG equipment. Future research could further validate the reliability of the SOPMI index in more diverse and dynamic VR environments and explore its potential application in wearable devices.