Navigating the Virtual Gaze: Social Anxiety's Role in VR proxemics

Title of the Paper

Navigating the Virtual Gaze: Social Anxiety’s Role in VR Proxemics

Paper Information

• Subject Area: Social interaction in virtual reality, specifically the impact of social anxiety on gaze behavior and interpersonal distance perception with virtual human agents (VHAs).
• Keywords: Social anxiety, virtual reality, virtual human agents, gaze behavior, interpersonal space

Research Background and Problem

• Identified Issues or Challenges:
  • Social anxiety (SA) affects how individuals process social cues, including avoiding direct eye contact and maintaining greater interpersonal distances (IPD). Existing studies have primarily focused on either gaze behavior or interpersonal distance, but lack a comprehensive exploration of both.
  • For virtual human agents (VHAs), designing gaze behavior and interpersonal space that better aligns with individual needs, particularly for socially anxious individuals, remains a challenge.
• Significance:
  • With the increasing prevalence of virtual reality (VR) and social VR applications, it is crucial to design more user-friendly human-computer interaction experiences, especially for users with social impairments.
  • The sensitivity of socially anxious individuals to social cues in both real and virtual environments may impact their overall sense of engagement in virtual settings.
• Research Motivation and Related Work:
  • Previous studies present conflicting findings: some suggest that direct eye contact increases discomfort and IPD, while others emphasize that direct gaze can foster intimate interaction.
  • This study aims to resolve these academic discrepancies by integrating dynamic and static gaze animations and considering individual characteristics such as levels of social anxiety, thereby providing new guidance for the design of virtual agents.

Solution

• Proposed Methods or Solutions:
  • Compare the effects of different gaze animations (static vs. dynamic, direct gaze vs. averted gaze) on preferred IPD, combined with social anxiety (SA) scores to analyze participants' behavioral patterns.
  • Use Bayesian parameter estimation to analyze the data and quantify the interaction effects between static and dynamic gaze animations and SA.
• Innovations:
  • Explore the role of gaze behavior in regulating interpersonal distance through the integration of dynamic and static gaze interactions.
  • Incorporate the dynamic influence of ET (Equilibrium Theory) to reveal how SA alters the balance of gaze and IPD in human-computer interaction.
• Implementation Steps and Key Techniques:
  1. Experimental Design: Participants perform a stop-distance task in VR, recording their IPD when approaching a VHA.
  2. Controlled Variables: Four gaze conditions (dynamic direct gaze, dynamic averted gaze, static direct gaze, static averted gaze).
  3. Measurement Tools: Use the Liebowitz Social Anxiety Scale (LSAS) to measure participants' SA levels and eye-tracking devices to record gaze behavior.
  4. Data Analysis: Apply Bayesian linear mixed models to analyze the effects of gaze conditions, SA scores, and their interactions on IPD.

Research Findings

• Specific Findings:
  • For participants with average SA levels, VHAs exhibiting direct gaze (both dynamic and static) led to shorter preferred IPD, while averted gaze (whether static or dynamic) increased IPD.
  • Participants with higher SA levels preferred larger IPD when VHAs maintained static direct gaze but chose smaller IPD with static averted gaze.
  • Dynamic direct gaze was perceived as a signal of affiliation, significantly reducing IPD and enhancing overall comfort and engagement in interactions.
• Comparison with Existing Solutions:
  • This study partially supports and refines Bailenson et al.'s findings on high-SA individuals' tendency to avoid direct gaze, uncovering more nuanced SA response patterns: high-SA individuals exhibit distinct IPD adjustments under different gaze conditions.
  • Provides a more comprehensive theoretical framework, integrating ET theory and highlighting the unique advantages of dynamic gaze.
• Experimental or Evaluation Results:
  • Bayesian analysis revealed that IPD adjustments varied under static and dynamic gaze conditions depending on participants' SA levels.
  • Confirmed the influence of VHA gender and participant gender on IPD, with male participants exhibiting shorter IPD toward female virtual agents.
• Limitations and Future Directions:
  • Cultural factors may significantly influence perceptions of interpersonal distance and gaze, warranting further investigation in future studies.
  • The potential mediating role of physiological arousal (e.g., skin conductance response) between SA and IPD remains unexplored and is a critical area for future research.
  • The experiment did not fully control for the effects of the initial position of dynamic gaze, and future studies should more precisely isolate the differences between initial and final gaze effects.

Conclusion

This study reveals the complex relationship between SA, virtual gaze behavior, and IPD, providing theoretical support for the design of social VR applications. It underscores the necessity and feasibility of incorporating individual psychological traits into virtual agent design. The findings offer valuable guidance for creating more inclusive virtual social experiences.