Virtual Selves, Real Emotions: The Impact of Photorealistic Avatar Personalization on Emotion in Virtual Reality

Paper Title

Virtual Selves, Real Emotions: The Impact of Photorealistic Avatar Personalization on Emotion in Virtual Reality

Publication Info

• Topic area: The effects of photorealistic personalized avatars on emotional responses in virtual reality.
• Keywords: Virtual reality, photorealistic avatars, personalization, emotion, embodiment, presence, physiological measures, arousal, valence, self-regulation.

Background and Problem

• Problem / challenge: While photorealistic personalized avatars are now technologically feasible, their high production cost necessitates understanding their added value, particularly their impact on emotions in VR, which remains underexplored.
• Significance: Emotions are central to VR applications across domains like gaming, healthcare, and education. Understanding how personalized avatars influence emotions can inform design decisions and justify their costs.
• Motivation and related work: Prior research has examined generic or stylized avatars and their effects on emotions, but studies on photorealistic personalized avatars have focused on emotional reactions to avatars rather than emotions experienced through them. Additionally, previous work has not systematically explored the full arousal-valence spectrum of emotions or employed comprehensive physiological metrics.

Solution

• Proposed approach: Systematic investigation of the effects of photorealistic personalized avatars on emotional responses in VR, using validated virtual environments (VEs) and both subjective and physiological measures.
• Novelty:
  1. Development and validation of four emotion-inducing VEs based on Russell’s Circumplex Model of Affect (CMA).
  2. Creation of photorealistic personalized avatars using advanced photogrammetry and motion capture techniques.
  3. Comprehensive measurement of emotions using psychometric scales and physiological markers.
  4. First evidence of how photorealistic personalized avatars affect emotions across the full arousal-valence spectrum.
• Procedure and key techniques:
  • Designed four VEs targeting high/low arousal and valence combinations.
  • Created photorealistic personalized avatars using a 144-camera photogrammetry rig and motion capture.
  • Conducted a within-participants study (n=51) where participants experienced the VEs through three avatar conditions (no avatar, generic avatar, personalized avatar).
  • Measured emotions using self-reports and physiological metrics (e.g., heart rate variability, skin conductance, pupillometry).

Results

• Concrete findings:
  • Personalized avatars increased subjective emotional arousal and valence compared to generic and no avatars.
  • HRV, a marker of self-regulation, was significantly higher for personalized avatars.
  • Personalized avatars enhanced embodiment, presence, and avatar perception metrics (e.g., visual appeal, realism).
  • Embodiment (generic and personalized avatars) increased physiological markers of arousal (e.g., skin conductance, pupil dilation).
• Advantage over baselines:
  • Personalized avatars consistently outperformed generic avatars and no avatars in emotional engagement, embodiment, and presence.
  • Generic avatars were perceived as more eerie than personalized avatars.
• Experiments / evaluation:
  • Validated four VEs (n=302) for inducing emotions across the CMA dimensions.
  • Main study (n=51) used a counterbalanced design to compare avatar conditions across subjective and physiological measures.
• Limitations and future work:
  • Potential novelty effects due to participants’ limited prior VR experience.
  • Generic avatars lacked diversity in skin tone customization.
  • Future research should explore longitudinal effects, interactive VR scenarios, and additional physiological measures (e.g., EEG).

Summary

This study systematically investigated the effects of photorealistic personalized avatars on emotions in VR, demonstrating that such avatars enhance emotional engagement, embodiment, and presence compared to generic or no avatars. Personalized avatars uniquely increased HRV, suggesting potential applications in emotion regulation and self-regulation interventions. The findings provide actionable insights for VR developers and researchers, emphasizing the value of photorealistic personalization in emotionally engaging VR applications. Future work should address avatar diversity, explore interactive scenarios, and examine long-term effects.