Get Real With Me: Effects of Avatar Realism on Social Presence and Comfort in Augmented Reality Remote Collaboration and Self-Disclosure

Research Background and Problem

• What problems or challenges did the authors identify?
  Current augmented reality (AR) systems enhance remote social interactions by displaying realistic user representations, thereby increasing social presence. However, the impact of highly realistic avatars in intimate interactions remains unclear and may lead to discomfort or anxiety. Some studies suggest that reducing avatar realism in intimate scenarios could mitigate the negative effects of social presence.

• Why is this problem important?
  As AR technology becomes increasingly prevalent in contexts such as remote collaboration, psychotherapy, and telemedicine, understanding how avatar realism affects user comfort and social presence is crucial for optimizing the design of future applications.

• Research Motivation and Related Work
  Existing research primarily focuses on the impact of avatar realism on social presence in collaborative scenarios, often exploring specific levels of visual or behavioral fidelity. However, there has been a lack of systematic analysis on how to optimize avatar design in sensitive scenarios (e.g., self-disclosure). The authors aim to extend this research by exploring the adaptability of avatar realism in various social interactions and its correlation with individual comfort levels.

Solution

• What methods or solutions did the authors propose?
  The authors employed a two-phase research approach to systematically analyze the effects of avatar realism on social presence and user comfort:

  1. Online Survey (N=157): Evaluated user preferences for seven different levels of avatar realism.
  2. Laboratory Experiment (N=42): Designed two scenarios (collaborative game task and self-disclosure task) to compare the effects of four types of avatars (ranging from abstract to realistic point-cloud models).

• What are the innovative aspects of this solution?

  • Expanded the research domain of AR and realism, particularly in sensitive self-disclosure scenarios.
  • Systematically selected avatar samples covering a wide range of realism, from highly abstract to fully realistic.
  • Combined established psychological scales (e.g., Networked Minds Social Presence Inventory [NMSPI], Positive and Negative Affect Schedule [PANAS]) with qualitative data to evaluate user experience comprehensively.

• What are the implementation steps and key technologies used?

  • Online Survey: Users ranked the suitability of different avatars across various contexts.
  • Laboratory Experiment:
    • Improved point-cloud avatar technology, such as background removal.
    • Users interacted in self-disclosure and collaborative scenarios using customized devices.
    • Participants wore AR headsets, and data on social presence and emotional responses were recorded.
  • Quantitative Analysis: Used the Aligned Rank Transform method for statistical validation of task-avatar conflicts.

Research Findings

• What specific findings were obtained?

  • Higher avatar realism (e.g., point-cloud and realistic photo models) significantly increased users' social presence, particularly in emotional understanding (PAU) and emotional interdependence (PEI).
  • Avatar realism positively correlated with user comfort, significantly reducing psychological discomfort compared to abstract avatars, while enhancing positive emotions.
  • In self-disclosure scenarios, although social presence was higher, overall comfort was slightly lower, reflecting potential anxiety in intimate interactions.

• What advantages does this solution have compared to existing ones?

  • Provided a more comprehensive multi-scenario analysis and cross-avatar type comparison of user experiences.
  • Offered actionable insights into user preferences and sensitive scenarios, optimizing AR avatar design.
  • Thoroughly evaluated the complex relationship between social presence and emotions through psychological scales and qualitative feedback.

• What were the experimental or evaluation results?

  • Point-cloud avatars performed best across all scenarios, but technical issues (e.g., image quality, headset occlusion of eyes) reduced their effectiveness.
  • Photo-realistic avatars were the second-best option, offering more consistent visual impact and higher technical controllability.
  • Abstract avatars (e.g., particle clouds) provided psychological comfort for some users (those preferring anonymity and emotional distance) in sensitive scenarios like self-disclosure.

• Limitations and Future Directions

  • Limitations: Technical issues with point-cloud avatars, such as pixelation and tracking limitations; participant sample bias in terms of gender and personality preferences; lack of testing in broader application scenarios.
  • Future Work:
    1. Improve technical performance to enhance point-cloud avatar transmission quality and resolution.
    2. Explore customizable avatar models to meet individual needs.
    3. Expand task types and interaction scenarios (e.g., remote workplace negotiations).
    4. Refine research on social factors, such as the impact of self-presentation on interaction experiences.

This study significantly advances the understanding of future AR remote interactions, providing empirical guidance for avatar design while identifying optimization directions to better accommodate the complexities of individual needs in diverse social scenarios.