The (Anti-)Affordance Problem: Effects of Physical Context on Collaborator Placement in Augmented Reality Meetings

Paper Title

The (Anti-)Affordance Problem: Effects of Physical Context on Collaborator Placement in Augmented Reality Meetings

Publication Info

• Topic area: Avatar placement preferences in AR collaboration influenced by physical context and task type.
• Keywords: Augmented Reality, avatar placement, proxemics, social layouts, affordance, anti-affordance, collaboration, conversational tasks, physical context, f-formations.

Background and Problem

• Problem / challenge: Existing AR collaboration platforms rely on preset configurations for avatar placement, often defaulting to face-to-face (F2F) arrangements. They lack flexibility and fail to account for the influence of physical context and task type on user preferences.
• Significance: Understanding how physical objects and task types influence avatar placement is critical for designing AR platforms that support effective and immersive remote collaboration in real-world environments.
• Motivation and related work: Prior studies have explored proxemics, f-formations, and social layouts in physical and virtual settings, but few have investigated how physical objects in AR environments affect avatar placement preferences. The interplay between task type (cooperative vs. competitive) and physical context remains underexplored.

Solution

• Proposed approach: Two mixed-methods within-subjects studies investigating avatar placement preferences in AR environments during cooperative (brainstorming) and competitive (negotiation) tasks, with varying physical contexts (e.g., empty table, box, chairs, tree).
• Novelty:
  1. Identification of the "(Anti-)Affordance Problem," where physical objects are perceived variably as affordances or anti-affordances.
  2. Insights into user preferences for frontal and slightly off-centre avatar placements across task types.
  3. Exploration of the nuanced effects of physical objects like boxes, chairs, and trees on avatar placement.
  4. Recommendations for designing context-aware AR collaboration platforms.
• Procedure and key techniques:
  • Study 1: Compared avatar placement preferences in an empty room versus one with a large box, across brainstorming and negotiation tasks.
  • Study 2: Examined the impact of chairs and a faux tree on avatar placement, using the same task types.
  • Participants manually placed avatars using AR headsets, and preferences were recorded through questionnaires and interviews.
  • Data analyzed using thematic coding and statistical tests.

Results

• Concrete findings:
  • Frontal (F2F) and slightly off-centre (SideF2F) placements were most preferred, with F2F chosen by 47% (brainstorming) and 61% (negotiation) in Study 1.
  • Physical objects influenced placement: 65% of participants in Study 1 and 79% in Study 2 changed their preferred avatar location when an object was present.
  • Chairs were universally seen as affordances, while the tree was consistently viewed as an anti-affordance.
  • The box elicited mixed reactions, seen as both an affordance and an anti-affordance depending on user interpretation.
• Advantage over baselines:
  • Unlike prior studies that focused on preset configurations or limited task types, this research provides a comprehensive understanding of how physical context and task type interact to shape avatar placement preferences.
• Experiments / evaluation:
  • Study 1: 18 pairs (36 participants) evaluated avatar placement in empty and box conditions.
  • Study 2: 12 returning pairs (24 participants) assessed placement in chairs and tree conditions.
  • Tasks: Brainstorming (cooperative) and negotiation (competitive).
  • Metrics: Preferred f-formations, qualitative interview insights, and statistical analysis of placement changes.
• Limitations and future work:
  • Limited to circular table setups; real-world collaboration often involves irregular table shapes.
  • Subjective preference data prioritized over objective position tracking.
  • Visual fidelity of AR devices (e.g., Meta Quest 3) may have influenced results; future studies could use higher-resolution devices.
  • Future research should explore naturalistic settings, content-based tasks, and the impact of different table shapes.

Summary

This paper investigates how physical context and task type influence avatar placement preferences in AR collaboration. Two studies revealed a strong preference for frontal and slightly off-centre placements, with physical objects like chairs acting as affordances and trees as anti-affordances. The "(Anti-)Affordance Problem" highlights the subjective and context-dependent interpretation of physical objects, such as boxes being seen as both affordances and anti-affordances. These findings inform the design of AR platforms that support flexible and context-aware avatar placement, enhancing remote collaboration in diverse environments. Future work should address limitations related to table shapes, device fidelity, and naturalistic settings.