Published March 4, 2025 | Version v1
Journal article Open

The Role of Sensorimotor Contingencies and Eye Scanpath Entropy in Presence in Virtual Reality: A Reinforcement Learning Paradigm

  • 1. Institute of Neurosciences, University of Barcelona
  • 2. University of Barcelona
  • 3. ARTANIM Foundation

Description

Sensorimotor contingencies (SC) refer to the rules by which we use our body to perceive. It has been argued that to the extent that a virtual reality (VR) application affords natural SC so the greater likelihood that participants will experience Place Illusion (PI), the illusion of ‘being there’ (a component of presence) in the virtual environment. However, notwithstanding numerous studies this only has anecdotal support. Here we used a reinforcement learning (RL) paradigm where 26 participants experienced a VR scenario where the RL agent could sequentially propose changes to 5 binary factors: mono or stereo vision, 3 or 6 degrees of freedom head tracking, mono or spatialised sound, low or high display resolution, or one of two color schemes. The first 4 are SC, whereas the last is not. Participants could reject or accept each change proposed by the RL, until convergence. Participants were more likely to accept changes from low to high SC than changes to the color. Additionally, theory suggests that increased PI should be associated with lower eye scanpath entropy. Our results show that mean entropy did decrease over time and the final level of entropy was negatively correlated with a post exposure questionnaire-based assessment of PI.

Files

paper10-ieee.pdf

Files (766.6 kB)

Name Size Download all
md5:4144cd8bec849514114f3fced30b086c
766.6 kB Preview Download

Additional details

Funding

European Commission
GuestXR - GuestXR: A Machine Learning Agent for Social Harmony in eXtended Reality 101017884