Systematic misperceptions of 3-D motion explained by Bayesian inference.

Area of Science:

• Visual perception
• Computational neuroscience
• 3-D motion perception

Background:

• Human perception exhibits systematic, reliable errors.
• Understanding three-dimensional (3-D) motion perception is complex.
• Previous models have not unified explanations for diverse 3-D motion misperceptions.

Purpose of the Study:

• To provide a unified explanation for systematic errors in 3-D motion perception.
• To develop a computational model for 3-D motion inference.
• To test model predictions experimentally.

Main Methods:

• Characterized binocular retinal motion signals for 3-D movement.
• Developed a Bayesian model of 3-D motion perception as optimal inference.
• Conducted perceptual experiments using virtual reality and stereoscopic displays.

Main Results:

• Model predictions for perceptual errors were confirmed experimentally.
• Errors in 3-D motion perception were found to depend on stimulus contrast, distance, and eccentricity.
• Observed systematic errors include lateral motion bias and confusion between approaching/receding motion.

Conclusions:

• A Bayesian framework offers a parsimonious explanation for diverse 3-D motion misperceptions.
• Sensory noise and stimulus properties systematically influence 3-D motion perception.
• The findings advance our understanding of visual inference in naturalistic environments.