A psychophysically-tuned computational model of human primary visual cortex produces geometric optical illusions

Area of Science:

• Neuroscience
• Computational Vision
• Psychophysics

Background:

• Geometric optical illusions (GOIs) highlight the gap between physical reality and visual perception.
• Understanding the distinct or shared perceptual mechanisms underlying different GOIs is limited.
• The influence of specific physical properties on these mechanisms requires quantitative investigation.

Purpose of the Study:

• To quantitatively investigate the interplay between physical properties and perceptual mechanisms in GOIs.
• To determine if different GOIs rely on similar or distinct perceptual mechanisms.
• To model the neural underpinnings of GOIs using computational approaches.

Main Methods:

• Conducted psychophysical experiments with 30 healthy adults, measuring perceptual biases across parametrically varied GOI properties.
• Developed and tuned a computational model of the primary visual cortex.
• Integrated parameters including orientation selectivity, receptive field characteristics, and intra-cortical connectivity.

Main Results:

• Individual physical properties had varied effects on perceptual biases depending on the specific GOI.
• Interactions between different physical properties were observed within a given GOI.
• Computational model simulations mirrored human psychophysical results when tuning receptive field properties and long-range interactions.

Conclusions:

• Physical properties synergistically contribute to GOI generation, rather than acting independently.
• Computational modeling, guided by psychophysics, can elucidate the mechanistic basis of visual perception.
• This research provides a framework for understanding the neural basis of geometric optical illusions.