Visual illusions via neural dynamics: Wilson-Cowan-type models and the efficient representation principle

Area of Science:

• Computational neuroscience
• Visual perception modeling

Background:

• Wilson-Cowan (WC) models are foundational for neuronal dynamics.
• Replicating visual illusions like brightness and orientation-dependent phenomena tests model fidelity.
• The efficient representation principle guides optimal neural activity.

Purpose of the Study:

• To investigate Wilson-Cowan models' ability to reproduce visual illusions.
• To assess WC model compliance with the efficient representation principle.
• To develop an improved model for enhanced visual illusion reproduction.

Main Methods:

• Reproducing visual illusions using Wilson-Cowan (WC) type models.
• Formally proving the absence of a minimized energy functional for WC dynamics.
• Employing variational modeling and local histogram equalization (LHE).
• Extending the model for V1 architecture with orientation dependence.

Main Results:

• WC equations can reproduce brightness and orientation-dependent visual illusions.
• WC dynamics do not minimize an energy functional, indicating suboptimality.
• A modified variational model (LHE) shows superior illusion reproduction compared to WC.
• A cortical extension of LHE effectively reproduces complex orientation-dependent illusions.

Conclusions:

• Wilson-Cowan models demonstrate capacity for reproducing certain visual illusions.
• The efficient representation principle is crucial for accurate neural modeling.
• Variational modeling, particularly LHE and its cortical extension, offers a more effective approach to modeling visual perception and illusions.