Evolutionary constraints on visual cortex architecture from the dynamics of hallucinations

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual System Modeling

Background:

• Normal vision in the primary visual cortex (V1) relies on external stimuli driving neural excitation.
• V1 can fail when spontaneous, self-organized excitation patterns override normal visual input, leading to hallucinations.
• Statistical mechanics suggests the hallucinatory state is highly robust, posing a challenge to understanding V1's stability.

Purpose of the Study:

• To investigate the role of long-range connections between inhibitory neurons in V1 models.
• To understand how V1 maintains normal visual processing and avoids spontaneous excitation patterns.
• To explore the functional significance of sparse long-range inhibition in V1.

Main Methods:

• Development of a computational model of V1 incorporating physiologically realistic long-range connections.
• Simulation of neural excitation patterns under different connectivity conditions.
• Analysis of model behavior to identify mechanisms preserving normal vision and generating hallucinations.

Main Results:

• Incorporating sparse long-range inhibition into the V1 model was found to be crucial for maintaining the normal vision state.
• The sparsity of long-range inhibition also contributes to the regularity of geometric visual hallucinations.
• This architectural feature appears to be an evolutionary adaptation optimizing V1 for stable visual processing.

Conclusions:

• Sparse long-range inhibition is a key factor in V1's ability to sustain normal vision and resist spontaneous pattern formation.
• The same inhibitory connections are vital for the development of orientation preference maps in V1.
• V1's distinct long-range connectivity (patchy excitatory, sparse inhibitory) is shaped by the dual needs of visual state stability and developmental map formation.