Modular inhibitory coding in binary networks

Area of Science:

• Computational neuroscience
• Neural network modeling
• Memory systems

Background:

• Biological neural networks feature distinct excitatory and inhibitory neurons.
• Understanding the specific roles of these populations in memory is essential.

Purpose of the Study:

• To investigate the roles of separated excitatory and inhibitory neuronal populations in memory storage.
• To analyze the network's capacity and performance based on its structure.

Main Methods:

• Developed a binary neural network model with segregated excitatory and inhibitory layers.
• Simulated memory storage via training inhibitory modules that interact with the excitatory layer.
• Analyzed network behavior under varying connection densities and memory loads.

Main Results:

• The inhibitory layer is critical for memory storage and management, with capacity scaling with inhibitory neuron count.
• Network performance degrades gradually with loss of excitatory-to-excitatory (E-E) connections but critically depends on inhibitory-to-excitatory (I-E) connections.
• The proposed coding scheme offers advantages in memory capacity and expansion.

Conclusions:

• Inhibitory neurons play a vital role in memory storage and recall, aligning with experimental observations in brain networks.
• The model provides insights into the sparse connectivity patterns observed in biological neural systems.