Bistable, irregular firing and population oscillations in a modular attractor memory network

Area of Science:

• Computational neuroscience
• Systems neuroscience

Background:

• Attractor neural networks are models for working memory in the cerebral cortex.
• Existing models often fail to replicate the irregular firing patterns observed in experimental data.

Purpose of the Study:

• To develop a large-scale network model that reproduces bistable activity and irregular neural firing observed in vivo.
• To investigate the role of modular network structure in generating these firing patterns and population oscillations.

Main Methods:

• A large-scale modular neural network model with connected hypercolumns was developed.
• The model simulated neuronal activity under a high-conductance regime with balanced excitation and inhibition.
• Analysis focused on firing regularity, bistable activity, and population oscillations (beta and gamma bands).

Main Results:

• The model successfully generated irregular single-cell firing and bistable activity, consistent with in vivo observations.
• Population oscillations in the beta band (ground state) and gamma band (active state) were observed.
• The modular structure maintained oscillatory and irregular firing in the active state without fine-tuning.
• Irregular firing emerged from a high-conductance regime driven by population oscillations.

Conclusions:

• The proposed modular network model offers a mechanistic explanation for irregular neural firing in cortical populations during working memory.
• The transition from beta to gamma oscillations is linked to the emergence of irregular firing during memory retrieval.
• This model reconciles firing properties with population dynamics in working memory networks.