Optimizing information capacity in modular neural networks through excitatory and inhibitory connectivity

Modularity is a key principle in complex networks like the brain, enhancing scalability, flexibility, and robustness. This study investigates how modularity affects information capacity in neural networks, focusing on the balance between excitatory and inhibitory connectivity. Using a computational...

Full description

Saved in:
Bibliographic Details
Published in:Neurocomputing (Amsterdam) Vol. 640; p. 130313
Main Authors: Khanjanianpak, Mozhgan, Valizadeh, Alireza
Format: Journal Article
Language:English
Published: Elsevier B.V 01.08.2025
Subjects:
Balanced networks
Biological neural networks
Entropy coding
Information processing
Modular networks
Balanced networks
Entropy coding
Modular networks
Biological neural networks
Information processing
ISSN:0925-2312
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:Modularity is a key principle in complex networks like the brain, enhancing scalability, flexibility, and robustness. This study investigates how modularity affects information capacity in neural networks, focusing on the balance between excitatory and inhibitory connectivity. Using a computational model, we analyzed modular architectures with varying connection probabilities within and between modules, while preserving global excitation–inhibition balance. We assessed information capacity across different dynamical states induced by external inputs. Results show that global long-range excitation leads to periodic states with low information content. In contrast, mixed excitatory and inhibitory inter-module connections produce correlated activity, constraining global information scaling. However, purely inhibitory inter-module connections promote uncorrelated, intermittent activity, maximizing information capacity locally and network-wide. We also explored a Hopfield-like network with global inhibition, where excitatory intra-module connections foster partial correlations within assemblies while preserving inter-assembly independence. These findings underscore the critical role of distinct excitatory and inhibitory connectivity patterns and modular organization in optimizing neural information processing. This study offers insights into the mechanisms driving neural dynamics and their contribution to efficient information processing in the brain, highlighting the interplay between structure and function in complex networks. [Display omitted] •Modular networks’ information capacity varies with excitation/inhibition patterns.•Global inhibition maximizes capacity by reducing activity regularity/correlation.•Excitatory assemblies in global inhibition act like Hopfield functional units.•Intra-assembly excitatory links boost modules’ correlation within, reduce between.
ISSN:0925-2312
DOI:10.1016/j.neucom.2025.130313