Geometry and dynamics of representations in a precisely balanced memory network related to olfactory cortex.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Biological memory relies on synaptic changes in neuronal assemblies.
• Emerging models propose excitatory and inhibitory neuron (E/I) assemblies for balanced neural activity.
• Understanding E/I assembly computation is crucial for memory network function.

Purpose of the Study:

• To investigate the computational effects of E/I assemblies in a biologically realistic spiking network model.
• To compare E/I assembly dynamics with traditional excitatory-only assemblies and global inhibition.
• To explore how E/I assemblies influence information representation and processing in neural networks.

Main Methods:

• Developed a spiking network model using experimental data from zebrafish telencephalic area Dp.
• Simulated network dynamics under conditions with E/I assemblies versus excitatory assemblies and global inhibition.
• Analyzed firing rate distributions, network dynamics, and information encoding within neuronal subspaces.

Main Results:

• E/I assemblies stabilized firing rate distributions, unlike excitatory assemblies with global inhibition.
• Networks with E/I assemblies exhibited continuous attractor dynamics, not discrete attractors.
• Learned inputs were mapped onto manifolds that focused activity, supporting pattern classification via covariance structure.

Conclusions:

• E/I assemblies transform neuronal coding space, creating continuous representations reflecting input relatedness and experience.
• These continuous representations facilitate rapid pattern classification and continual learning.
• E/I assemblies offer a potential neural basis for higher-order learning and complex cognitive computations.