The variable discharge of cortical neurons: implications for connectivity, computation, and information coding

Area of Science:

• Computational Neuroscience
• Neural Coding
• Synaptic Integration

Background:

• Cortical neurons display significant variability in spike discharge patterns.
• This spike variability is conserved across cortical processing stages.
• Understanding the principles of this statistical homogeneity is crucial.

Purpose of the Study:

• To investigate the principles underlying conserved spike variability in cortical neurons.
• To analyze a simplified model of synaptic integration in a high-input regime.
• To determine how neurons represent information and maintain signal integrity.

Main Methods:

• Analysis of a simplified integrate-and-fire neural model with decay.
• Simulation in a high-input regime with balanced excitation and inhibition.
• Mathematical derivation for neural spike count variance in networks.

Main Results:

• Balanced excitation and inhibition in the model produce highly variable interspike intervals, matching experimental data.
• Temporal patterns of synaptic input are not recoverable from output spikes, suggesting rate coding.
• Ensembles of 50-100 neurons provide reliable rate estimates within 10-50 ms.
• Derived conditions for stable signal and noise propagation in neural networks.

Conclusions:

• Cortical neurons likely use rate coding in ensembles, not precise spike timing, for information representation.
• Single neurons perform averaging-like computations; complex processing emerges from network convergence.
• The model explains how neural networks maintain reliable signal processing and noise stability.