Internally generated population activity in cortical networks hinders information transmission

Area of Science:

• Systems Neuroscience
• Computational Neuroscience
• Neural Coding

Background:

• Neuronal variability's role in sensory coding is complex and model-dependent.
• Existing models often use parametric structures, limiting analysis of synaptic circuitry's role.
• Understanding how neural networks generate variability is crucial for deciphering sensory information processing.

Purpose of the Study:

• Investigate how synaptic circuitry in spiking neuron networks shapes stimulus coding.
• Analyze the impact of spatially ordered connectivity on neuronal variability and information transmission.
• Determine the specific circuit mechanisms that lead to information-limiting noise correlations.

Main Methods:

• Simulated networks of spiking neuron models with spatially organized excitatory and inhibitory connectivity.
• Analysis of population-wide shared neuronal variability.
• Utilized a spatial neural field model to link circuit properties to noise correlations and information content.

Main Results:

• Spatially ordered connectivity generates realistic population-wide neuronal variability.
• Both feedforward and recurrent projections influence noise correlations.
• Inhibitory recurrent projections were identified as a key factor in creating correlations that reduce stimulus information.

Conclusions:

• Recurrent inhibitory connectivity plays a critical role in shaping noise correlations that can limit sensory information.
• The spatial structure of neural circuits, particularly recurrent inhibition, dictates the impact of variability on neural codes.
• This work provides a link between circuit architecture, population activity patterns, and the fidelity of sensory coding.