Circuit Models of Low-Dimensional Shared Variability in Cortical Networks.

Area of Science:

• Computational neuroscience
• Systems neuroscience
• Neural circuit dynamics

Background:

• Cortical circuits exhibit complex wiring but low-dimensional population-wide spiking variability.
• Existing models fail to explain this intrinsic variability, often attributing it to external factors.
• Understanding the origins of shared neuronal variability is crucial for comprehending brain function.

Purpose of the Study:

• To investigate if realistic biophysical parameters of inhibitory coupling can internally generate low-dimensional shared variability in spiking neural networks.
• To explore a parsimonious mechanism for attentional modulation of neuronal variability within cortical circuits.

Main Methods:

• Development of a spiking neural network model with parameters reflecting known physiological scales of inhibitory coupling.
• Simulation of network activity to analyze population-wide shared spiking variability.
• Introduction of top-down modulation to mimic attentional effects on inhibitory neurons.

Main Results:

• The model successfully generated low-dimensional shared variability that matched in vivo recordings of population activity.
• The model demonstrated that realistic inhibitory coupling is sufficient to produce intrinsic, low-dimensional variability.
• Simulated top-down modulation of inhibitory neurons replicated the differential modulation of shared variability observed during spatial attention.

Conclusions:

• Realistic biophysical properties of inhibitory coupling in neural networks can intrinsically generate low-dimensional shared variability.
• Top-down modulation of inhibitory neurons offers a plausible mechanism for attentional effects on neuronal variability.
• This work bridges the gap between cortical circuit structure and observed neuronal variability dynamics.