Cortical state contributions to neuronal response variability in the early visual cortex: A system identification...

Area of Science:

• Neuroscience
• Computational Neuroscience
• Visual Processing

Background:

• Neurons in the early visual cortex exhibit inconsistent responses to repeated stimuli, often attributed to random noise.
• This trial-to-trial response variability may stem from non-sensory factors, particularly fluctuating cortical states.

Purpose of the Study:

• To investigate the role of cortical state in neuronal response variability in the early visual cortex.
• To develop and validate a computational model that distinguishes stimulus-driven activity from cortical state-driven response variability.

Main Methods:

• Recorded neuronal spiking activity, local field potentials (LFPs), and multi-unit activity (MUA) in cat visual areas 17 and 18.
• Quantified response variability using a variability ratio (VR) and cortical state using global fluctuation index (GFI) and synchrony index (SI).
• Developed a compact convolutional neural network with parallel pathways to model stimulus-driven and cortical state-driven responses, fitting parameters to predict neuronal activity.

Main Results:

• Single neurons displayed significant trial-to-trial response variability, with varying degrees of correlation to cortical state indicators.
• The proposed model, incorporating cortical state information, significantly improved prediction accuracy of neuronal responses and receptive field estimation.
• Neurons with higher response variability showed greater benefit from the cortical state-driven pathway.

Conclusions:

• Cortical state fluctuations significantly contribute to neuronal response variability in the early visual cortex.
• Accounting for cortical state dynamics enhances the accuracy of system identification models for neural responses.
• Individual neurons differ in their response variability and its dependence on cortical state indicators.