Modeling the diverse effects of divisive normalization on noise correlations

Area of Science:

• Computational Neuroscience
• Systems Neuroscience
• Neural Coding

Background:

• Divisive normalization is a key model for neural activity across brain regions.
• Its impact on population-level neural response statistics, like noise correlations, is under-explored.
• Existing models of neural covariability often neglect normalization's influence.

Purpose of the Study:

• To develop a pairwise stochastic divisive normalization model to explain neural response covariability.
• To investigate how normalization affects noise correlations, considering shared vs. unshared normalization signals.
• To apply the model to empirical data and assess its performance against alternatives.

Main Methods:

• Development of a pairwise stochastic divisive normalization model.
• Theoretical analysis of how normalization modulates noise correlations.
• Application and validation of the model using calcium imaging data from mouse primary visual cortex (V1).

Main Results:

• The proposed model accurately fits V1 calcium imaging data, often outperforming alternative correlation models.
• Normalization's effect on noise correlations depends on whether normalization signals are shared between neurons.
• Analysis suggests that normalization signals are frequently shared among V1 neurons in the studied dataset.

Conclusions:

• The developed model provides a framework for quantifying the relationship between normalization and neural covariability.
• This work offers new insights into the circuit mechanisms underlying normalization.
• The findings highlight the role of normalization in neural information processing and representation.