Neuronal couplings between retinal ganglion cells inferred by efficient inverse statistical physics methods

Area of Science:

• Computational neuroscience
• Statistical physics
• Retinal neurophysiology

Background:

• Neural system complexity hinders direct measurement of constituent interactions.
• Inverse statistical physics methods for inferring neural couplings are computationally intensive.

Purpose of the Study:

• To present computationally efficient inverse algorithms for inferring neural couplings.
• To analyze neural recordings from the salamander retina under different stimuli.

Main Methods:

• Developed two complementary inverse algorithms based on Ising and leaky integrate-and-fire models.
• Applied algorithms to reanalyze multielectrode recordings from salamander retina in darkness and under visual stimulus.

Main Results:

• Identified strong positive couplings between nearby retinal ganglion cells, consistent across stimuli.
• Observed emergence of long-range couplings specifically under random visual stimulus.
• Discussed uncertainties in coupling inference due to recording limitations.

Conclusions:

• The developed algorithms offer computationally efficient solutions for inferring neural couplings.
• Findings highlight stimulus-dependent changes in retinal network connectivity.
• Methods facilitate real-time evaluation of couplings in large neuronal assemblies.