Linear noise approximation for oscillations in a stochastic inhibitory network with delay

Grégory Dumont, Georg Franz Josef Northoff, André Longtin

研究成果: 雜誌貢獻文章同行評審

7 引文 斯高帕斯(Scopus)

摘要

Understanding neural variability is currently one of the biggest challenges in neuroscience. Using theory and computational modeling, we study the behavior of a globally coupled inhibitory neural network, in which each neuron follows a purely stochastic two-state spiking process. We investigate the role of both this intrinsic randomness and the conduction delay on the emergence of fast (e.g., gamma) oscillations. Toward that end, we expand the recently proposed linear noise approximation (LNA) technique to this non-Markovian "delay" case. The analysis first leads to a nonlinear delay-differential equation (DDE) with multiplicative noise for the mean activity. The LNA then yields two coupled DDEs, one of which is driven by additive Gaussian white noise. These equations on their own provide an excellent approximation to the full network dynamics, which are much longer to integrate. They further allow us to compute a theoretical expression for the power spectrum of the population activity. Our analytical result is in good agreement with the power spectrum obtained via numerical simulations of the full network dynamics, for the large range of parameters where both the intrinsic stochasticity and the conduction delay are necessary for the occurrence of oscillations. The intrinsic noise arises from the probabilistic description of each neuron, yet it is expressed at the system activity level, and it can only be controlled by the system size. In fact, its effect on the fluctuations in system activity disappears in the infinite network size limit, but the characteristics of the oscillatory activity depend on all model parameters including the system size. Using the Hilbert transform, we further show that the intrinsic noise causes sporadic strong fluctuations in the phase of the gamma rhythm. © Published by the American Physical Society.
原文英語
期刊Physical Review E - Statistical, Nonlinear, and Soft Matter Physics
90
發行號1
DOIs
出版狀態已發佈 - 2014
對外發佈Yes

指紋 深入研究「Linear noise approximation for oscillations in a stochastic inhibitory network with delay」主題。共同形成了獨特的指紋。

引用此