Frequency filter interactions in network of non-oscillatory cells

ANDREA L. BEL; HORACIO G. ROTSTEIN

Online

Congreso; 30th Annual Computational Neuroscience Meeting; 2021

Resonance refers to the ability of dynamical systems toexhibit a peak in their amplitude response to oscillatoryinputs at a preferred (resonant) frequency. In neuronal cir-cuits, resonance is typically measured by using the imped-ance amplitude profile Z defined as the absolute value ofthe quotient of the Fourier transforms of the output and theinput. Resonance has been investigated in single neuronsby many authors both experimentally and theoretically(Cichon & Gan, 2015 Apr; Sajikumar et al., 2014 Aug 19).Network resonance has received much less attention. Twoimportant questions are (i) whether and under what condi-tions a network of neurons exhibits resonance in one or moreneurons in response to inputs to one or more neurons, and(ii) whether and under what conditions the information iscommunicated between neurons in a frequency-dependentmanner.In this project we address these issues by using a minimalnetwork consisting of two passive cells (linear, non-reso-nant neurons) recurrently connected via graded synapticinhibition or excitation and receiving oscillatory inputs ineither one or the two nodes (Sezener et al., 2021). In orderto investigate how network resonance emerges we extendthe concept of impedance to nonlinear systems by comput-ing the peak-to-trough amplitudes normalized by the inputamplitude. In order to investigate the communication offrequency-dependent information across neurons in thenetwork we borrow the concept of the coupling coefficientfrom the gap junction literature. The coupling coefficient K,defined as the quotient between the postsynaptic and pre-synaptic membrane potentials of two electrically coupledneurons, is used to measure the strength of the connectionin the presence of constant (DC) inputs. Here we extend thismetrics to synaptically connected neurons and to the fre-quency domain. Linear networks (linear neurons and linearconnectivity) can only show a low-pass filter K profile (K asa function of the input frequency). We show that the pres-ence of the more realistic nonlinear synaptic connectivitycan produce band-pass K profiles. We note that the conceptof communication of information we use here is differentthan the standard one used in information theory.