Abstract:The propagation of slowly-varying firing rates has been proved significant for the development of the central nervous system. Recent reports have shown that the membrane passive properties of dendrites play a key role in the computation of the single neuron, which is of great importance to the function of neural networks. However, it is still unclear how dendritic passive properties affect the ability of cortical networks to propagate slowly-varying spiking activity. Here, we use two-compartment biophysical models to construct multilayered feedforward neural networks (FFNs) to investigate how dendritic passive properties affect the propagation of the slow-varying inputs. In the two-compartment biophysical models, one compartment represents apical dendrites, and the other compartment describes the soma plus the axon initial segment. Area proportion occupied by somatic compartment and coupling conductance between dendritic and somatic compartments are abstracted to capture the dendritic passive properties. A time-varying signal is injected into the first layer of the FFNs and the fidelity of the signal during propagation is used to qualify the ability of the FFN to transmit wave-like signals. Numerical results reveal an optimal value of coupling conductance between dendritic and somatic compartments to maximize the fidelity of the initial spiking activity. An increase of the dendritic area enhances the initial firing rate of neurons in the first layer by increasing the response of neurons to slow-varying wave-like input, resulting in a delay of attenuation of the firing rate, thus promoting the transmission of signals in FFN. Using a mean-field approach, we examine that changes in area proportion occupied by somatic compartment and coupling conductance between dendritic and somatic compartment affect the signal propagation ability of the FFN by adjusting the input–output transform of a single neuron. With the participation of external noise, a wide range of initial firing rates maintains a unique representation during propagation, which ensures the reliable transmission of slow-varying signals in FFNs. These findings are helpful to understand how passive properties of dendrites participate in the propagation of slowly varying signals in the cerebellum.

Artificial Neuron with Somatic and Axonal Computation Units: Mathematical and Neuromorphic Models of Persistent Firing Neurons

Modeling Neuromorphic Persistent Firing Networks

Axonal slow integration induced persistent firing neuron model

Axonal Computations

Single-Neuron Criticality Optimizes Analog Dendritic Computation

Artificial Neuronal Devices Based on Emerging Materials: Neuronal Dynamics and Applications

Dendritic computations captured by an effective point neuron model.

Active Dendrites Enable Robust Spiking Computations despite Timing Jitter

Nonlinear Multiplicative Dendritic Integration in Neuron and Network Models

Highly Bionic Neurotransmitter-Communicated Neurons Following Integrate-and-Fire Dynamics

The Connection of Neuromorphic Architecture and Bio-Plausible Neurons and Synapses

Extending Integrate-and-Fire Model Neurons to Account for the Effects of Weak Electric Fields and Input Filtering Mediated by the Dendrite

Modeling somatic computation with non-neural bioelectric networks

The Passive Properties of Dendrites Modulate the Propagation of Slowly-Varying Firing Rate in Feedforward Networks.

Compact hybrid type electronic neuron and computational model of its dynamics

Silicon Modeling of Spiking Neurons with Diverse Dynamic Behaviors

A soma‐synapses neuron model and FPGA implementation

Design of Artificial Neurons of Memristive Neuromorphic Networks Based on Biological Neural Dynamics and Structures

Mathematical Modeling and Analysis of Spatial Neuron Dynamics: Dendritic Integration and Beyond

Dendritic Integration Regulation and Neuronal Arithmetic Implemented in a Proton-Coupled Neuron Transistor

Computational Modeling of Neuronal Networks