Neural System Identification With Spike-Triggered Non-Negative Matrix Factorization.

Shanshan Jia,Yonghong Tian,Tiejun Huang,Zhaofei Yu,Arno Onken,Jian K Liu

doi:10.1109/tcyb.2020.3042513

Abstract

Neuronal circuits formed in the brain are complex with intricate connection patterns. Such complexity is also observed in the retina with a relatively simple neuronal circuit. A retinal ganglion cell (GC) receives excitatory inputs from neurons in previous layers as driving forces to fire spikes. Analytical methods are required to decipher these components in a systematic manner. Recently a method called spike-triggered non-negative matrix factorization (STNMF) has been proposed for this purpose. In this study, we extend the scope of the STNMF method. By using retinal GCs as a model system, we show that STNMF can detect various computational properties of upstream bipolar cells (BCs), including spatial receptive field, temporal filter, and transfer nonlinearity. In addition, we recover synaptic connection strengths from the weight matrix of STNMF. Furthermore, we show that STNMF can separate spikes of a GC into a few subsets of spikes, where each subset is contributed by one presynaptic BC. Taken together, these results corroborate that STNMF is a useful method for deciphering the structure of neuronal circuits.

Highlights

N EURONAL circuits in the brain are highly complex
We set up a minimal model of the retinal Ganglion cells (GCs) as in Fig. 1(C) in order to investigate how upstream bipolar cells (BCs) affect spiking activity of the target GC
These two spike trains from ON and OFF BCs are not correlated. They are decorrelated due to the uncorrelated stimuli. These results show that correlations between subunit spikes are not driven by stimuli, but by the same BC identified by spike-triggered non-negative matrix factorization (STNMF)

Summary

Introduction

Even for the retina, a relatively simple neuronal circuit, the underlying structure, in particular, functional characteristics, are still not completely understood. The retina serves as a typical model for both deciphering the structure of neuronal circuits [1], [2], [3], [4], [5], [6] and testing novel methods for neuronal coding [7], [8], [9], [10], [11]. Ganglion cells (GCs), as the only output neurons of the retina, send visual information via the optic tracts and thalamus to cortical areas for higher cognition. Each ganglion cell receives inputs from a number of excitatory bipolar cells (BCs) as driving force to generate spikes (Fig. 1(B))

Methods

Results

Conclusion