Spatiotemporal dynamics and reliable computations in recurrent spiking neural networks

TL;DR

This paper demonstrates that incorporating distance-dependent connectivity in recurrent spiking neural networks enables reliable spatiotemporal pattern generation and computation, overcoming traditional unreliability issues.

Contribution

It introduces spatially extended networks with distance-dependent connections that produce trainable spatiotemporal patterns for dynamical computations.

Findings

Distance-dependent connectivity induces symmetry-breaking bifurcations.

Networks generate trainable spatiotemporal patterns.

Improved computational reliability in spiking neural networks.

Abstract

Randomly connected networks of excitatory and inhibitory spiking neurons provide a parsimonious model of neural variability, but are notoriously unreliable for performing computations. We show that this difficulty is overcome by incorporating the well-documented dependence of connection probability on distance. Spatially extended spiking networks exhibit symmetry-breaking bifurcations and generate spatiotemporal patterns that can be trained to perform dynamical computations.