Emergent computation in simple model of neural activity

TL;DR

This paper explores how simple neural network models with local interactions can produce complex, automaton-like activity patterns that may serve as computational templates in neural systems.

Contribution

It introduces a neural network model with antisymmetric local interactions that exhibits emergent automaton-like dynamics, suggesting a new mechanism for neural computation.

Findings

Network activity patterns resemble cellular automata dynamics

Emergent states may serve as computational templates

Dynamics are robust across various parameters

Abstract

We investigate the dynamics of a network consisting of an array of identical cortical units with nearest neighbor interactions under periodic arousal. Each unit consists of two interconnected populations of neurons tuned to a state in which many nonlinear resonances are available. The network is critically balanced due to short-ranged antisymmetric connections between units. For wide ranges of the network parameters, the patterns of activity resemble the dynamics of cellular automata. It is argued that these dynamical states may provide a template in which computation can be implemented.