Optimal Region of Latching Activity in an Adaptive Potts Model for Networks of Neurons

Abstract

In statistical mechanics, the Potts model is a model for interacting spins with more than two discrete states. Neural networks which exhibit features of learning and associative memory can also be modeled by a system of Potts spins. A spontaneous behavior of hopping from one discrete attractor state to another (referred to as latching) has been proposed to be associated with higher cognitive functions. Here we propose a model in which both the stochastic dynamics of Potts models and an adaptive potential function are present. A latching dynamics is observed in a limited region of the noise(temperature)-adaptation parameter space. We hence suggest noise as a fundamental factor in such alternations alongside adaptation. From a dynamical systems point of view, the noise-adaptation alternations may be the underlying mechanism for multi-stability in attractor based models. An optimality criterion for realistic models is finally inferred.

References in corpus (4)

• Weak pairwise correlations imply strongly correlated network states in a neural population
• The Ising Model for Neural Data: Model Quality and Approximate Methods for Extracting Functional Connectivity
• Ising models for networks of real neurons
• The storage capacity of Potts models for semantic memory retrieval

Cited by in corpus (1)

• Density-based clustering: A 'landscape view' of multi-channel neural data for inference and dynamic complexity analysis