Information Entropy Production of Spatio-Temporal Maximum Entropy Distributions

TL;DR

This paper introduces a framework to quantify the irreversibility of maximum entropy distributions in neural spike train data using transfer matrix techniques, linking statistical properties to causal interactions.

Contribution

It develops a novel method to measure irreversibility in maximum entropy models of neural activity, connecting entropy production with causal and memory effects.

Findings

Framework successfully quantifies irreversibility in spike train models

Transfer matrix approach links entropy production to neural causal interactions

Examples demonstrate applicability to neural spike train data

Abstract

Spiking activity from populations of neurons display causal interactions and memory effects. Therefore, they are expected to show some degree of irreversibility in time. Motivated by the spike train statistics, in this paper we build a framework to quantify the degree of irreversibility of any maximum entropy distribution. Our approach is based on the transfer matrix technique, which enables us to find an homogeneous irreducible Markov chain that shares the same maximum entropy measure. We provide relevant examples in the context of spike train statistics