Quantifying Spacetime Integration across a Partition with Synergy

TL;DR

This paper introduces four synergy-based measures of integration rooted in partial information decomposition, demonstrating their suitability for IIT and potential usefulness for complexity measurement in dynamical systems.

Contribution

The paper develops and compares new synergy-based integration measures based on partial information decomposition for IIT and dynamical systems.

Findings

Synergy-based measures outperform current IIT practices in simple networks.

The new measures are more suitable for the use-case of IIT.

Potential applicability of these measures to complexity in discrete dynamical systems.

Abstract

In service to the mathematical underpinnings of the Information Integration Theory of Consciousness (IIT), we introduce four measures of integration based on the partial information decomposition framework. We compare our measures to current IIT practice in simple deterministic networks. We find synergy-based measures more suitable for IIT's use-case than current practice. Outside IIT, these measures could also be useful as measures of complexity for discrete dynamical systems.