On the interpretability and computational reliability of frequency-domain Granger causality

TL;DR

This paper discusses the limitations of traditional frequency-domain Granger causality in neuroscience, advocates for an updated state space approach, and demonstrates its advantages through replication of previous simulations.

Contribution

It highlights the issues with outdated formulations of Granger causality and demonstrates the effectiveness of a modern state space implementation to improve reliability and interpretability.

Findings

Outdated formulations of Granger causality have significant pitfalls.

Updated state space methods mitigate or solve these pitfalls.

Replication shows improved reliability with the new approach.

Abstract

This is a comment to the paper 'A study of problems encountered in Granger causality analysis from a neuroscience perspective'. We agree that interpretation issues of Granger Causality in Neuroscience exist (partially due to the historical unfortunate use of the name 'causality', as nicely described in previous literature). On the other hand we think that the paper uses a formulation of Granger causality which is outdated (albeit still used), and in doing so it dismisses the measure based on a suboptimal use of it. Furthermore, since data from simulated systems are used, the pitfalls that are found with the used formulation are intended to be general, and not limited to neuroscience. It would be a pity if this paper, even written in good faith, became a wildcard against all possible applications of Granger Causality, regardless of the hard work of colleagues aiming to seriously address the methodological and interpretation pitfalls. In order to provide a balanced view, we replicated their simulations used the updated State Space implementation, proposed already some years ago, in which the pitfalls are mitigated or directly solved.