Resolve difficulties applying observational causality methods to cross-frequency interactions

Resolve the methodological difficulties that currently prevent the application of observational causality approaches to interactions between frequency components in neurophysiological data, accounting for nonlinearity, nonstationarity, and time–frequency resolution constraints.

Background

The paper argues that directionality and causality assessments would substantially improve the analysis of cross-frequency relationships, but current observational causality tools face major obstacles in this context. Linear Granger causality is blind to cross-frequency effects, nonparametric measures like transfer entropy require large datasets, and time–frequency uncertainty and filtering introduce temporal biases.

Addressing these issues is necessary to enable reliable causal inference between spectral components and to distinguish direct interactions from common-drive confounds in CFC analyses.