Dynamic functional brain connectivity results depend on modeling assumptions: comparing frequentist and Bayesian hypothesis tests

TL;DR

This paper compares frequentist and Bayesian methods for detecting dynamic functional brain connectivity, showing how modeling assumptions influence results and emphasizing the importance of careful model choice in neuroscience studies.

Contribution

It introduces a Bayesian hypothesis testing framework for dynamic connectivity that incorporates prior knowledge and compares it to traditional frequentist methods.

Findings

Modeling assumptions significantly affect dynamic connectivity detection.

Bayesian approach provides evidence for both static and dynamic connectivity.

Group-level results are more robust to modeling choices than individual-level results.

Abstract

Understanding the temporal dynamics of functional brain connectivity is important for addressing various questions in network neuroscience, such as how connectivity affects cognition and changes with disease. A fundamental challenge is to evaluate whether connectivity truly exhibits dynamics, or simply is static. The most common frequentist approach uses sliding-window methods to model functional connectivity over time, but this requires defining appropriate sampling distributions and hyperparameters, such as window length, which imposes specific assumptions on the dynamics. Here, we explore how these assumptions influence the detection of dynamic connectivity, and introduce an alternative approach based on Bayesian hypothesis testing with Wishart processes. This framework encodes assumptions through prior distributions, allowing prior knowledge on the time-dependent structure of connectivity to be incorporated into the model. Moreover, this framework provides evidence for both dynamic and static connectivity, offering additional information. Using simulations, we compare the frequentist and Bayesian approaches and demonstrate how different assumptions affect the detection of dynamic connectivity. Finally, by applying both approaches to an fMRI working-memory task, we find that conclusions at the individual level vary with modeling choices, while group-level results are more robust. Our work highlights the importance of carefully considering modeling assumptions when evaluating dynamic connectivity.