Resting-State fMRI Dynamics and Null Models: Perspectives, Sampling Variability, and Simulations.

Area of Science:

• Neuroimaging
• Computational Neuroscience
• Brain Dynamics

Background:

• Resting state functional MRI (rs-fMRI) studies increasingly focus on brain signal "dynamics".
• Developing hypothesis testing frameworks and null models for dynamical rs-fMRI is an active area of research.
• Existing null models have crucial shortcomings that limit their effectiveness in analyzing rs-fMRI dynamics.

Purpose of the Study:

• To critically evaluate recently proposed null models for rs-fMRI dynamics.
• To identify fundamental limitations in current null models for separating relevant brain signal dynamics from noise.
• To provide guidance on constructing more scientifically meaningful null hypotheses and models for rs-fMRI dynamics.

Main Methods:

• Critical analysis of existing null models for rs-fMRI dynamics.
• Discussion of the implications of distributional stationarity in time series analysis for rs-fMRI.
• Evaluation of statistical measures (e.g., kurtosis) for capturing functionally relevant brain dynamics.

Main Results:

• Current null models are often poorly formulated relative to the hypotheses they test.
• Distributionally stationary time series may not be ideal for building rs-fMRI dynamic null models.
• Measures like kurtosis may be unsuitable for capturing key features of functionally relevant brain dynamics.

Conclusions:

• Careful construction of null models is essential for analyzing rs-fMRI dynamics.
• Future research should focus on developing null models that better capture epochal signal variations indicative of brain responsiveness.
• Scientifically meaningful null hypotheses and well-tuned null models will emerge with further understanding of brain dynamics.