Brain Synchronization and Multivariate Autoregressive (MVAR) Modeling in Cognitive Neurodynamics.

Area of Science:

• Cognitive Neuroscience
• Computational Neuroscience
• Systems Neuroscience

Background:

• Long-range synchronization between brain networks, like the frontoparietal network (FPN) and forebrain subcortical systems, is crucial for cognitive functions.
• This synchronization manifests as measurable activity in electroencephalography (EEG) and other neural signals.
• Understanding these large-scale neural dynamics requires advanced analytical methods.

Purpose of the Study:

• To review the application and advantages of Multivariate Autoregressive (MVAR) modeling in cognitive neurodynamics research.
• To explore how MVAR methods analyze long-range synchronization and causality in neural networks.
• To highlight recent advancements in MVAR modeling for analyzing continuous neural signals like EEG and fMRI.

Main Methods:

• Review of Multivariate Autoregressive (MVAR) modeling techniques.
• Analysis of synchronization in neurocognitive networks using EEG, local field potential (LFP), and fMRI time series.
• Exploration of Granger causality, Directed Transfer Function (DTF), and Partial Directed Coherence (PDC) within the MVAR framework.

Main Results:

• MVAR modeling is highly effective for analyzing long-range synchronization in brain regions.
• MVAR methods, including DTF and PDC, accurately identify causality and directed propagation in neural activity.
• Non-linear multivariate analysis models offer superior accuracy and speed compared to univariate methods for neuronal communication.

Conclusions:

• MVAR modeling provides a powerful framework for understanding complex brain dynamics and functional connectivity.
• The reviewed methods offer significant advantages for analyzing synchronized neural activity and inferring causal relationships.
• Advancements in MVAR modeling continue to enhance our ability to study cognitive processes through neural signal analysis.