Behavioral Studies Using Large-Scale Brain Networks - Methods and Validations.

Area of Science:

• Cognitive Neuroscience
• Neuroimaging
• Computational Neuroscience

Background:

• Mapping human behaviors to brain activity is a central goal in cognitive neuroscience.
• Functional MRI (fMRI) advances enable investigation of neural activity via functional connectivity and brain networks.
• Noise in neural signals complicates the association between behaviors and specific neural signals.

Purpose of the Study:

• To review and facilitate understanding of data-driven approaches for analyzing brain networks and human behaviors.
• To explore methods beyond traditional graph theory for quantifying neural connections related to behavior.
• To provide examples of applying these methods to real-world neuroimaging data.

Main Methods:

• Graph theory for interpreting brain network topology.
• Connectome-based predictive modeling.
• Multivariate pattern analysis (MVPA).
• Network dynamic modeling.
• Deep learning techniques.

Main Results:

• Data-driven methods offer a novel perspective on whole-brain relationships between brain networks and behaviors.
• These approaches quantify meaningful networks and connectivity related to cognition.
• The review discusses pros and cons of each method with practical examples.

Conclusions:

• Advanced data-driven methods, particularly machine learning and deep learning, significantly expand the study of brain-behavior relationships.
• These techniques provide powerful tools for building models of human behavior based on neural network architecture.
• Further application of these methods is crucial for advancing our understanding of neural mechanisms.