Network and State Specificity in Connectivity-Based Predictions of Individual Behavior

Area of Science:

• Neuroscience
• Cognitive Science
• Behavioral Science

Background:

• Predicting individual behavior from brain functional connectivity (FC) is crucial for understanding human brain function.
• Task-based FC data may offer better behavioral predictions than resting-state FC.
• Interpretability can be enhanced by using features from predefined functional networks.

Approach:

• Examined the influence of matching functional network priors and task states to behavioral domains on predicting individual performance.
• Utilized Human Connectome Project data for out-of-sample predictions of working memory (WM), social cognition (SOCIAL), and emotion processing (EMO).
• Assessed predictions using FC from corresponding and non-corresponding states (WM/SOCIAL/EMO/resting-state) and networks (WM/SOCIAL/EMO/whole-brain).

Key Points:

• Predictive performance for individual abilities was generally poor across all conditions.
• Whole-brain FC predictions slightly outperformed task-specific network FC.
• A minor benefit of task-based FC over resting-state FC was observed for working memory performance.

Conclusions:

• No significant effects of matching network, task state, and performance domains were found.
• Multivariate FC patterns, during both task and resting states, contain limited information about individual performance levels.
• Reconsideration is needed regarding how the brain mediates individual differences in mental abilities.