Masked features of task states found in individual brain networks

Area of Science:

• Neuroscience
• Cognitive Science
• Functional Magnetic Resonance Imaging (fMRI)

Background:

• Complex task completion relies on flexible integration of information across brain regions.
• Previous research primarily used cross-subject analyses, focusing on shared features of functional brain networks during tasks.
• Individual differences in brain network organization during cognitive tasks remain underexplored.

Purpose of the Study:

• To investigate the extent to which task-specific brain states generalize across individuals versus remaining individually specific.
• To compare the efficacy of cross-subject versus individual-focused approaches in decoding brain states.

Main Methods:

• Utilized high-resolution "precision" functional Magnetic Resonance Imaging (fMRI) data with extended sampling.
• Trained machine learning classifiers to decode cognitive states from functional network data within single-person datasets across five diverse tasks.
• Tested classifier generalization on both independent data from the same individual and data from new participants.

Main Results:

• Individualized classifiers demonstrated generalization to new participants, indicating shared patterns of brain activity.
• Classification performance was significantly higher when tested within the same individual compared to across individuals.
• This within-person advantage was consistent across various analytical models, tasks, feature sets, and even for distinguishing highly similar task conditions.
• Findings were robust and replicated in an independent dataset.

Conclusions:

• Task-based brain states exhibit both generalizable and individually specific features.
• Individual-focused analytical approaches can reveal robust neural features of brain states, potentially uncovering information missed by traditional cross-subject analyses.
• Personalized neuroscience approaches hold significant potential for advancing our understanding of brain dynamics during complex cognition.