Noise is the Signal: Why Weak Brain Connections Predict Behavior

Key Points:

• A new study from Yale reveals that the 90% of brain connections typically dismissed as “noise” in neuroimaging can predict behavior with accuracy equal to or better than the strongest 10% of signals.
• The research shows multiple, non-overlapping brain networks can predict the same behavior, highlighting significant redundancy and functional flexibility in brain connectivity.
• Findings suggest that psychiatric conditions like depression may involve different neural pathways across individuals, indicating that treatment should target a broader range of brain circuits beyond the traditionally focused networks.
• The study challenges the assumption that strongest brain signals have the highest biological relevance, proposing that overlooked connections could be crucial for developing precision medicine and improving treatment for “treatment-resistant” patients.
• Using data from over 12,000 participants across four major datasets, the researchers emphasize that focusing only on top brain connections risks oversimplifying brain complexity and may contribute to reproducibility issues in neuroscience.