Towards Understanding Human Functional Brain Development with Explainable Artificial Intelligence: Challenges and Perspectives

The "Black Box" Problem in Brain Research: Current AI techniques used to analyze brain imaging data can identify patterns but cannot explain why they found them or what they mean for understanding how brains develop. This limits scientists' ability to validate findings against established theories of how the brain actually works.

Multivariate Analysis Reveals Brain Interactions: Moving from studying single brain areas to examining how multiple brain regions work together is scientifically important, but this complexity often makes AI models even less interpretable—creating a tension between analytical sophistication and scientific understanding.

XAI as a Solution for Neuroscience: The paper proposes that Explainable AI (XAI) techniques could bridge this gap by making AI models transparent enough to align with established developmental cognitive neuroscience frameworks, allowing researchers to understand not just what the AI found, but why it matters for brain development.

Accountability in Medical AI: For research that could eventually inform clinical decisions about child development and brain health, having AI systems that can explain their reasoning is essential—not just for scientific validity, but for responsible use of these technologies in healthcare contexts.

Framework Alignment Matters: The paper emphasizes that AI techniques should be evaluated based on whether they can explain their findings in terms that match real neuroscience theory, not just on prediction accuracy alone.