Early Brain Functional Segregation and Integration Predict Later Cognitive Performance

Abstract

The human brain in the first 2 years of life is fascinating yet mysterious. Whether its connectivity pattern is genetically predefined for neonates and predictive to the later cognitive performance is unknown. Numerous neurological/psychiatric diseases in adults with impaired cognitive functions have been linked with deteriorated “triple networks” that govern the high-level cognition. The triple networks are referred to salience network for salient event monitoring and emotion processing, default mode network for self-cognition and episodic memory, and executive control network for attention control, set maintenance and task executions. We investigate the infancy “triple networks” and their development in the pivotal period of the first two years of life with longitudinal resting-state fMRI from 52 term infants (24 having cognitive performance scores tested at 4 years old). We found that the triple networks harbor at the medial prefrontal cortex, an ideal brain region for unveiling early development of the high-level functions. Further parcellation of this area indicates consistent subdivisions from 0 to 2 years old, indicating largely predefined functional segregation in this highly heterogeneous region. Interconnectivity among the mediofrontal subdivisions reveals a significant invert U-shape curve for modularity, with the inter-network functional connectivity (FC) peaking at 6–9 months, manifesting a developing functional integration within the frontal region. Through long-range FC, we found the development of the high-level functions starts from salience monitoring, followed by self-cognition, then to executive control. We extract both within-frontal modularity index (reflecting short-distance FC), and outreaching index (measuring long-distance FC) for the newborns. Interestingly, these connectomics features for the newborns well predict their later cognitive performance at 4 years old. These results converge to favoring a predefined genetic dominance in the development of triple networks’ FC, which is essential for understanding early high-level neuro-cognitive development and promising for early abnormality detection.