Reduced information sharing within the orienting network predicts intelligence in late childhood.

Understanding how intelligence emerges from the developing brain remains a central question in developmental cognitive neuroscience. While prior work has linked intelligence to regional activation and structural maturation, less is known about how the efficiency of distributed functional networks contributes to cognitive ability during childhood. We examined 38 children aged 11-14 years as they performed the Attention Network Test (ANT) while undergoing electroencephalography (EEG). Attentional network engagement was characterised using cue-locked event-related potentials (ERPs), frontal midline theta power during conflict processing, and weighted Symbolic Mutual Information (wSMI) to quantify nonlinear information sharing across distributed networks. Canonical electrophysiological signatures of alerting, orienting, and conflict processing were robustly replicated. However, only distributed nonlinear connectivity within the orienting network predicted Full-Scale IQ (FS-IQ). Specifically, reduced information sharing during spatial orienting was associated with higher intelligence scores, whereas local ERP and theta measures showed no reliable associations. These findings suggest that intelligence in late childhood is linked not to greater neural activation, but to more efficient network-level coordination within task-relevant attentional systems. Our results extend the neural efficiency hypothesis into a developmental framework and highlight nonlinear connectivity as a sensitive marker of cognitive maturation.