Learning-based cognitive control: developmental trajectories in children aged 5- to 14- years.

Learning-based cognitive control (CC)-the ability to implicitly adapt control based on contextual regularities-has been studied in young, typically developing children, yet its developmental trajectory remains underexplored. This study examined how learning-based CC develops across childhood under increasing cognitive demands. Overall, 149 children aged 5-14 years (79 females, M = 9.1 ± 2.6) completed a modified Flanker task and a cued go-noGo task ("Addy"). Both tasks included a List-Wide Proportion Congruency (LWPC) manipulation contrasting predictable (mostly congruent/valid) and unpredictable (50%) contexts. Reaction times (RTs), accuracy, and delta scores (incongruent/invalid-congruent/valid RTs) were analyzed. In the Flanker task, LWPC effects were similar across ages, suggesting that learning-based CC emerges early and remains stable in low-demand contexts. In contrast, in the Addy task-requiring greater attentional control and motor inhibition-developmental differences emerged. Younger children adapted behaviour by favoring speed over accuracy, while from around 9 years of age children displayed a more balanced speed-accuracy trade-off and improved accuracy, indicating greater efficiency in managing competing task demands. These findings suggest that learning-based inhibitory CC efficiency under complex, multi-demand conditions continues to develop across childhood and highlight the value of ecologically valid paradigms.