The Development of Hemispheric Lateralization to Repetition-Based Rule-Learning With Linguistic Stimuli: A Meta Analysis of Infant and Adult fNIRS Studies

In most adults, language processing is lateralized to the left hemisphere of the brain, but what are the developmental origins of this lateralization? To address this question, we conducted a comprehensive meta-analysis combining 27 fNIRS studies examining brain responses to repetition-based artificial grammar sequences across 5 age groups: 0m, 6m, 7m, 9m and Adults. Our meta-analysis included 525 infants (0-9 months) and 42 adults, focusing on hemispheric activation patterns in temporal and frontal regions of interest. We examined the effects of age, hemisphere (left vs. right), and brain region (temporal vs. frontal) on hemodynamic responses measured via fNIRS. We employed random-effects meta-analytic models to estimate subgroup-level effect sizes and conducted moderated analyses to examine interactions between developmental and neural factors. While our hypothesis anticipated progressive hemispheric specialization, hemisphere effects were consistently non-significant across all analyses (p > 0.05). Instead, robust age-related effects emerged as the dominant pattern, consistently approaching or achieving statistical significance (HbO all-data: p = 0.0532; HbR speech-only: p = 0.0284). Developmental trajectory analysis revealed peak activation at 6-7 months across both hemispheres, followed by declining activation toward adulthood. Adults demonstrated significant effects in both hemispheres rather than the predicted left-temporal specialization, with unexpected subtle right-lateralization patterns (lateralization index = -0.3) that remained statistically non-significant. Regional analyses showed temporal areas exhibited slightly stronger developmental patterns than frontal regions, though these differences did not reach significance either. The absence of significant lateralization effects suggests that repetition-based rule-learning may engage bilateral neural networks rather than the strongly lateralized language systems typically observed in higher-order linguistic processing.