Mechanisms of interactive specialization and emergence of functional brain circuits supporting cognitive development in children

Abstract

Cognitive development is thought to depend on the refinement and specialization of functional circuits over time, yet little is known about how this process unfolds over the course of childhood. Here we investigated growth trajectories of functional brain circuits and tested an interactive specialization model of neurocognitive development which posits that the refinement of task-related functional networks is driven by a shared history of co-activation between cortical regions. We tested this model in a longitudinal cohort of 30 children with behavioral and task-related functional brain imaging data at multiple time points spanning childhood and adolescence, focusing on the maturation of parietal circuits associated with numerical problem solving and learning. Hierarchical linear modeling revealed selective strengthening as well as weakening of functional brain circuits. Connectivity between parietal and prefrontal cortex decreased over time, while connectivity within posterior brain regions, including intra-hemispheric and inter-hemispheric parietal connectivity, as well as parietal connectivity with ventral temporal occipital cortex regions implicated in quantity manipulation and numerical symbol recognition, increased over time. Our study provides insights into the longitudinal maturation of functional circuits in the human brain and the mechanisms by which interactive specialization shapes children's cognitive development and learning.

Fig. 1

Longitudinal sampling of 30 study participants over time. Black circles represent single fMRI scan acquisitions, and multiple scans for one child are represented by interconnecting lines

Fig. 2

Performance improvements over time. A hierarchical linear model (HLM) was used to determine changes in a reaction time and b accuracy with age in arithmetic task performance. Dark lines show group model fits, while lighter lines show individual data from each of the 30 children

Fig. 3

Longitudinal changes in IPS connectivity over time. Increases in a parietal-fusiform gyrus, b intra-parietal, and c inter-hemispheric IPS connectivity with age. Age-related decreases in d–f parietal-prefrontal cortex connectivity with age. Results of whole-brain voxelwise hierarchical linear modeling (HLM) showing brain regions with significant age-related changes in left IPS connectivity. Line plots show HLM fits for target regions extracted from the connectivity analysis. Dark lines show model fits, while lighter lines show individual participant data. IPS intraparietal sulcus, FG fusiform gyrus, SPL superior parietal lobule, dlPFC dorsolateral prefrontal cortex, vlPFC ventrolateral prefrontal cortex

Fig. 4

Inter-hemispheric IPS connectivity is related to individual differences in mathematical ability. Higher math ability was associated with higher inter-hemispheric IPS connectivity. Line plots show HLM fits for target regions extracted from the connectivity analysis. Colored lines show individual participant data, with light blue to purple colors indicating individual differences (low to high) in mathematical ability as measured by the NumOps subtest of the WIAT-II. Black lines show models for a hypothetical participant with NumOps score of 105 (which was roughly the mean NumOps score in the studied sample)

Fig. 5

Intraparietal sulcus (IPS) region of interest (ROI). The ROI (shown in red) was derived by computing the center (x = −28, y = −58, z = 46) of the cytoarchitectonically-defined subdivision hIP3. This ROI shows maximal overlap with meta-analytic activation maps generated using the search term ‘arithmetic’ restricted to pediatric and adolescent subjects (shown in blue) in Neurosynth.