Neural Underpinnings of Academic Skills: Cross-Domain and Cross-Generational Influences
Abstract
This PhD thesis explores the shared neural mechanisms between arithmetic and phonological processing and examines intergenerational influences on these abilities through a series of studies utilizing neuroimaging data from children and their mothers. The overlap between mathematics and reading has been largely hypothesized, bur very rarely tested directly within participants. During Study 1, I provided direct evidence of the association of arithmetic and phonological abilities in the brain. Conjunction analyses showed significant overlap in adults along the inferior frontal gyrus, inferior temporal gyrus and the cerebellum. In children, overlap was observed in multiple regions of the frontal cortex. Notably, our results indicated that overlapping activation was different when considering small or large arithmetic problems separately and therefore, the associations between each problem type and phonological processes required further exploration. In Study 2, I examined of these areas of significant overlap also displayed similar multivariate patterns of brain activity. Moreover, I hypothesized that verbal retrieval strategies would be more frequent during small problems, and therefore the neural mechanisms supporting the solution of these problems would be more similar to phonological processing mechanisms than large problems. Against our prediction, large problems consistently displayed greater pattern similarity with the rhyming condition, suggesting that arithmetic and phonology may share other mechanisms of verbal processing in addition to the retrieval of phonological representations. Finally, in Study 3, I examined the intergenerational influences in the neural correlates of mathematics and reading. Familial concordance was observed in several brain regions of the frontal cortex during arithmetic tasks, and in the hippocampus during phonological processing. I also investigated if similarities in whole-brain functional connectivity between parents and children were strong enough to accurately predict familial relationships. I found that connectivity matrices contained information that identified participants across multiple tasks; however, parent-child dyads could not be identified using this approach. Critically, this study emphasized the need for better methods of feature selection that can be informative of specific parent-child relationships.