Connecting Brain Networks to the Cognitive Components of Reading Comprehension

Vanderbilt University, Nashville TN

Investigators

Linked publications & trials

Abstract

PROJECT SUMMARY Skilled reading comprehension (RC) is a chief aim of education, yet 34% of fourth graders and 25% of eighth graders read below basic achievement levels. Poor comprehension is a major societal concern, as it negatively impacts educational outcomes, making these Americans less competitive for jobs and at greater risk of poverty. Remediation often focuses on mitigating problems with word recognition (WR), but RC also requires the integration of oral language comprehension (OLC) abilities (e.g. vocabulary and syntax) and executive function (EF), especially working memory. Understanding how these skills interact during RC is critical for predicting which children are most vulnerable to RC difficulty. Recent work using resting-state fMRI (rs-fMRI), task-based fMRI (t-fMRI) and diffusion-weighted MRI (dw- MRI) has described how the brain regions underlying WR, OLC and EF form neural networks capable of complex behavior. An efficient network architecture is thought to be central to integrative processes such as RC, and is disrupted in individuals with psychiatric disorders. However, how network properties relate to specific, educationally-relevant behaviors is unexplored. We propose to investigate how the network architecture of the brain is fundamental to the cognitive skills underlying RC in children. Specifically, we will investigate the brain's intrinsic resting-state networks (RSNs), which span areas responsible for component skills of RC, using rs-fMRI and dw-MRI from a longitudinal study of elementary- aged children, as well as task-fMRI from a to-be-collected sample of fourth graders. For our first aim, we hypothesize that individual differences in distributed processes such as WR, OLC and EF will be correlated with differences in specific RSN properties. For example, higher WR skill is associated with increased connectivity within visual areas (including the putative visual word form area), suggesting that a distinct visual RSN would also correlate with WR. However, we expect that differences in RC, which integrates these processes, will be most reflected in the interactions between RSNs. Our second aim is to determine if RSN properties in first grade can predict behavioral RC outcomes in fourth grade, especially in relation to extant behavioral metrics. Since decoding skill in first grade is a limiting factor for reading, RC metrics at that age may heavily reflect WR. However, network architecture differences may already be present and able to predict RC skill at an age where greater integration is required. In our final aim, we use t-fMRI to track how the network properties of the WR, OLC and EF RSNs dynamically change during the reading comprehension process. The results of these analyses will fill gaps in our understanding of how brain network structure is related to and predictive of behavioral indices of cognitive abilities, particularly RC. They will also lead to a better understanding of the systems-level neural mechanisms underlying proficient reading.