1299 Network-Wide Morphological Dynamics Predict Functional Cognitive Performance

INTRODUCTION: Learning-related morphological changes in brain connectivity remain poorly understood. Network analysis techniques provide robust interpretation of high-dimensional neural data during skill acquisition. This insight may reveal novel anatomical and personalized targets for restoring performance in individuals with intellectual disabilities, who today possess sparse therapeutic offerings. METHODS: Task-paired electrode data were collected from 23 subjects undergoing sEEG epilepsy evaluation. Each subject performed several trials of a temporal expectancy task paradigm in which reaction time was assessed following visual stimulus color-change. Subjects were categorized into “Learners” (n = 17; improved at the task) or “Non-Learners” (n = 6; did not improve) based on trial reaction time slopes. During analysis, each sEEG channel represented a node in a graph. Edge weights between each pair of channels were assigned with phase-locking value to form functional connectivity matrices, which were calculated in four frequency bands (alpha, beta, low-gamma, high-gamma). Changes in modularity (a network-wide measure of community segregation) and efficiency (a measure of cross-network navigation ease) over a task session were compared between Learners and Non-Learners for periods prior to the presentation stimulus (pre-trial), go-cue, and response. RESULTS: Learners demonstrated increased high-gamma modularity in the pre-trial period (p < 0.05), increased alpha efficiency in the go-cue period (p < 0.05), and increased alpha and high-gamma modularity in the response period (p < 0.05) relative to Non-Learners. CONCLUSIONS: Dynamic learning was associated with increases in modular community segregation and brain-wide functional efficiency over the duration of a cognitive task. These network analytical findings demonstrate the potential impact of graph network decoding on understanding dynamic learning and could lead to novel therapeutic control signals for future restoration strategies.