Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural Correlates of Speech Motor Sequence Learning in Primary Progressive Aphasia
Poster Session E, Friday, October 2, 11:00 am - 1:00 pm, Wangari Maathai
Hilary Miller1, Michael Brickhouse2, Daisy Hochberg2, Jason Tourville3, Alfonso Nieto-Castanon3, Bradford Dickerson2, Frank Guenther3; 1University of Connecticut, 2Massachusetts General Hospital, 3Boston University
Rationale: This study uses a non-native cluster learning paradigm to investigate the neural correlates of speech learning in primary progressive aphasia (PPA), with the aim of dissociating the impacts of brain atrophy in key phonological and speech motor regions—specifically left posterior inferior frontal sulcus (pIFS) and left ventral premotor cortex (vPMC)—on learning outcomes. Methods: A cohort of 23 participants with PPA or a related dementia diagnosis completed a two-day non-native consonant cluster learning paradigm, in which participants practiced pseudowords (e.g., “kpud”) containing consonant clusters that are illegal in English but permissible in another language. Learning was characterized through measures of error rate and duration, with separate measures calculated for each participant for (1) improvements across the training phase and (2) comparison of trained and novel items during a post-test phase. Cortical thickness was measured from each participant’s T1-weighted structural MRI scans using FreeSurfer software to obtain a mean cortical thickness for each region of interest. Separate regression analyses tested for relationships between each learning outcome with cortical thickness, controlling for age, handedness, and baseline performance on the task. A priori analyses tested first for correlations with cortical thickness in left pIFS and left vPMC, followed by an exploratory whole-brain analysis. Results: Regression analysis for the test-phase error rate learning measure (i.e., trained-novel comparison) revealed a significant correlation with left pIFS ( standardized beta = .443, p = .025), but not left vPMC (p = .24). Improvement in error rate across the training phase was not significantly correlated with either brain region, although borderline for vPMC (vPMC: standardized beta = .363, p = .058; pIFS: p = .15). For duration, neither region was significantly correlated with the test-phase measure, but improvement across training was significantly correlated with left vPMC (standardized beta = .479, p = .019). Whole-brain analyses also identified unique neural correlates of speed and accuracy: sensorimotor regions were predictive of duration, while error rate was associated with a broader working memory network. Discussion: This study aimed to determine the neural correlates of speech motor sequence learning in PPA and related dementias. Results partially supported the primary hypothesis that left pIFS and left vPMC are key cortical areas for speech motor sequence learning. Left pIFS findings were specific to improvements in accuracy, but only for the test-phase learning measure. Atrophy in left vPMC was associated with worse learning outcomes for both speed and accuracy, but only across the training phase. Interpreted within the GODIVA model of speech sequencing, these findings suggests patients with thicker cortex in left pIFS were better able to acquire a new representation for the trained cluster as a single unit in the phonological content buffer; while correlations with left vPMC indicate that concatenation of smaller motor gestures into a larger motor program in the models’ speech sound map allows for faster and more accurate transitions between sounds.
Topic Areas: Speech Motor Control, Disorders: Acquired