Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Structural circuit correlates of functional reorganisation during non-adjacent dependency learning
Poster Session D, Thursday, October 1, 4:30 - 6:30 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Brian Quintero-Manes1,2, Ana B. Chica3, Stephanie J. Forkel4,5, Estela Càmara6, Ruth De Diego-Balague1,2,6,7; 1Dept. of Cognition, Development and Educational Psychology, University of Barcelona, 2Institute of Neuroscience, University of Barcelona, 3Dpt. Experimental Psychology, University of Granada, 4Donders Institute for Brain, Cognition and Behaviour, Radboud University, 5Max Planck Institute for Psycholinguistics, Nijmegen, The Netherlands, 6Bellvitge Biomedical Research Institute (IDIBELL), 7Catalan Institution for Research and Advanced Studies (ICREA)
Statistical learning (SL), defined as the ability to extract regularities from environmental input, is considered a fundamental mechanism for language acquisition (Saffran et al., 1996). This ability is particularly relevant in the extraction of non-adjacent dependencies (NAD) characterizing grammatical relations between temporally distant elements in speech (Newport & Aslin, 2004). Previous work using artificial grammar learning paradigms has shown that NAD learning unfolds through at least two functionally distinct stages within the SL process. Early learning is characterized by the initial extraction of regularities, typically indexed by progressive improvements in behavioral performance, whereas late learning is associated with behavioral stabilization. Importantly, these stages have been linked functionnally to a shift from ventral to more dorsal frontoparietal brain networks. The main goal of the present work was to investigate the white matter circuits associated with these two learning stages. To achieve this, we applied the Functionnectome framework (Nozais et al., 2021, 2023), which projects task-related fMRI signal onto probabilistic white matter pathways derived from tractography, thereby enabling the identification of structural circuits contributing to functional processing. Participants were exposed while fMRI scanned to non-adjacent dependencies embedded in three-word auditory sentences from an artificial language and after each sentence, they indicated whether a predefined target word was present or absent in the sequence. Each sentence either contained a predictive dependency between the first and last word (AXC, Rule condition; e.g. sipa nabi runi; sipa meno runi) or consisted of non-predictive combinations lacking systematic dependencies (XXC, NoRule condition; e.g. nabi meno runi; meno sipa runi). The behavioral and neuroimaging results partially replicated previous findings (Orpella et al., 2020), showing a distinction into learning stages with a shift from progressive improvement of reaction times (RTs) to behavioural stabilisation. When the slope of the RT improvement in the Rule condition was used as a regressor, the early learning stage related to an early left-lateralized ventral network involving middle temporal gyrus (MTG), inferior parietal regions (IPL), specifically in the SupraMarginal gyrus (SMG), and inferior frontal gyrus (IFG). In the second stage, the stable difference in RTs from Rule and NoRule was used as a regressor, indicating a shift to a more dorsal and bilateral network extending to superior parietal (SPL) and prefrontal regions. Preliminary Functionnectome analyses suggest distinct white matter circuit involvement across learning stages. The early learning stage was associated primarily with projections involving the left arcuate fasciculus (AF) and cingulum pathways, whereas the late learning stage showed greater involvement of bilateral superior longitudinal fasciculus (SLF) projections. Together, these findings provide preliminary evidence that the functional reorganisation observed during non-adjacent dependency learning may rely on distinct structural circuitries. More broadly, this work highlights the potential of the Functionnectome framework to link functional dynamics with underlying white matter architecture in the study of language learning and cognitive network organisation.
Topic Areas: Control, Selection, and Executive Processes, Speech Perception