Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Disentangling the specific networks related to statistical learning from those for rule generalization
There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.
Poster A104 in Poster Session A, Tuesday, October 24, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Brian Quintero-Manes1, Ana B. Chica2, Estela Càmara3, Ruth de Diego-Balaguer1; 1Universitat de Barcelona, 2Universidad de Granada, 3Institut d'Investigació Biomèdica de Bellvitge (IDIBELL)
An important part of language learning is the identification of rules embedded in words and phrases. Usually, rules are characterized by sequential co-occurrences between elements (e.g., “These cupcakes are unbelievable”) thus tracking the statistical relationship between these non-adjacent dependencies is fundamental. Previous studies have showed that the tracking of these statistical relations are supported by a ventral fronto-parietal and basal ganglia network. As rules become consolidated, they can be abstracted and transferred to a new language but generalisation is not possible when left parietal lobe is impaired. The objectives of this study are twofold: 1) replicating the behavioral and neural findings which identified the left frontoparietal network involved in both learning and generalisation of linguistic rules and the left parietal lobe (LPL) being necessary for generalisation, and 2) disentangling the specific neural networks associated with the abstract generalization of linguistic rules from those related to statistical learning, involved in early learning of non-adjacent linguistic rules. Procedure: We conducted an experiment using fMRI and TMS in three sessions. Three artificial languages (L1, L2, L3) were used, each with different words presented auditorily. Rule sequences followed a three word A-X-C structure (e.g. cofa male runi), where "A" predicted "C," while no-rule sequences had an X-X-C structure. Implicit learning was measured through a target detection task of “C” (predictable in Rule but not in the No Rule condition). In the first session (L1), participants learned the A-X-C rule. In the second and third sessions, generalization was tested by presenting the same participants with a new language (L2 or L3) with new words but following the same A-X-C structure one week later. In these sessions Theta Burst Stimulation (TBS) was applied to the maximum BOLD signal peak (LPL) from the first session or the vertex (Control area) during fMRI scanning while learning was performed. Results: In the first learning session, results showed the engagement of the left frontoparietal network for rule learning. These findings replicated the role of the LPL in rule learning. Importantly, the TBS results replicated the impaired generalization observed under stimulation of the LPL, while generalization remained maintained under vertex stimulation. Additionally, compared with the results of the first session, TBS at the vertex (i.e. when generalisation was possible), also engaged the frontoparietal network, but with a more bilateral pattern compared with the first session. In contrast, when the LPL was interfered and re-learning was forced due to impaired generalisation, subcortical areas, including the hippocampus, cerebellum and the brainstem , which have previously been associated with statistical learning, take a more important role than in the first session. Conclusion: These results confirm again the key role of the LPL in generalization and highlight the core function of subcortical regions in statistical learning, particularly when cortical areas are not fully available.
Topic Areas: Language Development/Acquisition,