Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Dynamic Interactions Between Sensorimotor and Language Networks in Rhythm-Scaffolded Non-Adjacent Dependency Learning
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
Alba Nuez-Vilà1,2, Tomas Berjaga-Buisan3, Gustavo Deco3,4,5,6, Estela Camara-Mancha2, Ruth de Diego-Balaguer1,2,6; 1Institute of Neurosciences, Department of Cognition, Development, and Educational Psychology, University of Barcelona, Barcelona, Spain, 2Cognition and Brain Plasticity Unit, Institut d’Investigació Biomèdica de Bellvitge (IDIBELL), Barcelona, Spain, 3Center for Brain and Cognition, Computational Neuroscience Group, Department of Engineering, Universitat Pompeu Fabra, Barcelona, Spain, 4Centre for Eudaimonia and Human Flourishing, Linacre College, University of Oxford, Oxford, UK, 5International Centre for Flourishing, Universities of Oxford (UK), Aarhus (Denmark) and Pompeu Fabra (Spain), 6Institució Catalana de Recerca i Estudis Avançats (ICREA), Barcelona, Spain
Grammar acquisition requires linking elements separated across intervening material, such as "the boy" and "got an ‘A’" in “the boy that helped the girl got an ‘A’ on the test” (Hilton & Goldwater, 2021). The rhythmic structure of speech prosody may scaffold the learning of these non-adjacent dependencies (NADs) by exogenously capturing attention and directing it toward critical NAD-bearing windows. Supporting this view, recent behavioral evidence shows that NAD learning is enhanced when dependent syllables coincide with predictable beat positions (Franzoia & de Diego-Balaguer, 2025). However, how brain networks related to rhythmic processing work in coordination with attentional and language systems to modulate NAD learning remains largely unknown. In this fMRI study, we leverage the INSIDEOUT framework (Deco et al., 2022) to characterize whole-brain network dynamics across distinct rhythm-scaffolded NAD learning conditions. INSIDEOUT quantifies functional hierarchy by measuring temporal irreversibility across brain regions: specifically, it compares time-shifted correlation matrices of the forward fMRI time series against an artificially reversed backward time series, where higher irreversibility reflects stronger breaking of detailed balance, indicating a more pronounced functional hierarchy. Participants (n = 37) were passively exposed to a 10-minute continuous auditory stream across three artificial language-rhythm conditions, during which fMRI data were acquired: RB, a beat-inducing rhythm with on-beat NADs; R, a beat-hindering rhythm containing NADs; and B, a beat-inducing control without NADs. Following exposure, an explicit familiarity rating task (all three conditions) and an implicit target detection task (RB only) assessed post-learning NAD discrimination, with RT advantages serving as an individual learning index for the fMRI analyses. Behavioral results revealed explicit NAD learning in both RB and R (RB+: t = 2.58, p = .044; R: t = 3.72, p = .002) but not in the B control (t = 1.76, p = .263). RT measures in the RB target detection task further revealed a strong group-level advantage in anticipating both the first NAD element (β = −0.033, p = .049) and, crucially, the second element (β = −0.163, p < .001) relative to baseline filler syllables. To map the neural mechanisms underlying this behavioral pattern, our ongoing fMRI analyses test the hypothesis that rhythm-scaffolded NAD learning recruits a hierarchical cascade across sensorimotor, attentional, and language networks. We predict that sensorimotor-language network coupling will be selectively enhanced in RB over B or R, reflecting the unique advantage conferred by the convergence of rhythmic and statistical regularities. The B condition, providing rhythmic structure without linguistic dependencies, is expected to engage sensorimotor networks without driving equivalent integration with language systems. The R condition, conversely, should engage language-related processing without the motor entrainment afforded by a beat-inducing rhythm. Additionally, considering the proposed role of endogenous orienting of attention in NAD learning (e.g., Orpella et al., 2020), we will study the differential recruitment of the dorsal attention network across conditions.
Topic Areas: Prosody, Multisensory or Sensorimotor Integration