Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Procedural Memory Predicts Early Language Development: Domain-General and Production-Selective Effects with Preliminary fMRI Evidence
Poster Session D, Thursday, October 1, 4:30 - 6:30 pm, Wangari Maathai
This poster is part of the Sandbox Series.
Jueyao LIN1, Jiayi LU1, Xiaocong CHEN1, Zhengqin LIU1, Lu LI1, Changsheng LI1, Michael T. Ullman2, Han ZHANG3, Caicai ZHANG1; 1Hong Kong Polytechnic University, 2Georgetown University, 3ShanghaiTech University
Introduction. Procedural memory (PM), which underlies implicit sequence learning through repeated exposure, has been linked to grammatical development (Conti-Ramsden et al., 2015; Ullman, 2016). Yet it remains undercharacterized whether PM–language associations extend to vocabulary, which also involves systematic phonological-articulatory sequencing (Gupta & Dell, 1999), and whether they are stronger for production than comprehension given production’s greater active sequencing demands, or vary across learning phases. Further, existing pediatric evidence for PM–language links is largely behavioral or based on school-aged children, leaving open how procedural learning is instantiated in the developing brain of preschoolers, when language abilities are rapidly emerging. The current study therefore examined whether phase-specific procedural learning predicts fine-grained grammar and vocabulary abilities across comprehension and production in Cantonese-speaking preschoolers, integrating Day 1 and Day 2 behavioral trajectories with task-based fMRI during Day 1. Method. Fifty-six Cantonese-speaking 4-year-olds (M = 4.54 years) completed a two-day Serial Reaction Time task (Thomas et al., 2004) with interleaved Sequence-Random blocks (8-item sequence; Day 1: six Sequence-Random block-pairs; Day 2: two Sequence-Random block-pairs). A PM index was computed at each block-pair as the z-scored Random–Sequence reaction time difference, with individual learning trajectories (slopes) extracted for Day 1 (initial acquisition) and Day 2 (post-learning re-engagement). Language outcomes included grammar comprehension, grammar production, vocabulary production from Hong Kong Test of Preschool Oral Language (Wong et al., 2019), plus receptive vocabulary from the PPVT (Dunn & Dunn, 2007). A mixed-effects model examined Day 1 and Day 2 slopes as predictors of language across modality (comprehension vs. production) and domain (grammar vs. vocabulary), controlling for age, nonverbal IQ, and gender. The final three Sequence-Random block-pairs on Day 1 were performed inside the MRI scanner (n = 19 after motion scrubbing). Preliminary whole-brain group-level analysis examined the Sequence-Random contrast across all block-pairs (voxel-wise p < .02, cluster-level FWE α < .05, k ≥ 219 voxels). Results. Behaviorally, the Day 1 trajectory predicted all four language outcomes (b = 2.79, p = .01), with no significant interactions, indicating broad cross-domain and cross-modality effects. The Day 2 trajectory showed no significant main effect but a significant modality dissociation (modality × Day 2 slope: b = 0.25, p = .04), with post-learning re-engagement more predictive of production than comprehension. No domain-related effects emerged. Preliminary whole-brain group-level analysis identified one significant cluster in right sensorimotor cortex (paracentral lobule, postcentral gyrus, precentral gyrus; k = 224, peak t = −4.97), showing reduced activation for Sequence relative to Random blocks during Day 1 learning. Conclusion. These findings support a process-based account in which PM supports early language through sequencing demands rather than specific domains: initial procedural encoding broadly predicted grammar and vocabulary abilities, whereas post-learning re-engagement was more sensitive to production. Preliminary fMRI evidence showed reduced sensorimotor activation for predictable relative to unpredictable sequences during procedural learning, possibly reflecting incipient predictive processing and reduced reactive demands for structured input. Ongoing analyses will test whether individual PM trajectories predict neural differentiation and language outcomes, extending the current findings toward individual-level brain-behavior-language mapping.
Topic Areas: Language Development/Acquisition, Language Production