Praveen Prem

Purpose: Statistical learning (SL) is the implicit process of extracting statistical regularities from the environment and has been found to be strongly linked with language ability in both children and adults. Notably, SL has been shown to activate brain regions that are implicated in language/reading; however, the extent to which this brain activity is related to reading-related tasks is unknown. Here, we used fMRI to explore the brain-behavior relationships between SL and reading processes. Methods: 44 Adult participants (28 Female, mean age 21 years old) completed a Visual SL (VSL) task inside the MRI that included two phases: 1) familiarization, wherein participants were presented with novel triplets of shapes varying in transitional probabilities, and 2) testing, which involved selecting best-fitting triplets that assessed their learning of these probabilities. Outside the scanner, participants completed three Rapid Automatized Naming (RAN) tasks consisting of quickly naming strings of digits, letters, and objects as a measure of basic reading ability. Results: Using a whole-brain correlational approach with RAN performance as covariates, significant brain-behavior clusters were identified between VSL task-evoked neural activity with RAN digits and objects, but not letters. The regions were predominantly left-lateralized in visuospatial-motor network as well as the parietal regions, caudate, and cerebellum, with the precuneus as a common hub. Conclusion: These findings link to a more multimodal understanding of the neurocognitive processes, indicating that SL-related neural activity in skilled readers is preferentially linked to naming of symbol sets with inherent semantic content (digits, objects) rather than arbitrary orthographic forms (letters).

Statistical Learning (SL) is the ability to implicitly extract and learn statistical regularities from the environment. SL processes are ubiquitous across the lifespan through their proven role in language acquisition and adaptive behaviors. While recent neuroimaging research has identified networks of active brain regions during SL processes, less is known about whether structural differences in these regions may contribute to SL abilities. Here, we explore the extent to which visual SL (VSL) and auditory SL (ASL) performance are linked to cortical vs. subcortical brain volume. 61 Participants (43 Female) completed ASL and VSL tasks, each involving a familiarization and a testing phase. During familiarization, participants observed a sequence of structured triplets of unfamiliar and non-linguistic shapes (for VSL), and environmental sounds (for ASL). During testing, participants identified correct triplets/pairs on a series of alternative-forced-choice questions for both VSL and ASL. High-resolution T1-weighted Magnetic Resonance Imaging (MRI) scans were taken of all participants, from which volumes were extracted for each cortical (planum temporale, fusiform gyrus, middle temporal gyrus, inferior frontal gyrus) and subcortical (caudate, nucleus accumbens, putamen, globus pallidus) regions of interest. Hierarchical regression analyses revealed that performance during the ASL task was significantly related to the volume of the IFG, whereas VSL performance was significantly associated with volumes of all examined subcortical ROIs (p < 0.05). These findings suggest that differences in the IFG and subcortical volumes are differentially associated with SL abilities across visual and auditory modalities, further advancing our understanding of the domain-general and domain-specific aspects of SL processes.