Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural-Differentiation Mapping Reveals Experience-Dependent Tuning of Sublexical Units
Poster Session A, Friday, September 12, 11:00 am - 12:30 pm, Field House
Jeremy J. Purcell1, Michaela Brooks1, Cory McCabe2, Lucia Andrea Zepeda Rivera1, Robert W. Wiley3, William W. Graves2, Donald J. Bolger1; 1University of Maryland, 2Rutgers University – Newark, 3University of North Carolina at Greensboro
A central question in written language processing is understanding how experience with sub-lexical units shapes the neural representations that support skilled reading. Prior studies of sub-lexical processing have primarily focused on orthographic regularities, such as letter sequences, but there has been little investigation into the role of phonologically informed units or their experience-dependent tuning in the brain. This leaves open key questions about how the frequency of graphemes and onset-rime structures influences the quality of their neural representations. One study (Purcell & Rapp, 2018) provided initial evidence that higher-frequency words are associated with greater spatial differentiation in the neural response, particularly within the left ventral occipitotemporal cortex (vOTC). This finding suggests that increased experience leads to more efficient, well-differentiated neural coding, which is an interpretation that goes beyond earlier findings of reduced activation for high-frequency words observed in mean BOLD signal studies (e.g., Kronbichler et al., 2004). However, the Purcell & Rapp (2018) findings still await independent replication, and it remains unknown whether similar differentiation effects extend to sub-lexical units that are phonologically informed. The present study addresses both of these gaps. First, we aimed to replicate the original finding that word frequency predicts local neural differentiation in the vOTC. Second, we tested the novel hypothesis that sub-lexical frequency at both fine (i.e., grapheme) and coarse (i.e., onset-rime) levels of granularity are positively associated with the degree of neural differentiation during single word reading. Using the Sublexical Toolkit (Wiley et al., 2024; 2025), we computed experience-based frequency measures for 160 words and 40 pseudowords. Twenty-nine neurotypical, above-average literate adults performed single-word and pseudoword reading aloud during fMRI. We applied a local heterogeneity regression (Hreg) method to quantify stimulus-specific neural differentiation and conducted parametric analyses to identify voxels where differentiation varied as a function of word, grapheme, and onset-rime frequency. Two main findings emerged. First, in the left vOTC, we replicated the positive relationship between word frequency and neural differentiation such that higher frequency words are associated with greater local neural differentiation. Second, we found that in the mid-vOTC - adjacent to regions tuned to whole-word frequency - there were distinct subregions selectively associated with the frequency of graphemes and onset-rimes. This suggests that the mid-vOTC is sensitive to phonologically informed, experience-dependent features of written words. This study is the first to replicate the association between word frequency and neural differentiation and the first to show that phonologically informed sub-lexical units exhibit distinct tuning for graphemes and onset-rimes within the reading network. These findings underscore the potential of neural-differentiation mapping as a powerful tool for charting the functional architecture of reading and other experience-dependent cognitive systems.
Topic Areas: Reading, Methods