Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Experience-Dependent Tuning of Graphemic Vowel Units in the left vOTC
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
Jeremy Purcell1, Michaela Brooks1, Cory McCabe2, Lucia Z-Rivera1, Robert Wiley3, William Graves2, Donald Bolger1; 1University of Maryland, 2Rutgers University – Newark, 3University of North Carolina at Greensboro
A central challenge in reading neuroscience is understanding how experience shapes neural representations of sublexical graphemic units. Previous mean BOLD studies have shown reduced activation for higher frequency words in left ventral occipitotemporal cortex (vOTC) (e.g., Kronbichler et al., 2004), but activation magnitude offers limited insight into the sparseness of local neural codes. Applying a neural differentiation measure motivated by sparse coding theory, Purcell and Rapp (2018) found the reverse profile: higher frequency words were associated with greater local differentiation in left vOTC. This study aimed to determine if similar effects extend to phonologically informed graphemic units. Using the English Sublexical Toolkit (Wiley et al., 2024; 2025), we computed token frequency measures spanning multiple levels of grapheme granularity (onset, nucleus, oncleus, rime, syllable) for 160 words (70 monosyllabic, 90 multisyllabic) ranging from 1 to 4 syllables and 3 to 9 letters, extending beyond the monosyllabic stimuli typical of prior work. Sixty one neurotypical, literate adults performed single word reading aloud during fMRI. We applied local heterogeneity regression (Hreg) to obtain stimulus specific neural differentiation maps for each word and examined the parametric relationship at each voxel between sublexical graphemic frequency measures and local neural differentiation, while controlling for stimulus length, spelling to sound consistency and whole word frequency. All tests were restricted to a meta analytic map of reading to maximize sensitivity. First, we replicated the positive relationship between word frequency and neural differentiation, though the effect was small and selective to left vOTC, suggesting whole word experience has a circumscribed influence on local representational differentiation. Second, the left vOTC was generally more sensitive to sublexical graphemic unit frequency than whole word frequency, suggesting experience dependent tuning is driven more by sublexical than lexical statistics. Third, vowel containing graphemic units (nucleus, oncleus, rime, syllable) drove the strongest effects while consonant only units showed little influence. Fourth, among the vowel-containing graphemic units, nucleus frequency effects were more prominent laterally while oncleus and rime effects localized medially. These findings are in part consistent with the medial graphemic form area (GFA) proposed by previous work (Bouhali et al, 2019) but further indicate that this region is particularly sensitive to CV and VC graphemic clusters. These findings provide the first evidence that sublexical graphemic unit frequency predicts local neural differentiation during natural reading aloud, without task manipulations or specially designed stimuli. The results reveal previously uncharacterized functional organization within left vOTC: a dissociation between vowel containing and consonant only graphemic units in driving experience dependent differentiation and a medial-to-lateral topography reflecting graphemic grain and composition. Collectively, these findings suggest a level of sensitivity to phonologically informed structure in the visual word form area (VWFA, a functional area within the left vOTC) that is under-appreciated in accounts primarily focused on orthography (e.g., Dehaene et al., 2005; 2011). Generally, neural differentiation mapping can uncover fine grained, experience dependent representational structure invisible to mean BOLD analyses. This approach may extend beyond written language to any cognitive domain in which experience dependent features can be estimated or modeled.
Topic Areas: Reading, Methods