Slide Slam N3
The involvement of bilateral supramarginal gyrus in orthographic processing during Braille reading
Yun-Fei Liu1, Judy Kim1,2, Brenda Rapp1, Marina Bedny1; 1Johns Hopkins University, 2Yale University
Learning to read leads to cortical specialization for orthographic representations of letters, letter combinations, and words. An outstanding question is whether the sensory modality of reading (visual print vs. tactile Braille) influences its neural basis. According to the task-based hypothesis, visual and tactile reading depend on the same visual word form area (VWFA) mechanism within the ventral occipitotemporal cortex (vOTC) (Amedi, Hofsetter, Maidenbaum et al., 2017; Heimler, Behor, Dehaene et al., 2015). On the other hand, based on the connectivity hypothesis, we predicted that Braille reading would recruit regions of posterior parietal cortex that are connected to both the somatosensory and language networks (Saygin, Osher, Norton et al., 2016; Li, Osher, Hansen et al., 2019). To distinguish between these hypotheses, twelve congenitally blind fluent Braille readers were recruited for functional MRI scans, during which they read real words and pseudowords on a Braille display. Participants were asked to press one button if the word was a pseudoword, a second button if it was an animate real word and a third button if it was an inanimate real word. The task ensured attention to both the form and the meaning of the words. All stimuli consisted of four Braille cells (e.g., ⠉⠕⠗⠝ [corn]) and were presented in Grade 2 Universal English Braille (UEB). Grade 2 UEB contains “contractions”, such that one Braille cell may stand for more than one letters (e.g., ⠜ [ar]; ⠬ [ing]). This enabled us to study the effects of letter length on neural activity, independent of the physical length of the stimuli (e.g., ⠎⠥⠛⠜ [sugar]; ⠒⠉⠻⠞ [concert]). We also tested which cortical areas distinguish between real words and pseudowords based on univariate and multivariate responses (MVPA). Parietal cortex showed several signatures of orthographic sensitivity. First, a whole-brain univariate correlational analysis revealed a positive effect of letter length of real words in bilateral parietal cortex (anterior supramarginal gyrus, SMG) and prefrontal cortex. Crucially this letter length effect in contracted Braille is independent of physical word length and is therefore unlikely to reflect low level tactile responses. Second, similar parietal SMG areas responded more to pseudo words than real words in univariate analysis. By contrast, larger univariate responses to words were observed in classical semantic areas, including the bilateral precuneus and bilateral angular gyri (AG). Finally, multivariate activity patterns in parietal/parieto-occipital regions distinguished between words and pseudowords. Larger responses to psuedowords were also observed in medial vOTC but not lateral. The present results suggest that fronto-parietal cortices, specifically the SMG, may become specialized for form-based orthographic processing of Braille. Our findings do not rule out the possibility that vOTC contributes to Braille reading, however, they suggest that tactile reading of Braille relies on partially distinct neural mechanisms as compared to visual print.