Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
An fMRI study of phonological decoding: Learning to read in an artificial script
Poster E125 in Poster Session E, Thursday, October 26, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Anna Chrabaszcz1, Kailee M Lear1, Corrine Durisko1, Julie A Fiez1; 1University of Pittsburgh
Introduction. Skilled reading is characterized by fast and efficient visual word recognition and relies on an extensive network of dorsal fronto-temporo-parietal and ventral occipito-temporal brain regions, supporting phonology-based and orthography-based processes, respectively[1-4]. In contrast, emergent reading is characterized by heavy reliance on phonological decoding, or the process of transforming letters into speech sounds[5-6]. We used an artificial grapheme-phoneme training paradigm to emulate the initial stage of reading acquisition to characterize the brain network that supports this process in novice reading. Methods. Thirty-two neurotypical participants (M age=26, 9M/21F/2Other) learned a rune-like artificial alphabet consisting of 24 letters that mapped onto English language sounds. Each of the 10 consonant sounds was represented by 2 homophonous letters (like K and C can be both pronounced as /k/ in English). After 10 sessions of training to decode words written in the artificial script (160 words per session), participants completed an fMRI scan during which they read target words (not seen during training) (16 targets x 16 runs) interspersed with visual patterns (control condition). They were instructed to first decode the word silently and name it after the button press. Participants’ memory of target words was tested on the next day with the spelling, word recognition and lexical decision tasks. Imaging data were processed and analyzed in the AFNI software. Brain areas showing significant activation were compared to the canonical reading network generated via the Neurosynth. Additionally, all participants completed a standardized assessment of their reading-related and general cognitive functions. Results. Analysis of behavioral training effects showed improvement in word decoding accuracy and reading speed which generalized to novel words (92% accuracy). Participants also demonstrated orthographic learning of the target words in that they responded to them faster than to homophonous spellings of the same words (p = 0.006) in the word recognition task and the LDT (p = 0.0016). Analysis of imaging data showed that regions with significant activation during decoding of the target words vs. visual patterns largely overlapped with the Neurosynth’s reading network, including left SMA, bilateral clusters in the Middle/Inferior Occipital gyrus, Inferior/Superior Parietal lobule, fusiform area, and clusters in the right cerebellum (Lobules VI, VIII, Crus 1 and 2). Distinct brain activations not present in the Neurosynth’s reading network were found in the left cerebellum (Lobules VI, IX, X, Crus 1) and bilateral calcarine sulci (which correlated with the WAIS vocabulary measure and passage reading performance). Despite existing literature implicating fronto-temporal areas in phonological processes and letter-to-sound conversion[2,3,7] and in contrast with the Neurosynth’s reading network, we observed activation only in superior, but not inferior, portions of the left IFG (p.triangularis and p.opercularis) and no significant activation in p.orbitalis or in any of the temporal areas. Conclusion. Our findings highlight differences in the brain networks supporting skilled reading vs. novice reading and are consistent with the ideas that the nature of decoding changes over reading development[6]. Thus, the lack of significant activation in the frontal and the temporal areas may be indicative of decoding that has not yet become lexicalized.
Topic Areas: Reading,