Presentation

Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Poster Slams

Examining visual word recognition and composition in natural reading with eye movements: A co-registered MEG and eye-tracking study

Poster C18 in Poster Session C, Friday, October 7, 10:15 am - 12:00 pm EDT, Millennium Hall

Graham Flick^1,2, Liina Pylkkänen^1,2; ¹New York University, ²NYUAD Research Institute

Eye movements during reading are influenced by the linguistic and cognitive demands of what is being read, indicating that the brain networks controlling eye movements, and those controlling the recognition and integration of individual words, must cooperate. Historically, however, neuroscientific studies of reading have tended to present stimuli in rapid serial visual presentation (RSVP), eliminating eye movements. This makes it unclear if neurocognitive accounts of reading, when informed by RSVP studies, generalize to the natural behavior of interest. Co-registered eye-tracking and electroencephalography (EEG, e.g., Kretzschmar et al., 2015), or functional magnetic resonance imaging (fMRI, e.g., Henderson et al., 2015), studies have begun to address this issue, but each suffers from a weakness in spatial (EEG) or temporal (fMRI) resolution. Magnetoencephalography (MEG), when source estimation is informed by structural MRIs, is a promising alternative, providing high temporal and relatively high spatial resolution to characterize neural activity during reading. In this work, we are using co-registered MEG and eye-tracking to examine visual word recognition in natural reading and directly contrasting it, within the same individuals, with word recognition in RSVP. In primary analyses, we are examining the influence of predictability and lexical frequency on neural responses to words, localized to occipitotemporal areas that support visual word recognition. Participants in this study complete three components while concurrent eye-tracking and MEG data are collected: (1) a localizer to isolate stages of visual word recognition in occipitotemporal cortex (Gwilliams et al., 2018; Flick et al., 2021); (2) a factorial design manipulating frequency and predictability of words within 432 target sentences, read naturally and in RSVP; and (3), a story-reading task consisting of short stories from the Natural Stories Corpus (Futrell et al., 2021), presented one paragraph at a time and read naturally. With synchronized pulses and a photodiode to ensure accurate timing, we are extracting MEG responses following the onset of individual fixations and using distributed minimum-norm estimation to localize these responses to each individual’s cortical surface. Each component of the MEG protocol is then designed to test specific hypotheses concerning these responses. First, the initial localizer will be used to examine the consistency of evoked responses and assess how these may differ between sentence reading in RSVP and reading with eye movements. Second, the factorial manipulation of predictability and frequency will be used to examine correlates of top-down and bottom-up influences on visual word processing, with a particular focus on whether activity localized to the left fusiform gyrus, proposed to house a bottleneck on visual word recognition (Woolnough et al., 2021), can explain variability in fixation times. Lastly, the story reading task will provide a large sample of fixation-related response data, localized to the cortical surface, which will be used to compare different operational definitions of predictability (e.g., word number, cloze, linguistic surprisal) and their correlates. In this way, our approach aims to provide a highly informative and comprehensive dataset of brain responses during natural reading, which we hope can be used to test numerous extant hypotheses concerning the brain basis of reading.

Topic Areas: Reading, Methods