Search Abstracts | Symposia | Slide Sessions | Poster Sessions
Neural ramping dynamics scale with relative progress in sentences, not elapsed time
Poster Session F, Friday, October 2, 2:45 - 4:45 pm, Wangari Maathai
Leonardo zeine1, Peter Donhauser1, Sonja Kotz2, David Poeppel3; 1Ernst Strüngmann Institute for Neuroscience, 2Maastricht University, 3New York University
Sentence comprehension unfolds over highly variable timescales. It is unclear whether sentence-level neural dynamics track elapsed time or the relative progress of sentential structure building. Prior work reported ramping activity during sentence processing, but most studies relied on fixed sentence length or presentation time, which does not allow distinguishing between absolute and relative temporal coding. Here, we analysed MEG recordings from 71 participants listening to natural Dutch sentences and wordlists (MOUS dataset; Schoffelen et al., 2019). Using a data-driven encoding approach, we isolated sustained and transient components spanning the full trial and tested whether sentence-long dynamics were better explained by absolute duration or by relative sentence progress. A dominant sustained component showed logarithmic ramping throughout sentences, was absent in wordlists, and reached a similar maximum amplitude regardless of sentence length. Word-level analyses showed that this component was driven primarily by the relative position in a sentence rather than lexical information or prosodic boundaries. Other sustained components were more strongly related to syntactic and prosodic features. Model comparison confirmed that a relative-time account explained the MEG data better than an absolute-time account. These findings indicate that sentence comprehension is organized by a temporally scaled neural trajectory that tracks progress toward sentence completion rather than elapsed time.
Topic Areas: Speech Perception, Methods