Not so fast! Top-down predictions do not affect the earliest stages of visual word processing
Poster Session E - Monday, March 31, 2025, 2:30 – 4:30 pm EDT, Back Bay Ballroom/Republic Ballroom
Trevor Brothers1, Tamara Swaab2,3, Matthew Traxler2,3; 1North Carolina A&T State University, 2Univeristy of California, Davis, 3C Davis Center for Mind and Brain
Prior electrophysiological studies have suggested that linguistic prediction can influence the earliest stages of visual word processing, within 200ms of word onset. However, the timing and polarity of these prediction effects have varied widely across studies, leading some researchers to question the validity of these findings (Nieuwland, 2019). Here, we investigated ERP effects of linguistic prediction in four reading comprehension experiments (N = 155) and an active prediction task (N = 51). In all experiments, predictable and unpredictable words triggered robust visual responses over occipito-temporal sites (P1, N1), but the amplitude of these components did not differ as a function of predictability. Furthermore, mass univariate analyses revealed no significant predictability effects in early windows (0-200ms), despite the considerable statistical power of the current experiments. Instead, the neural effects of word predictability emerged at a slightly later time point (200-300ms), likely reflecting the benefits of contextual predictions on lexico-semantic access. Our findings suggest that earlier stages of visual decoding are impermeable to linguistic context, which places important constraints on predictive coding accounts of visual word recognition.
Topic Area: LANGUAGE: Other