Is prediction automatic or adaptive during speech comprehension? An EEG decoding study

Poster Session B - Sunday, April 14, 2024, 8:00 – 10:00 am EDT, Sheraton Hall ABC

Melissa G. Jacuinde¹, Timothy G. Trammel¹, Tamara Y. Swaab¹, Matthew J. Traxler¹; ¹University of California, Davis

Prior studies suggest that speech processing is facilitated by prediction of imminent input. But it is unclear if predictive processing is automatic, as assumed by Bayesian models, or adaptive, as assumed by rational models. We examine this by manipulating the predictability of animacy of critical nouns in spoken sentence contexts (high or low constraint) and the proportion of sentences that reliably predict animacy (80% or 20%). If prediction is automatic, then animacy information should be retrieved in the high constraint condition prior to the critical nouns, regardless of the proportion manipulation. Activation of animacy features will be examined by decoding EEG prior to critical nouns. Facilitated processing of critical nouns as a function of (pre)-activation will be examined by comparing ERP animacy effects. EEG will be decoded in a 1000ms silence period prior to the critical nouns. If prediction is automatic, we expect significant above-chance decoding accuracy of animacy features for high constraint sentences in both probability conditions. If pre-activation of animacy features automatically facilitates processing of critical nouns, the constraint manipulation will modulate the size of the N400 animacy effect in both proportion conditions as well. However, if facilitated processing of the target nouns depends on the reliability of prediction in the global context of the experiment, we expect a reduced N400 animacy effect for high constraint sentences in the 80% relative to the 20% condition. This latter pattern of results would provide evidence that pre-activation of animacy features is automatic, but facilitation from prediction is not.

Topic Area: LANGUAGE: Semantic