Presentation
Search Abstracts | Symposia | Slide Sessions | Poster Sessions | Lightning Talks
Can Prediction Error Explain Predictability Effects on the N1 during Picture-Word Verification?
There is a Poster PDF for this presentation, but you must be a current member or registered to attend SNL 2023 to view it. Please go to your Account Home page to register.
Poster A119 in Poster Session A, Tuesday, October 24, 10:15 am - 12:00 pm CEST, Espace Vieux-Port
Jack Taylor1,2, Guillaume Rousselet2, Sara Sereno2; 1Department for Psychology, Goethe University Frankfurt, 2School of Psychology and Neuroscience, University of Glasgow
Predictive coding models posit that brain activity scales with prediction error: the difference between ascending (bottom-up) input and descending (top-down) predictions. Prediction error captures two key variables: magnitude and certainty of the ascending-descending difference. Greater violations of expectations should elicit greater prediction error signals, and this effect should become larger as certainty increases. We asked whether such a simple predictive coding account could describe neural activity indexed by the N1 (~170 ms) event-related potential component elicited by words. Indeed, findings have shown that the word-related N1 is sensitive to predictions, with unpredicted words generally eliciting greater-amplitude N1s than predicted words. However, effects of error magnitude and certainty have mostly been investigated in isolation, providing incomplete tests of predictive coding. In our pre-registered study, we tested the account via the interaction between prediction congruency (error magnitude) and predictability (certainty). We recorded electroencephalograms for 68 participants while they completed a picture-word verification paradigm. PICTURE-word pairs were congruent (e.g., ONION-onion) or incongruent (e.g., ONION-torch), while predictability was manipulated continuously based on norms of picture-name association (% name agreement). Pre-registered analyses failed to find evidence that the direction of the congruency-predictability interaction matched that expected under a simple predictive coding account. Exploratory Bayesian analyses found strong evidence against the account, with the congruency-predictability interaction 59.98 times more likely in the opposite direction. Specifically, higher predictability elicited larger N1s for picture-congruent words, and smaller N1s for picture-incongruent words. We argue that a simple predictive coding account of the N1 is either incorrect or requires elaboration.
Topic Areas: Reading, Control, Selection, and Executive Processes