Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Sketchpad Series 

Two wrongs make a right? Electrophysiological and behavioural comparisons of reward prediction error and prediction error in memory consolidation

Poster Session C - Sunday, March 8, 2026, 5:00 – 7:00 pm PDT, Fairview/Kitsilano Ballroom

Annaliese Anesbury¹ (), Alex Chatburn¹, Karsten Rauss², Ina Bornkessel-Schlesewsky¹; ¹Adelaide University, Adelaide, Australia, ²University of Tuebingen, Germany

Previous research has examined reward prediction errors (RPEs) in reinforcement learning and semantic prediction errors (PEs) in language and perception, yet no study has directly compared their EEG-derived indices or their relationship to declarative memory encoding and consolidation within a common framework. While both types of prediction errors have been independently linked to modulations in theta oscillations, and changes in negative event-related potential amplitudes (feedback-related negativity, FRN, and N400, respectively), the extent to which they reflect distinct or overlapping neural mechanisms and how they influence declarative memory remains unclear. In this study in progress, participants (planned N=80) are completing a novel combined semantic-reward oddball-like roving paradigm (based on Lindborg et al., 2023), in which category shifts in both nouns and reward values are expected to independently elicit PEs and RPEs. EEG and behavioural data are being recorded during learning, immediate recognition and a delayed recognition phase, occurring 2 hours and 45 minutes later, testing recognition accuracy for the nouns using an old/new paradigm. It is expected that RPEs and PEs will elicit FRN and N400 components, respectively, and that stimuli inducing prediction errors will show downstream memory advantages. Linear mixed models will use single-trial FRN and N400 amplitudes, along with theta phase synchrony, as predictors of delayed recognition accuracy to determine whether the effects of RPEs and PEs on memory are additive, indicating separable systems, or interactive, suggesting overlapping systems. Preliminary findings are expected to clarify the degree of separability versus shared nature within predictive processing systems.

Topic Area: LONG-TERM MEMORY: Semantic