Poster F90, Tuesday, March 28, 8:00 – 10:00 am, Pacific Concourse
Signed reward prediction errors drive declarative learning
Esther De Loof1, Kate Ergo1, Lien Naert1, Clio Janssens1, Filip Van Opstal2,3, Tom Verguts1; 1Ghent University, Belgium, 2Université Libre de Bruxelles, Belgium, 3University of Amsterdam, Netherlands
Reward prediction errors (RPEs) are thought to drive learning. This has been firmly established in procedural learning paradigms (e.g., classical and operant conditioning). However, empirical evidence on whether RPEs drive declarative learning – a quintessentially human form of learning – remains surprisingly absent. In this study, we used a declarative learning paradigm in which RPEs were coupled to the acquisition of Dutch-Swahili word pairs. The occurrence of signed RPEs (SRPEs; “better-than-expected” signals) during declarative learning improved recognition performance in a follow-up recognition test, with increasingly positive RPEs leading to better recognition. In addition, we demonstrate that classic declarative memory mechanisms such as time-on-task fail to explain recognition performance. The beneficial effect of SRPEs on recognition performance was subsequently affirmed in an EEG replication study. Moreover, we found oscillatory (high-beta and high-alpha) signatures for SRPEs during reward feedback, similar to SRPE signatures found in earlier procedural learning paradigms. Importantly, these results offer a powerful reinterpretation of the testing effect, with key implications for education.
Topic Area: LONG-TERM MEMORY: Episodic