Reward Prediction Error and Declarative Memory. Issue 5 (May 2020)

• Record Type: Journal Article
• Authors: Ergo, Kate; De Loof, Esther; Verguts, Tom
• Is Part Of: Trends in cognitive sciences. Volume 24:Issue 5(2020)
• Page Start: 388
• Page End: 397
• Publication Date: 2020-05
• Subjects: reward prediction error -- declarative memory -- reinforcement learning; Cognitive science -- Periodicals; Cognitive neuroscience -- Periodicals; 153.05
• DOI: 10.1016/j.tics.2020.02.009
• Languages: English
• ISSNs: 1364-6613

Abstract

Abstract : Learning based on reward prediction error (RPE) was originally proposed in the context of nondeclarative memory. We postulate that RPE may support declarative memory as well. Indeed, recent years have witnessed a number of independent empirical studies reporting effects of RPE on declarative memory. We provide a brief overview of these studies, identify emerging patterns, and discuss open issues such as the role of signed versus unsigned RPEs in declarative learning. Highlights: The concept of reward prediction error (RPE) learning has been extremely influential in the nondeclarative memory literature, but the role of RPEs in declarative memory has only recently begun to be explored. Two main approaches to measure RPE in declarative memory have been developed. In the reward-prediction approach, reward is iteratively sampled by the experimenter from a statistical distribution and predicted by the subject. In the multiple-repetition approach, subjects see the same memoranda repeatedly and estimate their probability of success. Behaviorally, the reward-prediction approach has mainly yielded signed RPE signatures, whereas the multiple-repetition paradigm has mainly yielded unsigned RPE signatures. However, there are exceptions, and neural (electroencephalography, fMRI) paradigms have yielded both in single studies. An RPE perspective provides testable hypotheses on why reward and reward prediction manipulations may improve or impair declarative learning.