Prediction Errors Disrupt Hippocampal Representations and Update Episodic Memories

Abstract The brain supports adaptive behavior by generating predictions, learning from errors, and updating memories to accommodate new information. Prediction error, or surprise, triggers learning when reality contradicts expectations. Prior studies have shown that the hippocampus signals prediction errors, but have never linked this neural signal to memory updating. Here, we uncover new mechanisms that reveal this missing link. In a human fMRI study, we elicited mnemonic prediction errors by interrupting familiar narrative videos immediately before the expected endings. We found that the same amount of hippocampal activation could exert opposing effects on memory: hippocampal activation preserved memories after expected endings, but updated memories after prediction errors. In contrast to previous studies, we showed that univariate activation was insufficient for understanding hippocampal prediction error signals. We explained this surprising finding by tracking the evolution of hippocampal activation patterns, and connectivity between the hippocampus and neuromodulatory regions. We found that hippocampal activation patterns stabilized as each narrative episode unfolded, sustaining episodic representations. Prediction errors disrupted these sustained representations, and the degree of disruption predicted memory updating. The relationship between hippocampal activation and subsequent memory depended on concurrent basal forebrain activation, providing new evidence about how cholinergic modulation may regulate attention and memory. We conclude that prediction errors create conditions that favor memory updating, prompting the hippocampus to abandon ongoing predictions and render memories malleable.