Prediction and encoding are coupled, not competitive

Forming new memories requires a focus on the external world; retrieving existing memories requires attention to our internal world. Computational models propose that the hippocampus plays a central role in resolving the tension between encoding and retrieval by alternating between states that prioritize one over the other. We tested two opposing hypotheses about whether encoding and retrieval states in the hippocampus are associated with a behavioral trade- off or behavioral coupling between encoding and retrieval, with retrieval operationalized as the use of existing memories to make predictions about the future. Across 3 Experiments (N = 197), participants viewed a series of scene categories that contained structure (e.g., beaches are followed by castles); this structure enabled memory retrieval to guide prediction. After structure learning, participants completed a simultaneous prediction and encoding task. They were shown trial-unique category exemplars and asked to make predictions about upcoming scene categories. Finally, they were given a surprise memory test for the trial-unique images. This allowed us to test whether there was competition vs. coupling between encoding of trial-unique images and prediction of upcoming images. We consistently observed coupling between encoding and prediction: accurate predictions were associated with better encoding, and factors that hurt prediction also hurt encoding. This coupling could not be explained by generic on- vs. off-task states. Together, these findings suggest that distinct encoding and retrieval states in the brain need not be associated with a tension between encoding and prediction in behavior; instead, these cognitive states may be cooperative rather than competitive.