The context repetition effect: predicted events are remembered better, even when they don't happen.

A key function of the medial temporal lobe (MTL) is to generate predictions based on prior experience (Bar, 2009). We propose that these MTL-generated predictions guide learning, such that predictions from memory influence memory itself. Considering this proposal within a context-based theory of learning and memory leads to the unique hypothesis that the act of predicting an event from the current context can enhance later memory for that event, even if the event does not actually occur. We tested this hypothesis using a novel paradigm in which the contexts of some stimuli were repeated during an incidental learning task, without the stimuli themselves being repeated. Results from 4 experiments show clear behavioral evidence in support of this hypothesis: Participants were more likely to remember once-presented items if the temporal contexts of those items were later repeated. However, this effect only occurred in learning environments where predictions could be helpful.