Abstract
<p>Schemas, knowledge structures that represent regularities across episodes, can enhance memory for events that are either congruent or incongruent with their content. Across four experiments including 193 university students, we investigated whether these schema effects emerge when chronology forms a core aspect of a schema’s content, enabling temporal predictions of future events. After pre-training on schematically defined four-item sequences, participants studied novel sequences that either affirmed, violated, or were unrelated to these schemas. Memory benefits were observed for both schematically congruent and incongruent sequences relative to sequences that were schema-unrelated. Critically, incongruent memory benefits only emerged for violations of content, and not for violations of temporal order, despite behavioural evidence of prediction error at encoding for both violation types. Our findings show that schema effects depend on the type of schematic violation and that incongruent benefits are not the inevitable consequence of prediction error at encoding.</p>