Episodic events are flexibly encoded in both integrated and separated neural representations

Abstract Remembering everyday events involves noticing what different experiences share and preserving the details that set them apart, yet the neural processes supporting this balance remain unclear. Here, we record EEG while participants view naturalistic movie scenes that introduce episodic events with overlapping elements. Using time-resolved representational similarity analysis, we find that these events evoke both similarities and dissimilarities in neural patterns as new information unfolds. Similarities predict successful inference of information across separate episodes, consistent with integrative encoding. Dissimilarities, by contrast, predict accurate memory for individual events, indicating the formation of distinct event-specific traces. Together, these findings indicate that the brain encodes both integrated and separated neural representations to flexibly support different mnemonic goals and to balance relational inference with detailed recollection.