Does prediction error play a differential role in updating recent versus remote episodic memories?

Poster Session E - Monday, April 15, 2024, 2:30 – 4:30 pm EDT, Sheraton Hall ABC

Catalina Mengyao Yang¹, Katherine Duncan¹, Morgan Barense¹; ¹University of Toronto

Mnemonic prediction errors (PE), or the mismatch between retrieved memories and reality, are thought to trigger memory updating. Yet, it remains unclear whether PE is required to update memories of all ages, and whether PE has distinct effects depending on the remoteness of a memory. There is also debate regarding whether recent or remote memories are more readily updated upon reactivation. We address these questions by investigating how reactivation procedures (with or without PE) and memory age (delay between acquisition and reactivation) interact to determine the updating of naturalistic episodic memories. In Session-1, we showed participants a series of videos featuring salient action-outcome events. In Session-2, participants saw the videos again when they returned after one of three delays (1-hour, 24-hours, or 2-weeks). Here, we elicited PE in half of the videos by abruptly interrupting the action-outcome contingency. Immediately following this, we introduced new interfering videos. In Session-3, we conducted a structured memory interview for Session-1 videos. We operationalize memory updating as the number of details from the new interfering videos incorporated into the original video memories. Preliminary data (n=34) demonstrate that our design is positioned to answer our question: 1) memory intrusions rates within each condition are sufficient for meaningful comparisons and are similar to previous reports, and 2) despite some forgetting, participants were able to remember most videos with high confidence during reactivation. Our study integrates theories of memory storage dynamics and prediction errors, and we expect our final results to elucidate interacting factors that influence memory updating.

Topic Area: LONG-TERM MEMORY: Episodic