In sum: It’s not surprising changes, but contextual stability, that determines how we segment continuous experience into discrete events.

Across three experiments, we manipulated contextual stability while keeping prediction errors constant. In a separate experiment, we manipulated prediction errors while holding the context stable. To assess event segmentation, we used temporal distance and temporal order tasks.

Across 4 experiments, we both manipulated contextual stability and prediction error. Findings consistently showed that contextual stability, not prediction error, better accounts for how people segment continuous experience into memory units.

Corrected link: osf.io/preprints/ps...

Our study examined how working memory (WM) supports event segmentation—whether it accumulates information during event comprehension or reactivates items at boundaries. We found evidence for both, suggesting that WM plays a functional role in both processes

So proud of you✨✨

Many thanks to @erengunseli.bsky.social for providing great mentorship, and to Zeynep Adıgüzel and Bilge Uysal for their support and roles in developing super cool ideas ✨

We also propose possible underlying mechanisms and experimental predictions to test these models