The role of context in segmentation and continuity

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

Shira Baror¹ (baror.shira@gmail.com), Aya Ben-Yakov¹; ¹Edmond and Lily Safra Center for Brain Sciences, The Hebrew University, Jerusalem, Israel

Human experience intertwines an ongoing sense of continuity with spontaneous segmentation into discrete events in memory. Nevertheless, what governs the fundamental interplay between continuity and segmentation is currently unknown. To answer this question, we leverage two well-established yet separately studied phenomena that showcase these ends: On the continuity end, serial dependence: the observed bias of current perceptual decisions towards preceding ones. On the segmentation end: the emergence of event boundaries following contextual change which are found to shape memory. From a Bayesian standpoint, serial dependence is associated with predictions, and segmentation with prediction failure, thus a shared account may underlie them both, and govern the continuity-segmentation balance. Here, we optimized a paradigm that tests serial dependence and event-boundary effects on memory in a single setup. Participants viewed tilted everyday objects, each surrounded by a colored circle frame which served as the objects’ context, and performed an object-orientation task, which classically exhibits serial dependence. This was followed by two memory tasks- associative memory for the color of the frame that surrounded each object, and temporal-order memory for object pairs. The results show that context change triggered event boundaries, leading to better associative memory at boundaries, as well as improved temporal order memory for pairs of items within events compared with across events. Complementarily, serial dependence decreased at event boundaries. These results provide evidence for a potentially shared context-based account for perception and memory, by which context stability promotes continuity, and context change triggers segmentation.

Topic Area: LONG-TERM MEMORY: Episodic