Discrete memories of a continuous world: A working memory perspective on event segmentation

Abstract

We perceive the world in a continuum but remember our past as discrete episodic events. Dominant models of event segmentation suggest that prediction errors or contextual changes are the driving factors that parse continuous experiences into segmented events. These models propose working memory to hold a critical role in event segmentation, yet the particular functioning of working memory that underlies segmented episodic memories remains unclear. Here, we first review the literature regarding the factors that result in the segmentation of episodic memories. Next, we discuss the role of working memory under two possible models regarding how it represents information within each event. Lastly, we summarize the role of the hippocampus in segmenting and integrating events. Clarifying the contributions of working memory to event segmentation is important to improve our understanding of the structure of episodic memories.

Top-down attention shifts behavioral and neural event boundaries in narratives with overlapping event scripts

“…For example, events are described as arising from "contextual stability in perceptual features" 78 , "the spatiotemporal characteristics of the environment" 79 , or "how fast various aspects of information from the environment tend to evolve" 39 . Although the dynamics of the external environment are certainly a primary input to the event segmentation process, our findings instead favor a view of event boundaries as actively constructed in the mind and dependent on the prior knowledge and current goals of an individual 13,80 . Rather than simply inheriting the temporal dynamics of the environment, event segmentation is a cognitive process that optimizes the organization of continuous experience into the most relevant units.…”

“…At the top of this hierarchy, in regions traditionally characterized as the Default Mode Network (DMN), neural representations exhibit periods of relative stability (for tens of seconds) punctuated by moments of rapid change 5,6 . These dynamics align with a long-standing idea in cognitive psychology: that continuous experiences are segmented into discrete, meaningful events [7][8][9][10] , with boundaries between events serving as critical moments for updating mental models in working memory [11][12][13] and for determining the structure of episodic memory [14][15][16][17] .…”

Working memory capacity for continuous events: The root of temporal compression in episodic memory?

“…We intentionally used continuous events that do not contain event boundaries to investigate at what point the stream of experience starts to be temporally compressed in the current event model maintained in WM. Our results suggest that the information accumulation process underlying event model formation (see Güler et al, 2023;Lu et al, 2022) can be "lossy" because of WM capacity limits in representing continuous events. This, in turn, would lead to the formation of incomplete episodic memory representations when the time elapsed between two event boundaries is too long (Jeunehomme & D'Argembeau, 2020, 2023.…”

Effects of context changes on memory reactivation

“…We experience the world continuously but perceive and store events that are segmented into meaningful parts (Zacks & Swallow, 2007). This segmentation in episodic memory reflects changes in contextual features such as background color, sound, location, or task rules (Clewett et al, 2017;Heusser et al, 2018;Horner et al, 2016;Raccah et al, 2022;Wang et al, 2022;Güler et al, 2023;Güler et al, 2024;Nolden et al, 2024 ).…”

“…For example, retrieval is more successful when it shares the same context with encoding (Godden & Baddeley, 1975;Robin et al, 2016;Tulving & Thomson, 1973). Changes in context also determine the temporal structure of episodic memories by generating event boundaries at context transitions (e.g., Clewett et al, 2017;Wang et al, 2022;Güler et al, 2023;Güler et al, 2024;Nolden et al, 2024). These findings suggest that context strongly influences how an item is preserved in LTM.…”