Event segmentation and encoding rely on information accumulated over different temporal scales.

People spontaneously divide experience into discrete units (event segmentation) that shape episodic memory. Although it is generally assumed that both event segmentation and episodic memory reflect the generation and maintenance of event models (mental representations of the ongoing situation), the degree to which these two aspects of cognition reflect the same processes is unclear. We examined this question by disrupting the ability to accumulate coherent information over time and evaluating the effect on both processes. We asked participants to segment and afterwards recall video excerpts that were presented uninterrupted, interrupted by a white-noise display every minute, or interrupted and jumbled by playing the 1-min-long video clips out of order. Interrupting or jumbling videos did not significantly change overall segmentation rates, patterns, or the tendency to hierarchically organize events. However, a finer-grained temporal analysis indicated that the segmentation rate briefly increased following a jumbled clip, stabilizing after 5 s. These findings suggest that content-based event segmentation stabilizes quickly following jumps in the narrative. In contrast, jumbling the videos every minute significantly reduced the number of events recalled and their organization with respect to time and semantics. Thus, event segmentation and memory encoding are differently impacted by disruptions in an ongoing narrative. Whereas event segmentation appears to rely on information accumulated over a relatively short timescale, episodic memory depends on the ability to integrate information over longer time periods.