The role of context in continuity and segmentation

“The role of context in continuity and segmentation”

Human experience intertwines continuity, the seamless flow of events, with segmentation, the spontaneous partitioning of experience into discrete units.
Despite their cognitive significance, it is unclear whether these processes operate independently or share a common mechanism. Here we explored this question by examining the link between serial dependence (SD)—the tendency to align perceptions with prior decisions—and the impact of event boundaries, prompted by contextual shifts, on memory. Considering a Bayesian perspective that associates SD with predictions, and segmentation with prediction errors, a common mechanism may govern both processes.
Across three experiments, we tested how contextual changes affect SD and segmentation-related memory effects. The results indicate that both processes are context-sensitive. Contextual boundaries can reduce SD even in the absence of sensory change, and boundaries modulate memory in ways consistent with event segmentation. Yet, across experiments we observed dissociable patterns of boundary effects on SD and memory, which are slightly more consistent with distinct contextual influences on perception and memory than with a single unified predictive system.
Boundary effects on SD and memory did not covary across participants, but given the low within-participant reliability of these measures, this absence of correlation cannot be taken as strong evidence for independence.
Overall, our findings show that both SD and memory are shaped by context, but clarifying whether they reflect a shared or partly distinct mechanism will require further research.

Our findings reveal that SD can be modulated by contextual boundaries even in the absence of sensory change, and that temporal order memory is modulated by boundaries but not by sensory change, whereas associative memory shows the opposite pattern.
Boundary-related changes in SD and memory did not correlate across participants, but given the limited within-participant reliability of these measures, this lack of correlation cannot be taken as strong evidence regarding whether their contextual modulation relies on shared or distinct mechanisms.
Taken together, the results indicate that context shapes both the maintenance of perceptual and conceptual stability and the segmentation of experience into discrete, memorable events, while leaving open the question of whether a single mechanism or distinct mechanisms underlie these contextual influences.