Event Segmentation and Memory: Optimising Episodic Encoding within a Virtual Environment

Abstract

This thesis describes ten experiments exploring the role event boundaries play in the transfer of information from short- to long-term memory. In doing so the thesis explores the interaction between working memory and episodic memory, asking whether it is possible to optimise encoding using experimentally imposed event segmentation. The studies were inspired by the Method of Loci, where participants are trained to use visuo-spatial imagery as a strategy to enhance memory. Here, however, we employed an innovative self-made event sequencing virtual environment that allows the boundaries between to-be-remembered words to be imposed and manipulated during learning. Previous work suggests that the presence of doorways may signal an event boundary, providing structure that enhances memory transfer. An initial set of experiments identified a memory improvement effect when word lists were segmented via the presence of doorways, but similar improvements were found when segmentation was achieved solely via gaps in space or time. A second set of experiments explored the possibility that memory can be optimised by manipulating the quantity and domain of information presented between event boundaries. Findings revealed that both overloading (by presenting highly imageable words between boundaries) and under loading (by limiting visual-spatial information between boundaries) working memory had a detrimental impact on memory. Taken together the results suggest there is a time- and item limited ‘Goldilocks zone’ that optimally supports the transfer of information into long-term memory. More broadly, the data confirms that working memory is a distinct system, and the pattern of boundaries experienced between events directly affects the likelihood of successfully transferring information into long-term memory. Findings are discussed in relation to current theories of working memory and event segmentation, highlighting that long-term learning involves the simulation of events within working memory, and demonstrating that virtual learning environments can be used to create event boundaries and enhance memory.