Optimising episodic encoding within segmented virtual contexts

The encoding of episodic memories depends on segmentation; memory performance improves when segmentation is available and performance is impaired when segmentation is absent. Indeed, for episodic memories to be created, the encoding of information into long-term memory requires the experience of event boundaries (i.e., context-shifts defined by salient moments of change between packets of to-be-learned stimuli). According to this view episodic encoding, and therefore learning, is critically dependent on the nature of working memory. Motived by this theoretical framework, here we explore the effects of segmentation on long-term memory performance, investigating the possibility of optimising learning by aligning the presentation of stimuli to the capacity of working memory. Across two experiments, we examined whether manipulating the boundaries between events influences memory. Participants travelled within a virtual environment, with spatial–temporal gaps between virtual locations providing context-shifts to segment sequentially presented lists of words. Both accurate recall and memory for temporal order improve and the number of falsely recalled words reduces when reducing the quantity of information presented between boundaries. Taken together, the present results suggest that closely matching the quantity of information between boundaries to working memory capacity optimises long-term memory performance.