Malleability of event boundaries in autobiographical memory

Action sequences are typically perceived as events with clear beginning and end points. From person to person, there is general agreement as to what is perceived as the boundary of a given event. Some have theorized that humans use event models to make constant predictions about ongoing action (Reynolds, Zacks, & Braver, 2007). When these predictions fail, an event boundary is marked. In memory, it has been found that information that occurs at a perceived event boundary is better recalled than non-boundary information. Yet, given the reconstructive nature of remembering, very little work has gone into understanding how stable event boundaries are in memory. The current project looked at how events are perceived in autobiographical memory, and whether boundary locations could be altered when participants were specifically probed about these boundaries. Results indicated that event boundaries were quite flexible. Flexibility scores were highest for events that were far enough from encoding that participants could not retrieve all the specific details of the event, yet close enough to recall that all contextual detail had not yet been lost. We suggest that this may indicate that boundary locations are not as important in memory as the gist understandings of events that guide reconstruction.