Did that just happen? Event segmentation influences enumeration and working memory for simple overlapping visual events

For working memory to be efficient, it is important not only to remember, but also to forget-thus freeing up memory for additional information. But what triggers forgetting? Beyond continuous temporal decay, memory is thought to be effectively 'flushed' to some degree at discrete event boundaries-i.e. when one event ends and another begins. But this framework does not readily apply to real-world visual experience, where events are constantly and asynchronously beginning, unfolding, and ending all around us. In this rush of things always happening, when might memory be flushed? In a series of experiments, we explored this using maximally simple visual events. A number of dots appeared, a subset moved at random speeds in random directions, and observers simply had to estimate the number of dots that moved. Critically, however, these motions could begin and end asynchronously. In general, asynchronous motions led to underestimation, but further experiments demonstrated that this was driven only by endings: regardless of whether dots started moving together or separately, animations with asynchronous endings led to underestimation-even while carefully controlling for both the overall amount of motion and average starting and ending times. (In contrast, no such effect occurred for asynchronous beginnings.) Thus, the ends of events seem to have an outsize influence on working memory-but only in the context of other ongoing events: once a motion ends amidst other unfinished motions, it seems more difficult to recall that particular motion as having occurred as a distinct event.