Abstract
Based on the large number of objects in any visual scene, and our ever-changing goals, it is ecologically natural for humans to dynamically adjust which information and items are stored in Visual Working Memory (VWM) across views. This means that effectively using VWM in context requires observers to e.g., load and purge items from VWM, switch attention to new items, compare and make decisions, and so forth. Here, we explore these dynamic aspects of VWM, and we identify independent limits on the dynamic manipulation of information in VWM - distinct from VWM storage limits (K). In contrast to the more typical One-Shot change-detection task that has been used to estimate VWM storage capacity (K), we developed a Flicker method that separately estimates storage and dynamic processes. Participants viewed alternating displays of many colored squares separated by a blank where one square changed colour on each iteration. To find the changing target, participants had to not only store items, but had to employ dynamic processes, such as loading and purging items from memory, switching attention, making decisions, etc. Response times were transformed into an estimate of storage capacity (K), as well as an additional component representing these dynamic processes (∆). In a series of experiments, we demonstrate that Flicker K estimates are more reliable than One-Shot K estimates (Exps 1-6), that Flicker K correlates with One-Shot K, while Flicker ∆ remains independent (Exp 7), and that Flicker K and ∆ increase (i.e., improve) during the early elementary school years (Exps 8-9). Our approach and results place a renewed focus on the dynamic requirements of using VWM in context, and demonstrate the importance of incorporating both storage (K) and manipulation (∆) in models of VWM.
Meeting abstract presented at VSS 2014