The Nature of Working Memory: Separate, Flexible Resources for Location- vs. Feature-based Representations in Visual Short-term Memory

• Main Author: Adamo, Maha
• Other Authors: Ferber, Susanne
• Language: en_ca
• Published: 2012
• Subjects: visual short-term memory; attention; 0633
• Online Access: http://hdl.handle.net/1807/33844

Working memory, or the ability to maintain and manipulate information in mind when it is no longer physically present, is a pervasive yet severely capacity-limited component of cognition. Visual working memory, also known as visual short-term memory (VSTM), is limited to three or four items on average, with individual differences that range from roughly two to up to six items. Despite agreement that capacity is functionally limited, the current literature is split with respect to the nature of VSTM representations on two key questions: (1) What information is maintained in VSTM? (2) How is information stored in VSTM? The studies presented here address these questions using an event-related potential (ERP) task and a series of behavioural experiments that incorporate attentional selection via retrospective cueing (retro-cues). Experiment 1 manipulated both the number of features and the number of locations to be remembered in a lateralized change-detection task, with differences in the amplitude and topography of the resulting contralateral delay activity (CDA) indicating separate stores for features and locations. Experiments 2a-c established the basic effects of retro-cues on change-detection tasks, showing that attentional selection operated on one system at a time, with overall shorter response times and increased capacity estimates once baseline capacity was exceeded. Experiments 3a-b demonstrated that retro-cues biased VSTM resources to the cued item at the expense of representational strength of the other, non-cued items, showing flexible reallocation of resources. Experiments 4a-b presented multiple retro-cues to further examine the flexible reallocation of resources in VSTM, showing that capacity benefits depended on spatial specificity of retro-cues and that VSTM weights could be reallocated multiple times before probe comparison. Experiment 5 discounted the potential role of a general alerting mechanism in boosting capacity estimates, showing again that the retro-cue benefit required specificity of the cue. Experiment 6 showed that flexible reallocation of resources within one system did not change the online maintenance of representations within the other system. Thus, the studies collectively address the questions of (1) what and (2) how information is stored by supporting a two-system model of VSTM in which (1) locations and features are stored (2) independently via flexibly allocated resources.