ABSTRACT
Visual working memory (VWM) maintains a subset of visual representations that are relevant to ongoing tasks. Visual attention selects a subset of currently visible stimuli for further perceptual processing, memory consolidation, and action. Interactions between VWM and visual attention are central to theories of attention and visual search, with VWM maintaining template properties for the guidance of attention. Here, I discuss three lines of research probing the architecture of this interaction. First, I review work suggesting that the attentional guidance provided by VWM is inherently positive: VWM can enhance the salience of matching stimuli, but it cannot be configured to reduce the salience of matching stimuli. Second, I review evidence regarding representational states in VWM that do and do not interact with attention, concluding that guidance can be flexibly limited to a subset of items in VWM and that this “template” state can encompass multiple objects. Finally, I review recent work indicating that although VWM is structured by objects and locations, these structural properties do not appear to manifest in the guidance of attention, suggesting an architectural division between object coding in VWM and mechanisms of attention guidance.
