Abstract
During scene viewing, the eyes and attention are directed serially to objects of interest, and visual short-term memory (VSTM) is used retain information from recently attended objects. In the present study, we examined whether people can strategically control the contents of VSTM during scene viewing, retaining task-relevant objects in VSTM even as attention and the eyes are directed to subsequent objects. Stated more generally, is VSTM a passive storage system in which new objects automatically displace older objects, or can people strategically protect task-relevant objects from subsequent interference? Participants viewed a set of real-world objects presented serially within a 3-D rendered scene. One object in the sequence was cued by a tone as “to-be-remembered”. At the end of the sequence, memory for the visual form of one object was tested, with the cued object six times more likely to be tested than an uncued object. First, there was a general recency effect, with highest memory performance for the most recently attended objects. Second, objects at a particular serial position were retained more accurately when cued than when uncued, demonstrating that participants could indeed prioritize task-relevant objects for retention. Third, the advantage for a cued object was observed at the expense of memory for subsequent objects in the sequence, consistent with the strategic use of limited VSTM resources. Finally, cued objects early in a sequence were remembered less accurately than cued objects late a sequence, demonstrating some degree of interference. Thus, participants could prioritize task-relevant objects for retention in VSTM, but protection from subsequent interference was not perfect. Strategic maintenance of objects in VSTM could play an important role in real-world visual behavior, especially when object information must be maintained across shifts of attention and the eyes to other objects (such as when comparing two spatially separated objects).