Abstract
Often, visual search experiments involve single-target (ST) search: Observers look for one target embedded among distractors. But multiple-target (MT) search is ubiquitous; consider collecting your keys and wallet before departing from home. Importantly, MT search incurs speed and accuracy costs, relative to ST search: Observers’ RTs are slowed, they make more (and longer) fixations, and they are more likely to miss and false-alarm (Hout & Goldinger, 2011a). Paradoxically, this increased workload creates more robust incidental memory for distractors encountered while viewing (Hout & Goldinger, 2010). In the current investigation, we employed a rapid-serial visual presentation task (Williams, 2010), wherein participants maintained a varying number of targets in working memory (WM) while passively viewing streams of 24 briefly presented images (centrally displayed, one at a time). This task allowed us to ensure equal encoding opportunities across conditions. Moreover, the target appeared in each quartile of the stream (early, mid-early, mid-late, late) equally often. After search, we administered a surprise, 2AFC recognition memory test for items previously seen; foils were semantically matched. Search accuracy was better under ST conditions, relative to MT, and people located the target more accurately when it appeared late in the stream, relative to earlier. With respect to incidental memory, recognition performance was better for items encountered under MT conditions, and was better for items that tended to appear after the target, relative to before it. Finally, we explored pupil diameters in order to examine attentional vigilance and its relation to subsequent memory formation (Papesh, et al., 2011). The results are discussed with respect to our preliminary model of learning in visual search (SQuEaL; Hout & Goldinger, 2011b). We suggest that MT search may involve a strategy shift, wherein cognitive resources are transferred from maintaining potential target representations in WM to the faithful encoding of incoming search items.
Meeting abstract presented at VSS 2012