Abstract
Many activities depend on directing attention to several goal-relevant objects (the butter, the knife, etc.), requiring flexible configuration of attentional guidance. Visual working memory (VWM) fulfills this role by maintaining features of the current target object. Although this basic framework is well accepted, there is debate over the architecture of guidance. Several researchers have claimed that only one feature from a dimension can interact with selection, whereas others hold that multiple features can interact simultaneously with selection. Supporting the former view, van Moorselaar et al. (2014) reported that memory-based capture (the delay produced by a memory matching distractor during visual search) was eliminated when participants held more than one color in VWM. They argued that multiple features fail to guide attention simultaneously and that such maintenance might eliminate memory-based guidance entirely. In the present study, participants held one or two colors in memory, searched for an orientation-defined target, then responded to a memory probe. In the search display, two colored distractors were present on some trials, with either both, one, or neither distractor matching a remembered color. The critical contrast was between a) one remembered color, one matching distractor and b) two remembered colors, two matching distractors. If only one feature can interact with selection, capture magnitude should be limited to that generated by a single color match; if multiple features can interact with selection, capture could plausibly increase with an increasing number of color matches. We found robust memory-based capture when either one or two colors was held in VWM. Critically, the magnitude of the capture effect was larger with two matching distractors than with one, indicating that both features interacted with selection. In contrast with van Moorselaar et al. (2014), these data provide clear evidence that multiple feature values can interact with and guide attention simultaneously.
Meeting abstract presented at VSS 2015