Abstract
When a visual search target is both salient (i.e., a singleton) and has a known defining feature (e.g., red), observers often search for the singleton rather than adopting a narrower attentional set tuned specifically to the target feature (e.g., redness) — even when the narrower set (feature search mode; Bacon & Egeth, 1994) would prevent interference from irrelevant distractors. One possible explanation for this bias is that the feature mode is difficult to maintain under conditions of distractor homogeneity. However, recent work by Leber and Egeth (OPAM 2002, 2003) has argued against this maintenance account. Here, we evaluate whether feature search mode is difficult to configure in homogeneous displays. To examine the configuration process, we asked observers to switch between two tasks in alternating runs of two trials (i.e., AABB, etc.): A) search for the “oddball” color in a rapid serial visual presentation (RSVP) of homogeneously colored non-targets (singleton search); B) search specifically for red among heterogeneously colored non-targets (feature search). In this design, performance for both tasks is worse on switch trials, reflecting the cost of reconfiguration. (Leber, 2003). Our study focused on the feature search trials (Task B), where we compared the switch cost in two conditions. On 50% of the feature search trials, non-target items were all heterogeneous in color; on the remaining feature search trials, the non-targets were initially homogeneous and did not become heterogeneous until 300 ms prior to the target. If configuration of feature search mode is more difficult with homogeneous displays, then task-switch recovery should be impaired in the homogeneous-onset trials. Results confirmed this expectation, suggesting that feature search is indeed difficult to configure under conditions of homogeneity. These results may explain why observers often search for singletons rather than specific features when both strategies are available.
Supported by NIH/NIMH MH070115