Exploring mode selection in visual search

Purpose: Bacon & Egeth (1994) have shown that subjects performing visual search tasks on static displays operate in one of two modes: 1) singleton detection mode (SDM), where salient items in the display gain attentional processing, or 2) feature search mode (FSM), where only items sharing a defining feature of the target (e.g., specific color or shape) are granted attentional processing. Folk, Leber, & Egeth (2000) also observed these search modes in a temporal selection task using rapid serial visual presentation (RSVP). The present experiments examined whether mode selection is established via executive control or if it is stimulus dependent. Methods: An RSVP stream containing a single target was presented at fixation in all experiments. RSVP was used in order to stimulate the visual system with non-targets for more than 1000ms prior to the target on each trial, allowing us to evaluate a short-term bottom-up account. To index the subjects' mode of search, the effect of irrelevant, peripheral, target- colored or non-target-colored singleton distractors presented prior to the target was examined. Previous work has shown that when non-target stream letters are homogeneous in color, subjects use SDM, while subjects use FSM when the non-target letters are heterogeneous (Folk et al., 2000). The current experiments used the homogeneous and heterogeneous trial types in within-subject designs, to determine if and when mode switching would occur. Results/Conclusions: When trial types were mixed within blocks, not only did subjects use FSM in the heterogeneous trials, they also used it in the homogeneous trials. Additionally, when a homogeneous block followed a heterogeneous block, subjects persisted in using FSM for both blocks. A stimulus-driven explanation was unable to account for these results. Additional experiments explored why subjects did not take advantage of SDM; it appears that subjects preferred staying in one mode to avoid switching costs.