Abstract
Rajsic, Wilson, & Pratt (2015; 2017) recently discovered a visual form of "confirmation bias" wherein people are biased to seek cued objects in visual search, even when this strategy is inefficient. Participants searched for a target letter in simple circular displays of colored letters of varying proportions. Critically, only one color was initially cued (e.g., "Press Z if the p is green, otherwise press M."), and all letters occurred in either the cued or in an uncued color. The ostensibly most efficient strategy was to restrict search to the minority color (e.g., green if 25% of letters were green; red if 25% were red) and use inference where necessary, obviating the need to actually inspect the target if it did not occur in the minority color, since a target was present in every trial. People rarely adopted this strategy, and prioritized inspections of cue-colored objects. However, when Rajsic et al. (2017) increased the cost of each inspection (e.g., with occluded, gaze-contingent letters), people adopted the efficient strategy. In the current paradigm, we increased the cost of inspections by presenting (1) randomly scattered "traditional" visual search displays and (2) increasing the set size. We also investigated whether varying the prevalence of cue-colored targets would encourage a more efficient strategy. We observed typical prevalence effects in search accuracy. However, search RTs suggested that people adopted a more efficient search strategy, especially as set size increased. These results suggest that while the "default" strategy of simply searching for what is most cognitively available is typically efficient, increasing the cost of inspections encourages more careful and deliberate deployment of attention.
Meeting abstract presented at VSS 2018