Errors in visual search: How can we reduce them?

Abstract

Observers routinely make errors in almost any visual search task. In previous online experiments, we found that indiscriminately highlighting all item positions in a noisy search display reduced errors. Here, we conducted two eye-tracking studies to investigate the mechanics of this error reduction: Does cueing direct attention to previously overlooked regions or enhance attention/processing at cued locations? Displays were presented twice. In Experiment 1, for half of the displays, the cue was only presented on the first copy (Cue – noCue) and for the other half, only presented on the second copy (noCue – Cue). Cueing successfully reduced errors but did not significantly affect reaction times (RTs). This contrasts with the online experiment where the cue increased RTs while reducing errors. In Experiment 2, we replicated the design of the online experiment by splitting the displays into noCue – noCue and noCue – Cue pairs. We now found that the cue reduced errors, but increased RTs on trials with high-contrast targets. The eye-tracking data show that participants fixated closer to items and fixation durations were shorter in cued displays. The smaller fixation-item distance reduced search errors, where observers never fixated the target, for low-contrast targets and the remaining low-contrast errors seemed to be recognition errors, where observers looked at the target but too quickly looked away. Taken together, these results suggest the main reason that errors were reduced was because attention was more properly directed to overlooked regions by the cues. Enhancement of attention at the cued areas may have played an auxiliary role.