Evidence for salience-guided search in a fine-localization task

Abstract

When is search for a target interrupted by unusual distractors? Recent work on this question is equivocal, with some studies finding distraction for both task-relevant and task-irrelevant stimulus dimensions, and other studies finding distraction effects only for task-relevant dimensions. The present study seeks to clarify this discrepancy by utilizing a novel point-and-click task asking participants to indicate the precise location of a target (an oblique line amongst vertical lines) using a computer mouse. Using spatial accuracy as a dependent variable, and displaying stimuli for only short durations, results from Experiment 1 show that accuracy is disrupted significantly more when a unique distractor is defined by a feature in a task-relevant (horizontal line) rather than task irrelevant (red vertical line) dimension. Experiments 2 and 3 looked at the effect of vertical red-line distractors when modulated in terms of frequency and salience, respectively, and found no effect of these modulations, but a significant effect of unique distractor presence versus absence. On the face of it, these results are irreconcilable with a dual-route account that assumes selection of a single dimension prior to search, and instead support an attention-contingent search guided by target salience (a la Wolfe's Guided Search Theory). In other words, it appears that fine localization not only requires focal attention but occurs through a summed activation map which takes input from multiple feature dimensions. Meeting abstract presented at VSS 2016