Abstract
Highly salient objects, such as flashing, brightly colored hazard signs, typically attract attention. Salience-based models of attention predict that the probability of object selection increases as a monotonic function of perceptual salience. Accordingly, previous visual search studies have demonstrated that high-salience distractors lengthen target selection times compared to low-salience distractors. However, with this discrete measurement, it is unclear how distractor suppression evolves over time and whether perceptual salience of distractors modulates this time course. To address these questions, we developed a novel chronometry paradigm to probe the temporal evolution of distractor suppression for varying salience. Participants localized a shape-defined target with a keypress while trying to ignore salient color singleton distractors appearing on approximately half of all trials. There were two singleton distractor colors: low feature contrast (LFC, pink) and high feature contrast (HFC, blue), while all other objects were rendered in red. On each trial, three consecutive beeps were presented at intervals of 500 ms. The first beep signaled the start of a trial, followed by a search array at various stimulus-onset asynchronies (SOA, 100-700 ms, sampled every 100 ms). Participants were asked to respond synchronously with the third beep that occurred at 1000 ms. Proportion correct served as a measure of the effectiveness of distractor suppression as a function of SOA. We demonstrated that while distractor salience did not differentially interfere with target selection at short SOAs (< 600 ms), HFC distractors produced greater attentional interference than LFC distractors in mid SOA trials (~700 ms). Interestingly, the interference effect of HFC distractors decreased at longer SOAs (> 900 ms) while that of LFC distractors increased. Our results suggest that distractor suppression for LFC distractors occurs and decays faster compared to HFC distractors. Thus, paradoxically, it makes weakly salient distractors more disruptive when the stimulus-response latency is long.
Meeting abstract presented at VSS 2016