Abstract
Recent work has shown that the efficiency of attentional selection can be facilitated by not only the enhancement of target features but also the active distractor suppression (Geng, 2014). In this study, we explored the role of probabilistic expectations in the suppression of salient distractors. In three experiments, participants were asked to report a bar's orientation inside a gray shape singleton (i.e., target) among gray distractors. Critically, on some trials, one distractor was a color singleton, which captures attention (i.e., singleton capture; Theeuwes, 1994). In Experiment 1, we manipulated the color singleton's variability (fixed or various), and the likelihood of the occurrence (80% or 20%). The results showed the color variability (3 colors in Experiment 1 and 192 colors in Experiment 3) does not affect singleton capture. In contrast, the likelihood of occurrence strongly modulates singleton capture (RT and first fixation). When the singleton has a low-likelihood of occurrence, attention is strongly captured compared to when a high-likelihood of occurrence. In Experiment 2, we tested whether the difference is due to better proactive or reactive suppression by randomly inserting a probe display on some trials that contained a letter inside each shape. Participants were asked to report the letters they saw (Gaspelin et al., 2015). The probe could occur either just before the search display (pre-probe trials) or after (post-probe trials). We found a high probability of report for the letter on the singleton distractor in post-probe trials in the low-likelihood condition suggesting attention is captured by the color singleton, but a low report in the high-likelihood condition, which suggests the capture is proactively suppressed when the color singleton occurs frequently. Our findings demonstrate attentional capture is more sensitive to the frequency than specific features, and that frequency determines whether proactive suppression mechanisms can be exploited to facilitate performance.
Meeting abstract presented at VSS 2018