Behavioral Evidence of Top-Down Suppression of Attention Capture with the Letter-Probe Technique

Abstract

There is considerable debate about how visual attention is involuntarily drawn to salient stimuli (called attention capture). Stimulus-driven theories posit that certain super salient stimulus features automatically capture attention, whereas goal-driven theories posit that only stimuli matching our attentional set (i.e., what we are looking for) capture attention. In the current study, we test a hybrid account (called the signal suppression theory), which posits that salient stimuli produce an automatic “attend-to-me” signal but that this signal may be suppressed if the observer exerts strong top-down control. To test this hypothesis, we used a modified additional singleton paradigm in which participants searched for a target shape and attempted to ignore an irrelevant color singleton. We manipulated search mode (singleton-detection mode vs. feature search mode) to encourage or discourage attention capture. On a small subset of trials, participants were instead asked to perform a probe task. On these trials, letters appeared briefly at each search location and participants reported all letters they saw. When singleton detection mode was encouraged, attention capture was observed on the no-probe trials. On probe trials, participants were more likely to report the singleton probe letter than distractor or target probes. When feature search mode was encouraged, capture was not observed on the no-probe trials. On probe trials, participants were less likely to report the singleton probe letter than the target or distractor probes. This behavioral evidence corroborates previous ERP-based evidence of top-down suppression of singleton distractors when participants are in feature search mode. In addition, the finding that singleton detection mode led to improved probe processing at the singleton location provides strong evidence against the hypothesis that attention capture effects in the additional singleton paradigm reflect “filtering costs” (a generalized slowing of responses) rather than true capture of attention to the singleton’s location. Meeting abstract presented at VSS 2015