Unpredictable singleton distractors in visual search can be subject to second-order suppression

Recent evidence suggests that attentional capture by salient-but-irrelevant distractions can be avoided via suppression, thereby improving performance in visual search. Initial evidence suggested it is only possible to suppress salient distractors with constant and predictable features (first-order suppression). We show that previous failures to find evidence for second-order suppression of unpredictable feature singletons may have been due to low feature variability: If it is probable that the salient distractor colour is the target colour on another trial, suppressing this item might hinder performance. We first validated a new multiframe letter-probe paradigm, where observers counted the search displays with a target shape and always reported as many letter probes as possible from the final display. When target and singleton colours were constant (Experiment 1), a singleton suppression effect was observed, with probe letters at the singleton distractor location reported less frequently than those at non-singleton distractor locations. When two randomly swapped target/singleton colours were employed (Experiment 2), no suppression effect was observed, replicating previous findings. Critically, when target-colour items and the singleton could have one of eight different random colours (Experiment 3), a robust suppression effect reappeared. These observations demonstrate that first-order suppression is not universal, and that second-order suppression can be applied to singleton distractors under some circumstances. Suppression effects were observed for displays with and without targets, suggesting that they are not a product of direct target-singleton competition.