Abstract
An accompanying three-minute video abstract can be found here: https://vimeo.com/650625000. Through statistical learning, humans can learn to suppress visual areas that often contain distractors. Recent findings suggest that this form of learned suppression is insensitive to context, so that the suppression generalizes over contexts. This puts into question its real-life relevance, because it would require a constant re-learning of attentional biases for contexts that have already been encountered, and those biases would persist when their relevance has clearly evaporated. We investigated both context-dependent learning and the flexibility of applying statistically learned distractor suppression, processes which have not been dissociated in previous context-based suppression experiments. Participants searched for a unique shape, while ignoring a uniquely colored distractor item. Crucially, we created two contexts, each of which was assigned its own high-probability distractor location such that the location where the distractor was most likely to appear depended on the context. In Experiment 1 two different search tasks were each assigned to a specific context. In Experiment 2 participants performed a search task while a central letter (A or B) indicated the context for that specific trial. Both experiments were divided into training blocks in which each block had one context which was associated with a specific high probability distractor location, and testing blocks in which both contexts were intermixed and all distractor locations were equiprobable. Response times in Experiments 1 and 2 showed that participants quickly stored away a suppressed location upon encountering a new context, and revived the stored-away location upon encountering its context again. We conclude that participants can learn to suppress a location in a context-dependent way, but they do not apply this suppression flexibly when contexts are intermixed. Thus, so long as contexts are not continuously changing, statistically learned suppression can have important real-life consequences.