The fate of distractors in working memory: No evidence for their active removal.

Not all the information processed in working memory (WM) must be retained. Due to the strict limitations of its capacity, the mechanisms that prevent WM from being cluttered and choked by no longer relevant information are of paramount importance. The present study tested the hypothesis put forward by the SOB-CS model of an active and attention-demanding mechanism that would remove no-longer relevant items from WM. Such a mechanism has been advocated to account for the well-known fact that, in complex span tasks, processing distractors at a slower pace results in better recall of memory items. According to the SOB-CS model, slow pace would free more time for removing distractors, thus alleviating the interference they create on target items. In direct contradiction with this hypothesis, a first experiment demonstrated that distractors are not less, but more accessible at the end of complex span task trials in which they have been processed at a slow rather than a fast pace. Using the repetition priming effect occurring in a lexical decision task inserted as processing component within a complex span task, a second experiment established that distractors processed at a slower pace do not elicit weaker, but stronger repetition priming effects, indicating that they have not been removed. Along with previous findings, the present study not only shows that there is no trace of distractor removal in the long term, in the short term, nor immediately after processing, but demonstrates that memory traces of distractors are stronger in situations assumed to involve a more complete removal by the SOB-CS model. These empirical evidence suggests that distractors are not actively removed from working memory after having been processed.

An empirical test of the independence between declarative and procedural working memory in Oberauer's (2009) theory.

It has recently been suggested that working memory could be conceived as two symmetrical subsystems with analogous structure and processing principles: a declarative working memory storing objects of thought available for cognitive operations, and a procedural working memory holding representations of what to do with these objects (Oberauer, Psychology of learning and motivation 51: 45-100, 2009). Within this theoretical framework, the two subsystems are thought to be independent and fueled by their own capacity. The present study tested this hypothesis through two experiments using a complex span task in which participants were asked to maintain consonants for further recall while performing response selection tasks. In line with Oberauer's conception, the load of the procedural working memory was varied by manipulating the number of stimulus-response mappings of the response selection task. Increasing the number of these mappings had a strong detrimental effect on recall performance. Besides contradicting Oberauer's proposal, this finding supports models that assume a resource-sharing between processing and storage in working memory.