Compensation mechanisms that improve distractor filtering are short-lived.

We investigated possible compensation mechanisms for improving filtering of distractors from entering visual Working Memory (WM). Participants preformed a change-detection task in which three targets, six targets, or three targets along with three distractors (the filtering trial) were randomly presented. In six experiments, we tried to reduce the filtering cost, calculated as the difference in accuracy between the three targets and the filtering condition, by either cueing the possible locations of the distractors using placeholders (that could be either fixed throughout the experiment or change every trial; i.e., location cue), or by providing the location cue coupled with a warning cue singling the upcoming filtering trial. Results revealed that the filtering cost was not reduced by a fixed location cue (Experiment 1 and Experiment 5). However, the fixed location cue coupled with a warning cue (Experiment 2 and Experiment 5) or a location cue that changed positions every trial (Experiment 6), were sufficient to reduce the filtering cost. Additionally, longer preparation interval for filtering trials did not further reduce the filtering cost (Experiment 3). We argue these findings support that in the context of visual WM, spatial filtering settings can only be held for a limited amount of time. Thus, these filtering settings must be reactivated in order to be effective and to reduce the filtering cost.