In a task switching setting, participants are required to maintain two different task sets in working memory but to engage in only one of the tasks in each trial. Curiously, if one of the tasks is labeled irrelevant, performance in the remaining task is still worse than when the same task is performed in a single-task context. This so-called fade-out cost demonstrates that the suddenly irrelevant task cannot be discarded directly (Mayr & Liebscher, 2001).
In the present study, we used list-wide proportion manipulations to induce two different control states (a) at the task-set level by manipulating the proportion of task switches (Experiment 1) and (b) at the response-conflict level by manipulating the proportion of response congruency (Experiment 2) in a task switching experiment with a subsequent fade-out part.
If fade-out costs originate at the task-set level, fade-out costs should be smaller following many repetitions compared to many switches in Experiment 1. If fade-out costs originate at the response-conflict level, fade-out costs should be smaller following many incongruent compared to many congruent trials in Experiment 2.
Preliminary data of Experiment 1 indicate reduced fade-out costs after blocks with many repetition trials compared to many switch trials suggesting that fade-out costs derive at the task-set level. If this is the sole source of fade-out costs, we expect that in Experiment 2 the congruency manipulation will not affect the size of fade-out costs. Data collection is still ongoing, but complete data will be presented at the conference.