Working memory related to different subtasks can be maintained in separate, non-interfering stores

How can WM maintain more information when its capacity is only around 4 items? Here, we explore the possibility that information related to separate subtasks do not count towards this limit, perhaps, because they are maintained in non-interfering stores. Across the two experiments, we investigated if increasing the WM load related to one subtask interfered with the execution of a concurrent but distinct second subtask. In Experiment 1, participants first saw pictures that were to be kept in mind and used for a later subtask B. They then executed subtask A, while keeping in mind these subtask B pictures. Although subtasks A and B involved maintaining and updating identical sets of pictures, increasing the number of subtask B pictures did not interfere with subtask A execution, which was only affected when the number of pictures relevant to it was increased, suggesting that subtask A and B pictures were maintained in separate non-interfering stores. An objection might be that in Experiment 1, subtask B pictures were passively and not goal-directedly maintained. In Experiment 2, participants executed a more complex subtask that forced participants to maintain and update two separate sets of subtask B pictures while executing subtask A and subtask A also involved WM maintenance and updating of identical pictures. We found that even here increased load of subtask B pictures did not affect subtask A performance. Thus, at least in some multitasking situations, information related to distinct subtasks can be maintained in separate non-interfering stores.