Switching between tasks requires individuals to inhibit mental representations of the previous task demands and to activate representations of the new demands. To date, investigators have identified only one way to measure task set inhibition-that is, through a backward inhibition (BI) paradigm. In this paradigm, participants take more time to return to a task set that was recently abandoned (e.g., "A" in an ABA task sequence) than to a nonrecently abandoned task set (e.g., CBA), and investigators have demonstrated that this time cost reflects time needed to overcome the inhibition of the recently abandoned task set. To date, however, investigators have not been able to use this paradigm, or any other, to isolate brain activity related to task set inhibition. For example, contrasting the brain activity elicited by ABA and CBA trials will not isolate activity related to task set inhibition, because inhibition occurs during the initial switch away from task A (i.e., ABA). Given that there is currently no way to directly isolate the brain activity related to task set inhibition, we decided instead to examine how brain activity during task switching varies in individuals who are better than others at inhibiting the previous task set. We found that participants who were good at inhibiting previous task sets, as measured with the BI paradigm, exhibited more activity in the basal ganglia and supplementary motor area/premotor area when task switching, as measured via functional magnetic resonance imaging. These findings suggest that activity in these regions plays a role in task set inhibition.Keywords Task switching . Backward inhibition . Basal ganglia . Putamen . Executive control To engage in goal-driven behavior, individuals must maintain in working memory higher-order mental representations of the ongoing goal and of the procedures needed to implement that goal. These higher-order representations, which together form a "task set," will configure cognitive resources so that lower-level systems (e.g., the visual system) will be biased toward task-relevant information and the motoric system will implement the correct sequence of responses to such information. For example, to correct all of the misspelled words in a document, an individual must maintain a task set maintained in working memory that will bias the visual system to search for and identify misspelled words and to then enable the motoric system to enact the appropriate set of responses on the keyboard to correct those words when they are found.Investigators have demonstrated that although individuals must exert intentional "executive" control to implement a task set, even more control is needed if a change in demands (e.g., the doorbell rings during editing) requires individuals to switch tasks (see, e.g., Monsell, 2003). Task switching requires individuals to activate the new task set and also to deactivate the previous task set so that the new task set can gain dominance over behavior. Although the deactivation of no-longer-relevant task sets can...