In so-called conflict tasks (e.g., the Stroop) cognitive conflict is induced by the use of stimuli that simultaneously elicit a task-relevant and a task-irrelevant response. Under such circumstances, cognitive control is recruited to reduce response conflict and allow goal-directed behavior. Importantly, the extent to which control is recruited depends on the environmental circumstances. For example, control is recruited more strongly if response conflict has just occurred, or if it is likely to occur. Although much research has been devoted to characterizing such local and global control-adaptation mechanisms in response-conflict paradigms, it is currently unknown whether similar mechanisms are found when dealing with other conflict types. In particular, in the present studies we focused on control adaptation to the occurrence of task conflict. In four task-switching experiments, we measured task conflict as the performance costs associated with bivalent congruent trials (i.e., the employed stimulus afforded two tasks, but both stimulus features elicited the same response) and univalent trials. We found that such valency costs were reduced both following a bivalent trial and when bivalent trials were frequent, suggesting that local and global adaptations to task conflict occur similarly to what observed with response conflict. Furthermore, local control adaptation did not generalize when switching task, suggesting that the observed mechanisms may be task specific. Surprisingly however, global adaptation mechanisms displayed the opposite pattern, being present more strongly (or exclusively) in switch trials. Taken together, our results suggest that control recruitment is modulated similarly by the occurrence of response conflict and task conflict.