AbstractAn integral aspect of human cognition is the ability to inhibit habitual responses in order to initiate complex, rule-guided actions. Moreover, humans have also the ability to alternate between different sets of rules or tasks, at the cost of degraded performance when compared to repeating the same task, a phenomenon called the ‘task switch cost’. While it is recognized that switching between tasks requires often to inhibit habitual responses, the interaction between these two forms of cognitive control has been much less studied than each of them separately. Here, we use a computational model to draw a bridge between inhibitory control and voluntary action generation and thereby provide a novel account of seemingly paradoxical findings in the task switch literature. We investigated task switching in the mixed antisaccade task, in which participants are cued to saccade either in the same or in the opposite direction to a peripheral stimulus. Our model demonstrates that stopping a habitual action leads to increased inhibitory control that persists on the next trial. However, enhanced inhibition affects only the probability of generating habitual responses, and, contrary to previous accounts, cannot be characterized as proactive task interference. In addition, our model demonstrates that voluntary actions (but not habitual responses) are slower and more prompt to errors on switch trials compared to repeat trials. We conclude that precisely the interaction between these two effects explains a variety of contradictory findings reported in the literature.