Survival of species in an ever-changing environment requires a flexibility that extends beyond merely selecting the most appropriate actions. It also involves readiness to stop or switch actions in response to environmental changes. Although considerable research has been devoted to understanding how the brain switches actions, the computations underlying the switching process and how it relates to the selecting and stopping processes remain elusive. A normative theory suggests that switching is simply an extension of the stopping process, during which a current action is first inhibited by an independent pause mechanism before a new action is generated. This theory was challenged by the affordance competition hypothesis, according to which the switching process is implemented through a competition between the current and new actions, without engaging an independent pause mechanism. To delineate the computations underlying these action regulation functions, we utilized a neurocomputational theory that models the process of selecting, stopping and switching reaching movements. We tested the model predictions in healthy individuals who performed reaches in dynamic and uncertain environments that often required stopping and switching actions. Our findings suggest that unlike the stopping process, switching does not necessitate a proactive pause mechanism to delay movement initiation. Hence, the switching and stopping processes seem to be implemented by different mechanisms at the planning phase of the reaching movement. However, once the reaching movement has been initiated, the switching process seems to involve an independent pause mechanism if the new target location is unknown prior to movement initiation. These findings offer a new understanding of the computations underlying action switching, contribute valuable insights into the fundamental neuroscientific mechanisms of action regulation, and open new avenues for future neurophysiological investigations.