Previous research has demonstrated that cognitive offloading (i.e., externalizing mental processes) is useful for immediate problem solving. However, long-term effects of cognitive offloading on subsequent problem solving without offloading are remarkably understudied. Our main goal was to investigate the effects of representation control (i.e., adjusting the format of the representation to task requirements) on incidental procedural learning in an information extraction task with interactive visualizations. More specifically, we tested how the availability of representation control for solving tasks in a practice phase affects procedural learning measured in a testing phase in which representation control is no longer available. In both phases, we explored time on task as well as proportion correct as proxies of task performance and analyzed how often participants used representation control. We conducted three experiments in which participants could modify and reorganize information displays in the practice phase, whereas in the testing phase, they had to solve equivalent and near transfer problems without this offloading opportunity. We show that representation control is beneficial for immediate task solution, particularly for problems that require a spatial transformation. This benefit was more pronounced for automated (i.e., system-controlled) types of control (Experiment 3). In contrast, in subsequent problem solving without representation control (up to 24 hr later), participants who had been using representation control previously fell back to the level of an untrained baseline condition (Experiments 1 and 2). However, the detrimental effect of representation control was confined to equivalent tasks and did not generalize to a near transfer task.