Abstract
Learning is facilitated when practice schedules are adapted to a learner's performance compared to equivalent amounts of random practice (Choi et al. 2008). Task switching could either be sensitive to the learner's actual performance or act as a reward for successful goal achievement. We examined this by comparing two groups that were either rewarded for goal achievement with a task switch or a task repetition. Participants learned to perform four spatially distinct key-press patterns as fast as possible and without error through a discovery process. Their goal was to beat their best movement time (MT) for each specific pattern. For example, when a learner in the WinShift (WS) schedule achieved success (a "win") they switched to a different pattern; failure to beat their best time resulted in immediate repetition of the same pattern. The opposite contingency was used in the WinRepeat (WR) group. The results for both MT and errors in retention tests performed on the same day as practice and the next day were significantly better for the WR group, where a "win" was rewarded with a task repetition. Analysis of the acquisition data revealed that both groups "won" frequently early and less frequently later in practice. This meant that later in acquisition the WS group had mostly blocked practice while the WR group had mostly random practice. These findings reveal that the effectiveness of adaptive practice depends on the nature of contextual interference promoted by the algorithm.Acknowledgments: This study was funded by NSERC.