Alternating between actor and observer roles during practice in a dyad practice condition has proven to be an efficient, and in some cases, a more effective method of facilitating motor learning compared to individual practice. Further, providing individuals control over their own (self-control) or another learner's (peer-control) practice environment has also been shown to be superior for skill learning relative to individuals without control. The purpose of the present experiment was to examine the learning of a speed cup-stacking task under individual versus dyad practice conditions. In dyads, partners either alternated actor and observer roles on a trial-to-trial bases, or under novel learner-controlled conditions wherein either the actor or the observer role was given control over when the partners should switch roles (n=22/group). Switching behaviours of the dyad conditions were the primary variables of interest (i.e., switch frequency, switch strategies). Participants practiced a speed cup-stacking task and learning was assessed in a 24-hour retention test. Although there were no performance differences between dyad conditions, dyad learners effectively chose when to switch roles with their partner during practice, without undermining learning. These results also highlight the dynamic nature of dyad practice such that the observers chose to switch roles during practice more frequently than the actors. Overall, this experiment provides further support for dyad practice as an efficient and effective method of skill learning.