This thesis compared the effectiveness of traditional KR manipulations to another condition where the subject assumed an active role in the learning process. It was expected that increasing the participants' autonomy would enable them to develop transferable knowledge regarding the provision of feedback within learning, which would enhance their transfer of learning from one motor task to another. Some recent studies have indicated that reduced frequencies of KR seem to facilitate motor learning by helping subjects to develop their own error detection capabilities (Winstein & Schmidt, 1990). This thesis takes the view that both the development of error detection capabilities and the timing of feedback provision could be optimised by focusing more closely on the role of the learner, and seeking ways to enhance their cognitive involvement in the feedback process. The hypothesis of the experiments was that the expected superiority in retention of a reduced frequency KR group, in relation to a 100% KR control group, could be matched by a cognitive strategy group. The reduced frequency KR groups (bandwidth, relative frequency and summary KR) in the first four experiments failed to support the experimental hypotheses. The fifth experiment was designed to rectify the shortcomings of these experiments. Three summary KR conditions (1-trial, 15-trial, & strategy groups) performed a linear slide task. In retention, the 3 group one-way ANOVAs for I CE I and VE were significant, revealing that the 1-trial summary group's scores were poorer than either the 15-trial summary and strategy groups. In conclusion, these findings provide some evidence that informing subjects of the importance of problem solving activities during practice can reduce the need for supervision of feedback provision, without risking impaired retention. This research is a first step towards demonstrating that cognitive factors involved in learning motor skills can be incorporated in the learning session to increase the autonomy of the subject.