Motor deficits in the affected arm following stroke often lead to poor performance in bimanual tasks, which require the coordination of both arms. Bimanual tasks require the control of multiple degrees of freedom. One of the challenges in rehabilitation is to present feedback from multiple degrees of freedom without overwhelming a patient's perception. This study examines how to decrease the dimensionality of the feedback by combining data from multiple degrees of freedom to form a more compact feedback structure. It is hypothesized that reducing the dimensionality of the feedback may enhance motor learning and lead to a more effective training paradigm.