Statistical learning of movement

Movement related sensory feedback is not necessary for learning to execute a motor skill

Sensory feedback has traditionally been considered critical for motor learning. While it has been shown that motor learning can occur in the absence of visual or somatosensory feedback, it is thought that at least one must be present. This assumption contrasts with literature demonstrating that motor imagery (MI) - the mental rehearsal of a movement - is capable of driving motor learning even though the lack of actual execution precludes sensory feedback related to movement. However, studies of MI typically employ simple tasks that do not require improvements in motor execution per se, suggesting that MI might improve task performance primarily through perceptual mechanisms. To avoid this limitation, we designed a novel motor task requiring the repeated execution of unfamiliar kinematic trajectories where learning was assessed through changes in the speed-accuracy function (SAF) across five sessions. General task performance was controlled for by assessing performance on randomly generated trajectories. Groups included physical practice (PP; with and without added visual feedback), MI, and perceptual control (PC), the latter of which only observed the trajectories. All groups performed physically on the final session. Upon the final session, the MI group performed better than the PC group, and better than initial session PP performance. These results suggest that motor learning occurred in the MI group despite the lack of sensory feedback related to the movement, and that this learning was not simply the result of perceptual learning. Our results question long-standing assumptions about MI based learning and the necessity of feedback in motor learning generally.