Recognition memory for human motor learning

Almost everything that is presently known about how we retain motor skills, like a golf swing or a tennis serve, comes from studies in which participants are asked to reproduce from memory previously learned movements. From work on human verbal memory, it is well known that this kind of retention test, a test of recall memory, substantially underestimates how much learning is actually retained. But, could it be that we also retain information about previously learned movements like a particular tennis serve which we can no longer re-produce and which appears to be forgotten? If motor memory behaves at all like other forms of human memory, it may be possible to access this information if we replay the somatic cues that accompanied the original learning. In the present study, we introduce a procedure in which, following motor learning, we probe retention, using tests of recognition, in which participants hold the handle of a robot arm as the robot replays candidate movements. Participants are asked to indicate whether or not the displacement of the arm produced by the robot matches the movement they themselves produced at the end of learning. The key tests are done 24 hours following initial training to eliminate transient retention and assess consolidated memory. We show that recognition memory for previously learned movements substantially exceeds (by a factor of greater than two) estimates of retention for the same movements based on recall testing, that is, based on reproduction of learned movements from memory. To the best of our knowledge these tests have never been previously undertaken. This is the first documented evidence of recognition memory for human movement. These studies show for the first time that substantial information regarding previously learned movements is retained that is uncovered using recognition tests. There is a memory retrieval failure when testing relies on active movement reproduction. Using somatosensory inputs to access previously learned movements has potential benefits both in training athletic skills and in rehabilitation. To specifically address how this might work, we conducted additional tests to assess whether recall for previously learned movements can be improved by using retrieval cues (passive playback of a small number of movements). Using this technique, we find that we can restore recall memory for movements to levels characteristic of recognition memory performance. Thus, appropriate movement retrieval cues enable us to rescue otherwise forgotten movements.