Scaling down motor memories: de-adaptation after motor learning.

Although adaptation to novel motor tasks is sometimes a very slow process, de-adaptation is usually extremely rapid. Such rapid de-adaptation is seen in dynamic learning in which subjects can take hundreds of movements to learn a novel force environment but only a few movements to de-adapt back to a normal or "null" force environment. We investigated whether this effect is unique to the null environment or reveals a more general rapid adaptation mechanism by studying how subjects behave when their dynamic environment changes. We observed that after learning a dynamic force field, subjects took longer to de-adapt when the forces were turned off than to adapt to a novel scaled-down version of the experienced field. This demonstrates that rapid adaptation is not unique to the "null" force environment. Moreover, we examined subjects' ability to adapt from a learned field to either a scaled down field or to a field in which the sign of the forces changed. Even though in both conditions the required change in force output was identical, subjects were significantly faster at adapting to the scaled down field. The result suggests that rapid de-adaptation reflects a capacity to scale down the relative contribution of existing control modules to the motor output.