Adaptation to dynamic environments displays local generalization for voluntary reaching movements.

The shape of the directional generalization function for adaptation to a viscous force-field environment has been controversial. Some studies have suggested wide, essentially global generalization and others have suggested narrow, local generalization. Here, we show definitively that motor adaptation displays narrow generalization with a minimal global component and a peak at the trained movement direction for both single-trial and asymptotic adaptation. Furthermore, we find that reaching movements in opposite directions do not interfere with one another during force-field learning.