Figure 3. Generalization under different variance conditions.

A) Baseline and generalization of the mean (±SEM) of a perturbation for a typical subject as measured by the absolute angle of final hand position relative to the target. Solid lines are generalization patterns after learning and dashed lines denote the pre-training (baseline) results. B) Average generalization (±SEM) across subjects. Solid lines denote generalization patterns after learning and dashed lines denote the pre-training (baseline) results. C) Percent adaptation (±SEM) in the generalizing directions relative to the learning direction. D) Percent adaptation (±SEM) in the generalizing directions relative to the learning direction after correcting for the sign of the mean of the perturbation; blocks with −30° mean have the target directions (x-axis) reflected relative to the learning direction. E) Diagram illustrating the direction of an asymmetry caused by used-dependent learning. The blue curve denotes a symmetric, local generalization pattern - without used-dependent learning. If there is used-dependent learning, hand movements in trials towards other targets would be attracted towards the direction to which the hand moved during the learning block (dashed red line). This effect would predict an asymmetry with the same side as the one observed in Fig. 3D.