Figure 1.

Experimental design. (A) Participants manipulated two joysticks on the gamepad using both thumbs to control the onscreen cursor. (B) Cursor position corresponding to the tip of the virtual two-link arm in the frontal plane. The up/down inputs from the right and left joysticks determined the torque applied to the proximal and distal joints, respectively. Therefore, the up/down inputs were directly relevant to the cursor movements (task-relevant). Conversely, the left/right inputs from both joysticks were irrelevant to the cursor movements (task-irrelevant). (C) In the training task, the participants were required to quickly move the cursor from the starting position to the target. The participants were divided into two groups based on the target shapes: small and wide. In the small target condition, the target was a small circle with a 10 mm diameter, whereas in the wide target condition, the target was an arched and wide shape with a curvature radius of 68 mm, width of 79 mm, and center angle of 60°. (D) The test task involved moving the cursor towards a target selected pseudo-randomly from 1 of 5 directions: −60°, −30°, 0° (above the starting position), 30°, 60°. (E) Sequence of tasks across the training and test phases. During the training phase, the participants performed 240 trials divided into eight blocks (30 trials per block). During the test phase, the participants performed 120 trials divided into four blocks (30 trials per block). (F) Quantification of the movement direction. The black square represents a bin. The black circles represent cursor points in a trajectory. The green arrow indicates a vector within the same bin. The red arrow indicates a vector in the adjacent bin.