Fig 2. Task order.

Top row: First session was the baseline session, where the cursor was aligned with the hand position in reach training tasks. Bottom row: Second session involved training with a rotated cursor. Each session began with reach training and consisted of four tasks (e.g. reach training, active localization, passive localization, and no cursor reaches). Each set of four tasks represents a block, and each block was completed four times, starting with training. This was followed by two hand localization tasks, during which either the participants moved their own unseen hand to the arc (active localization) or the robot guided their hand (passive localization). “Top-up” reach training followed each of these localization tasks. Each block ended with a no-cursor reach task. For the no-cursor reaches following training with a visuomotor rotation in the second session (bottom), participants completed this task twice, once when told to not use their strategy (no-cursor reaches without strategy) and once when told to use it (no-cursor reaches with strategy). The order of these two tasks were counterbalanced within and between participants.