When people try to move their arm with a robot and can’t see or feel where they’re supposed to go, they miss the target by a lot—much more than when they get visual or force cues.

When healthy people move their arm with a robot using just one kind of feedback (like seeing a target or feeling resistance), they make more mistakes and use more muscle effort than when they get both kinds of feedback at once.

When people get both a visual target and a gentle push from the robot at the same time, they hit the target more accurately and with less muscle strain than when they get only one kind of cue.

When people use a robot to move their arm along a curved or wiggly path instead of a straight line, their muscles work harder and they make more mistakes, no matter if the robot moves their arm or they move it themselves.

When both arms are moved together by a robot while watching a screen, healthy people’s muscles work harder than when only one arm is moved, even if they’re not trying to move it themselves.

When people’s arms are moved by a robot without seeing where they’re supposed to go, their muscles don’t work as hard as when they can see the target path—even if they’re not moving on their own.