I consider it more error-based (especially what was shown in the YouTube video) where the patient is getting feedback on whether he hits the target for each trial.
There is some terminal feedback available (more reinforcement learning) at the end of a training session where the patient can see the percentage of targets hit, etc.