This might be true for learning mechanical and motion related tasks, which seems to be all they've tested here. For cognitive learning I doubt this is true, how

This might be true for learning mechanical and motion related tasks, which seems to be all they've tested here. For cognitive learning I doubt this is true, however. If I tried to teach someone arithmetic by giving them problems where the answer was always four he wouldn't learn at all. Learning a general theory requires finding out what does work and what doesn't work. Also, I'm concerned with the correlation and causation here. Someone who naturally bowls well is likely to have strikes and also more likely to finish the game well. I think the real result here is that people who suck at bowling are less consistent, more of the time.