And this is what we call "learning". When dog (or human) tries to catch a ball, it (he/she) fails many times. Brain memorizes trajectories and classify dog's response into two categories (fail/pass), little by little building databases of "right" and "wrong" responds. Is not it what we call "skill building"?

Next time if a trajectory is reasonably close to some "familiar" one, those neurons which hold that one in memory, respond most and similar successfull motion is recalled to catch the ball.