Lol you are correct! At least training them becomes smoother the faster you administer reward. Learning happens at different timescales in the brain, and immediate feedback (about <300 ms) yields the most reliable neural updates.