reward timing

Choose one observable action, mark it clearly, and place the reward where the next successful repetition can begin.