A Proposal and an Application of Reinforcement Learning to Anticipatory Strategy