Choice As A Function Of Reinforcement Ratios In Delayed Matching-to-sample.