Reinforcement Learning Model With Dynamic State Space Tested on Target Search Tasks for Monkeys: Self-Determination of Previous States Based on Experience Saturation and Decision Uniqueness.