Constraints on Adaptive Networks for Modeling Human Generalization