A modular structure for modeling problem solving strategy development, selection and use in humans and machines is outlined. A typical experiment performed with human subjects is outlined first. Then we turn to model details. The most typical module used in this context is a neural network (NN), though probability, logic (standard and fuzzy) modules have also been employed. Selected modules are described. The way individual models articulate with the modular structure is an important subordinate theme. Emphasis is placed on central associative mechanisms in these models that complement one another and also compete, replete with potential for substitute mechanisms, alternatives and parallel solutions. An earlier model, the Strategy Abstractor, is experiencing having its name preempted for and used by the modular structure; other models we present are increasingly being viewed as variations on some theme within the SA modular structure. More recent work on individual performances, robot models, and animation is mentioned.