I would like to build a cognitive model that is able to learn how to solve a complex puzzle such as the tower of Hanoi based off of limited feed-back (set of valid moves, closer to goal, farther from goal). I have found some references for learning the Tower of Hanoi and the performance of schizophrenics, but I was wondering if there was a particular name for the type of learning I'm looking for (advanced reinforcement learning?) and if there are further studies on how people learn these types of puzzles.