Heuristics in Problem Solving: The Role of Direction in Controlling Search Space

Abstract

Isomorphs of a puzzle called m+m resulted in faster solution times and an easily reproduced solution path in a labeled version of the problem compared to a more difficult binary version. We conjecture that performance is related to a type of heuristic called direction that not only constrains search space in the labeled version, but also facilitates solution by presenting an "easy to keep in mind" destination as the goal state that does not tax working memory. Using direction makes it possible to solve a problem by building a path toward the solution rather than eliminating unpromising paths. The latter always involves search, which slows down the solution process and requires storing a large number of intermediate states in memory. Direction allows for smaller search. We speculate that discovering direction in a given search space enables operation selection and guidance in the solution path.