MacGregor(Psychological Review, 1987, 94, 1, 107-108)suggested that the limitation of chunk size of short term memory can be interpreted as the optimization of memory search if the memory network is considered as a hierarchical tree. On the other hand, some recent empirical studies suggest cognitive maps have a sort of hierarchical structure. The main purpose of this note is to reconsider the structure of human cognitive maps as location search network from a viewpoint of optimization. Some mathematical propositions on search efficiency under certain conditions are briefly reviewed. Then, the hierarchies of human cognitive maps, obtained by Hirtle and Jonides(1985, Memory&Cognition, 13, 3, 208-217)'s experiment, are analyzed and copmpared with the efficient structure based on those mathematical propositions. It is shown that, the cognitive maps, as dynamically changing and self-organizing structures, may be considered efficient search networks.