The Cognitive Map: Could It Have Been Any Other Way? the Cognitive Map

Could the human cognitive map, with all of its peculiarities, be structured in any other way and still perform the useful functions it does? We can perform a thought-experiment by imagining that we must design the cognitive map for a robot, operating under limited cognitive resources, which must assimilate and use knowledge about its large-scale environment acquired from observations during travel. By taking this design perspective, we can determine that some familiar proposals for the structure of the cognitive map are inadequate, by themselves, to meet the constraints of the task. We develop a design consisting of separate representations for relative-position information, topological connections, and knowledge of routes, each rich in states of partial knowledge. Each step of the derivation is motivated by the pragmatic needs of the task, and the result bears a strong resemblance to the human cognitive map. We also discuss the relationship between this type of argument from design and the more usual approach of psychological explanation. When we observe the peculiarities of the human cognitive map, we are tempted to ask, \Could it have been any other way?" We can conduct a thought-experiment to study this question, by viewing it from the point of view of the computer scientist and system designer rather than that of the psychologist.