Understanding human behavior in the context of exploration and navigation is an important but challenging problem. Such understanding can help in the design of safe structures and spaces that implicitly aid humans during evacuation or other emergency situations. In particular, the role that memory plays in this process is something that is crucial to understand. In this paper, we develop a novel serious game-based experimental approach to understanding the non-randomness and the impact of memory on the human exploration process. We show that a simple memory model, with a depth of between 6 and 8 steps, is sufficient to approximate a 'human-like' level of exploration efficiency. We also demonstrate the advantages that a game-based experimental methodology brings to these kinds of experiments in the amount of data that can be collected as compared to traditional experiments. We feel that these findings have important implications for 'safety-by-design' in complex infrastructural structures.
Keywords: Agent based modeling; Exploration; Indoor navigation; Markov models; Memory; Serious games.