Toward A Movement Paradigm For Artificial Human Agents
Thomas Clemen, Nima Ahmady-Moghaddam, Daniel Glake, Ulfia Lenfers, Florian Ocker, Daniel Osterholz, Jonathan Ströbele · 2022
The behavioral logic of agents in agent-based simulation models is becoming increasingly complex. This intrinsic development calls for new approaches to understanding and analyzing agent behavior. For example, when applying an agent-based model (ABM) without knowing the full model specifications, its agents – particularly their decision-making processes – are a black box to the user. Similarly, when designing an ABM, the individual steps of an agent’s decision-making trajectories might cease to be fully traceable when the model becomes sufficiently complex. Under such circumstances, ABM users and designers are forced to infer underlying reasons for behaviors from observation. This process is comparable to ecologists observing the behavior of organisms. In this contribution, we focus on the movement behavior of spatially explicit human agents. We reflect on the Movement Ecology Paradigm (MEP) and transfer some of its conceptual results to the world of human agents. Notably, we outline the MEP’s Why-Where-How complex and assess whether these three questions are sufficient for describing human agents’ underlying motivations for movement behaviors. A first, basic version of the Human Agent Movement Paradigm (HAMP) is proposed. Like the ODD protocol in ecology, the HAMP can be seen as an initial step toward a standard protocol to describe the movement of spatially explicit agents.