It is very challenging for policymakers and other decision makers to make any kind of decision on a new policy, as the reaction of a person to that policy (policy as one form of a norm) in a given situation is highly individual and based on their own subjective perspective. This becomes even more challenging in environments with a high degree of uncertainty (as is usually the case for policymakers).

Social simulations are a powerful tool for policymakers and other decision makers to support them in their decision-making process. To build agent-based social simulations that provide this support two main challenges exist: norm (policy) realistic behavior and the usability of the simulation.

Norm realistic behavior includes differentiated norm engagement as well as seeing norms as more than just restrictions on behavior. Situated norm engagement means that people react differently to norms and focus only on the parts that are relevant for them. Seeing norms as more than just restrictions on behavior means that people can also violate norms and be motivated to circumvent norms. To address these two parts, we formalize different perspectives on norms and develop a novel agent deliberation architecture, called the Perspective-Based Agent Deliberation Architecture (PBADA) that can represent different perspectives on norms. Another key element of our agent deliberation architecture is that norms are explicit objects.

Having norms as explicit objects is crucial for addressing the challenge of usability of the simulation. It allows policymakers to modify them interactively in the simulation. In general, we see usability as empowering the policy maker to use the simulation in a - for them - meaningful way. Policymakers need to understand how a norm (policy) is influencing the behavior of the agents and in what way. Furthermore, policymakers need to be able to modify existing norms and add new ones on the fly. This requires interaction tools and visualization capabilities necessary to support them in this process. To address this challenge, we present preliminary work on such an interaction tool.