Agent-based models do allow implementing heuristics that take into account a
private world-model and own reasoning on (possible) actions and behavior of
other agents. Crucially, these heuristics need not be perfectly optimizing, can well
cope with imperfect or missing information, and may even consider (perceptions
of) interests of other agents. As such, agent-based modeling allows for a variety of
rationalities not just a 'perfect, self-interested rationality' (cf. Gintis, 2009) and is a
supplement to behavioral economic experiments (Pyka and Müller, 2016).
Moreover, policy makers can experimentally derive interventions that either
implicitly account for or even explicitly exploit such behavioral economic
heuristics, e.g. by providing a 'nudge' (cf. Shafir, 2013).