Abstrakt
in this talk I propose to construe agent-based models (ABMs) as mechanistic models. This then invites the question whether ABMs provide mechanistic explanations.
I review the most popular account of mechanistic explanation, the model-to-mechanism mapping account, together with the "the more details the better" assumption, and argue that there are important issues that have not been addressed properly. I propose an account of hypothetical mechanistic explanation which answers various objections to the standard account of mechanistic explanation, and according to which simple and idealized ABMs provide explanations to the extent that the model mechanism is similar in relevant respects and to relevant degrees to the actual mechanism.