What do moral rules mean?

People often communicate social and moral expectations to one another in terms of rules. While these rules often seem simple (e.g. "No walking on the grass") people understand that much more is being communicated than it at first appears. In this paper we present a novel theory of how people understand moral rules and instantiate that theory in a computational cognitive model. We argue that moral rules are a special kind of speech act, directed at a group and expressing a summary of the agreement that rational actors would arrive at when navigating an interdependent choice problem. Rules, on this conception, are therefore closely tied to their reasons, which reflect the interests of the parties to the agreement. This structure is very powerful: it allows people to interpret and apply rules flexibly in novel situations and edge cases. Specifically, we argue that when judging whether an action violates a rule, people reflect on the reason for the rule and harness mental models of agreement (and their heuristic approximations) to determine whether the action upholds or undermines the reason. Put simply, people ask whether the purpose of the rule would still be upheld if the rule permitted everyone to act in this way.