[Traum & Allen1994] criticize the
approach to plan recognition based on a pure recognition of intentions: they
point out that, in many ordinary interactions, the collaboration among
agents can be explained by assuming that they are adhering to social
obligations, more than working at a common higher level goal.
Although the idea of obligations is reasonable, we do not think it is in
contrast with our approach.
In particular, nothing prevents from having plans addressing conventional
obligations, whose decomposition can specify how the related obligations can be
met in the reply.