A combined logic of expectation and observation: a generalization of BDI logics

Although BDI logics have shown many advantages in modelling agent systems, the crucial problem of having computationally ungrounded semantics poses big challenges when extending the theories to multi-agent systems in an interactive, dynamic environment. The root cause lies at the inability of modal languages to refer to the world states which hampers agent reasoning about the connection of its mental attitudes and its world. In this paper, following ideas in hybrid logics, we attempt to readdress the computational grounding problem. Then, we provide a formalism for observations - the only connection between mind and worlds - and expectations - the mental states associated with observations. Finally, we compare our framework with BDI logics.