Abstract We present a formalism that allows to distinguish between, on the one hand, expectations about future states (agents’ a priori beliefs about those states), and, on the other hand, what will be known and believed by the different agents in those future states ( a posteriori knowledge and belief). We use plausibility models within Dynamic Epistemic Logic (DEL) to model beliefs, expectations and judgments of plausibility. Having such a formalism in place, we can reason about an agent’s false beliefs and false expectations, and when to make belief updates ( relevant announcements) so future undesirable situations may be avoided. A potential application of our framework is human-robot interaction. Based on reasoning about the human’s false expectation, a proactive robot can autonomously decide when and what to announce to help avoiding that the human ends up in an undesirable state.
Bolander et al. (Sat,) studied this question.