Belief Revision Postulates Generalized for Multi-Agent Systems
A recent preprint on arXiv (2605.02249v1) broadens traditional AGM belief revision principles to accommodate multi-agent scenarios in epistemic planning. This research clarifies how the belief acquisition of one agent regarding a state property influences the beliefs of all agents within a system, employing a unified multi-agent Kripke model. The authors introduce a straightforward operator called generalized full-meet multi-agent revision that adheres to all established postulates and present a generalization for iterated revision alongside an operator based on event models. The paper also addresses potential challenges in defining epistemic operators within Kripke models. Ultimately, the study aspires to establish a formal framework for assessing dynamic epistemic reasoning systems.
Key facts
- arXiv:2605.02249v1 is a new announcement type: new
- The paper investigates belief revision in epistemic planning
- It generalizes classical AGM postulates to multi-agent settings
- Agents' beliefs are represented via a single multi-agent Kripke model
- Generalized full-meet multi-agent belief revision satisfies all postulates
- A generalization of standard postulates for iterated revision is defined
- An event model-based revision operator is presented
- Potential issues in defining an epistemic operator on Kripke models are discussed
Entities
Institutions
- arXiv