Belief Contraction in Dynamic Epistemic Logic

2026-06-30 · Source: Artificial Intelligence · Field: Science & Research — Artificial Intelligence & Machine Learning, Mathematics & Computational Sciences · Depth: Expert, quick

Summary

A new mechanism for belief contraction in Dynamic Epistemic Logic (DEL) is introduced, addressing limitations of existing approaches. The standard DEL formulation (Baltag, Moss, Solecki, 1998) effectively models belief expansion but struggles with factual belief contraction. While enriching Kripke models with plausibility orderings offers a classic response, this method has expressive limitations, particularly in modeling beliefs that violate positive introspection or contraction dynamics from hedged public announcements. The proposed mechanism operates directly on standard Kripke models, free from doxastic accessibility relation constraints. It satisfies some standard contraction properties, with conditions for unsuccessful contraction analyzed. The authors provide a sound and complete axiomatization using reduction axioms. Furthermore, a more general dynamic logic, extending standard DEL for private or semi-private announcements, is defined and also receives a complete and sound axiomatization. This work was published on 2026-06-30.

Key takeaway

For research scientists working on formalizing belief change in AI systems, this work offers a novel approach to belief contraction within Dynamic Epistemic Logic. You should consider this new mechanism, defined directly on standard Kripke models, as an alternative to plausibility ordering methods, especially when modeling complex scenarios involving introspection violations or hedged announcements. This could improve the accuracy and expressiveness of your epistemic models.

Key insights

This research presents a novel mechanism for belief contraction within Dynamic Epistemic Logic, overcoming prior expressive limitations.

Principles

• Standard DEL struggles with factual belief contraction.
• Plausibility orderings have expressive limitations.
• New mechanism works directly on Kripke models.

Method

The paper defines a mechanism for belief contraction directly on standard Kripke models and provides a sound and complete axiomatization via reduction axioms. It also axiomatizes a general dynamic logic for private/semi-private announcements.

Topics

• Dynamic Epistemic Logic
• Belief Contraction
• Kripke Models
• Axiomatization
• Plausibility Orderings
• Epistemic Events

Best for: AI Scientist, Research Scientist

Related on AIssential

• Belief Engine: Configurable and Inspectable Stance Dynamics in Multi-Agent LLM Deliberation — The Belief Engine provides auditable, parameter-controlled stance dynamics for LLM agents in deliberative simulations.
• On Strong Equivalence Notions in Logic Programming and Abstract Argumentation — A new strong equivalence notion restores compatibility between logic programming and abstract argumentation in dynamic contexts.
• R-Mod: Minimal Structural Revision of S5 Epistemic Models — R-Mod minimally structurally revises S5 epistemic models to accommodate new knowledge, addressing limitations of belief revision.
• State Representation and Termination for Recursive Reasoning Systems — Explicit state representation and an order-gap termination criterion enhance recursive reasoning systems' efficiency and reliability.
• Better Understanding, Understanding Better — A comparative epistemic logic formalizes understanding as a graded concept, allowing for formal comparisons between agents' understanding levels.
• Project NIKA: Unlocking Epistemic Agency in 4-Bit Quantized Models — Constraining LLMs with a deterministic governor can induce "epistemic agency" and superior reasoning, even in small, quantized models.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by Artificial Intelligence.