# Dynamic epistemic logic Dynamic epistemic logic is a modal logic framework for reasoning about knowledge, belief, and information change, modeling how agents' epistemic states update when public announcements, private communications, or other events occur. Dynamic epistemic logic (DEL) is a family of formal frameworks that extends epistemic logic to model how the knowledge and beliefs of multiple agents change as a result of actions or events. It provides a rigorous mathematical language for representing not only what agents know about the world, but also what they know about each other's knowledge, and how that higher-order knowledge evolves over time. DEL is used in computer science, game theory, and philosophy to analyze scenarios involving communication, observation, and information update. The core idea of DEL is to combine a static epistemic model, which describes the possible worlds and the agents' indistinguishability relations among them, with a dynamic component that specifies how the model is transformed when an event occurs. The most basic dynamic operation is the public announcement, where all agents simultaneously learn a proposition. More complex operations include private announcements, where only a subset of agents receives information, and action models that can represent a wide range of communicative events, including lies, secrets, and simultaneous actions. ## Historical Development The roots of DEL trace back to the 1980s, with early work by philosophers and logicians such as Jaakko Hintikka and Robert Stalnaker on epistemic logic and its dynamic extensions. The formalization of public announcement logic (PAL) emerged in the late 1980s and early 1990s, notably through contributions by Jan Plaza and others. The general framework of action models, which allows for arbitrary event types, was introduced by Alexandru Baltag, Lawrence Moss, and Slawomir Solecki in a 1998 paper, establishing the modern DEL paradigm. Subsequent developments by Hans van Ditmarsch, Wiebe van der Hoek, and Barteld Kooi expanded the theory and its applications, culminating in their 2007 book 'Dynamic Epistemic Logic'. ## Key Concepts and Formal Machinery DEL is built on a multi-agent epistemic language that includes modal operators for knowledge (often denoted K_i for agent i) and common knowledge. The static semantics is given by Kripke models, where each possible world is assigned a valuation of atomic propositions, and each agent has an equivalence relation (or a more general relation for belief) that captures which worlds they consider possible. A formula like K_i φ is true at a world if φ holds in all worlds that agent i considers possible. The dynamic part of DEL introduces event models, which are Kripke-like structures whose worlds represent possible events. Each event has a precondition, a formula that must be true for the event to occur, and a relation for each agent indicating which events they can distinguish. The product update rule then combines the original epistemic model with the event model to produce a new model that reflects the information change. For public announcements, the event model is simple: there is one event whose precondition is the announced formula, and all agents can distinguish it from nothing else, so the update simply restricts the model to worlds where the formula holds. ## Applications in Multi-Agent Systems DEL has found significant applications in multi-agent systems, a subfield of [artificial-intelligence](https://www.wikiprompt.org/wiki/artificial-intelligence) concerned with interactions among autonomous agents. In this context, DEL provides a precise semantics for communication protocols, allowing designers to specify what agents know after messages are exchanged. For example, in distributed systems, DEL can model the knowledge that processes have about the state of other processes, which is crucial for tasks like consensus and coordination. The framework also supports reasoning about security protocols, where agents may have private information and need to maintain secrecy or authenticate each other. In game theory, DEL is used to model the dynamics of information during a game. Players' strategies often depend on what they know about other players' knowledge and beliefs. DEL can represent the effects of moves that reveal information, such as bidding in an auction or drawing a card in a card game. The logic allows for the analysis of epistemic conditions for equilibrium, such as common knowledge of rationality, and how these conditions change as the game progresses. ## Relationship to Other Logical Frameworks DEL is closely related to other dynamic logics, such as propositional dynamic logic (PDL) and dynamic doxastic logic, which extend the static picture with program-like operations. However, DEL is distinguished by its focus on epistemic and doxastic notions, and by its use of event models that can represent simultaneous and private actions. The field also connects to epistemic temporal logic, which combines knowledge with time, and to the study of belief revision, which deals with how rational agents should change their beliefs when receiving new information. Recent research has explored connections between DEL and [machine-learning](https://www.wikiprompt.org/wiki/machine-learning), particularly in the context of [large-language-model](https://www.wikiprompt.org/wiki/large-language-model)s and their reasoning about knowledge. While DEL traditionally assumes perfect logical agents, there is growing interest in applying its principles to model the limitations and uncertainties of AI systems. The formal tools of DEL can help in designing agents that maintain consistent beliefs and update them appropriately in dynamic environments, a challenge that is also relevant to fields like [neural-network](https://www.wikiprompt.org/wiki/neural-network)-based decision making. ## Computational Aspects and Extensions Model checking and satisfiability for DEL are generally decidable but computationally complex. The complexity depends on the specific variant; for example, public announcement logic has a PSPACE-complete satisfiability problem, while full DEL with action models can be undecidable in some formulations. Researchers have developed tableau methods and other proof systems to handle practical reasoning tasks. Extensions of DEL include probabilistic DEL, which assigns probabilities to worlds and events, and DEL with factual change, which allows the update to alter the truth values of atomic propositions, not just the agents' knowledge. Another active area is the combination of DEL with [curriculum-learning](https://www.wikiprompt.org/wiki/curriculum-learning) and other training paradigms in AI, where the idea of gradually exposing agents to more complex information structures mirrors the incremental updates in DEL. The logic also informs the design of communication protocols in multi-robot systems and human-robot interaction, where agents must coordinate based on partial and changing information. ## See Also - epistemic-logic - modal-logic - [multi-agent-system](https://www.wikiprompt.org/wiki/multi-agent-system) - belief-revision --- Source: https://www.wikiprompt.org/wiki/dynamic-epistemic-logic License: CC BY-SA 4.0 (https://creativecommons.org/licenses/by-sa/4.0/) Last updated: 2026-09-14T06:27:19.152942+00:00