## A Model-Theoretic Analysis of Knowledge (1988)

Venue: | in Proc. 25th IEEE Symposium on Foundations of Computer Science |

Citations: | 55 - 11 self |

### BibTeX

@ARTICLE{Fagin88amodel-theoretic,

author = {Ronald Fagin and Joseph Y. Halpern and Moshe Y. Vardi},

title = {A Model-Theoretic Analysis of Knowledge},

journal = {in Proc. 25th IEEE Symposium on Foundations of Computer Science},

year = {1988},

volume = {91},

pages = {382--428}

}

### Years of Citing Articles

### OpenURL

### Abstract

Understanding knowledge is a fundamental issue in many disciplines. In computer science, knowledge arises not only in the obvious contexts (such as knowledgebased systems), but also in distributed systems (where the goal is to have each processor "know" something, as in agreement protocols). A general semantic model of knowledge is introduced, to allow reasoning about statements such as "He knows that I know whether or not she knows whether or not it is raining." This approach more naturally models a state of knowledge than previous proposals (including Kripke structures). Using this notion of model, a model theory for knowledge is developed. This theory enables one to interpret the notion of a "finite amount of information". A preliminary version of this paper appeared in Proc. 25th IEEE Symp. on Foundations of Computer Science, 1984, pp. 268--278. This version is essentially identical to the version that appears in Journal of the ACM 38:2, 1991, pp. 382--428. y Part of th...

### Citations

1483 | Some philosophical problems from the standpoint of artificial intelligence - McCarthy, Hayes - 1969 |

710 |
A New Introduction to Modal Logic
- Hughes, Cresswell
- 1996
(Show Context)
Citation Context ...gies and all of the consequences of his knowledge, and each knows that all of the other agents are equally wise. It is well-known that Axiom 3 can be derived from the other axioms and inference rules =-=[HC]-=-. It is convenient to refer sometimes to Axiom 3 as describing positive introspection, and to Axiom 4 as describing negative introspection. Before we formally define knowledge structures, let us discu... |

654 |
Concurrency and automata on infinite sequences
- Park
- 1981
(Show Context)
Citation Context ... We note that if ssi;j 0 t for all agents i 2 P, then f s = f t . This notion of equivalence between states in Kripke structures is closely related to the notions of equivalence [Ho] and bisimulation =-=[Pa]-=- between states in finite-state automata. Example 4.25: We now consider a very interesting situation where both no-information extensions and least-information extensions enter the picture. Suppose we... |

522 | Reaching agreement in the presence of faults - Pease, Shostak, et al. - 1980 |

491 | Knowledge and common knowledge in a distributed environment
- Halpern, Moses
- 1990
(Show Context)
Citation Context ... interaction. (A fact p is common knowledge if everyone knows that everyone knows that everyone knows ... that p. For a discussion of the significance of common knowledge for distributed systems, see =-=[HM1]-=-.) We study two model-theoretic constructions, no-information and least-information extensions, that capture the notion of finite information, and finite information in the presence of common knowledg... |

439 | Semantical considerations on nonmonotonic logic - Moore - 1985 |

433 |
Knowledge and Belief
- Hintikka
- 1962
(Show Context)
Citation Context ...unds for Byzantine Agreement [DS]. The need to formally model this type of reasoning is our motivation for constructing a semantic model for knowledge. The first attempt to do so was made by Hintikka =-=[Hi]-=-, using essentially the notion of possible worlds. Hintikka's idea was that someone knows ' exactly if ' is true in all the worlds he thinks are possible. Possible-world semantics has been formalized ... |

374 | Agreeing to disagree - Aumann - 1976 |

315 | A formal theory of knowledge and action - MOORE - 1985 |

183 | e Computational Complexity of Provability in Systems of Modal Propositional Logic - Ladner - 1977 |

183 |
Formulation of Bayesian analysis for games with incomplete information
- Mertens, Zamir
- 1985
(Show Context)
Citation Context ...llary of this investigation is that finite Kripke structures cannot model lack of common knowledge. Approaches similar to ours have been taken by van Emde Boas et al. [VSG] and by 3 Mertens and Zamir =-=[MZ]-=-. In [VSG] an epistemic model is used to analyze the Conway Paradox. Like ours, that model captures an infinite hierarchy of knowledge levels, but it does not have the expressive power of knowledge st... |

145 |
The complexity of reasoning about knowledge and time
- Halpern, Vardi
- 1986
(Show Context)
Citation Context ...ions (K1), (K2), and (K3) generalize naturally. One can also add other natural semantic restrictions; e.g., a restriction that says that each agent's knowledge increases monotonically with time. (Cf. =-=[Le, HV]-=- for the Kripke semantics of knowledge and time.) The above examples suggest that the methodology described in this paper is quite general. This line of thought is pursued in [FV1], which investigates... |

138 |
Authenticated algorithms for Byzantine agreement
- Dolev, Strong
- 1983
(Show Context)
Citation Context ...ean knows that Nixon knows about the Watergate break-in" is much harder. Yet this latter sentence precisely captures the type of reasoning that goes on in proving lower bounds for Byzantine Agree=-=ment [DS]-=-. The need to formally model this type of reasoning is our motivation for constructing a semantic model for knowledge. The first attempt to do so was made by Hintikka [Hi], using essentially the notio... |

130 |
Knowledge and common knowledge in a Byzantine environment i: crash failures
- Dwork, Moses
- 1986
(Show Context)
Citation Context ...f joint knowledge, where S is the set of all agents. Joint knowledge is important in situations where some agents are reasoning about the knowledge shared by certain groups of agents (see for example =-=[DM]). Thus, w-=-e extend our language by adding a new modality C S , for each nonempty set S of agents. Let E S ' be an abbreviation for V i2S K i ' (i.e., "everyone in S knows "'). Furthermore, let E S sup... |

129 | The synthesis of digital machines with provable epistemic properties - Rosenschein, Kaelbling - 1986 |

91 | Programming simultaneous actions using common knowledge - Moses, Tuttle - 1988 |

79 | Approximating common knowledge with common beliefs - Monderer, Samet - 1989 |

65 | Using reasoning about knowledge to analyze distributed systems - Halpern - 1987 |

59 |
Semantical analysis of modal logic. Zeitschrift für Mathematische Logik und Grundlagen der
- Kripke
- 1963
(Show Context)
Citation Context ...ossible worlds. Hintikka's idea was that someone knows ' exactly if ' is true in all the worlds he thinks are possible. Possible-world semantics has been formalized (cf. [Sa]) using Kripke structures =-=[Kr]. In a Kripke structure for kn-=-owledge, the "possible worlds" can be viewed as nodes on a graph that are joined by edges of various colors, one corresponding to each "knower" or "agent". Two possible w... |

41 |
Three views of common knowledge
- Barwise
- 1988
(Show Context)
Citation Context ... E sup ffi + 1' is true in s. Since / is true in all states of M , it follows that / is common knowledge in every state of M . 2 There are other interpretations of the notion of common knowledge. See =-=[Ba]-=-. 17 Remark 4.10: Interestingly, the theorem does not hold if there is only one agent. The intuitive reason is that in that case there are only finitely many distinct knowledge states (given our assum... |

31 | Reasoning about knowledge: An overview - Halpern - 1986 |

31 | On the model theory of knowledge - McCarthy, Sato, et al. - 1979 |

28 | On the Complexity of Epistemic Reasoning - Vardi - 1989 |

19 |
An axiomatic characterization of common knowledge
- Milgrom
- 1981
(Show Context)
Citation Context ...icular, f ! j= '. Again, the proof of the converse is similar. We now describe some axioms for joint knowledge. These are generalization of axioms for common knowledge due to Lehmann [Le] and Milgrom =-=[Mi]-=-. 1. C S ' ) ' 2. C S ' ) C S C S ' 3. :C S ' ) C S :C S ' 4. C S ' 1sC S (' 1 ) ' 2 ) ) C S ' 2 5. C fig ' j K i ' 6. C S ' ) C T ' if T ` S 7. C S (' ) E S ') ) (' ) C S '). Axioms 1-4 are analogous... |

18 |
Formal Theories of Knowledge
- Rosenschein
- 1985
(Show Context)
Citation Context ...inguishable to a processor if it has the same local state in both. This is the situatedautomatasapproach, where knowledge is ascribed on the basis of the information carried by the state of a machine =-=[Ro]-=-. This approach has been used in a number of papers on distributed systems, including [HM1, DM, FHV, MT, RK]. However, there are situations where it is not clear how to use Kripke structures to model ... |

17 | Knowledge and implicit knowledge in a distributed environment - Fagin, Vardi - 1986 |

13 |
An internal semantics for modal logic
- Fagin, Vardi
- 1985
(Show Context)
Citation Context ...ive a complete axiomatization for knowledge structures. Corollary 3.3: The problem of deciding if a formula is satisfiable is PSPACE-complete (provided there are at least two agents). We note that in =-=[FV1]-=-, it is shown how to obtain an elegant, constructive proof of Corollary 3.2, by working only with knowledge structures and not making use of the completeness theorem for Kripke structures. Theorem 3.1... |

12 | Reasoning about knowledge and probability: preliminary report - Fagin, Halpern - 1988 |

11 | On Aumann’s Notion of Common Knowledge, an Alternative Approach - Tan, daWerlang - 1988 |

10 |
Knowledge, common knowledge and related puzzles (Extended Summary
- Lehmann
- 1984
(Show Context)
Citation Context ... all is0. In particular, f ! j= '. Again, the proof of the converse is similar. We now describe some axioms for joint knowledge. These are generalization of axioms for common knowledge due to Lehmann =-=[Le]-=- and Milgrom [Mi]. 1. C S ' ) ' 2. C S ' ) C S C S ' 3. :C S ' ) C S :C S ' 4. C S ' 1sC S (' 1 ) ' 2 ) ) C S ' 2 5. C fig ' j K i ' 6. C S ' ) C T ' if T ` S 7. C S (' ) E S ') ) (' ) C S '). Axioms ... |

10 |
A Study of Kripke-type Models for Some Modal Logics by Gentzen’s Sequential Method. Publications of the Research
- Sato
- 1977
(Show Context)
Citation Context ...ng essentially the notion of possible worlds. Hintikka's idea was that someone knows ' exactly if ' is true in all the worlds he thinks are possible. Possible-world semantics has been formalized (cf. =-=[Sa]) using Kripke -=-structures [Kr]. In a Kripke structure for knowledge, the "possible worlds" can be viewed as nodes on a graph that are joined by edges of various colors, one corresponding to each "know... |

10 |
A model-theoretic analysis of monotonic knowledge
- Vardi
- 1985
(Show Context)
Citation Context ...e knowledge at one level induces positive knowledge at the next level. This would not happen if we modified the definition of knowledge structures by eliminating negative introspection, as is done in =-=[Va1]-=-. Let w = hf 0 ; : : : ; f k i be a (k + 1)-world, and let w + = hf 0 ; : : : ; f k ; f k+1 i be the onestep no-information extension. Note that for each (k + 2)-world hf 0 ; : : : ; f k ; f 0 k+1 i t... |

7 |
Definite reference and mutual knowledge, in Elements of Discourse Understanding
- Clark, Marshall
- 1981
(Show Context)
Citation Context ...only what a processor knows to be true about the world, but also its knowledge about what the other processors know. Such reasoning can, however, get very complicated. As Clark and Marshall point out =-=[CM], while it-=- may be somewhat difficult to keep straight a pipeline of gossip such as "Dean knows that Nixon knows that Haldeman knows that Magruder knows about the Watergate break-in", making sense out ... |

7 |
private communication
- McCarthy, Prescott
(Show Context)
Citation Context ...amma1 (i). From now on, whenever we have to verify that (K3) holds, we will check only the nontrivial direction. It is not obvious that every world is the prefix of some knowledge structure. McCarthy =-=[Mc]-=- posed essentially this question as an open problem in 1975 (in a different framework, of course). In fact, it may not be obvious to the reader that there are any knowledge structures at all. As we sh... |

7 |
The Conway Paradox: Its Solution in an Epistemic Framework
- Boas, Peter, et al.
- 1981
(Show Context)
Citation Context ...on knowledge. An interesting corollary of this investigation is that finite Kripke structures cannot model lack of common knowledge. Approaches similar to ours have been taken by van Emde Boas et al. =-=[VSG]-=- and by 3 Mertens and Zamir [MZ]. In [VSG] an epistemic model is used to analyze the Conway Paradox. Like ours, that model captures an infinite hierarchy of knowledge levels, but it does not have the ... |

5 | What can machines know? on the epistemic properties of machines - Fagin, Halpern, et al. - 1986 |

4 |
An nlog n algorithm for minimizing the states in a finite-automaton
- Hopcroft
- 1971
(Show Context)
Citation Context ...of (mn + 2)-worlds. 25 We note that if ssi;j 0 t for all agents i 2 P, then f s = f t . This notion of equivalence between states in Kripke structures is closely related to the notions of equivalence =-=[Ho]-=- and bisimulation [Pa] between states in finite-state automata. Example 4.25: We now consider a very interesting situation where both no-information extensions and least-information extensions enter t... |

3 | Hierarchies of beliefs and common knowledge. Research Paper No - Brandenburger, Dekel - 1985 |

2 |
Common knowledge with probability one
- Brandenburger, Dekel
- 1987
(Show Context)
Citation Context ...knows a fact p if p holds in all the worlds that the agent consider possible. In a Bayesian setting, an agent knows a fact p if the probability that p holds according to the agent's distribution is 1 =-=[BD1]-=-. (See also [FH, MS] for an approach that mixes Bayesian and non-Bayesian approaches). Mertens and Zamir describe a Bayesian analogue to knowledge structures [MZ], which they call infinite hierarchies... |

2 |
A guide to modal logics of knowledge
- Halpern, Moses
- 1984
(Show Context)
Citation Context ... the axioms of the previous section, with modus ponens and knowledge generalization as the rules of inference, give a complete axiomatization for the Kripke structure semantics of knowledge, while in =-=[HM2]-=- it is shown, again with respect to Kripke structure semantics, that the question of deciding if a formula is satisfiable is PSPACE-complete (provided there are at least two agents). From Theorem 3.1 ... |