## Causality in Bayesian Belief Networks (1993)

Venue: | In Proceedings of the Ninth Annual Conference on Uncertainty in Artificial Intelligence (UAI--93 |

Citations: | 45 - 18 self |

### BibTeX

@INPROCEEDINGS{Druzdzel93causalityin,

author = {Marek Druzdzel and Herbert Simon},

title = {Causality in Bayesian Belief Networks},

booktitle = {In Proceedings of the Ninth Annual Conference on Uncertainty in Artificial Intelligence (UAI--93},

year = {1993},

pages = {3--11},

publisher = {Morgan Kaufmann Publishers, Inc}

}

### Years of Citing Articles

### OpenURL

### Abstract

We address the problem of causal interpretation of the graphical structure of Bayesian belief networks (BBNs). We review the concept of causality explicated in the domain of structural equations models and show that it is applicable to BBNs. In this view, which we call mechanism-based, causality is defined within models and causal asymmetries arise when mechanisms are placed in the context of a system. We lay the link between structural equations models and BBNs models and formulate the conditions under which the latter can be given causal interpretation. 1 INTRODUCTION Although references to causality permeate everyday scientific practice, the notion of causation has been one of the most controversial subjects in the philosophy of science. Hume's critique that causal connections cannot be observed, and therefore have no empirical basis, strongly influenced the empiricist framework and refocused the concept of causality to scientific models as opposed to reality. A strong attack on ca...

### Citations

7074 |
Probabilistic Reasoning in Intelligent Systems
- Pearl
- 1988
(Show Context)
Citation Context ...nherent part of scientific practice. The confusion about the meaning of causality is clearly seen in the domain of probabilistic and decisiontheoretic models based on Bayesian belief networks (BBNs) (=-=Pearl, 1988-=-) and influence diagrams (Howard & Matheson, 1984). On one hand, the directionality of arcs brings into a model asymmetric relations among variables, which some theoreticians have associated with caus... |

1493 |
The Sciences of the Artificial
- Simon
- 1981
(Show Context)
Citation Context ...e rest of the world, one can make a strong philosophical argument that they usually consist of strongly interconnected elements, but that their connections with the outside world are relatively weak (=-=Simon, 1969-=-). This property allows them to be successfully studied in isolation from the rest of the world. Abstractions of systems, used in science or everyday thinking, are often called models. There is a larg... |

1290 | Local computation with probabilities on graphical structures and their application to expert systems - Lauritzen, Spiegelhalter - 1988 |

497 | Causation, Prediction and Search - Spirtes, Glymour, et al. - 2000 |

351 | Influence diagrams - Howard, Matheson - 1981 |

205 | A Theory of Inferred Causation - Pearl, Verma - 1991 |

112 |
The epistemology of a rule-based expert system: A framework for explanation
- Clancey
- 1983
(Show Context)
Citation Context ...in, have indicated that diagnostic rules alone are not sufficient for generating understandable explanations and that at some level a model incorporating the causal structure of the domain is needed (=-=Clancey, 1983-=-; Wallis & Shortliffe, 1984). Usefulness for human interfaces is not the only reason for capturing causality in probabilistic models. As long as the only goal of using a model is prediction of a proba... |

108 | Some practical issues in constructing belief networks - Henrion - 1987 |

94 |
Causality in device behavior
- Simon
- 1986
(Show Context)
Citation Context ..., the method has been extended to non-linear and dynamic models, involving first-order differential equations (Iwasaki, 1988) and was shown to provide a sound basis for qualitative physics (Iwasaki & =-=Simon, 1986-=-) and non-monotonic reasoning (Simon, 1991). 3 CAUSALITY IN BAYESIAN BELIEF NETWORKS It is often the case that, although something is known about the qualitative and statistical properties of a system... |

79 |
Models of discovery
- Simon
- 1977
(Show Context)
Citation Context ...has other clues or well-developed and empirically tested theories of interactions in the system that are based on elementary laws like "no action at a distance " or "no action without c=-=ommunication" (Simon, 1977-=-, page 52). Structural equations may be formed entirely on the basis of a theory or consist of principles derived from observations, knowledge of legal and institutional rules restricting the system (... |

72 |
Causal Ordering and Identifiability
- Simon
- 1953
(Show Context)
Citation Context ...graph of the system by representing each variable in the system by a node and drawing directed arcs to each node from its direct causal predecessors. For the formal description of the procedure, see (=-=Simon, 1953-=-). Example: The causal ordering procedure applied to the model described by (5) will first identify equations e 1 and e 4 as two self-contained structures of degree one. Both equations contain only on... |

40 |
1954) Spurious correlation: a causal interpretation
- Simon
(Show Context)
Citation Context ...nitially shown in the context of deterministic linear models (Simon, 1953). It was demonstrated to apply equally well to logical equations models (Simon, 1952) and linear models with error variables (=-=Simon, 1954-=-). It was shown to provide an interesting basis for treatment of the counterfactual conditional (Simon & Rescher, 1966). Recently, the method has been extended to non-linear and dynamic models, involv... |

29 | Cause and counterfactual - Simon, Rescher - 1966 |

18 | Causal schemata in judgments under uncertainty - Tversky, Kahneman - 1980 |

15 |
On the definition of the causal relation
- Simon
- 1952
(Show Context)
Citation Context ...l ordering originated in econometrics, where it was initially shown in the context of deterministic linear models (Simon, 1953). It was demonstrated to apply equally well to logical equations models (=-=Simon, 1952-=-) and linear models with error variables (Simon, 1954). It was shown to provide an interesting basis for treatment of the counterfactual conditional (Simon & Rescher, 1966). Recently, the method has b... |

12 | A backwards view for assessment - Shachter, Heckerman - 1986 |

7 |
Nonmonotonic reasoning and causation
- Simon
- 2008
(Show Context)
Citation Context ...ar and dynamic models, involving first-order differential equations (Iwasaki, 1988) and was shown to provide a sound basis for qualitative physics (Iwasaki & Simon, 1986) and non-monotonic reasoning (=-=Simon, 1991-=-). 3 CAUSALITY IN BAYESIAN BELIEF NETWORKS It is often the case that, although something is known about the qualitative and statistical properties of a system's mechanisms, the exact functional form o... |

3 | Studies in econometric method. Cowles Commission for Research in Economics - Hood - 1953 |

2 |
Model Based Reasoning of Device Behavior with Causal Ordering
- IWASAKI
- 1988
(Show Context)
Citation Context ...g basis for treatment of the counterfactual conditional (Simon & Rescher, 1966). Recently, the method has been extended to non-linear and dynamic models, involving first-order differential equations (=-=Iwasaki, 1988-=-) and was shown to provide a sound basis for qualitative physics (Iwasaki & Simon, 1986) and non-monotonic reasoning (Simon, 1991). 3 CAUSALITY IN BAYESIAN BELIEF NETWORKS It is often the case that, a... |