Causality in Bayesian Belief Networks

By Marek J Druzdzel, Herbert A. Simon. 1993

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.

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

Citations in the corpus

Learning Bayesian networks: The combination of knowledge and statistical data (1994, methodology): “An efficient polynomial algorithm for finding a maximum branching was first described by Edmonds (1967), later explored by Karp (1971), and made more efficient by Tarjan (1977) and Gabow et al. (1984). The general case (l > 1) was treated by Camerini et al.”
Learning Bayesian networks: approaches and issues (2011, background): “There has been much confusion over when utilizing causal networks is applicable (Druzdzel & Simon, 1993).”
Explanation in Probabilistic Systems: Is It Feasible? Will It Work? (2001, unknown): “Simon and I [13] tied the work on structural equation models in econometrics and AI to probabilistic models and formulated the semantic conditions under which a Bayesian belief network is causal.”
Axiomatizing Causal Reasoning (1998, unknown): “…prediction, explanation, and counterfactual reasoning—has been argued eloquently in a number of recent papers and books (Chajewska & Halpern, 1997; Heckerman & Shachter, 1995; Henrion & Druzdzel, 1990; Druzdzel & Simon, 1993; Pearl, 1995; Pearl & Verma, 1991; Spirtes, Glymour, & Scheines, 1993).”
Caveats for Causal Reasoning with Equilibrium Models (2001, background): “Since this graph is Markovian through time i.e., the variables in the future are d-separated from variables in the past by variables in the present, it can be represented by a convenient shorthand graph for an infinite sequence of time steps.”
Defining Explanation in Probabilistic Systems (1997, methodology): “We suggest an approach that combines what we feel are the best features of these two definitions with some ideas from the more re cent work on causality (Balke and Pearll994; Druzdzel and Simon 1993; Beckerman and Shachter 1995; Pearl 1 995).”
Counterfactuals and Policy Analysis in Structural Models (1995, background): “[Druzdzel and Simon, 1993], for example, state: "A vari able is considered exogenous to a system if its value is de termined outside the system, either because we can control its value externally (e.g., the amount of taxes in a macro economic model} or because we believe that this variable is…”
Counterfactual Probabilities: Computational Methods, Bounds and Applications (1994, background): “…is exogenous to the other relationships consideredin the analysis.Causal theories speci ed in functional form (as in[Pearl and Verma, 1991, Druzdzel and Simon, 1993,Poole, 1993]) are su cient for evaluating counterfac-tual queries, whereas the causal information embed-ded in Bayesian…”