Causality in Device Behavior

Abstract

This paper shows how formal characterizations of causality and of the method of comparative statics, long used in economics, thermodynamics and other domains, can be applied to clarify and make rigorous the qualitative causal calculus recently proposed by de Kleer and Brown (1984). The formalization shows exactly what assumptions are required to carry out causal analysis of a system of interdependent variables in equilibrium and to propagate disturbances through such a system. 2 Whether causal connections among events can be perceived and verified in the real world is a question much debated in philosophy. Formal treatments of the foundations of the sciences for a long time avoided notions of causation and spoke only of functional relations among variables. Nevertheless, in informal descriptions of real-world phenomena, statements of the form, "A causes B," are exceedingly common. It is the purpose of this paper to show what such statements might mean, and how they can be useful in describing the behavior of physical devices.The foundations of our formulation have already been laid in a substantial literature that grew, some thirty years ago, out of concerns with the causal relation in the field of econometrics. 1 The formal definition of the causal relation employed in that literature will serve our purpose of describing physical devices in causal terms. In the course of our analysis, we will show that this definition is nearly the same as a recent explication of the causal relation, by de Kleer ancTBrown (1984); and we will both compare and contrast our own formulation wittvttieirs. We will also show that the procedures proposed by de Kleer and Brown for the propagation of causal disturbances are closely akin to classical methods of comparative statics used widely in thermodynamics and economics. 2

Function in Device Representation

“…In this form of explanation, we display how the behaviours in F causally depend upon the properties of the components or the relations in D, and if F is environment-centric, also on M(D,E) and E. The general way to show this is by use of a technique known as causal ordering [6,7]. Causal ordering identifies which variables causally precede which other variables, and in what order.…”

Model Based Diagnostic Module for a FCC Pilot Plant

“…In this paper, the causal ordering framework of Iwasaki and Simon (1986) later extended within a graph theoretic framework (Porte et al, 1988) and in (Travé-Massuyès and Pons, 1997) for multiple mode systems has been adopted, due to its practical feasibility. This approach makes the most of its advantage by operating from the equations structure, hence only requiring a structural relation model (SRM) as initial knowledge.…”

Agent-based simulation of open source evolution

“…The system's behaviour is abstracted into a set of differential equations which define the structure of the causal links between the state variables in the model (Iwasaki & Simon 1986); this describes the mechanisms that are believed to operate in the referent system. The set of equations is solved to produce a behaviour, which is compared to the empirical data.…”