Safety, Reliability and Risk Analysis
Controversial discussions on causality have been present in ancient philosophy since the days of Aristotle. Despite the use of this concept in numerous subjects, there is no consensus on the definition of causality and its possible mathematization. Many authors have analyzed the relation between causes and effects; the predominant school of thought reduces causation to a physical relation (either deterministic or probabilistic) between two events. The distinction between causes and consequences is not always clear and meaningful as different “layers of understanding” may be applied to the notion of causality. From this point of view the cause-effect implication relation can be thought of as a first level representation of causation. By “double-clicking” the link between events, the in-depth layers of causality surface, allowing a better comprehension and distinction of the causality nature. It is then important to understand how causality can be incorporated in an accident model. Several accident models have been recently employed to interpret different contributing factors to an adverse event and to help improve our chances on accident prevention. However, accident models do not focus on the nature of the causal relationship. The causal relationship is always limited to “cause(s) imply effect(s)”, but it is never analyzed as to understand the mechanism that lead to such implication. Our language in itself is limited in the ways of describing causal relationships. We will see how the application of the effective metaphor of Agonist and Antagonist actions from the Force Dynamics framework will help analyzing the roles of different actors along the chain of causation. The use of this metaphor, enhanced by the introduction of the Inverse Agonist concept, will provide new insights on the interactions among those actors and will yield the insightful idea of primitives of causality. These primitives will be primal and fundamental notions at the base of a more general concept of causation. We illustrate the use of primitives of causality through an accident example, and we highlight the absence of relevant antagonist and inverse agonist actions that failed to block and de-escalate the accident sequence respectively. We argue that the primitives of causality here introduced allow a deeper understanding of causal mechanisms involved in system accidents and provide a richer basis for conceiving and articulating accident prevention strategies.
Loïc Brevault, Francesca Favaro, and Joseph Saleh. "On the Primitives of Causality: from the Semantics of Agonist and Antagonist to Models of Accident Causation and System Safety" Safety, Reliability and Risk Analysis (2013). https://doi.org/10.1201/b15938-10