Causal reasoning with continuous uncertain quantities
Jacob Pearl · 2003
The author considers an idealized model of causal schemata that facilitates the estimation of uncertain continuous variables by local and unsupervised computations. The updating rules dictated by this model reveal general strategies of reasoning, suitable for combining estimates from diverse, unreliable sources. The rules admit flexible control strategies and facilitate the generation of intuitively meaningful explanations, with a minimum of arithmetic machinery.>