Causal Chaining

Abstract

In Chapters 3 – 5, we focused on the simplest type of diagnostic problems where the underlying causal networks consist of only two layers (sets M and D) and developed problem-solving algorithms for these problems. It was assumed there that disorders are directly causally-associated with measurable manifestations. In contrast, in many real-world diagnostic problems indirect causal associations between disorders and manifestations occur through causal chaining of intermediate states: “d causes s” and “s causes m” may be two existing causal associations which, during problem- solving, may be chained together to form “d causes m”. For example, in diagnosing a plumbing system, the manifestation m = “no water pressure at faucet 6” might be caused by abnormal state s = “pipe 17 is blocked”, and this in turn might be caused by the disorder d = “frozen water in pipe 17”. In medicine, manifestation m = “left hemiparesis” (weakness on the left side) might be caused by s = “right cerebral hemisphere damage”, which in turn might be caused by disorder d = “right intracerebral hematoma” (bleeding into the cerebrum or brain).

“Where there is a casual sequence or chain of several events. A causing B, B causing C, C causing D, and D causing E, we can regard E as the effect of any or all of the preceding events.”

I. Copi, Introduction of Logic