Causal Direction in Causal Bayes Nets
Eva, Benjamin
Some authors maintain that we can use causal Bayes nets to infer whether or by consulting a probability distribution defined over some exogenous source of variation for or . We raise a problem for this approach. Specifically, we point out that there are cases where an exogenous cause of ( ) has no probabilistic influence on no matter the direction of causation—namely, cases where and are probabilistically indistinguishable. We then assess the philosophical significance of this problem and discus
