Critical Notice: Causality by Judea Pearl

This is a wonderful book; indeed, it is easily one of the most important and creative books I have ever read on the subject of causation and causal inference. Causality is impressive on many levels and should be of great interest to many different audiences. Philosophers will find of particular interest Pearl’s defense of what might be described as a broadly manipulationist or interventionist treatment of causation: Causal claims have to do with what would happen under ideal, suitably surgical experimental manipulations (‘interventions’). But Pearl moves far beyond philosophical generalities on this theme: He makes the case for the approach that he favors by developing a formal apparatus (‘a calculus of interventions’, as he calls it) for talking about interventions and how they relate to causation and information about probabilities that is at once both intuitively compelling and genuinely useful for purposes of calculation and estimation. He also actively engages the philosophical literature on causation at many points – for example, there are illuminating discussions of evidential decision theory, of probabilistic theories of causation of the sort favored by Eells and Cartwright, of Lewisian counterfactual theories of causation, and of the relationship between so-called type and token causal claims, as well as a quite novel treatment of the latter. For those interested in foundational questions at the intersection of econometrics, causal inference and statistics, Pearl offers an account, in terms of claims about the outcomes of hypothetical experiments, of what systems of structural equations mean (or should be understood to mean) when interpreted causally. This account draws on ideas that go back to such econometricians asHaavelmo, Frisch, Strotz andWald, amongothers, but Pearl expresses these ideas in a way that seems to me to be clearer and analytically sharper and relates them successfully to much more