Known Errata for "Causation" (2014) Polity Press The second question on p. 51-52 should be replaced with the following paragraph. In the atomic decay example depicted in Fig. 2, the presence of B lowered the probability of D from what it would have been if B had not existed (holding A's presence fixed). This example is commonly believed to refute the theory that one event c causes another event e just in case c raises the probability of e. After all, B is intuitively a cause of D yet lowers D's probability. One maneuver to evade this counterexample is to say that causation can proceed through causal links. On this revised theory, one event c causes another event e just in case there is a sequence of events starting at c and ending at e where each member of the sequence raises the probability of the next event. The atomic decay example is no counterexample to this revised theory because there exists a chain of probability-raising from A to B to D. The presence of A raised the probability of B above what it would have been without A, and B raised the probability of D above what it would have been without B (holding fixed the absence of C). So the revised theory implies the seemingly correct judgment that A is a cause of D. Can you construct a counterexample to this revised theory of causation as probability-raising? On pp. 76-77, the paragraph beginning with "Let's construct a specific counterexample …” should be replaced with the following 3 paragraphs. For a second example, let us attend to a different peculiarity of counterfactual logic: that its truth value is binary and crucially depends on what happens in \emph{all} of the relevant counterfactual worlds. Suppose at some past instant of time, the actual state of the world s included a tiny event c---like the presence of some particular particle at some location. By way of the fundamental laws, s entailed that the fundamental chance of some later decay event e was equal to 99 out of 100. By happenstance, e did actually occur. Suppose also that the fundamental laws dictate that for any state of the world exactly like s except without the event c, the fundamental chance of e is equal to 1 out of 100. It strikes me that the natural way to think about this situation is that the presence of c made e much more likely than it would have been otherwise, so that c should count as a cause of e. But the counterfactual account of causation appears to imply that c is not a cause of e. Because of the way the standard logic of counterfactuals is structured, we must reckon the truth value of "If c had not occurred, e would not have occurred," by ascertaining whether in absolutely \emph{all} the most-similar possible worlds where c does not occur, e also does not occur. And in this scenario, it seems that while e does not occur in 99 out of every 100 of the relevant worlds, e does occur in 1 out 100 of the relevant worlds. The non-existence of c does not \emph{ensure} the non-existence of e.\footnote{As an exercise, try to make explicit any background assumptions I made when inferring that some fraction of the relevant counterfactual worlds have event e occurring in them. A defender of the counterfactual account of causation could question these assumptions.} Because the counterfactual account of causation requires that e's non-existence be strictly ensured rather than be rendered extremely probable, it wrongly categorizes c as not being a cause of e. What the counterfactual account of causation is missing, by having a binary true or false value for each counterfactual, is a more finely-grained value that can track the degree to which the chances of events can depend on each other. On pp. 66, the paragraph beginning with "In trying to understand this procedure …” should be replaced with the following paragraph. In trying to understand this procedure, it helps to make two observations about the conceptual setup. First, we are supposed to evaluate all relevant counterfactual conditionals by examining them as stand-alone statements without presupposing any causal information. Because the goal of counterfactual accounts of causation is to analyze (or define) causation in terms of counterfactuals, one must avoid circularly defining causation in terms of our pre-theoretical judgments about causation. Second, I have followed Lewis in distinguishing between an event and the proposition (or statement) that that event occurred. This is noteworthy because in Lewis’s theory, the logic governing counterfactual conditionals in ordinary language provides a structure that is intended to restrict which counterfactual dependencies hold. (In the same year as his article on causation, Lewis published a book characterizing a family of counterfactual logics, and defending several logical rules that apply to ordinary language counterfactuals.)