David Lewis Causation, in: Papers II p. 160: Causation is not the only causal relation. Regularity analyses have failed; it is time to give up and try something else: a counterfactual analysis. p. 161: Lewis only analyzes token causation. p. 162: Worse still, Only C-events cause E-events ought to mean (d) For every c, if there is some e in E such that c causes e, then c is in C. if only has its usual meaning. But no; it unambiguously means (b) instead! where (b) is For every e in E, there is some c in C such that c causes e. - How can Lewis say that a sentence ought to mean something but means something else? By what standard? Compare: Only humans laugh = for every animal a, if a laughs, then a is human. This is parallel to: Only C s cause E s = for every event a, if a causes an E, then a is a C. And this is (d). There is no reason whatsoever to parse it as (b). p. 163: The vagueness of similarity does infect causation, and no correct analysis can deny it. p. 164: The B s depend counterfactually on the A s iff for all B s b, b holds at all the closest worlds at which an A holds. p. 166: roughly, an event A depends causally an event B iff the propositions that B
occurs / that B does not occur (under any true description) are counterfactually dependent on the proposition that A occurs / that A does not occur (under any description). Then e depends causally on c iff the family O(e), ~O(e) depends counterfactually on the family O(c), ~O(c). Causal dependence is not yet causation. p. 167: Causal dependence among actual events implies causation. If c and e are two actual events such that e would not have occurred without c, then c is a cause of e. But I reject the converse. Causation must always be transitive; causal dependence may not be; so there can be causation without causal dependence. p. 167: A causes B iff there is a causal chain that leads from A to B. A causal chain is causal dependence extended to a transitive relation in the usual way. p. 170: one problem is that if c causes e, c would not have occurred if e had not. Lewis solution is to flatly deny the counterfactuals that cause the trouble ; that is, to deny that c would not have occurred had e not occurred. That is: Drinking a bottle of arsenic caused the death of A. Had A not died, it could also not have been true that A had consumed a bottle of arsenic. Lewis denies the latter. Compare the three scenarios: (1) A drinks a bottle of arsenic and dies (2) A drinks a bottle of arsenic and does not die. (3) A does not drink a bottle of arsenic and does not die. Lewis says that (2) is more similar to (1) than (3); which it is not. p. 171: Under determinism any divergence, soon or late, req[u]ires some violation of
the actual laws. Causation, Postscript p. 172: Lewis account works only if the relation of dependence is assumed to hold between distinct events. p. 173: the farther we depart from actuality, the more we lose control over our counterfactuals. p. 174: problem is that (a) my action of raising the flag causes the rising of the flag (b) the rising and my raising are the same event. Lewis deals with this by distinguishing between events being parts of other events, events being names in terms of other events. He says: My action of raising the flag does not include its rising, it is only characterized in terms of its causal consequence, which it does not include. p. 179: if I change the course of events so that an event gets less likely, I might still cause it by doing so. p. 186: Lewis distinguishes between sensitive and insensitive causation; which is a matter of degree. Causation is sensitive to the extent to which the effect of the cause in question also depends on other things. If the effect of a cause does not depend on many other things, the causation is not very sensitive. This is supposed to explain the difference between - killing / causing to die - making / causing to come to be - breaking / causing to be in pieces etc.
... the causing to die in killing must be of a comparatively insensitive kind. p. 196: an event is more fragile if its identity depends on more variables. We have names for robust events, and such robust events are internally causally structured such that the causation that structures them is sufficiently insensitive. But note that insensitivity concerns types, and fragility concerns token events. An event is fragile to the degree that it would not survive changes in the circumstantial setting. p. 203: late preemption (where the occurrence of an effect of cause A deactivates cause B, which would have cause to the same effect) is a real problem to Lewis analysis. For here the effect would have ensued even if the actual cause would not have gone through (p. 200). According to a counterfactual analysis, A is not a cause, then. p. 206: Lewis introduces a notion of quasi-dependence: e quasi-depends on c if e is a later part of a process P of which c is an earlier part; where P follows a generally specifiable pattern. And he says that c is a cause of e if there is a chain of such dependences from c to e. Lewis claims that such a construal would avoid cases of late preemption. p. 211: It is fair to discover the appropriate standards of similarity form the counterfactuals they make true, rather than vice versa. See also Counterfactual Dependence and Time s Arrow. Events, in Papers II p. 243-4: If an event occurs in a space-time region, - it does not occur in any proper part of this region, - it is occurring in all of its proper parts, and - it occurs within every region that includes the region in which it occurs.
p. 244-5: Lewis first says that for each event there is a corresponding property that belongs to all and only those spatiotemporal regions in which this event occurs. Then he claims that events are these properties. He thus needs an independent way of identifying the respective properties. To say that an event is the property that consists in its occurring in a certain space-time region would not help much. (I guess Lewis wants to say that events are what populates space-time regions such as to fill them with content; just as properties populate property-bearers and give them their shape. Then he should say that events are denizens of space-time regions.) p. 250: events are fragile if it is hard to change them without destroying them. p. 256: Saying Hello and saying Hello loudly are different events, since their causes and effects differ, but they are not distinct events. Causation as Influence (2000) p. 186: Besides whether-whether dependence, there is also when-whether dependence and more; that is whether B occurs depends counterfactually on whether A occurs when B occurs depends counterfactually on whether A occcurs when... when... where... whether... how... whether... how... when... etc. If we draw these distinctions, we do not need an elaborate theory about fragile events. We do not need to put times, places, manners etc. into the essence of an event, but can put them in the description of the causal relation.
p. 190: Where C and E are distinct actual events, let us say that C influences E if and only if there is a substantial range C1, C2... of different not-too-distant alterations of C (including the actual alteration of C) and there is a range E1, E2... of alterations of E, at least some of which differ, such that if C1 had occurred, El would have occurred, and if C2 had occurred, E2 would have occurred, and so on. Thus we have a pattern of dependence of how, when, and whether upon how, when, and whether. That is, very roughly, C is a cause of E if differences in C make a difference for E. p. 196: The problem with absences is that there are too many of them. But this is not really a problem, because in any case it is true that all these absences cause what they make possible. In most cases, it is just not worth mentioning. Also, absences are a special case because here we cannot influence the time, place, or manner in which they occur. They do not occur.