LAWRENCE BRIAN LOMBARD
CAUSES, ENABLERS, COUNTERFACTUAL
AND THE ANALYSIS
(Received in revised form 21 March, 1989)
In this paper, I discuss a problem for the counterfactual analysis of event causation. While it may well be that the counterfactual analysis is subject to insuperable difficulties,1 I wish to argue that the difficulty that is to be discussed here may n o t be insuperable and that the counterfactual analysis may be revisable in a way that avoids the difficulty. In addition, I try to explain an interesting asymmetry discovered by Jonathan Bennett. My reason for doing that is that what explains that asymmetry also points the way to the solution to the problem for the counterfactual analysis I shall be discussing. I. T H E C O U N T E R F A C T U A L
ANALYSIS OF EVENT CAUSATION
The cotmteffactual analysis of event causation (referred to here sometimes as 'CAEC'), whose most prominent champion is David Lewis, z proposes to give conditions necessary and sufficient for one event to be a cause of another. They are conditions under which an individual e v e n t would not have occurred. It is n o t concerned with statements like 'he failed the course because he did not study', insofar as they assert a causal relation between f a c t s and not events? Nor is it concerned with conditionals like 'if he had studied, he would have passed', which would, I suppose, analyze, according to a counterfactual analysis of f a c t causation, singular causal statements relating facts. 4 According to the counteffactual analysis of event causation, (C)
Event c is an immediate cause of an event e if and only if it is the case that if c had not occurred, e would not have occurred (e is counteffactually dependent on c). And c is a remote cause of e if and only if there is a sequence of events, c, c I . . . . , cn, e, such that c is an immediate cause of C l , . . . and cn is an immediate cause of e.
Philosophical Studies 59:195--211, 1990. 9 1990 Kluwer Academic Publishers. Printed in the Netherlands.
196
LAWRENCE BRIAN LOMBARD
Remote causation cannot be analyzed directly in terms of counterfactual dependence, since, while the causal relation is transitive, the counterfactual conditional is not. C A E C requires revision if it is to accommodate, and avoid refutation based on cases of, causal overdetermination. It might be the case that, while c is an immediate cause of e, there is some other event, c', which is such that, had c failed to occur, c' would have, and c' would have been an immediate cause of e. But, despite its being false that had c not occurred, e would not have occurred, we would not deny that c was a cause of e. Also, it has been claimed, for example, that my becoming an uncle was not caused by my brother's wife's giving birth, even though the counterfactual, 'Had my brother's wife's giving birth not occurred, my becoming an uncle would not have occurred', is true. 5 C A E C might be saved from counterexamples like this by some further revision, or by insisting that we do have here a cause of causation, or by denying that phrases like 'my becoming an uncle' and its ilk refer to events. 6 But these and the other standard difficulties for C A E C and the revisions and complications that may be introduced in an attempt to deal with them shall be ignored here. The problem with the counterfactual analysis I wish to discuss has a different genesis. lI. PROBLEMS FOR THE COUNTERFACTUAL ANALYSIS Consider the following case, adapted from a recent paper by Jonathan Bennett: 7 In April, there was a heavy rainstorm, and in May and June there were electrical storms. The lightning in June started a forest fire. If the rainstorm in April had not occurred, there would have been a forest fire in May. Now suppose we assume that no event can occur at a time other than the time at which it in fact occurs .8 It will follow that (a)
If the rainstorm in April had not occurred, the forest fire (the one in June) would not have occurred.
The reason for this is that, though there would have been a forest fire (in May), it would not have been the one that actually occurred, since it
CAUSES AND ENABLERS
197
would have occurred earlier than the actual one. But (a) and the counterfactual analysis of event causation together imply (b)
The rainstorm in April caused the June forest fire.
A n d (b) is unacceptable. It is a bit of good c o m m o n sense that heavy rains can put out fires, they don't start them; it is false to say that the rains caused the fire? David Lewis too has expressed concern about combining C A E C with the thesis that the time of occurrence of an event is essential to it: Who would dare to be a doctor, if [the time of an event is essential to it]? You might manage to keep your patient alive until 4:12, when otherwise he would have died at 4:08. You would then have caused his death. For his death was, in fact, his death at 4:12. If that time is essential, his death is an event that would not have occurred had he died at 4:08, as he would have done without your action. That will not do.~~ These arguments are valid. The counterfactual analysis of event causation, as currently formulated, is incompatible with the thesis that an event cannot occur earlier (or later) than it does. Which of these two claims should we give up? There are undoubtedly m o r e adherents of the counterfactual analysis of event causation than there are believers in the thesis that no event can occur earlier than it in fact does, and m a n y of those who have considered the latter thesis have rejected it. 11 However, if one thinks that this problem for C A E C - - that it is false if the time of events is essential - - can be avoided by rejecting this essentialist thesis, one is, I think, mistaken. 12 F o r C A E C , as currently formulated, must be rejected anyway. The counterfactual analysis seems to be in trouble from other cases similar to the one mentioned above, though these other cases do not involve acceptance of my essentialist thesis concerning the time of occurrence of an event. So, rejection of that essenfialist thesis will not save C A E C from the problem that the essentialist thesis raises. Supose that Jones lives in a very dangerous neighborhood, and that one evening Smith attempts to stab him to death. Jones is saved because of the action of Brown who frightens Smith off. However, a year later, Jones is shot to death by the persistent Smith. So, if Brown's action had not occurred, Jones's death due to the shooting would not have occurred, since he would have died of stab wounds a year earlier. But, I find it intuitively quite unacceptable to suppose that Brown's
198
LAWRENCE BRIAN LOMBARD
action was a cause of Jones's dying as a result of a gunshot a year later. 13 If so, there is something wrong with CAEC, and my essentialist thesis plays no role in the derivation the objectionable consequence. Again, suppose we assume that the time of an event is not essential to it. In that case, it seems true that had Caesar's birth not occurred his death would not have occurred, because in such a case Caesar would not have been born at all. A n d it surely seems true to the defenders of the counterfactual analysis that there is a chain of causal dependencies stretching from Caesar's birth to his death. But I am not the least inclined to think that Caesar's birth was among the causes of his death. Regardless of one's position on the thesis that events occur essentially when they do, there simply is a distinction with which any version of the counterfactual analysis of event causation will have to take note of and accommodate. And that essentialist thesis is in no difficulty from a version of the counteffactual analysis that accommodates that distinction. For it will turn out that if C A E C is revised to take account of that distinction, the revised version will be neither subject to difficulty from the last two cases, which do not assume t h a t essentialist thesis, nor incompatible with that essentialist thesis. The offensive claims, that the rain in April was a cause of the forest fire, that life-savers' actions cause their patients' deaths, and that births cause deaths, are derivable principally because the counteffactual analysis of event causation, as currently formulated, incorrectly rules certain events to be causes of others. And it so rules, because it fails to make and then to take account of a certain distinction between ways in which events may fit into the causal history of another event.
III. DELAYERS Despite my view that the forest fire that actually occurred is not the same fire that would have occurred earlier, had it not rained in April, there is a sense in which the rain in April "delayed" the forest fire by a month; it delayed the occurrence of a forest fire, even though there was no forest fire such that the rain made it occur later. The rain in April delayed the occurrence of the forest fire, though it did not cause the fire it delayed. We can formulate the idea of a delayer in the following way: (D)
An event, c, is a delayer of an event, e, just in case e is of
CAUSES AND ENABLERS
199
type P, and had c not occurred, an event, e', of type P would have occurred anyway, but at a time earlier than the time at which e in fact occurred. 14 Delayers are generally not causes of what they delay. That this is so is one half of a thesis that Bennett dubbed "the asymmetry fact", the other half of which is that "hasteners" generally are causes of what they hasten. 15 Bennett, however, did not explain why he thought that fact to be a fact. It is one of my goals here to provide at least part of the explanation. 16 (b) is not true; it seems absurd to say that delaying a forest's burning is causing it to burn. Why does it seem absurd? Perhaps because to delay is in part to prevent, and to prevent is to "uncause". But delaying is only in part preveming; to delay an event of a certain type is to make an event of that type occur at a time later than an event of that type would otherwise have occurred. But this seems to imply that to delay an event is to make it happen, and that is, it would appear, to cause it) 7 I have no doubt that (b) false. But to see what in the derivation of (b) needs to be rejected, we must see just what it is about (b), which says that a delayer of a certain event is a cause of that event, that is irksome. The previous paragraph suggests, however, that (b) /s true, and that delayers are causes of what they delay. So, how is it possible for delayers to fail to be causes of what they delay? Bennett's case involves the following. The rains made the forest wet; so another month had to pass for the forest to be dry enough to allow lightning to start a forest fire. Thus, the April rain actually caused the onslaught of a condition (the wetness of the forest) the presence of which prevented lightning in May from causing a fire and the absence of which in June made it possible for lightning in June to cause a fire. And this is enough to convince me that (b) is false; and it also suggests to me the reason why that is so. IV. H A S T E N E R S
AND DELAYERS
In a sense analogous to (19), an event may hasten the occurrence of another: (H)
An event, c, is a hastener of an event, e, just in case e is an event of type P, and had c not occurred, an event, e', of type
200
LAWRENCE
BRIAN LOMBARD
P would have occurred anyway, but at a time later than the time at which e in fact occurred. A n d it seems clearly right to suppose that an event that hastens the occurrence of another may well be a cause of the other. The fatal shooting of a person surely hastens the victim's death and is clearly a cause of it. 18 But, while hasteners may often be causes of the events they hasten, the delayers of events are generally not their causes. What explains why this is so? A patient is brought to the emergency room, having recently been poisoned. A doctor administers an antidote, thus saving the patient's life. But the patient is allergic to the antidote and dies of the allergic ,reaction to it. The doctor's action delayed the patient's death. What makes this case one in which a delayer of a certain event is also a cause Of it is that the giving of the antidote (the delayer) was an event that caused the allergic reaction, which, in turn, caused the patient's death. Thus, in the sense of (D), the delayer of the death was also a cause of it. If there are to be cases of delayers that are not causes of what they delay, the causal pattern will have to be different. Indeed, if it delayers are generally not causes of what they delay, it will have to be that the causal pattern exhibited by the case just described is an unusual one. But two ideas suggest that that pattern cannot be unusual. First, as mentioned, to delay an event of a certain type is to make it occur later than an event of that type would otherwise have occurred; and that seems to imply that to delay an event is to make it happen, that is, to cause it. The second idea that suggests that delayers must be causes of the events they delay involves the combination of C A E C and the thesis that events cannot occur earlier than they do. Let e be an event delayed by c; if c had not occurred, and if e could not have occurred earlier than it did, then, while an event, e', much like e, would have occurred (for c delayed e), e would not have occurred (for e could not have occurred at the earlier time of e'). So, if c had not occurred, e would not have occurred, Thus, C A E C rules the delayer, c, to be a cause of the delayed event, e. Therefore, delayers must be causes of what they delay. Since (b) is false, there is at least one delayer that is not a cause of the event it delays; so, this reasoning must be unsound. And it is
CAUSES AND ENABLERS
201
CAEC, as formulated by (C), that is responsible for the unsoundness, not the thesis of the essentiality of an event's time of occurrence. It can happen that an event, c, can prevent the occurrence of an event, e, and also be a cause of an event, e', that is in some salient respect like e. In such a case, c is a delayer and a cause of the delayed effect e'. But it can also happen that c prevents e from occurring and an event like e does later occur, due in part (in some sense) to the occurrence of c, but where c is n o t a cause of the delayed event. And such cases are not only common, they are, I think, usual. How are such cases possible? V. C A U S E S A N D E N A B L E R S
The solution to our problem requires that the counterfactual analysis find a way to cope with the distinction between the causes of events and those conditions or states the obtaining of which merely makes it possible for one event to cause another. One reason for making this distinction is that, according to the counterfactual analysis of e v e n t causation, causes and effects are events. States of objects and conditions, however, are n o t . Events are changes; 19 and an object's being blue, a piece of salt's being soluble, and the structural weakness of a bridge are n o t changes. Causes make things happen; they are bringings about. What is brought about must not have been there before; so, what is brought about, an effect, is a change. And it is hard to see how a change can be brought about except by another change. Causes and effects, then, are changes, that is, events. Of course, if the match hadn't been dry, it would not have lit. But from this, I do n o t want to infer that the match's being dry was a cause of the match's lighting, for the dryness of the match is not an event (for it is not a change) and only events are causes (according to the view we are considering). In addition, at least according to standard views of event-causation, there are to be no temporal gaps between causes and their effects that are not filled by intermediary causes and effects. But the match had been dry (and in sufficient oxygen, etc.) for quite some time before it lit. Why did the match not light when it became dry? Because, I contend,
202
LAWRENCE BRIAN LOMBARD
the cause of its lighting, the striking, had not yet occurred. The match's being dry was not a mere "non-salient" cause of its lighting; it was not a cause of the lighting at all. It was just one of those conditions the obtaining of which made it possible for an event, a striking of the match, to cause it to light. Any version of C A E C that ignores the distinction between causes and those conditions the presence of which make it possible for one event to cause another will be unable to distinguish between an event that is a cause of effect, e, and an event that causes a thing to be in a state that makes it possible for an event to cause e. And an event that causes a thing to be in a state that makes it possible for an event to cause e -- a n e n a b l e r - - is generally not, I contend, a cause of e. But both the causes of events and the causes of enablers of events make counterfactuals true: it is not only true that if the match's striking had not occurred the match's lighting would not have occurred, it is also true that the lighting would not have occurred if whatever caused the match to become dry had not occurred. CAEC's failure to incorporate this distinction leads it to rule falsely that the rain in April was a cause of the fire. But all the rain in April did was to make it impossible for lighming in May to cause a fire (it prevented, it d i s e n a b l e d , a forest fire in May) and to make it possible for the lightning in June to cause a fire. The April rain made it impossible for lightning in May to cause the fire, for the rain in April made the forest too wet to burn even after a month. And the rain in April made it possible for lightning in June to cause a fire then, since if it hadn't rained in April, the forest would have been dry enough in May for the lightning in May to cause the forest to burn; and a forest that is completely burnt in May cannot burn again in June. Delayers are generally not causes of the events they delay, for a delayer generally causes a thing to be in a state the obtaining of which makes it possible for another event to cause the delayed event at a time later than that at which an event similar to the delayed event would have occurred had the delayer not occurred. VI. D I S T I N G U I S H I N G
CAUSES AND ENABLERS
Suppose that substances with molecular structure S are soluble in water
CAUSES AND ENABLERS
203
and that one can make a substance have structure S by sprinkling Dust on it. Sprinkling Dust on a substance, then, is an enabler; it is a cause of something's coming to be in an enabling condition, a state (the state of being soluble) the obtaining of which makes it possible for some event (a putting of the substance in water) to cause a certain effect (the dissolving of the substance). And suppose that I sprinkle some Dust on a substance and then toss the substance in some water, whereupon it dissolves. Sprinkling Dust on the substance was not a cause of its dissolving; all it did was produce in that substance a capacity, a capacity to dissolve if dropped in water. And that capacity is a capacity to be affected in a certain way if acted on in a certain way. The notion of a capacity is "multi-layered". "Primary" capacities are capacities to be affected in certain ways if acted on in certain ways. "Secondary" capacities are capacities the activation of which makes it possible for other capacities to be realized by causing objects to have those other capacities. 2~ I am locked in my study; so, of course, I cannot get to the kitchen. Nevertheless, I can get to the kitchen: if the door were to be unlocked I could get to the kitchen. Similarly, a substance without structure S lacks the primary capacity to dissolve; but it has the secondary capacity to dissolve, since it has the capacity to have Dust sprinkled on it, and if that were done, it would acquire the primary capacity to dissolve.
Causes are, roughly speaking, activators of primary, not secondary, capacities. Sprinkling Dust on a substance that is to be dropped into water is not a cause of the substance's dissolving and unlocking the door is not a cause of my getting to the kitchen. And in the same sense, my poisoning of the tea you are about to drink is not a cause of your death by poisoning and my placing of an explosive device, which must still have its timer set, under the fibrary is not a cause of the library's blowing up. A primary capacity is a condition that makes it possible for one event to cause another; it is a disposition of a substance to be affected in certain ways. A secondary capacity is a condition that makes it possible for a thing to come to have a primary capacity. The solubility of salt is a primary capacity; it makes it possible for a dropping of some salt into water to cause it to dissolve; it has the disposition to dissolve if placed in water. But an otherwise insoluble substance can dissolve, in
204
LAWRENCE BRIAN LOMBARD
the sense that if Dust were sprinkled on it, it would become soluble; it would then acquire the disposition to dissolve if put in water. An enabler is a cause of a thing's acquiring of a primary capacity, of an enabling condition. A pure enabler is an event the occurrence of which makes it causally possible for a certain effect to occur, but which does not itself bring about that effect; it is an event that merely causes a thing to acquire a primary capacity, but does not also activate that capacity. A disenabler is an event that causes a thing to lose a certain primary capacity. The occurrence of a disenabler prevents the occurrence of an effect, a certain change, in a thing that would have occurred, had the disenabler not occurred, by causing the thing to lose the capacity to change in that way. VII. WHAT IS THE DIFFERENCE BETWEEN AN ENABLER AND A REMOTE CAUSE? The first domino in a row is pushed over, and its going over is a remote cause of the fourteenth domino's going over. And it is true that had the first domino's going over not occurred, the fourteenth domino's going over would not have occurred (barring overdetermination). The first domino's going over is a cause of, and is not merely an enabler of, the fourteenth domino's going over. But its going over seems to make the fourteenth domino's going over possible. Is it both a cause and an enabler? No, The first domino's going over does not make it possible for the thirteenth domino's going over to cause the fourteenth domino's going over. The thirteenth domino's going over, had it occurred, would still have caused the fourteenth to go over, even if the first domino's going over had not occurred. It is true, of course, that had the first domino not fallen over, the thirteenth would not have fallen over; and neither would the fourteenth. That fact notwithstanding, if the thirteenth domino's falling over had occurred, so would the fourteenth domino's falling over. The first domino's going over does not make it possible for the thirteenth domino's going over to cause the fourteenth domino's going over; rather, it causes the thirteenth domino's going over.
CAUSES AND ENABLERS
205
Remote causes are not enablers; they are remote activators of primary capacities. This contrasts with the case of killing by poisoning someone's tea. Even if the tea isn't poisoned, it still gets drunk; poisoning the tea does not cause the tea to be drunk. It is just that without the poisoning of the tea, the drinker does not get poisoned and die. Poisoning the tea, then, isn't a remote cause of the death; it is an event that causes the tea to be in such a state that drinking it causes the death. On the other hand, it is false to say the following: the first domino's going over isn't a remote cause of the fourteenth domino's going over, it merely causes the thirteenth domino to be in such a state that its falling over causes the fourteenth domino's going over. Rather, it causes the thirteenth domino's going over. There is a difference in the causal role played by a genuine cause of a remote effect such as the first domino's going over, and an enabler, such as the placing of the thirteenth domino in the proper position (so that when it falls it falls into the fourteenth). The difference seems to involve the idea that there is no causal chain of events leading from a pure enabler of a certain effect to that effect. There is a sequence of causes and effects that leads to the placing of the thirteenth domino in its proper position. This sequence terminates in that domino's being in the state of having a certain spatial location relative to the other dominos. And there the causal story (apart from irrelevant effects such as the disturbances of small particles) of that causal sequence comes to an end. Everything is now right for another sequence of events, involving the falling over of the first domino, to lead to the fourteenth domino's falling over. And in that sequence of events, the placing of the thirteenth domino in its proper position has no place. In a similar way, the causal story of the tea-poisoning ends when the tea has become poisonous. VIII. T H E A S Y M M E T R Y F A C T
An event can be a m e r e disenabler of a certain effect, by simply being the cause of a thing's losing of the primary capacity to undergo the change which is that effect. The rainstorm in April is a disenabler of the
206
LAWRENCE BRIAN LOMBARD
forest's catching fire in May, for it causes the loss of the forest's capacity to burn then by causing the loss of an enabling condition of the burning, the forest's dryness. But a single event can be both a disenabler of a certain effect, by being the cause of a thing's loss of the capacity to undergo the change which is that effect, a n d an enabler of a similar effect, by being a cause of the thing's acquiring of that capacity at a later time. Suppose, for example, that a substance is soluble if it has either molecular structure S or molecular structure S' and that sprinkling Gunk on a substance with structure S causes it to lose structure S and also causes it to acquire structure S' at a later time; and suppose that Gunk is sprinkled on a substance that is then placed in water and dissolves. Sprinkling Gunk on the substance is a disenabler of the dissolving that would have occurred had Gunk not b e e n sprinkled; and it is an enabler of the actual dissolving. In this case, the sprinkling of Gunk is a cause neither of the dissolving that would have occurred if Gunk had not been sprinkled on the substance (it already had structure S) nor of the dissolving that did occur. The April rain, in Bennett's case, counts as both enabler and disenabler. The rain disenabled the May fire by making the forest too wet to burn then. But it also enabled the June fire, for, in preventing the May fire, it made it possible for lightning in June to have something unburnt to ignite. One and the same event can be both a disenabler of a certain effect, by being the cause of a thing's loss of a certain capacity, a n d a cause of a similar effect. The administering of an antidote to a poison, an antidote which is also poisonous to the patient, prevents a death of the patient due to the activity of the first poison and causes a later death due to the activity of the poisonous antidote. The administering of the antidote is a disenabler of a death and it is also a cause of (a distinct) death. It is a delayer of the patient's death. Cases such as this last one, where an event is both a disenabler of a certain effect and also a cause of a later, similar effect, seem to me to be rather hard to come by. Insofar as this is so, we have an explanation of Bennett's insight that delayers are generally not causes of what they delay. But for a hastener to be a cause of an event it hastens, it is sufficient
CAUSES AND ENABLERS
207
that the hastener cause the occurrence of an event belonging to a certain sort, where another event of that sort would have occurred later anyway. And that is not at all rare. But hasteners are not always causes of the events they hasten. An event may hasten the occurrence of another by causing the loss of a disenabling condition that would have been removed later. For example, a certain damp log, which is next to a fire, wilt not ignite until it is dry; it is left out in the sun and its drying out in that manner will take two days. However, I take a hair dryer to the Iog, and it dries out in three hours, whereupon it ignites. My action hastens the log's catching fire, but is n o t a cause of it. Blowing moderately hot air on a thing is not a cause of its catching fire; it just makes is possible for it to catch fire, it makes it flammable. A delayer that is also a cause of what it delays would have to be an event that (i) causes the onslaught of a disenabling condition (which prevents a cause from having its effect), and (ii) activates a condition enabling the effect (that is, either by itself or by some other event, brought about after the disenabling condition is brought about) by causing an event whose occurrence was (in the sense of (D)) first disenabled and then enabled. That is quite a lot of work for a single event to do; and it is reasonable to think that events like that are rather rare. Delayers are usually not causes of the events they delay. A hastener that is not also a cause of what it hastens would have to be an event that (i) causes the toss of a disenabling condition that would have ceased obtaining anyway, though later, and (ii) causes the onslaught of an enabling condition. And while such cases are not uncommon, they seem less common that cases of hasteners that are causes of what they hasten. If this is right, then we have an explanation of this asymmetry: while hasteners generally are causes of what they hasten, delayers generally are not causes of what they delay. IX. T H E C O U N T E R F A C T U A L
ANALYSIS
Enabters are usually not causes of the events they enable. Their occurrence, though, makes certain counterfactuals relating them and the events they enable true; and the counterfactual analysis of event
208
LAWRENCE
BRIAN LOMBARD
causation takes that fact as evidence for the claim that they are causes. What this shows is that C A E C , as formulated by (C), must be given up. If a version of C A E C is to avoid this difficulty, it must be sensitive to the different ways in which events m a y fit into the causal history of other events. There is a chain of counterfactual dependencies that leads from a person's birth to his or her death. So, C E A C rules the birth to be a cause of the death. A n d such a ruling is surely false. The birth was not a cause of the death; it was merely a cause of an enabling condition of the death. Nothing has the capacity to die unless it is alive, for being alive is an enabling condition of dying, and the birth is a cause of a person's coming to be alive. Similarly, since all humans are mortal, a doctor who saves a patient's life only (though not merely) delays the patient's eventual death. This fact, however, does not by itself m a k e the life-saving actions of doctors causes of their fortunate patients' eventual deaths. They m a y merely make it possible for later events to cause those deaths; and that's not so bad. The physicians have nothing to fear from the metaphysicians. If a counterfactual analysis of event causation is to be saved from counterexamples like these, the analysis must not count as a cause of a given effect any event that is merely a cause of the onslaught of a condition that makes it possible for some event to cause that event. It must not count m e r e enablers as causes of the events they enable. The counterfactual analysis (C) asserts that an event, c, is a cause of an event e just in case either c is an immediate cause of e - - in which case, if c had not occurred, e would not have occurred - - or c is a remote cause of e - - in which case, there is a chain of immediate causes and effects connecting c and e. A mere enabler of e, however, cannot be an immediate cause of e for it only causes something to be able to undergo e. The event that makes a substance soluble cannot be an immediate cause of its dissolving, for it merely causes the substance to b e c o m e soluble. A n d nothing dissolves just because it is soluble? 1 Thus, while it is true that if e's enabler had not occurred, e would not have occurred, it is not true that e's enabler is an immediate cause of e. So, c is an immediate cause of e just in case, (i) if c had not occurred, e would not have occurred and (ii) c is not a m e r e enabler of e. A n event, c, is a remote cause of e
CAUSES AND ENABLERS
209
just in case there is a sequence of events, c, % . . . , cn, e, such that c is an immediate cause of cl,. 9 and cn is an immediate cause of e. Thus, an event is a cause of e just in case it satisfies the original version of the counterfactual analysis (C) and is not a mere enabler of e.
In Bennett's case, the combination of the thesis that the time of an event is essential to it and this revised version of the counterfactual analysis cannot be used to derive the objectionable claim that the rainstorm in April caused the June fire. The reason for this is that the rain was merely an enabler, and not a cause, of that fire. The revised version of the counterfactual analysis and thesis of the essentiality of an event's time of occurrence are not incompatible. This revision of the counterfactual analysis of event causation does not, of course, address other difficulties that can be raised against the idea that causation can be analyzed in terms of counterfactual dependence. And perhaps one of those other difficulties will eventually prove to be insuperable. However, I think that the difficulties that I l~ave discussed here are not insuperable, and that the counterfactual analysis can be revised, by taking account of the distinction between causes and enablers, so as to meet those difficulties. On the other hand, it is not clear that a counterfactual analysis so revised can retain its original goal of understanding causation strictly in terms of counterfactual dependence of one event on another, since effects depend counterfactually both on their causes and on the causes of the conditions that enable their causes to be effective.22
NOTES i In his paper, "Causes and Counterfactuals", Journal of Philosophy 70 (1973), pp. 570--572, Jaegwon Kim provides a summary of at least some of the (now) standard objections to the counterfactual analysis of event causation. I shall, in this paper, take no position on whether or not they can all be met; I do think, however, that at least some of them can be. 2 See, for example, Lewis's "Causation", reprinted in his Philosophical Papers, Volume / / (New York: Oxford University Press, 1986), pp. 159--172. This paper originally appeared in Journal of Philosophy 70 (1973), pp. 556--567. 3 Whether they actually do or do not will not be a matter of concern here. 4 Perhaps one good reason for distinguishing claims about fact-causation from claims about event-causation is that there is no match between such claims, unless the circumstances in which, e.g., the statements 'the house burned down' and 'the house's
210
LAWRENCE
BRIAN LOMBARD
burning down occurred' are true, are the same. But this is not so, since the former, but not the latter, would be true if the house burned down twice. This point is Davidson's; see "The Logical Form of Action Sentences", reprinted in his Essays on Actions and Events (New York: Oxford University Press, 1980), p. 114. 5 See Jaegwon Kim, op. cit., and his "Noncausal Connections", N o ~ 8 (1974), pp. 41--52. Kim holds this view, in part, because if the relation between the two were causal, it would be a case of instantaneous causal interaction at a distance. 6 I prefer the last mentioned way of dealing with such cases. I think that the case is not a case of causation, for reasons having to with my view that there are no events that are the changes that objects undergo when they merely change relationally or Cambridgely; and when I become an uncle that is how I change. Thus, there is no event involving me (my becoming an uncle) to be causally related in the way required to my sister-inlaw's giving birth. Thus, I do not see a problem for the counterfactual analysis created by such cases. See my Events: A Metaphysical Study (London: Roufledge & Kegan Paul, 1986), pp. 102--104. 7 Jonathan Bennett, "Event Causation: The Counterfactual Analysis", in James E. Tomberlin (ed.), Philosophical Perspectives, I: Metaphysics, 1987 (Atascadero, California: Ridgeview Publishing Company, 1987), pp. 367--386; Bennett's version of the case is on p. 373. [Hereafter, Bennett's paper will be referred to as 'Bennett'.] 8 I defend the claim that no event can occur at a time other than the time at which it in fact occurs in Events: A Metaphysical Study, pp. 212--216. 9 It appears that Bennett, who accepted that Co) is false (in Bennett, p. 373), is now prepared to accept (b) as true. After all, in saying that the rain caused the fire, we are merely pointing out a remote and non-salient cause; see his Events and Their Names (Indianapolis: Hackett Publishing Company, 1988), p. 70. I think the rain was not a cause, remote or non-salient, of the fire, and that Bennett was wrong to change his mind. 10 David Lewis, "Events", Philosophical Papers, Volume H (New York: Oxford University Press, 1986), p. 250. 11 I consider some objections to this essentialist claim about events in Events: A Metaphysical Study, pp. 206--212, and find them wanting. There is also an argument against the claim that an event cannot occur earlier than it in fact does in Bennett, p. 369; it too is, I think, unsound. ~2 Lewis, for example, takes his argument as a reason to reject this essentialist thesis about events. 13 Bennett, however, apparently believes that this conclusion is not unacceptable; for, to Bennett, all that 'Brown's action causes Jones's death' means is that that action is in the causal history of the death. See Events and Their Names, pp. 70--71. I too believe that the passerby's action was in the causal history of Jones's death; but I think that events and actions can be in the causal history of another without being a cause of the other. This indeed is a principal point of the present paper. 14 It must be noted that what I mean by 'a delayer of an event' and its companion, 'a hastener of an event', are not what Bennett means by such phrases. 15 See Bennett, pp. 374--375. Apparently, Bennett has now rejected the asymmetry fact as a fact about causes; see Events and Their Names, p. 71. ~6 Again, it is worth pointing out that Bennett and I do not mean the same things by 'hastener' and 'delayer'. And because of that, the asymmetry thesis Bennett has now rejected is different from the asymmetry thesis that I think is true. 17 One might suggest that the fact that some event delays the fire does not entail that that event makes the fire occur, on the grounds that a delayer does not cause the delayed event to occur; rather, it causes the delayed event to have a certain feature, that of occurring at a certain (later) time. Thus, delayers are causes, but not causes of the events thereby delayed. But this suggestion presupposes the idea that what are caused
CAUSES
AND ENABLERS
211
by causes are (at least sometimes) not events but features of events. However, such a view is at odds with the view that the causes and effects of events are themselves events, and thus cannot happily be combined with C A E C to circumvent the problem that (b) raises. 18 I hasten to note that the hastened effect is not the same event as that which would have occurred had the hastening cause not occurred. My commitment to the essentiality of an event's time of occurrence is not the only reason for holding that view here. 19 This is a principal thesis of Events:A MetaphysicalStudy. 20 See Aristotle, DeAnima, Bk. II, ch. 5, and Metaphysics, Bk. V, ch. 12. 21 If something dissolved just because it became soluble, it would be because the event that made the thing soluble was also a cause of its dissolving. The enabler, then, would not have been a pure enabler. 22 I am grateful to Pat Francken and Richard Baer for a helpful discussion with m e concerning some of the issues discussed in this paper, and to Larry Powers for reading an earlier draft.
Department of Philosophy, Wayne State University, Detroit, M148202, U.S.A.