Lewis’ Counterfactuals
In the debate over causation, one of the notable names is David Lewis. It is his work on counterfactuals, stemming from Hume, that is worthy of such attention. Two of Lewis’ works “Counterfactual Dependence and Time’s Arrow” and “Causation as Influence” are under consideration in this essay. The first essay “Counterfactual Dependence and Time’s Arrow” will be used to demonstrate Lewis’ original counterfactual analysis. After criticism towards this original theory, the second essay “Causation as Influence” shows Lewis’ revisions to his argument. It is from these two essays that one should have a comprehensive understanding of counterfactual theories of causation, at least as Lewis sees it, only to be followed by criticism from Michael Strevens’ essay “Against Lewis’ New Theory of Causation.”
To better understand Lewis’ theory, it is best to have an initial understanding of the counterfactuals that Lewis employs. A counterfactual is in regards to a proposition including event A and event Bwhere B follows from A that if A were (or had been) the case, B would be (or have been) the case. It is important to understand that event A and event B are separate events. Still in Lewis’ counterfactual theory, there exists a causal dependence between the two events. Causal dependence is a critical point in Lewis’ theory and consists of a set of qualifying conditions. First in order for there to be causal dependence is that causal dependence occurs between events. The second condition for causal dependence is that the two events be distinct from each other. What this means is that these two events cannot be related in a sense that they are identical, overlap, or imply the other. Lastly for causal dependence, Lewis claims that it should not include what he calls backtracking counterfactuals which hold the past to be fixed in time. It is the counterfactual and the causal dependence upon which Lewis’ theory rests.
The causal dependence that is critical to Lewis’ theory when taken into consideration requires the understanding of time. Lewis makes the point to show the asymmetry of time. This is especially apparent in “Counter Dependence and Time’s Arrow.” The asymmetry of time is that the past is fixed in place whereas the future seems to be ambiguous and open to possibility. This seems to be a matter of fact that is common sense. Despite this seemingly common knowledge, Lewis is careful to explain this as it is not a fact that is known in and of itself or a priori. It is only from empirical knowledge that we come to know the asymmetry. Lewis explains it that “The way the future is depends counterfactually on the way the present is. If the present were different, the future would be different; and there are counterfactual conditionals, many of them as unquestionably true as counterfactuals ever get, that tell us a good deal about how the future would be different if the present were different in various ways. Likewise the present depends counterfactually on the past, and in general the way things are later depends on the way things were earlier. Not so in reverse” (Lewis 455). What this quote demonstrates is better known as temporal asymmetry. It is because of this asymmetry in time that Lewis claims that causal dependence, and therefore causation, is asymmetrical as well. This is backed primarily by Lewis’ claim that there is asymmetry of overdetermination. Asymmetry of overdetermination is essentially that earlier events determine later events but that this does not happen in reverse.
Another asymmetry that Lewis presents is that considering the asymmetry of miracles. Consider the example of Nixon having possession of button that releases nuclear weapons, resulting in a nuclear holocaust. It is important to note that Lewis wants to consider the closest actual world. This is important as we approach the situation closer. The two possible worlds are that Nixon does not press the button and that Nixon presses the button. The one where he presses the button results in a nuclear holocaust which is a world far different from our own. Suppose now that Nixon presses the button and the button malfunctions in a way that it prevents a nuclear holocaust. What this shows is that the malfunctioning case is more similar the real world than is the one where Nixon pushes the button. This demonstrates that in order for the world to be similar to the real world is much more than the alternative world. As Lewis puts it that “Divergence from a world such as w0 is easier than perfect convergence to it” (Lewis 475). Since divergence is easier than convergence, Lewis is able to explain a fact important to causal dependence since “The asymmetry of miracles, and hence of counterfactual dependence, rests on a feature of worlds like w0 which very simple words cannot share” (Lewis 475).
A problem that Lewis, as well as older counterfactual theories, faces is that of preemption. Preemption is an attempt to dismantle the causal dependence that Lewis is arguing for. Preemption takes two different forms. There is early preemption and late preemption. For right now we shall only consider early preemption in the face of early counterfactual theories and Lewis’ rebuttal. An example of early preemption would be that of two assassins attempting to kill a primer minister. Event B would be that the first assassin throws a knife at the prime minister’s head and that event C is another assassin decapitating the prime minister. Event D where blood squirts from his neck follows from event B and so event E is the result of event C. Yet this is not necessarily true for D is a guarantee that event E happened. This seems to pose a problem for simple counterfactual theories. For after all D did cause E, they both occurred, C and E are wholly distinct; however, E does not counterfactually depend on C at all. Instead it is counterfactually dependent on D. This seems to stump original counterfactual theories, yet Lewis has a rebuttal. In addition to the four points found in original counterfactual theories, Lewis adds that in order for C to cause E there must exists a causal chain from C to E. What he means by chain is that each event counterfactually depends on the previous event. How this relates to the example of early preemption is as follows. Event E is not counterfactually dependent on C because if event C had not happened event E would in virtue of event B. Lewis solves this problem though that there exist a causal chain between event E and event C based on counterfactual dependence. It is from this transitive chain that we see that event C causes event E and therefore Lewis has overcome early preemption.
Looking back, there is a general understanding of Lewis’ original theory. His counterfactual theory relies on causal dependence which as of now is rather sturdy. In addition to causal dependence, Lewis has provided a logical backing that causal dependence is asymmetrical because there is no a priori evidence to prove such asymmetry. Lewis also includes the transitive causal chain as a way of responding to problems brought on by early preemption. It is from this original composition of Lewis that there are evident advantages to his theory.
Still although positive qualities exist in Lewis’ theory, there are still flaws in his original thesis.One of the issues with Lewis’ theory is that it is context sensitive. By context sensitive, it means that Lewis’ causation only pinpoints one cause to an effect in all situations. This absoluteness cannot be true as Lewis’ theory implies. Consider the case where a camper lights a fire, where the fire is then blown by a gust of wind such that it catches to the rest of the forest and the forest burns down.It seems commonsense to the reader and for Lewis that the cause of the fire was the camper lighting the match. However it is also possible to consider that the oxygen present in the environment was also the cause for the spreading of the fire. Lewis would say that the oxygen present is not a cause but merely a condition for the fire. Yet if we were to take the spread of a fire in a laboratory, where oxygen is meant to be excluded, then it seems as if oxygen would be the cause of the fire spreading. This demonstrates that Lewis’ theory does not account for different causes that result in a certain effect. Still, Lewis would argue that he is interested in the broad notion of cause in such a way that the camper lighting a fire is still responsible for the fire and the oxygen present is merely an additional condition to fire.
Context aside, Lewis’ theory is still under scrutiny when under the consideration of temporal asymmetry. Temporal asymmetry rests under the idea that the past is fixed and that the future is undetermined. Essentially it is that the past determines the future. However the laws of thermodynamics are symmetrical in time. Although it is very rare possibility, one can imagine a cooked egg returning to its raw state and hopping back into the shell. It is with this that Lewis’ temporal asymmetry seems to not be as strong.
Another issue with Lewis is over transitivity where it can be claimed that a certain event causes another event even though they are not directly related. This would be in such a case where event A causes event B and that event B causes event C. This in turn deduces that A causes C. This type of thinking runs into issue with Lewis’ theory. Take the example of a hiker in the woods. The hiker is walking along the trail when suddenly a boulder is dislodged and comes tumbling towards the hiker’s way. The hiker seeing this tumbling boulder, ducks preventing being hit by the boulder, and then considers to walk on the path. It is from this example of a causal chain that the boulder not only causes the hiker to stop and duck, but to also continue his stride. This proves to show some difficulty in transitivity for Lewis’ theory.
The issue of transitivity is a difficult one to answer. This is partly in fact that it is associated with the preemption, which Lewis’ theory also has difficulty dealing with. It was earlier mentioned during the example of the assassins and the prime minister that Lewis’ original theorywas able to side step such an objection. This is only possible under early preemption. There is early preemption and late preemption. Early preemption is when the preempted alternative is cut short in a process before the process reaches completion. Late preemption is when the main process has gone through completion and brought about the effect. Assume that we have two people, Billy and Suzy. Both are throwing rocks at a bottle. Event B is that Billy throws a rock at the bottle. Event C is that Suzy throws a rock at the bottle and is slightly ahead of Billy, and as a result event D where the bottle breaks. The example results with the bottle being broken regardless of Suzy. Even if she threw later or earlier, the bottle would have been broken. This shows that the shattering of the bottle is not counterfactually based on Suzy. This is an issue for Lewis for early preemption had the effect depend on a middle event that involves the action of the cutter. With late preemption, the focus is pushed to the last event. Because of this effect on the last event, Lewis is unable to side step the issue as he was with early preemption. Thus this is a problem for Lewis’ original theory.
Context sensitivity, temporal asymmetry, and transitivity are minor criticisms of Lewis’ original counterfactual theory. It is the issue of late preemption and the trumping argument which provide the most difficulty for Lewis. Late preemption is the issue where the ending event is of no concern for there are multiple causes possible for the effect. Still there is trumping, which uses simultaneous causes to result in the same effect. Consider the example of a sergeant and a general commanding a group of troops. The general has authority over the sergeant and so his word has more authority. Suppose then that at event B that the sergeant shouts “Advance” and at that same time that event C happens where the general shouts “Advance.” The soldiers in turn hear the general’s command at event D and so results in the soldiers advancing at event E. Although the troops are meant to obey the superior officer, it is not necessarily true that C caused E as E is not counterfactually dependent on C. This example is not counterfactually dependent because E would have happened because of the presence of event B. One possible objection would be to say had Billy’s rock or perhaps Suzy’s rock shattering the bottle at a different time would be a different shattering. This is not a sound objection as just because an event occurs at a slightly different time does not mean that it is a different event.
It is from the problems brought on by late preemption and trumping, that Lewis’ original theory no longer holds as being sound. It is from here that we shall turn to his recent theory explained partially by his essay “Causation as Influence.” Lewis returns to the issue mentioned at the end of the paragraph above regarding the time of event. Contrary to what is said, he in fact believes that time is a factor. The evidence that Lewis brings to the argument is that “Without the hanging that allegedly caused the death of Ned Kelly, for instance, he would sooner or later have died anyway, but he would have died a different death and the event that actually was Kelly’s death would have never occurred” (Lewis 185). What Lewis is demonstrating is that events are fragile. Lewis extends his points of the fragility of events to include that events are contingent on how, when, and whether. This leads Lewis to conclude that for every event C causing event E that thereis a series of possible C events as there are possible E events. Returning to the issue of Suzy and Billy Lewis is able to respond that “Suzy’s throw is much more of a cause of the bottle’s shattering than Billy’s. Even if the throws are so much alike that removing Suzy’s throw altogether, it is still true that altering Suzy’s throw slightly while holding Billy’s fixed would make a lot of difference to the shattering, but altering Billy’s throw slightly while holding Suzy’s fixed would not” (Lewis 191). This same kind of reasoning can be applied to the trumping example as well. Changing the shouting of the major’s command while keeping the sergeants fixed would make a difference whereas changing the sergeants command and keeping the major’s would do little to nothing at all. It seems from “Causation as Influence” that Lewis has revised his original theory to combat some old problems.
Looking at Lewis’ theory though shows that there are some modifications. One of the more notable ones is the action he takes towards causal dependence and transitivity. Lewis asks “Is it still necessary to take the ancestral, now that our definition of causal dependence has evolved from simple whether-whether dependence to a pattern of influence? Does our improved definition of dependence allow us just to identify causation with causal dependence? No. Influence is not invariably transitive” (Lewis 191). It seems as if Lewis is straying away from a transitive property. Lewis shows in his essay how different events can be the result of different causes or that the same cause can affect different events. With alterations and the fragility of events, it naturally follows the contemplation of events which are absent from a causal chain.Lewis mentions the existence of absences and how they affect causation. The fact that there is an absence of nerve gas is that we are able to live and breathe the air. Lewis does not want readers to get carried away with what he means by absence for “Absences are not events. They are not anything: where an absence is, there is nothing relevant there at all. Absences are bogus entities. Yet the proposition that an absence occurs is not bogus” (Lewis 195).
Having looked at Lewis’ original theory followed by his revisions in “Causation as Influence” there seems to be a well back support for counterfactual theories. As of now, Lewis has what appears to be successfully rebutted against major objections towards his theory. We will turn now to a recent objection posed by Michael Strevens. Strevens takes issue with the main component of Lewis new theory being that involving alterations. Strevens refers to these alterations as a concrete event. Strevens takes issue with the idea of fragility as “Had the jar shattered a moment later, or had it broken into a slightly different number of shards of glass, this concrete event would have never occurred” (Strevens 2). This seems to be a pickier interpretation of Lewis’ an argument. Strevensthen simplifies Lewis’ argument to the where if C is the cause of E then if C would not have occurred then E would not have occurred. This summary of Lewis’ theory comes off as being a bit of an over simplification of Lewis’ argument. This simplificationis even simpler thanhis “Counterfactual Dependence and Time’s Arrow” argument. Furthermore, this view does not reflect his later “Causation as Influence.”Strevens poses a counterexample where Sylvia and Bruno throw rocks. Sylvia throws a right angle from a jar while Bruno throws a rock directly at the jar in such a way that Sylvia’s rock deflects Bruno’s rock and in turn hits the jar. Strevens claims that this shows that “Where Lewis goes wrong is in equating the causal relation between events and the causal influence relation between their concrete realizers” (Strevens 10). What this means is that Strevens is against causal claims where event C is the cause of event E is now riddled with a multitude of causes that each take part in making event E. Strevens then moves on to his issue with causal influence by its own. Strevens explains that his counterexample shows dependence on both Sylvia and Bruno’s rock throw and that there is not a possible alternative. This appears to side with Lewis more than against it as it is sensitive to the how, when, and whether conditions that are important to Lewis. From my interpretation,Strevens seems to draw a lot of points that don’t promote his argument and ultimately weaken his original claims including his counterexample.