Michael Moore's Causation and Responsibility offers an integrated conception of the law, morality, and metaphysics, centered on the notion of causation, grounded in a detailed knowledge of case law, and supported on every point by cogent argument. This is outstanding work. It is a worthy successor to H.L.A. Hart and Tony Honoré's classic Causation in the Law, and I expect that it will guide discussion for many years to come.
Moore's main thesis is that there is a single notion of causation at work in the law, scientific explanation, and commonsense thought. I am convinced, and only dispute the details. In particular, I dispute Moore's claim that this single notion of causation does not allow absences as causes or effects. I argue that Moore's rejection of absence causation complicates and problematizes his otherwise plausible view. It complicates his view of legal and moral responsibility, since of course we attribute responsibility over absences (e.g., negligence). Moore is thus driven to a disjunctive account of the grounds of responsibility, with mere counterfactual dependence providing a second noncausal ground of responsibility, albeit responsibility of a diminished sort.
Two further problems then arise. The first is the beheading problem. Since the route from beheading to death involves the absence of blood flow to the brain, on Moore's view beheading someone cannot possibly cause them to die. The second problem is the correlation problem. Moore argues that there is counterfactual dependence between mere correlates of a common cause, and so, on his view, responsibility can arise from mere correlation with a crime.
Accepting absences as causal would allow Moore to simplify his view. He could treat causation as the sole ground of responsibility and so would no longer need a disjunctive account. He would avoid the beheading and correlation problems. Beheadings can count as causes, and correlates will no longer bear responsibility, since correlation is not causation.
I. SUBSTANTIATING UNITY
I begin with an overview of Moore's position in order to situate the issue of absence causation. As Moore says at the start: “The central idea that organizes the book is that causation as a prerequisite to legal liability is intimately related to causation as a natural relation lying at the heart of scientific explanation.”Footnote 1 What he means is that the core concept of causation at work in the law—once pruned of certain confusions—just is the core concept of causation at work in scientific explanation and commonsense thought.Footnote 2 Thus Moore's central idea can be characterized as:
Unity: There is a single core notion of causation at work in the law, scientific explanation, and commonsense thought.
One might think that the notion of causation has many roles in the law, serving as a placeholder for a range of more specific notions.Footnote 3 Or one might think that the notion of causation has no role in the law, or that it has a single unified role in the law but one that is special and distinctive to the law. Unity is the denial of all of these thoughts.
Moore's central project in Causation and Responsibility is to substantiate Unity by detailing this single core notion of causation. Perhaps his primary commitmentFootnote 4 is to a singularist conception of causation, which denies that token causal relations are grounded in laws of nature, counterfactuals, or any other general features of the world. The contrast is with generalism, as seen in nomic sufficiency and counterfactual dependency accounts. Moore remains neutral between versions of singularism, including those that reduce token causal relations to token physical processes such as energy flows and those that treat token causal relations as metaphysically primitive.
Moore notes that most versions of singularism are committed to the claim that absences cannot feature in causal relations: “One of the great strengths of most singularisms . . . is that they get it right in how they classify omissions, preventions, and double-preventions as noncausal.”Footnote 5 This can be readily seen on an energy-flow conception, assuming that absences are nothings (but see Sec. IV.B), as nothings cannot provide either sources or sinks of energy.Footnote 6 However, it is not immediately obvious that primitivist singularisms must deny absence causation. But in any case, whether or not it flows from singularism, Moore independently endorses:
No Absences: Absences cannot serve as causes or effects.
No Absences is one of Moore's main tenets, defended repeatedly.Footnote 7 And so Unity takes on substance. The one core notion of causation at work in the law, scientific explanation, and commonsense thought is said to be the notion of a singularist relation that does not allow absences as relata.Footnote 8
No Absences then drives Moore to a disjunctive account of legal liability and moral responsibility. He begins from the assumptions that legal liability must be grounded in moral responsibility and that moral responsibility must in turn be grounded in objective natural features of the world.Footnote 9 Causation, on his view, is the main natural ground of responsibility:
“[C]ause” is univocal . . . it refers to a natural relation that holds between events or states of affairs. Because moral responsibility is tied to such a natural relation, and because the law is tied to morality, the law also is tied to this natural relation. . . . It thus behooves us to enquire after the nature of such a relation.Footnote 10
But there is an immediate tension looming between No Absences and the idea that responsibility must be grounded in causation, since—as Moore is well aware—the law and morality both manifestly ascribe responsibility in cases of negligence and others involving absences. This tension forces Moore to retreat from the idea that causation is the one natural ground of responsibility, and move to a disjunctive account with mere counterfactual dependence providing a second natural ground for responsibility (albeit responsibility of a diminished sort):
Disjunctivism: Legal liability and moral responsibility are grounded either in causation, or in relations of mere counterfactual dependence without causation (in which case the level of responsibility is diminished).Footnote 11
The issue of whether No Absences is a plausible claim about causation is thus connected to the issue of whether Disjunctivism is a viable view of the grounds of responsibility.Footnote 12
I should note that No Absences on its own does not require Disjunctivism; some of Moore's other commitments come into play. Strictly speaking, No Absences itself requires only that responsibility can obtain in negligence and related cases, even without causation. It is thus compatible with a purely noncausal (nondisjunctive) account of responsibility, though that would conflict with Moore's guiding idea of the centrality of causation to responsibility.Footnote 13 No Absences is also compatible with disjunctive accounts on which the second disjunct is not counterfactual dependence but something else that covers negligence and related cases, although it is not obvious what that other relation might be.Footnote 14 Finally, No Absences is compatible with disjunctive accounts on which mere counterfactual dependence (or some substitute) grounds responsibility at the same or even at an increased level as does causation. Nevertheless, the idea that the level of responsibility is diminished fits with Moore's idea of causation as “the big dog”Footnote 15 when it comes to grounding responsibility, and this idea gets used by Moore to buttress the idea that there is a metaphysical difference between real causation and what happens when absences are involved (see Sec. V for further discussion).Footnote 16 So Disjunctivism is in principle separable from No Absences, but it is a natural concomitant given Moore's other commitments.
Neither Unity, nor causal singularism, nor Moore's realist-naturalist background assumptions about the grounds of responsibility is in dispute. I propose to accept all of these claims arguendo. (For the record, I endorse Unity, I am skeptical of causal singularism, and I am neutral as to the realist-naturalist assumptions about responsibility.) What are in dispute are No Absences and its concomitant Disjunctivism. The dispute concerns how best to substantiate Moore's own central thesis of Unity, for I argue that the single core notion of causation at work in the law, scientific explanation, and commonsense thought is a notion that embraces absence causation.
II. THE BEHEADING PROBLEM
My first argument against No Absences is that it mishandles paradigm cases of causation such as cases of beheadings. Decapitation can cause death if anything can. Yet death by decapitation belongs to a class of cases of disconnections (what Moore, following Hall, calls “double preventions”),Footnote 17 which involve absences as causal intermediaries. This class of cases turns out to include many other paradigm ways to cause death, including drowning, strangulation, and pulling the trigger on a gun. This class also turns out—when one looks at how muscles work—to include virtually all human actions.
The beheading problem is really a two-headed monster with both a causal and a moral aspect. The causal aspect stems from No Absences. Moore must deny that decapitation can cause death. Moreover, given that killing someone requires causing them to die,Footnote 18 Moore must deny that chopping off someone's head can kill them or can legally constitute homicide. The moral aspect stems from No Absences together with Disjunctivism. For once Moore denies that decapitation can cause death, he must then deny that the beheader can be fully responsible for the fate of the beheaded. At most the beheader can bear the diminished level of responsibility due to mere counterfactual dependence. Indeed, in cases where a preempted backup is afoot, Moore will find neither causation nor counterfactual dependence and so must grant the preempting axe-wielding maniac complete legal and moral impunity.
I fear I may be belaboring the obvious by arguing that decapitation can cause death. But this is what is at issue.Footnote 19
A. A Provisional Taxonomy
Say that positive causation relates a presence to a presence, omission relates absence to presence, prevention relates presence to absence, and omission of prevention relates absence to absence:
Disconnection then involves a step of prevention followed by a step of omission. That is, disconnection involves a process running from a presence to an absence and then on to a further presence: + – +. The absence in the middle—had it been present—would have prevented the effect, and so the cause prevents what would have prevented the effect; thus the name “double prevention.” Do not be confused: there is only one step of prevention in a double prevention.
This is a provisional taxonomy, for it assumes a real metaphysical difference between presence (+) and absence (–). I argue (in Sec. IV.B) that in the cases at issue there is merely a linguistic difference between an event positively described and an event negatively described, and I revisit the taxonomy at that point. But since Moore thinks that there is a real metaphysical distinction, I at least beg no questions against him by speaking in his terms.
To illustrate the difference between positive connective causation and disconnection, imagine a red button on a black box, with a wire leading out to a bomb. The button is a detonator button, but the wiring of the black box is left open. Perhaps pressing the button generates an electrical signal that travels to the bomb and triggers an explosion; that would be a connective wiring pattern:
Or perhaps pressing the button cuts off what had been inhibiting an explosion signal. That would be a disconnective wiring pattern:
Both wiring patterns can be found in actual causal mechanisms. For instance, as Lewis notes,Footnote 20 when Passenger pulls the brake cord to stop the runaway train, if the train happens to be outfitted with vacuum brakes, the train will stop in a connective way. But if the train happens to be outfitted with air brakes, the train will stop in a disconnective way. Could which way the mechanism is wired really matter for causation? Could Passenger cause the train to stop only if it has vacuum brakes but not if it has air brakes? If Terrorist pushes the detonator button, could she cause the bomb to explode only if it is wired by connection but not if it is wired by disconnection?
B. Beheadings and Other Disconnections
When Killer beheads Victim, consider the way in which this causes Victim to die. What happens is that the flow of oxygenated blood that was sustaining Victim's life has been disconnected. In other words, there is a step of prevention—the beheading causes the absence of blood flow to Victim's brain—followed by a step of omission—the absence of blood flow to Victim's brain causes death. In the terms I prefer (see Sec. IV.B.4), what is salient is not a line of energy flow from Killer's axe to Victim's brain. Rather, what is salient is that Killer's axe has interrupted an ongoing flow of energy (in the form of oxygenated blood) that was sustaining Victim's brain. Diagrammatically:
Once one understands how disconnecting the flow of oxygenated blood can cause death, one is in a position to see that such disconnections feature in many other paradigm causes of death. These include drowning, suffocation, strangulation, asphyxiation, bleeding to death, cardiac arrest, and stroke, all of which cause death by preventing oxygenated blood from preventing death. It is important to appreciate that these cases are not deemed causal merely because people are unaware of how they work. Everyone understands how strangling someone can cause them to die. Likewise the Centers for Disease Control and Prevention—in full understanding of the way that heart failure can lead to death—declare: “Heart disease is the leading cause of death in the United States and is a major cause of disability.”Footnote 21
Once one understands the general pattern of causation by disconnection, one is in a position to see that disconnections feature in all sorts of other paradigmatic causal mechanisms (not just air brakes and certain detonators). These mechanisms also include guns and muscles. With guns, pulling the trigger disconnects a barrier (the sear) that had been preventing the gun from firing (by blocking the spring, the release of which propels the hammer to strike the powder). In other words, there is a step of prevention—the pulling of the trigger causes the absence of the blockage—followed by a step of omission—the absence of the blockage causes the gun to fire. In the terms I prefer (Sec. IV.B.4), what is salient is not a line of energy flow from Shooter's finger to the bullet. Rather what is salient is that Shooter's finger has released a pent-up reservoir of potential energy (in the spring and also in the powder) that propels the bullet.
Muscles work in a similar way. According to the widely accepted sliding filament theory, muscle cells are composed of myosin and actin filaments. Contraction occurs when myosin binds actin and the filaments slide together. In the uncontracted state, myosin is ready to bind actin (the filaments are like coiled springs, ready to bind) but tropomyosin blocks the binding sites on the actin (the tropomyosin is like the sear in the gun). Contraction requires disconnecting the tropomyosin barrier. What happens is that an electrical impulse from the nervous system generates a calcium cascade through the muscle fiber. The calcium then binds to troponin—attached to tropomyosin—pulling tropomyosin off the binding sites of the actin. This unleashes myosin to bind actin, and so the muscle contracts. Essentially, there is a step of prevention—the electrical impulse from the nervous system causes the absence of the tropomyosin blockage—followed by a step of omission—the absence of the tropomyosin blockage causes muscle contraction. In the terms I prefer (Sec. IV.B.4), the energy flow involved in muscle contraction does not actually come from the nervous system. Rather the nervous system signal serves to release a reservoir of potential energy stored in the muscle fibers.
Once one understands that muscles work by disconnection, one need only appreciate that every voluntary human movement involves muscle contraction to appreciate that every human action involves disconnection. Hence, if we human beings are ever to be held causally responsible, disconnections must be causal. To think otherwise is to make a mockery of causation in the law, since, as Moore makes explicit: “[A]ll liability doctrines in criminal law and torts require that a defendant's act cause something.”Footnote 22 In short, Moore does not merely have a beheading problem, he has a human action problem.
C. Disconnections and the Theoretical Role of Causation
Disconnections are not merely features of many paradigmatically causal sequences (and virtually all considered by the law); they are also suited to play the theoretical roles that causation plays. Causation is often said to be among the most central of our concepts, and its centrality is found in its connections to a range of surrounding concepts, including counterfactuals, statistics, agency, explanation, and responsibility. Disconnections (and absence causation generally) sustain these connections and so deserve to count as causal.
To begin with, disconnections display counterfactual dependence. For instance, if Terrorist had not pushed the detonator button, then the bomb would not have exploded. Likewise, if Killer had not beheaded Victim, then Victim would not have died. Such counterfactual dependence is sometimes thought to be constitutive of causationFootnote 23 and features in the standard sine qua non test used in the law. Thus counterfactual dependence is at least an excellent indicator of causal relatedness (one need not endorse the identification of causation with counterfactual dependence to recognize this fact). So the fact that there is counterfactual dependence provides excellent reason to think that disconnections are causal.
In disconnections, there is also statistical relevance. For instance, the following probabilistic inequality holds: the probability of the bomb exploding given the circumstances plus Terrorist's pressing of the detonator button is greater than the probability of the bomb exploding given the circumstances but minus Terrorist's pressing of the detonator button. Likewise, the probability of Victim dying given the circumstances and Killer chopping off his head is greater than the probability of Victim dying given the circumstances but without Killer chopping off his head. Such statistical relevance is sometimes thought to be constitutive of causationFootnote 24 and serves as the basis by which statisticians infer causal relationships in the world. Thus statistical relevance is also an excellent indicator of causal relatedness (again, one need not identify causation with statistical relevance to appreciate the connection). So again, there is excellent reason to think that disconnections are causal.
Further, disconnections provide agential means for their outcomes. For instance, an agent such as Terrorist whose sole end is to explode the bomb may press the detonator button as a means to her end. Likewise an agent such as Killer may behead Victim as a means to achieve the end of Victim's death. Such agential means is sometimes thought to be constitutive of causationFootnote 25 and may be the original wellspring of our causal concept. So here is yet another excellent reason to think that disconnections are causal.
Disconnections also serve to explain their outcomes. For instance, if the FBI agents want to explain why the bomb exploded, they would surely cite the fact that Terrorist pressed the detonator button. Likewise if Detective wants to explain why Victim died, then surely she ought to cite the fact that Killer chopped off his head. These explanations are causal explanations and would not be possible unless disconnections were genuinely causal.
Indeed—to expand on the point about explanation—absences are routinely cited in actual scientific explanations, such as in the biological explanations of various diseases. For instance, absence of vitamin C causes scurvy, absence of vitamin D causes rickets, absence of insulin causes diabetes mellitus, and absence of growth hormone causes dwarfism. The way in which the human immunodeficiency virus (HIV) causes death is by disconnecting the immune system—an absence of functioning CD4+ T cells serves as causal intermediary, allowing opportunistic infections and cancers to spread unchecked. Thus consider the following discussion of how Duchenne muscular dystrophy causes death:
Muscular dystrophy in the mdx mouse has been described as a mutation in a colony of C57B1/10ScSn mice, which results in the absence of the 427kDa membrane-associated protein dystrophin. . . . A deletion on the human X-chromosome causes the absence of an analogous protein and leads to Duchenne muscular dystrophy (DMD). . . . The absence of dystrophin leads to the destabilization of [the transmembrane glycoprotein complex], yielding weaker muscle fibers that undergo progressive degeneration followed by massive necrosis. Ultimately, premature death of DMD patients occurs.Footnote 26
For one further example of absences in scientific explanation—this time from physics—consider the following discussion of electron–hole pair generation in semiconductors:
When an electron (which is a negative charge carrier) is freed from the atom, it leaves behind a hole, or the absence of an electron (which acts as a positive charge carrier). Free carriers are generated when electrons have gained enough energy to escape their bonds to the atom and move from the valence band to the conduction band. This process is called “electron–hole pair generation.” Electron–hole pairs can be created by any mechanism which delivers sufficient energy to an electron, including absorbing energy from light (as in a photo diode) and thermal excitation (absorbing heat energy).Footnote 27
Such electron–hole pair generation is understood to be causal: “The electron absence created by this process is called a hole,” and “These positively-charged holes can cause a catastrophic negative shift in the threshold voltage of the device.”Footnote 28 Evidently the notion of causation at work in scientific explanation has room for absences.Footnote 29
Finally, disconnections create (full-fledged) moral responsibility for their outcomes. For instance, Terrorist is (fully) morally responsible for any harmful consequences of the explosion, and Killer is (fully) responsible for Victim's death. Given that killing requires causing to die,Footnote 30 the commonsensical idea that beheadings, drownings, stranglings, and shooting (etc.) can all be killings presupposes that absences can be causal.
Indeed, if one looks at actual legal practice, one finds absences routinely recognized as causal in cases of negligence and breach of contract. These are aspects of the law that Moore suggests need “pruning”Footnote 31 but that a less revisionary approach must accept. In this vein, Hart and Honoré note:
There are frequent contexts when the failure to initiate or interrupt some physical process; the failure to provide reasons or draw attention to reasons which might influence the conduct of others; and the failure to provide others with opportunities for doing certain things or actively depriving them of such opportunities are thought of in causal terms.Footnote 32
Here is one of the many examples they then cite: “So a failure to deliver to a manufacturer on time a piece of machinery . . . may . . . be held the cause of the loss of the profits which would have been made by its use.”Footnote 33 It looks as though the notion of causation at work in the law has room for absences as well.
So I conclude that science, morality, and law converge on a picture of causation that is unified in allowing for causation by disconnection and absence causation generally. Any substantiation of Unity that upholds No Absences has thereby falsified the image of causation found in the law, scientific explanation, and commonsense thought. Indeed, my own opinion is that any view of causation that does not recognize disconnections as causal should be dismissed at the outset for having the absurd consequence that beheading someone cannot cause them to die. The rest of the argument is just belaboring the obvious.
D. Moore on the Beheading Problem
In response to my 2000 and 2004 presentations of the beheading problem, Moore grants that beheadings, drownings, strangulations, shootings, and indeed all human actions are deemed noncausal on his account.Footnote 34 But he tries at least to address the moral aspect of the problem. He begins by claiming that the real effect is “close” to the disconnective consequence.Footnote 35 For instance, with the gun case, he says: “[W]hat A has caused—the movement of the sear—is close to the state of affairs which caused the bullet to fire, viz the spring behind the hammer moving.”Footnote 36 He then suggests that full moral responsible can be regained by “getting sloppy” about event individuation:
Decapitations and deaths are close enough that we should say that to intend the first is to intend the second. Analogously, we should say that to do an act causing the first is to do an act causing the second. In each case this is true irrespective of the actual metaphysics of event-individuation. . . . Even if Herod did not literally either cause or intend John's death, what he did cause and intend is close enough that his moral responsibility is that of an intender and cause of death.Footnote 37
I have three replies, the first of which is that I find Moore's suggestion unclear. No explication of “closeness” is offered. Initially it sounds as though Moore might be using this term in a spatial sense. But in that sense, decapitation and death may be relatively far apart (depending on the distance from neck to brain), and many noncauses (such as the speck of dust right next to Victim's ear) may be much closer. Moreover, we seem quite able to distinguish spatially close events in other cases, so it is hard to see why spatial closeness should suddenly induce sloppiness here. Later it sounds as though Moore might be using “closeness” in a logical sense, such as when he suggests that there is some sort of logical connection between decapitation and death.Footnote 38 But obviously there is no such logical connection. Indeed it is not inconceivable that—with speedy action and sufficient technology—a decapitated head might be reattached and the person saved. So I do not think Moore's response is sufficiently articulated yet to take seriously.Footnote 39
It might help to consider the case where Terrorist blows up the air traffic control tower, preventing the needed signals from reaching the pilots and leading to a horrific midair collision with hundreds of innocents dead. I trust that Moore would admit that Terrorist is fully responsible for these deaths and would not say that Terrorist acts with impunity if there is a preempted backup afoot. Yet this case is also a disconnection, since what Terrorist is doing is preventing a signal from being sent that would have prevented the collision. But in what possible sense is the destruction of the air traffic control tower “close” to the collision? Spatially these may be miles apart. Logically there is no necessary connection. So I fail to understand what Moore means by “closeness” and I fail to see how any of the loose associations of this notion are applicable in the control tower case. Relatedly, I fail to see how “getting sloppy” about event individuation can lead us to conflate the event of the destruction of the tower with the collision miles up in the sky.Footnote 40
Second, I fail to understand how getting sloppy can in any way help Moore to resolve the moral aspect of the beheading problem. Morally, what is wanted is that Killer be fully morally responsible for Victim's death. But that Killer is not fully responsible for Victim's death follows directly from No Absences and Disjunctivism. For No Absences has the (already absurd) consequence that Killer's action does not cause Victim's death, and Disjunctivism has the consequence that causation is required for full responsibility. Which thesis is Moore abandoning? Perhaps he is allowing that Killer's action does cause Victim's death, because he is suggesting that sloppy event individuation alters the causal facts. But then accepting sloppy causation is just a surreptitious way to abandon No Absences. In the context of the present dispute, that would be surrender.
Or perhaps Moore is allowing that there can be full moral responsibility without real causation after all (perhaps what backs full moral responsibility is some legal surrogate “schmausation” that factors in sloppy event individuation). That is the way to abandon Disjunctivism. But then what becomes of Moore's earlier insistence on the role of real causation in morality?Footnote 41 And then what exactly is Moore's view of the natural ground for legal liability and more responsibility,? Perhaps he is really advocating:
Disjunctivism Revised: Legal liability and moral responsibility are grounded either in causation, or in relations of mere counterfactual dependence without causation (in which case the responsibility is diminished), or in relations of causation between the act and some outcome that is “close” to one of its effects (in which case the responsibility may return to the level of full responsibility).
I see little attraction to this further descent into disjunctivity, with added unclarity and ad hoccery.
Third, at most Moore is addressing the moral aspect of the beheading problem. He is not touching the causal aspect of the problem. Indeed he basically admits that by his lights the beheader “did not literally” cause death.Footnote 42 So even if Moore could somehow resolve the moral aspect of the beheading problem, he would still face a clear counterexample to his account of causation. By my lights, an account of causation that rules that decapitation cannot cause death does not deserve serious consideration.
III. THE PROBLEM OF CORRELATES
My second argument against No Absences is that it attributes spurious responsibility to mere correlates as well as to effects, at least when combined with Disjunctivism and Moore's views about counterfactual dependence. For Moore argues that counterfactual dependence can hold not just between cause and effect but also between correlates of a common cause and between effect and cause.Footnote 43 Granting arguendo that he is right about the extent of counterfactual dependence, he must be wrong about its moral implications, for evidently there is no moral responsibility due to correlation or effectuation.
To illustrate, imagine that Kingpin phones Thug to order him to kill Rival. This causes Thug's phone to ring, which wakes Neighbor and causes him to grumble. It also causes Thug to receive the order and set off to kill Rival:
Moore thinks (contra Lewis) that Thug kills counterfactually depends on Neighbor grumbles, and that Kingpin phones counterfactually depends on Neighbor grumbles.Footnote 44 But evidently Neighbor is not morally responsible in any way for the crimes of Kingpin or Thug. This shows that Disjunctivism is false. Mere counterfactual dependence does not ground responsibility.Footnote 45
One possible reply to the problem of correlates would be for Moore to retract the claim that counterfactual dependence can hold in cases of correlation or effectuation. But for Moore to make such a move would be self-serving. Plus he would owe us an explanation as to why his own arguments to the contrary went wrong. Of course, the philosopher who likes Moore's overall view but is unmoved by Moore's claims about counterfactual dependence is welcome to take this line, but Moore himself in good faith cannot.
A second and more plausible possible reply would be to add further necessary conditions for responsibility unmet by Neighbor. The questions will then arise as to how plausible the further necessary conditions are, and whether the case of Neighbor can still be tweaked to satisfy them. For instance, Moore requires that the omission be a breach of a positive duty and that the agent have the ability to satisfy the duty.Footnote 46 But the case of Neighbor can be tweaked to satisfy these requirements. We need only imagine that Neighbor had (for some independent reason, perhaps to honor a promise) a positive duty to keep quiet that grumbling breached; and that Neighbor had the ability to stifle the grumble.
Moore might also plausibly impose some sort of requirement of intention on Neighbor. But the case can easily be tweaked to satisfy this requirement as well. We need only imagine that Neighbor also had the intention to kill Rival and perhaps even intended to kill Rival by that act of grumbling (due to bizarre background beliefs about the magical powers of grumbling).
Of course further requirements can be considered. But it seems that what Moore needs is an in-principle argument to the effect that no case of mere correlation or effectuation could possibly satisfy such a requirement. I think it is obvious what the right requirement is. The right requirement is causation. The reason Neighbor cannot possibly be morally responsible for Kingpin's order or Thug's murder is that Neighbor did nothing to cause these outcomes. This provides an in-principle explanation for why mere correlates and effects cannot bear responsibility. But that is to abandon Disjunctivism's idea that mere counterfactual dependence can be an independent ground of responsibility and move to the view I am recommending, on which causation is the sole relevant requirement.
Indeed, it seems to me that even if some further requirements could somehow be added that could not in principle be met by mere correlates or effects, this would still miss the point. For at least one reason Neighbor is not responsible for Kingpin's order or Thug's murder is simply that Neighbor does not bear the right connection to these outcomes. This is true even if other reasons emerge that might also exonerate Neighbor. But that is just to say that where there is mere counterfactual dependence without causation, there is no legal liability or moral responsibility to be found. The law has it right, since as Moore notes, “Many of the liability rules of both criminal law and torts are framed explicitly in terms of someone (usually the defendant) causing something (usually a harm).”Footnote 47 Mere counterfactual dependence cannot go it alone.Footnote 48
IV. THE PROBLEM OF NOTHINGNESS
I now turn to consider Moore's arguments for No Absences, beginning with what I take to be Moore's primary argument, which is the metaphysical concern that absences are nothing and so cannot push anything around. Call this the problem of nothingness. I agree with Moore that there are metaphysical concerns about absence causation, but I doubt that these are sufficiently concerning to override the kind of paradigm cases and theoretical role considerations cited above (see Sec. II). Rather my view is that since it is obvious that absences can be causal, this fact can shed light on what absences are and how the metaphysics of causation works. In other words, while Moore argues from certain semantic and metaphysical premises to the conclusion that decapitation cannot cause death, I am reversing the argument and arguing from the premise that decapitation can cause death to the denial of some of Moore's premises.
G.E. Moore, in his discussion of skeptical paradoxes, teaches us that no argument can be persuasive whose conclusion is less plausible than the negation of any of its premises. So in effect I am applying G.E. Moore's lesson to Michael Moore's main argument for No Absences. In particular I am suggesting that if one sees an argument whose conclusion is that decapitation cannot cause death, then one should question the premises.
A. The Premises Involved
The problem of nothingness can be understood as arising from the conjunction of a semantic and a metaphysical premise. The semantic premise is that omission claims (claims about absences, omissions, and failings, etc.) do not describe actual existents. Rather they denote negations of existential quantifications over events. Thus Moore first introduces the notion of an omission with: “an omission is literally nothing at all.”Footnote 49 He continues by glossing the meaning of omission claims via negative existentials about event types:
An omission to kill is not some ghostly kind of killing. It is like an absent elephant, which is no elephant at all. An omission to kill is an absent event of killing; not the absence of any particular event of killing—it is not an absent act-token—for there are no “negative events.” There are negative propositions about events, such as “James omitted to kill Smith” meaning “it is not the case that James killed Smith.” Such negative statements are negative existentially quantified ones: if there is an omission to kill, then what is true is that it is not the case that some instance of the type of event, killing, existed.Footnote 50
And so:
Semantic Premise: Omission claims denote negative existential quantifications over events.
To illustrate, I assume—as per the Davidsonian tradition—that the positive “James killed Smith” serves to denote an event e such that e is a killing, with James as the agent and Smith as the patient of e:
(∃e) (Killing(e) & Agent(e) = James & Patient(e) = Smith)
Then Semantic Premise has it that omission claims such as “James's omission to kill Smith,” “James's failing to kill Smith,” and “the absence of James's killing Smith” come out as:
NOT (∃e) (Killing(e) & Agent(e) = James & Patient(e) = Smith)
This is the negation of an existential quantification over an event, and it says that there is no event that is a killing by James of Smith.
The metaphysical premise is that only actual concrete entities (such as actual events) are eligible to serve as causal relata. Or at least, negations of existential quantifications over events are ineligible. Thus Moore concludes: “[I]t is baffling how there could be relata for singular causal relations involving omissions.”Footnote 51 Or as he puts the point earlier:
[O]missions cause nothing. “Nothing comes from nothing, and nothing ever can” is good metaphysics, as well as catchy lyrics in musical productions. Absent elephants grow no grass by their absence; absent savings cause nothing, and certainly not the deaths they fail to prevent.Footnote 52
In short, Moore endorses:
Metaphysical Premise: Negative existential quantifications over events cannot be causes or effects.
After all, the fact that there is no event of a certain type hardly seems capable of pushing anything around.
I question both Semantic Premise (in Sec. IV.B) and Metaphysical Premise (in Sec. IV.C). My official view is that Semantic Premise is false and Metaphysical Premise is controversial though still probably true. My fallback position—if I am wrong about Semantic Premise—is that Metaphysical Premise should then be rejected. Overall it seems to me that there is something of a paradox brewing between the claim that decapitation can cause death, Semantic Premise, and Metaphysical Premise. One of these must go. I say Semantic Premise should be the first to go, and Metaphysical Premise is the fallback. Moore's position of denying that decapitation can cause death strikes me as the least plausible choice of the lot.
B. Against the Semantic Premise
Look again at the Davidsonian rendering of the positive “James killed Smith”:
(∃e) (Killing(e) & Agent(e) = James & Patient(e) = Smith)
There are many places where one could insert a negation. There is the option of giving the negation widest scope, as per Semantic Premise. But there are a range of alternatives in which the negation has narrower scope, such as:
(∃e) NOT (Killing(e) & Agent(e) = James & Patient(e) = Smith)
This says that there is an event (a positive presence) that is not a killing of Smith by James. Such is a rather truistic claim, though if we think of the existential quantifier as bearing a contextually given restrictor, it might in a given context impose stronger demands. A further example:
(∃e) (NOT Killing(e) & Agent(e) = James & Patient(e) = Smith)
This says that there is something done by James to Smith which is not a killing—such would be true if, for instance, James kissed Smith instead of killing him. Again, context might serve to restrict the domain of quantification further, making for even stronger demands about what James did to Smith.
On all of the options other than the one where the negation takes widest scope, the absence claim is denoting a positive presence but merely describing it negatively. For instance, the last example above describes a positive event of James doing something other than a killing to Smith. Any of these narrow-scope options—if sustainable—would undermine Semantic Premise. Absence talk can be a way to describe something, as Hart and Honoré advocate:
The corrective here is to realize that negative statements like “he did not pull the signal” are ways of describing the world, just as affirmative statements are, but they describe it by contrast not by comparison as affirmative statements do.Footnote 53
So which is the right semantics for omission claims? Does the negation take widest scope or narrower scope? I think the answer is: it depends. In particular, it depends on what the speaker intends. The widest-scope and the various narrower-scope readings are all perfectly legitimate semantic values that a speaker might intend to express through omission claims. This is just a familiar scope ambiguity in ordinary language. The metaphysical problem thereby disappears. When I truly claim that the gardener's failing to water my flowers caused them to wilt, I need only be read as describing the cause with the negation taking some narrower scope, to be speaking of something. So I am advocating, as against the idea in Semantic Premise that omission claims can only have widest-scope negation, the pluralistic idea that omission claims can have multiple interpretations with negations of various scopes, including both widest-scope and narrower-scope readings.
1. Evidence for Narrower-Scope Readings
To see how natural narrower-scope readings can be for some omission claims, consider “Tom's failing to pass the driving test.” On the widest-scope reading that Moore insist on, this yields:
NOT (∃e) (Passing(e) & Agent(e) = Tom & Theme(e) = driving test)
This does not require Tom to have ever taken a driving test, and indeed, this does not even require Tom to exist. Such a reading is possible but not very natural. A much more natural reading—indeed perhaps the most natural reading of this phrase without strong contextual cues to the contrary—has the following fairly narrow-scope negation:
(∃e) (NOT Passing(e) & Agent(e) = Tom & Theme(e) = driving test)
This tells us that Tom indeed took a driving test (an actual event) but merely did not pass it. Likewise consider “Mary's failing to score a goal,” which is naturally read as denoting an event in which Mary missed an attempt, and “Tom the tax cheat's omitting to provide true information to the government,” which is naturally read as denoting an event in which Tom lied.
As further evidence that there are narrower-scope readings available for omission claims, consider phrases such as “the best thing that ever happened to me” or “the most embarrassing event of my life.” These phrases obviously denote actual events. And it is perfectly natural to identify these with omissions, absences, and failings, as in “The most embarrassing event of my life was my failing to close my zipper.” Indeed, here are a few of the many examples Google turns up:
The best thing that ever happened to me was not getting into the Peace Corps in 1973.
http://www.nationalservice.gov/for_individuals/current/stories_detail.asp?tbl_stories_id=97
The worst thing that ever happened to me was not getting an ass-pile of free money.
The most critical event was the failure to close the bow doors before departure.
www.emsa.europa.eu/implementation-tasks/accident-investigation/download/499/516/23.html
The single most unfortunate event was the failure to raise interest rates during 2002–2003 after the Reserve Bank had engineered their fall as an antidote to the Dot.Com crash two years earlier.
Some subjects also received an analogous visual discrimination task, or auditory tasks in which the rare target event was the omission of a tone, or the repetition of a tone within a series of alternating tones.
The most frequent event was the failure of S trust beneficiaries to timely file the elections required by Sec. 1361(d)(2).
I'm in Atlanta at the 90th Annual Meeting of the American Meteorological Society, and the picture I'm getting from the presentations here is that the most significant weather event of 2009 was the failure of the summer rains in the Horn of Africa.
Indeed, it is amusing in this regard to note the following definition of “failure,” attributed to the author Steven K. Scott:
An event in which you did not achieve your desired outcome (definition of failure).
So I conclude that there are narrower-scope readings of omission claims. Thus omission claims can be used to describe actual events.
2. Moore's Three Arguments against Narrower-Scope Readings
Moore takes his semantics-of-omission claims—on which these must involve wide-scope negation—to be “in conformity with ordinary usage and the semantic intentions of ordinary speakers when they speak of omissions.”Footnote 54 Note that I need not contest that wide-scope negation readings are in conformity with ordinary usage and the semantic intentions of ordinary speakers in some cases. I am not contesting the claim that wide-scope negation represents a perfectly legitimate semantic value that a speaker could intend to express. All I am claiming is that there are also the narrower-scope readings that a speaker might also intend to express.
But Moore seems to think that only his wide-scope negation reading is possible. He does say: “theorists sympathetic to omissive causation seek to substitute a different generic meaning for ‘omission,’ a meaning that takes omission to be some actually existent particulars (and not just absence of types, as I contend).”Footnote 55 Indeed, he explicitly notes my suggestion of moving the negation inside the scope of the quantifier and offers three objections.Footnote 56
Moore's first objection is: “What we do not typically do is use a property an event does not have to form a description with which to pick out that event.”Footnote 57 By way of reply, I think this is just false (and I wonder what evidence Moore had in mind for such a general sociological claim). There are plenty of contexts in which it is perfectly natural to pick out an event by a property it does not have, as evidenced by the Google quotes above (in Sec. IV.B.1). Or think of how one might—in an appropriate context—refer to someone's wedding as “the event in which she failed to remember her vows,” or (more playfully) “the event in which she failed to remain sober,” or (imagining now a quite different wedding) “the event in which she failed to get drunk.” Any speaker of the language will understand how this works and be familiar with many examples of this sort.
The situation is no different with reference to individuals. For instance, I might describe a reliable friend as “the person who never let me down.” Likewise, the phrase “the person not wearing pants” will—in an appropriate context—denote a real present person, not some negative spook.Footnote 58 Overall, natural language has the expressive resources to pick out actual events and individuals via descriptions (with or without negations), and it would be shocking if ordinary speakers could not make use of these resources.Footnote 59
Moore's second objection to the possibility of narrower-scope readings is the objection from “indeterminacy,” since:
[W]hat is not going on at t when everything that is going on at t is going on, is very very large. True enough, there is no saving of Jones by me going on at t; but the Martians are not invading, the moon is not disintegrating, the flies are not buzzing at t too.Footnote 60
I offer two replies, the first of which is that language is sometimes indeterminate. The fact that a reading would make for indeterminacy simply cannot exclude that reading as a possibility.
My second reply is that this level of indeterminacy afflicts only the reading on which the negation is just inside the quantifier and the quantifier goes unrestricted, as in:
(∃e) NOT (Killing(e) & Agent(e) = James & Patient(e) = Smith)
This is indeed highly indeterminate in reference, since it can denote virtually any event in the universe, as long as that event is not a killing of Smith by James. But the indeterminacy diminishes to the extent we find material in the scope of the quantifier that is not in the scope of the negation, as in:
(∃e) (NOT Killing(e) & Agent(e) = James & Patient(e) = Smith)
This makes a relatively more determinate claim that can only denote something James did to Smith.
The indeterminacy also diminishes to the extent we find restrictions on the quantifier. In a more developed representation, the existential quantifier over events “(∃e)” would be replaced by a contextually restricted quantifer “[∃e: e has F],” where F is a property supplied by context. Then—given a suitably restrictive value for the restrictor—even the reading with the negation just inside the quantifier could have quite determinate reference:
[∃e: e has the property of being done by James to Smith at midnight last night in the alleyway] (NOT Killing(e) & Agent(e) = James & Patient(e) = Smith)
Likewise, “the person not wearing pants” is not condemned to indeterminacy in reference due to all the people not wearing pants in the universe. In a suitable context where the domain of individuals is restricted to those in a given room, it might determinately denote the one and only person in that room wearing a dress.Footnote 61
Moore's third objection to the possibility of narrower-scope readings is the objection from causal irrelevance:
[M]uch of what is going on (in the totality of states of affairs at t) is causally irrelevant to Jones’ death. . . . On no plausible theory of the causal relation is my conversing a cause of Jones’ death, even in the situation where that was what I was doing rather than saving Jones.Footnote 62
Thus consider “Moore's failing to save Jones” and suppose it has the reading (as I say it has) with the negation bearing on the event being a saving of Jones:
(∃e) (Agent(e) = Moore & NOT (Saving(e) & Patient(e) = Jones))
This just requires that Moore did something that was not a saving of Jones. (That there is such a reading may be highlighted by noting the propriety of claims such as “Moore's failing to save Jones was the worst thing Moore ever did.”) Context permitting, “Moore's failing to save Jones” might co-denote with “Moore's conversing with Schaffer.” In such a context—assuming that causal ascriptions are extensional—“Moore's failing to save Jones” could only describe a cause of Jones's death if “Moore's conversing with Schaffer” equally described a cause of Jones's death, and the objection says that the latter result is absurd.
In reply, I think that this latter result is completely fine and not at all absurd. It might help to shift the example slightly. Consider a context in which “Lifeguard's failing to save Sinker” and “Lifeguard's taking a nap” co-denote. I think it is quite clear that in such a context we may appropriately speak of “Lifeguard's taking a nap” when listing causes of Sinker's death. This is a straightforward case of negligence. It helps to shift the example to one for which it is evident how the cause could have made a difference to the effect. In the case Moore provides, it is not evident how Moore's conversing with Schaffer could have been among the causes of Jones's death (it almost sounds as if Jones must have died of boredom with the conversation—which is of course hardly conceivable). But once we move to a context where “Moore's conversing with Schaffer” and “Moore's failing to save Jones” are co-denoting, it becomes evident how Moore's conversing with Schaffer made a difference to Jones's fate. Indeed we need only imagine that Moore is Lifeguard and that his conversing with Schaffer was just the way he whiled away his time while poor Jones (a.k.a. Sinker) flailed around and screamed for help.
3. A Robust Metaphysical Distinction?
Not only does Moore think that omission claims must always involve the negation taking widest scope, he also thinks that some things that that look on the surface like straightforwardly positive event nominals are—metaphysically speaking—absences in disguise. Thus he notes that “Smith's survival just is the absence of his dying,” and continues, “We must penetrate the seemingly positive language (‘surviving,’ ‘starving,’ ‘keeping of the secret,’ ‘ignoring the pleas of the drowning man’, etc.) to the underlying reality: to survive is not to die.”Footnote 63 Later, he speaks of “the most troublesome of this nest of conceptual worries,” asking “Is death an event, but not-dying its absence? Or is surviving the event, and not-surviving (i.e., dying) its absence?”Footnote 64
I find Moore's claim of a robust metaphysical distinction implausible, especially when combined with No Absences. For on Moore's view, whichever one of dying/surviving is the disguised metaphysical absence cannot possibly have causes or effects. He opts to say that dying is the event and surviving the absence.Footnote 65 He is thus committed to the claim that surviving cannot be involved in causal relations. So he must deny that when Doctor takes bold action to save Patient from death, Doctor's action can cause Patient to survive. And he must deny that Patient's surviving can cause Patient's loved ones to rejoice. (Had Moore chosen surviving as the event, he would have been committed to denying that Killer's action can cause Victim to die, and committed to denying that Victim's dying can cause Victim's loved ones to mourn.)
Likewise, Moore opts to say that telling a secret is the event and keeping the secret is the absence. He is thus committed to the claim that keeping a secret cannot be involved in causal relations. So he must deny that when Gangster threatens Witness, Gangster's threat can cause Witness to keep the secret. And he must deny that Witness's keeping the secret can cause Gangster to be acquitted. (Had Moore chosen keeping the secret as the event, he would have been committed to denying that Gangster's threat could have backfired and actually caused Witness to tell the secret, and committed to denying that Witness's telling the secret could have caused Gangster to be found guilty.)
Things get worse. If we stick to the standard neo-Davidsonian regimentation of action sentences, “Mary kept the secret” comes out as:
(∃e) (Keeping(e) & Agent(e) = Mary & Theme(e) = the secret)
On Moore's view, it is then impossible for “Mary kept the secret” to come out true, no matter what Mary might have done or failed to do. For Moore countenances no such events. The absurdities mount: if Larry claims to know that Mary kept the secret, then—since knowledge is factive—Larry must be wrong (no matter what Mary might have done or failed to do, and no matter what evidence Larry might have).
Overall I would say that when Victim dies there is an actual event to be found involving Victim that—context permitting—can be denoted by “Victim's dying” or by “Victim's failing to survive.” These are just positively and negatively valenced descriptions of a single actual event. Likewise, when Patient survives, there is a single actual event to be found involving Patient that—context permitting—can be denoted by “Patient's surviving” or by “Patient's failing to die.” Similarly, “Mary's keeping the secret” and “Mary's not telling the secret” are both descriptions of a single event involving Mary, and “Mary's telling the secret” and “Mary's not keeping the secret” are both descriptions of a different sort of single event involving Mary. The key is recognizing the possibility of negative descriptions with narrower-scope negation. This allows dyings, survivings, secret tellings, and secret keepings to figure as possible causes and effects and allows “Victim died,” “Patient survived,” “Mary told the secret,” and “Mary kept the secret” to be possibly true.Footnote 66
Moore notes that one might be led to think that “the positive/negative valencing of events is a purely arbitrary feature of language, not a feature of the world itself.”Footnote 67 But he goes on to offer two main reasons for thinking that there is a robust metaphysical distinction in play. The first reason is an appeal to science: “[W]hat is an actual event or state of affairs, and what is an absence, is a matter for science. It is up to our best science to tell us whether there are really dyings, or whether surviving is the actual event.”Footnote 68 In reply, I must say that I find it hard to imagine a scientist wasting her time with this question or devising an experiment that bears on whether dying or surviving is the real metaphysical presence (what exactly would she look for?). I am all in favor of deferring to science in describing reality. I would add only that scientists—in my estimation—would disdain the distinction Moore needs.Footnote 69
Moore's second reason for thinking that there is a robust metaphysical distinction in play comes from a purported parallel with objects: “There is no room for any plausible skepticism about the distinction between actual elephants and absent ones.”Footnote 70 This is true but beside the point. I absolutely grant that there is a distinction between actual objects and absent ones, and that there is a distinction between actual events and absent ones. There is a world out there. That is not in dispute. What is in dispute is whether omission claims must always denote absent events (as Moore thinks) or may also denote actual events that are out there in the world (as I maintain).
4. Revisiting the Issues
If omission claims can denote actual events—and can do so in the context of causal ascriptions—what becomes of No Absences, the taxonomy of absence causation (see Sec. II.A), and the beheading problem? Two main revisions are in order. First—given that Moore and I disagree about what absences actually are—our dispute is best understood as a linguistic dispute over whether omission claims can denote causes and effects:
No Absence Talk: Omission claims cannot denote causes or effects.
Indeed there is a sense in which our earlier dispute over No Absences was partly verbal. For Moore understands this thesis as the claim that negative existential quantifications over events cannot be causes or effects (as per Metaphysical Premise), and I am noncommittal on this thesis (see Sec. IV.C). And I understand No Absences as the claim that events described in negative terms can be causes and effects, and Moore at least has no objection to events being causes and effects (even if he denies that they can be described in negative terms). What we most directly disagree about is the role of omission claims in causal ascriptions, as No Absence Talk properly captures.
Second, the taxonomy of absence causation and the statement of the beheading problem need revision. The distinction between positive and negative causation is not a difference in the metaphysical status of the relata but rather a difference in the relation. In connective causation there is a physical process (such as a line of energy flow) connecting the events involved. In disconnection there is no process connecting the events. Rather what is salient in these cases is that there was a preexisting process preventing the effect event, which the cause event disconnects.Footnote 71 Thus in beheading cases, there need be no energy flow from the beheading to the brain death of the victim. Rather what is salient is that there was a process preventing brain death—namely the process of oxygenated blood flow which had been sustaining the brain—and the beheading disconnects this preexisting process.
So the underlying metaphysical dispute between Moore and me concerns whether there must be a physical process connecting cause and effect:
Connectedness: One event can cause another only if the two events are connected by a physical process, such as an energy flow.
Disconnections show that there are other ways in which one event can cause another. Connectedness is a hasty generalization from colliding billiard balls. Connectedness is a consequence of those versions of singularism that would reduce causation to physical processes such as energy flows.Footnote 72 Though Connectedness is not automatically a consequence of primitivist versions of singularism, and in that sense singularism is compatible with the denial of Connectedness.
In what follows, I stick with my policy of speaking in Moore's terms and thus continue to treat our dispute as centering on No Absences. But in my own terms our dispute is a linguistic dispute over No Absence Talk, coupled with a metaphysical dispute over Connectedness.
C. The Metaphysics of the Causal Relata
I am arguing that Semantic Premise should go. But suppose that I am wrong; Semantic Premise is true, in that natural language is not able to express claims about actual events by negative descriptions (or at least natural language is not able to express such claims via anything like omission claims). Indeed, suppose that omission claims must always denote negations of existential quantifications over events, as Moore thinks. Even supposing all of this, I would still say that absences can be causal. For there is also Metaphysical Premise, and it would seem to me—at that point in the dialectic—that one would have to choose between thinking that decapitation cannot cause death and thinking that negations of existential quantifications over events cannot serve as causal relata. By my lights, the former claim deserves far more credence than the latter.
The metaphysics of the causal relata are highly controversial, and there are many respectable views available. For instance, perhaps the main alternative to thinking of the causal relata as events is thinking of the causal relata as facts.Footnote 73 On the fact view, the causal relata are abstract entities that can be identified with—or at least are in correspondence with—true propositions. There are negative facts, for there are true negative propositions. So even if omission claims must always denote negative existential quantifications over events, on the fact view these can indeed serve as causal relata. Indeed, as Moore points out,Footnote 74 one of the main arguments for the fact view over the event view is to accommodate absences. Thus D.H. Mellor maintains: “[T]he ‘C’ and ‘E’ in a true causal ‘E because C’ need not assert the existence of particulars. They may deny it. . . . They are negative existential statements.”Footnote 75
Moore's main objection to the fact view is the standard objection that facts are too otherworldly to stand in causal relations.Footnote 76 As Bennett—himself an advocate of the fact view—memorably puts the point:
Some people have objected that facts are not the sort of item that can cause anything. A fact is a true proposition (they say); it is not something in the world but is rather something about the world, which makes it categorically wrong for the role of a puller and shover and twister and bender.Footnote 77
Moore immediately suggests that the fact theorist should retreat from facts (understood as true propositions) and instead uphold the truthmakers for facts (what Mellor calls “facta”)Footnote 78 as the causal relata. But at that point, as Moore notes, the fact theorist has lost her claim to saving absence causation: “Because facta . . . are real entities, there is no obvious sense to be assigned to the view that an absent thing . . . can be a singular cause. The advantage here seems to be only with linguistic facts, that is, true propositions.”Footnote 79
By way of reply, I do not think that the fact theorist should retreat from facts so easily and lose her ability to accommodate absence causation so readily. I think that it is at least defensible for the fact theorist to stick to her view that the causal relata are indeed abstract proposition-like entities. To begin with, there are relations in the world that clearly look to involve propositions. For instance, if I believe that snow is white, then it looks as though I am standing in the belief relation to the proposition that snow is white. But my beliefs are clearly part of the world. So from the claim that causation relates propositions, it does not immediately follow that causation is otherworldly.
I think that the key to understanding the belief case is to appreciate that beliefs are not fundamental entities. Rather, people stand in belief relations to propositions in virtue of certain deeper features of reality. These deeper features might, for instance, be brain states that are physically interrelated in a way structurally isomorphic to the way certain propositions are informationally interrelated. It is then in virtue of their concrete and this-worldly brain states that a person may be said to believe a given proposition.
The fact theorist—to sustain a parallel idea about causation—need only maintain that causation is not a fundamental relation. Rather, she should think of causal relations as obtaining in virtue of certain deeper features of reality.Footnote 80 These deeper features might, for instance, be just the global history of events (the Humean mosaic); or they might involve fundamental laws of nature, or irreducibly singular processes, and so on. It is in virtue of these concrete and this-worldly deeper features that one fact may be said to cause another.Footnote 81
I am not defending the fact view but only saying that it is not indefensible. If it came to choosing between giving up the controversial metaphysical thesis that true propositions are ineligible to play the role of the causal relata and giving up the uncontroversial commonsensical thesis that decapitation can cause death, I would give up the controversial metaphysical thesis. (Of course, if I am right about the semantics of omission claims, then no such choice need be made. This is a fallback position.)
V. EXPLAINING THE INTUITIVE CONTRASTS
Alongside the problem of nothingness, Moore offers three further arguments for No Absences, all of which center on intuitive contrasts. First, Moore (following Phil Dowe and Helen Beebee)Footnote 82 claims a metaphysical “intuition of difference” between the causation in presence cases and what happens in absence cases. Second, Moore maintains that there is a moral difference between presence and absence cases. And third, Moore maintains that there is a causal difference between presence and absence cases that surfaces in overdetermination. No Absences is then claimed to provide the best explanation for these intuitive contrasts.
There is much to say about each of these arguments beyond asking: Could anything here really establish that beheading someone cannot cause them to die? But I will emphasize one line of reply, which is that all of Moore's “presence” cases are actually disconnections (his paradigm “presence” cases are drownings and shootings). So whatever his intuitive contrast might turn on, they could not possibly turn on the involvement of absences.
A. The Metaphysical, Moral, and Causal Contrasts
To begin with, Moore reports a metaphysical intuition that the causation in some cases differs from what happens in some absence cases.Footnote 83 As he explains:
[F]ailing to prevent something does not seem anything like causing that thing to exist. Suppose you hold someone's head under water until they drown and I do not stop you. Have we each caused the victim's death? The metaphysical intuition I find compelling: you caused the death, whereas I only failed to prevent it.Footnote 84
Moore then immediately offers No Absences as the best explanation of this metaphysical intuition of difference.
Second, Moore maintains that there is a moral difference between some cases and some absence cases. He does so by contrasting cases of drowning versus not stopping a drowning, noting that the drowner clearly bears a higher level of blame. He then rhetorically asks: “How are we to make sense of this moral distinction except with a metaphysical distinction between killing and not-saving, or more generally, between actions causing and omissions failing to prevent?”Footnote 85 Here No Absences and Disjunctivism are jointly offered as the best explanation for the diminished level of responsibility found in the not stopping a drowning case.
Moore also offers the real case of Judge Tally, in which Tally found out that the Skelton boys were riding into the next town to kill a man named Ross who had seduced their sister, and found out that a telegram had been dispatched to Ross warning him to flee.Footnote 86 Tally—siding with the Skelton boys in the dispute—ordered the warning telegram not to be delivered, and so the Skelton boys caught Ross and shot him dead. Moore rhetorically asks: “[C]ompare [Tally's] responsibility to that of Ross's actual killers, the Skeltons. Is Tally as much to blame as those who shot Ross to death?”Footnote 87 The idea again is that Tally's lesser level of blame can only be explained via No Absences plus Disjunctivism.
Third and finally, Moore maintains that there is a causal difference between presence and absence cases surfacing in overdetermination. Thus Moore considers cases in which two shooters inflict equally fatal wounds on the same victim at the same time, and notes that each individual shooter should be blamed for the death.Footnote 88 He contrasts such cases with cases in which two absences yield equally fatal results, such as a case in which a fatal bus accident is preceded by the mechanic failing to fix the brakes and the bus driver failing to apply the brakes.Footnote 89 He claims that in such cases neither the bus driver nor the mechanic has caused the accident, offering No Absences as the best explanation for the causal difference.Footnote 90 He also notes a concomitant moral difference, which he takes to be best explained by Disjunctivism: “[S]ince neither the bus driver nor the mechanic had the ability to prevent the accident, neither can be blamed for the accident.”Footnote 91
B. The Bearing on Absence Causation
I propose to accept arguendo the existence of the metaphysical, moral, and overdetermination contrasts. My question is whether the existence of these intuitive contrasts provides any support for No Absences. Moore makes two assumptions about these cases. First, he assumes that he is always contrasting cases of purely positive causation with cases involving absences. Second, he assumes that there is no other possible explanation of the contrasts other than the involvement of absences. I think both assumptions are manifestly false.
Starting with the first assumption, it might help to tabulate Moore's contrast cases:
But—as Moore himself admits when facing the beheading problem (see Sec II.D)—all the cases on his “positive” side of the ledger involves absences (Sec. II.B). None of these intuitive contrasts could possibly turn on the involvement of absences, so nothing here could possibly support No Absences. If anything, the fact that Moore has the metaphysical intuition of causation in the drowning case shows that by his own lights disconnections are causal.
Turning to the second assumption, there are many other possible explanations for these intuitive differences, and Moore does not consider any alternatives. So even if (counterfactually speaking) he were right that his “positive” cases are absence-free, he would still have hardly succeeded in making the case that No Absences was the best explanation of the intuitive contrasts observed. Indeed he would have hardly begun.Footnote 92
There are parallel intuitive contrasts between cases that are evidently independent of any presence/absence distinction. If White hands Black the weapon, and Black then uses the weapon to kill Red, there is a strong intuition that Black is more responsible than White. Likewise if White deals Red a minor blow at the same time as Black deals Red a major blow, and both blows together are just enough to kill Red, there is a strong intuition that Black is more responsible than White. So there must be intuitive contrasts explicable in terms other than No Absences.
It might help to display this wider array of moral contrasts:
Assuming that there is a single underlying explanation for these intuitive contrasts, it hardly looks as though the matter has anything to do with absences. Indeed it cannot have anything to do with absences, since every one of these cases is a human action and thereby involves disconnection (see Sec. II.B).
C. Explanatory Speculations
So what does explain the intuitive metaphysical, moral, and causal differences that Moore reports? I think this is an excellent question, though it is an optional question for me in my dispute with Moore. I am primarily concerned to rebut a style of argument for No Absences, which represents No Absences as the best explanation for these differences. Showing that No Absences cannot possibly explain these differences (since all the cases involve absences) suffices to rebut this style of argument. One need not offer a better explanation to rebut a bad one.
That said, I think that there are many, many differences between the cases Moore contrasts. They all involve absences, they are all disconnections, and by my lights they all involve causation. But the fact that all of these cases are similar in these three respects hardly entails that they are similar in every respect. Moreover, it is not obvious that there needs to be a single underlying explanation for this range of contrasts.
Still, I want to suggest two possible lines of explanation that might hold some promise. The first possible line builds on Moore's own idea that causal relations are scalar, coming in degrees.Footnote 93 He invokes this idea to address the issue of causal apportionment, suggesting that one way in which the law apportions responsibility is in terms of the degree of causal contribution (how substantial a given factor was in producing the effect). To illustrate, he offers the case of the shooter and the accomplice who drives the getaway car, saying: “[E]ven though the accomplice contributes to the victim's death in the sense that he makes it possible, his causal contribution is nowhere near that of the actual killer.”Footnote 94
Granting arguendo that Moore is right that causal relations come in degrees (or that there is some associated degree-based notion such as proportion of causal responsibility), and that the degree of causality corresponds to the degree of culpability, then Moore himself has provided a straightforward style of explanation for the intuitive contrasts (which has nothing to do with absences). The explanation is that he is contrasting more and less substantial causal factors. Drowning the victim, shooting the victim, and providing the major blow would all count as major causes, while not stopping the drowning, merely providing the weapon (or driving the getaway car), and providing the minor blow would all count as minor causes. Thus there is a rival explanation for the intuitive contrasts in Moore's own terms.
A second possible line of explanation for the moral contrasts is to note that the cases involve acts of different types, with morality being sensitive to act type. Some act types—such as drowning and shooting—are prohibited in the strongest terms, while other act types—such as not stopping a drowning and driving a getaway car—meet with prohibitions of lesser strength. This is to presume a morality that is sensitive not just to consequences but to the type of act that produced them. Some ways of causing death are worse than others, not because of any differences in the manner of death but rather due to differences in the way the agent acts. This fits the law's conception of a crime as defined by act type and thereby facing prohibitions of various strengths.Footnote 95 I should emphasize that this second line of explanation is primarily an attempt to explain the intuitive moral contrasts between the cases and does not speak to any intuitive metaphysical or causal contrasts. (For the record, the moral contrast is the only intuitive contrast I find clear.)
To summarize, the intuitive contrasts Moore notes simply cannot be contrasts between cases of positive causation and cases with absences, since all of his cases involve absences. This is no accident. Causation by disconnection features in all human actions and in virtually every case of causation considered by the law (see Sec. II.B). But I want to conclude this section by emphasizing that Moore has brought to light some important and seemingly connected contrasts. After one sees that No Absences cannot explain these contrasts, one is still left with the deep question of how to explain them.
VI. CONCLUSIONS
As Sir Frederick Pollock memorably advised, “the lawyer cannot afford to adventure himself with the philosopher on the metaphysics of causation.”Footnote 96 I think that Moore's quest to substantiate Unity is noble but that the adventure has ended badly, in causal and moral absurdities.
The causal absurdities stem from No Absences and emerge in the beheading problem, where Moore is committed to denying that decapitation can cause death. He is equally committed to denying that drowning or strangulation can cause death and to denying that pulling the trigger of a gun can cause death. Indeed he is committed to the result that human actions cause nothing. This is an outcome the lawyer surely cannot afford.
The moral absurdities stem from No Absences plus Disjunctivism, and manifest themselves in both the beheading problem and the problem of correlates. Moore is committed to denying that the beheader is fully morally responsible for the death of the beheaded, and he is committed to complete legal and moral impunity for the preempting beheader. He is also committed to attributing spurious moral responsibility to mere correlates as well as to effects. By his lights, the defense can plead for a lesser sentence for the beheader on grounds that it was only a beheading. And by his lights, the prosecution can argue for liability due to mere correlation with a criminal act on grounds of counterfactual dependence with the crime. Such adventures would be laughed out of court.
The fatal misstep that began this whole misadventure was No Absences. If absences are allowed to be causal, then the way is clear to admit the obvious truth that decapitation can cause death and so straightforwardly resolve the beheading problem. The way is then clear to maintain a unified causal account of legal liability and moral responsibility (no causation, no blame), without descent into disjunctivity. This yields a straightforward solution to the problem of correlates, since correlation is not causation.
In short, the defender of Unity who rejects No Absences need not suffer the beheading problem nor the correlation problem and thereby gains some opportunity to bring Moore's noble adventure to a happier conclusion.
