“Moore or Less” Causation and Responsibility
- First Online:
- Cite this article as:
- Alexander, L. & Ferzan, K.K. Criminal Law, Philosophy (2012) 6: 81. doi:10.1007/s11572-011-9127-8
- 266 Views
What makes someone responsible for a prohibited result? Traditionally, those who believe results matter morally and legally have looked at the matter this way: If the actor acts culpably vis-à-vis the result—the result is her purpose in acting, or she believes to a practical certainty that the result will occur if she acts, or believes that the result’s occurrence is sufficiently likely—and if the result does occur and would not have occurred but for her culpable act (counterfactual dependence), then she is responsible for that result.1
There were two qualifications attached to this view, however. First, the view did not produce the intuitively correct verdict in cases where the prohibited result was overdetermined—where there was another sufficient cause of the prohibited result in addition to the culpable act (86). So the requirement of counterfactual dependence (CD) was relaxed for such cases because of the strong intuition that being a sufficient cause should be, well, sufficient. And second, where the causal route from the culpable act to the prohibited result was a deviant or surprising one, holding the actor responsible for the result again contradicted the intuitions of those who believed results should matter. So a requirement of “proximate causation” was meant to limit the scope of CD responsibility for results.
The proximate causation requirement has turned out to be extremely problematic. No good theory has ever been advanced to explain and justify which causal chains satisfy proximate causation and which do not. The two most prominent tests were whether the harm caused was “foreseeable” and whether the harm was “within the risk.” But foreseeability seems related to the culpability of the act and not to causation (Kessler 1994). And “harm within the risk” misunderstands the nature of risks, which come in baskets of differing probabilities of different harms, and which are epistemic rather than ontic. (When harm of whatever kind results from the culpable act, then whatever the causal chain between the culpable act and the harmful result, the actual probability of that harm occurring in the way it did was 100%.)
Because we believe an actor’s culpability is all that should matter, morally and legally, and not what does or does not happen as a result of her culpable act, the conundrums presented by overdetermination and proximate causation are beside the point (Alexander and Ferzan 2009). For Michael Moore, however, results do matter morally and legally (29–33), so resolving these problems is crucial to responsibility ascriptions. Causation and Responsibility is Moore’s magisterial attempt at resolving the conundrums of causation.
In Causation and Responsibility, Moore seeks to meet the challenge of showing how moral (and therefore legal) questions turn on metaphysics. Moore thus must make successful metaphysical claims about what causation is, make successful moral claims that morality turns on these metaphysical claims, and then make successful claims that law ought to turn on morality.2
Moore’s books require a quiet room, a cup of tea, and well, the intellectual equivalent of a good hair day—if one is not feeling smart that day, one ought not bother to pick up something written by Michael Moore. On those focused days, though, the book pays tremendous dividends. Indeed, part of the joy of the book is learning about that in which one has little or no expertise. Moore is a master teacher, and one cannot help but learn something new in this book. As criminal law theorists, we walked away with a pretty good understanding of tropes. And metaphysicians will learn about accomplice liability and the evil deeds of one Judge Tally who prevented a warning telegram from being sent.
In terms of organization, we do have some quibbles. First, Moore spends some time “pruning the law’s demands for the concept of causation.” A couple chapters later, however, Moore cuts down the tree. This is more than a bit confusing because Moore presents arguments that he does not truly endorse. All of this might be well and good if one were not simultaneously trying to wrap one’s mind around the metaphysical machinery that Moore has brought to bear on the topic. Indeed, we worry that many of Moore’s conclusions are lost among the nuanced arguments. Second, given that Moore spends the beginning chapters pruning a tree he later chops down, it might have been nice to have a chapter or two in which Moore gives us a map of the new forest. Some of Moore’s important moves, such as the introduction of the “fit” relation discussed below, are relegated to passing mention and footnotes.
No review could possibly do justice to the many nuanced arguments in Causation and Responsibility. Our review will focus on two important moves within the book. First, in making causation matter while attempting to resolve proximate causation conundrums, Moore adds a new dimension to the analysis: scalarity. Scalarity introduces three quantitative determinations into the causation inquiry: (1) how much of a cause someone was; (2) how much a result counterfactually depended on an action or omission; and (3) how much of the result and the way the result came about was within the actor’s contemplation. Each of these shades of gray enters into the ultimate question of how blameworthy an actor was. In our view, this notion of scalarity does not resolve the problems presented by causation and often merely relocates them.
The second portion of our review focuses on the nature of deontological constraints. In Causation and Responsibility, Moore argues for the deontological relevance of intentions, as opposed to the appropriation of another’s body, labor, or talents as a means. And he has since argued that he sees no difference between “using as a means” and “intending harm to the one used” (Moore forthcoming). We believe these are different and explain, in the second part of this review, why the using view is neither redundant nor wrong.
Moore has been a strong critic of proximate cause doctrines (chs 7–10; Moore 1993). He does not, however, reject the intuitions on which they are based. His conclusion is that the error of prior theorizing in the “results matter” camp was the assumption that the culpable actor is either wholly responsible for prohibited results or not responsible at all for them. Moore believes in the scalarity of responsibility for results, that culpable actors are “more or less” responsible for them.
Moore’s general thesis is that “causation is a singular relation between token states of affairs” (54). Moore injects scalarity into responsibility in three ways. First, causation—which is not CD for Moore—can be scalar, a matter of degree (275). One can be a big cause, a medium-sized cause, or a small cause of a result. The amount of causation affects the amount of responsibility and blameworthiness (276). Moore’s rough cut is that one’s causal contribution cannot be de minimis for moral responsibility, which he takes to require that it be substantial (276).
Second, CD itself can be scalar. Although CD is not causation, CD is still a way of rendering one responsible for results, and a necessary way in cases of culpable omissions, which for Moore are causally inert (chs. 17–18). CD is not as blameworthy as big causation, but is more blameworthy than small (and perhaps medium-sized) causation (307, 468). However, if CD were not scalar, then because it is present in every case of causal responsibility other than cases of overdetermination, it would cancel out the scalarity of causation in many cases and render responsibility total even in cases where the causal contribution was minimal. Fortunately, as Moore sees it, CD is, like causation, a matter of degree. The degree is a function of the number and closeness of possible worlds in which the culpable actor’s act produces the same result (320). Where both causation and CD exist, the more serious relation governs (470).
Third, there is scalarity in the relation between what the actor believed her act would do and how it would do it, and what the act actually does and how it does it. Because no events transpire exactly as actors foresee them, this “fit” question is a loose requirement and matter of degree.3
There are serious theoretical problems with each of these attempts to inject scalarity into responsibility for results. The notion of big versus small causes seems to rely on a notion of force or energy that is inapplicable to many instances of presumably culpable causation. The notion of scalarity required to handle CD, with its counting of nearby possible worlds, is problematic. And the fit requirement appears merely to take the sloppiness of standard proximate cause analysis and stick it under a new label.
The Scalarity of Causation
Moore believes that one’s causal contribution to a prohibited result can be large, small, or somewhere in between. The principal who shoots the bank guard is a big cause of the guard’s death. The accomplice who gave the principal the murder weapon is also a cause, but less of a cause than the principal. So also is the emergency room doctor, who had she not been negligent would have saved the guard’s life.
Now notice that these judgments are not judgments about the level of culpability of the accomplice—who might have wanted the guard to be shot—or of the doctor, who is surely less culpable than the principal or the accomplice, but which fact seems irrelevant to her causal contribution to the result as a metaphysical matter.
Moore needs scalarity because he has abandoned proximate causation and thus has no other mechanism to limit our responsibility for the far-off results of our actions. So, if Caesar’s crossing the Rubicon is part of a chain of events that one day results in Alex murdering Betty, Moore needs a way to cut that tie. It is true that Caesar lacked the requisite mental state, but without some limit Caesar would still be responsible as a matter of the metaphysics because Moore accepts that mental states are causally inert. Without proximate cause to limit Caesar’s responsibility, scalarity serves as the new limiting principle.
How are we to understand this causal scalarity? At times, Moore seems to base the notion on the metaphor of ripples in a pond, which gradually peter out. That suggests that the length of the causal chain between the culpable act and the prohibited result, or perhaps the number of links in that chain, is what determines the size of the causal contribution. But the Unabomber might rely on many distinct acts between his mailing his letter bomb and the victim’s opening it. (Consider the number of postal employees who have to handle it, for example.)4
If the number of links in the causal chain does not seem to matter in all cases, what else might determine the size of the causal contribution? Surely not the kinetic energy of the culpable act. The killer who only needs to give the victim a tiny shove to get him to fall to his death—because the victim is standing at the edge of the cliff—is surely a big cause if big causes there be, despite the minuscule size of this shove. Nor does this killer seem distinguishable from the killer who must use all her might to push the victim over the cliff.
And how do we deal with the cases where the culpable act causes harm by supplying reasons rather than forces?5 Suppose defendant knows that the Bridge of San Luis Rey is in a dangerous condition and will collapse if walked on. Victim wishes to walk to town and can either go there by way of the bridge or by another route. He has reasons to go by way of the bridge—he particularly enjoys the breathtaking view of the chasm. But he has reasons to go by the other route—it takes less time. Suppose he is leaning ever so slightly towards going by the faster route, when defendant calls him up and says he will walk with him to town and will meet him on the other side of the Bridge. That reason—the prospect of defendant’s company for part of the trip—tips the balance of reasons in favor of taking the Bridge, with a fatal result. Surely defendant should be as fully responsible for victim’s death as any killer can be. But exactly how are we to assess the magnitude of the causal contribution of reasons?
Moore admits that the scalarity of causation needs a lot of further development.6 So what we have, perhaps, is not an account of causation’s scalarity but rather a promissory note that such an account will be forthcoming. We shall accept that promissory note for now, but we would not pay face value for it.
The Scalarity of CD
Except in cases of overdetermination, CD will always be present whenever causation is present. So if CD is nonscalar, the scalarity of causation will not matter most of the time. Defendant will be fully responsible for any prohibited result caused by her culpable act, for (excepting overdetermination cases) CD will always co-exist with causation.
Moore believes CD can be scalar (320). He also believes CD responsibility is never as great as causal responsibility (307), though that would mean that, for example, parents who refuse to feed their baby, with the purpose that the baby die, will not be as responsible for the baby’s death as they would be had they shot him. (Moore believes that omissions, which are causally inert and therefore can only be linked to results through CD, are less blameworthy than actions that cause the same results because the former are more easily justified by their consequences than are the latter (ch. 3). We disagree. Given an affirmative duty to act, as with parents vis-a-vis their children, the same consequences that do and do not justify violation of the duty as are applicable to negative duties are equally applicable to that affirmative duty.)
“[L]ess necessary’ means the relevant counterfactual holds true only in possible worlds closer to the actual world…” Something x being more strongly necessary to something else y, means that the counterfactual, “If no x, no y,” holds true in possible worlds more removed from the actual world; something being less necessary, means the counterfactual holds true only in possible worlds closer to the actual world. As long as the Lewisian similarity metric that I explore in the book is not completely bankrupt, this quality-space notion of degrees of necessity seems to me to make excellent sense. (Moore forthcoming).7
We find this analysis curious. Moore never commits to modal realism (385)—in the book, he mentions that the Lewisian similarity metric is “problematic” (389), noting there is “a real question whether the notion is up to the work assigned to it” (387). We sincerely doubt that Moore subscribes to the lush ontology of modal realism. Rather, we take it, Moore uses possible worlds analysis as a way of speaking about his mode of analysis.
We find Moore’s ambivalence troubling. If Moore’s project is to show the relevance of metaphysics to the law’s causation questions, and if in so doing, Moore condemns the harm within the risk test, as he has previously condemned foreseeability, because both tests lack the requisite conceptual rigor, then it is a bit curious to use possible worlds analysis as just a way at getting at his intuitions without any actual metaphysical grounding. We are simply being asked to swap the layman’s terms for Moore’s philosophically sophisticated ones—but the lack of grounding remains. If there aren’t possible worlds, then what does Moore mean by using this language?
Moreover, if one takes the metaphysical commitment to possible worlds seriously, then the analysis does not yield Moore’s desired result. For example, the motorman’s negligent speeding in Berry v. Borough of Sugar Notch, on which the tree’s falling on the motorman was causally dependent, was, for Moore, a tiny degree of CD. For, says Moore, “only in possible worlds very close to the actual world does the tree fall on the motorman. A little more speed, a little less speed, and he would not have been hit” (Moore forthcoming).
As one of us argued, the same is true of myriad cases where we believe defendant should be fully responsible for the result. The speeding defendant who hits a child who runs out from behind a parked car can say the same thing: a little more speed, or a little less speed, and the child would not have been hit (Alexander forthcoming).
Moreover, why are the nearest possible worlds those in which we hold everything constant except the motorman’s speed? Change the color of the motorman’s shirt and he still gets hit by the tree. Change the number of birds singing and he still gets hit. Change his speed a tiny bit, his time of departure a tiny bit, and he still gets hit. Indeed, one can change an infinite number of things without altering the result. If there are an infinite number of nearby possible worlds in which the same result occurs, then even if there are an infinite number in which it does not, infinity over infinity does not lead to a “small CD” conclusion.
And consider the parents who starve their child (death by omission). Suppose the child would not have died had Aunt Emma arrived in time with the food she planned to give the child. And why did she not arrive in time? Because she was riding on the trolley in Sugar Notch and got hit by a tree. Does that make the parents “small CD” responsible for the child’s death? If there is small CD in Berry, there is small CD here. But that verdict would be perverse.
The Scalarity of Fit
Moore believes that to be fully responsible for a prohibited result, one needs not only to have acted culpably and to have fully caused the result by one’s act; in addition, the way the result was caused by the act must have been to a large extent contemplated by the actor at the time of the act. There is room for sloppiness here, says Moore; the match of the plan as contemplated and how it actually unfolds need not be perfect, just close enough for government work (45–46). “How much of the causal route must be within the contemplation of the actor when he acts is a vague matter in morality, but this matches the vagueness in the idea of coincidence and abnormality.” (244 n. 92).
In the tale of Pierre and Monique that is in our book, there are two versions (Alexander and Ferzan 2009). In both, Pierre intends to bring about Monique’s death by shooting a gun in her direction. In the first version, Pierre intends to kill Monique with a bullet, but his aim is off and the bullet misses. However, Monique, fearing for her life, runs out into a thunderstorm and is struck and killed by a bolt of lightning. In the second version, everything is exactly like the first version, except that Pierre plans to miss with the bullet but scare her into running outside and risk getting struck by lightning. In that version, in which Monique again dies by being struck by lightning, Pierre’s killing occurs exactly as he had planned, unlike the first Pierre’s killing.
Moore believes the second Pierre is more responsible for Monique’s death than the first Pierre, and that is because the second Pierre’s killing occurred the way he planned it whereas the first Pierre’s did not. One of us also introduced a possible third Pierre, who had the first Pierre’s general plan to kill with a bullet but who also contemplated the possibility that if the bullet missed, Monique might still be killed by running out into the storm (Alexander forthcoming). Moore does not tell us how that Pierre should be treated—like Pierre 1 or like Pierre 2? Moreover, if this is a culpability question, is it that Pierre 3 did not purposefully kill Monique? If an actor hits a target he intends to but not quite exactly where he intended, are there fine-grained discounts to be made? Specifically, does Moore intend to add to the Model Penal Code’s “element analysis” a mens rea with respect to causal route, so that an actor who intends a harm through method 1, consciously disregards method 2, and is unreasonably unaware of method 3’s possible occurrence will be held guilty of murder, manslaughter, or negligent homicide depending upon how the harm occurs? Moore never quite tells us how “fit” will fit.
What makes this inquiry even more perplexing is that quite often the actor does not intend the harm and thus has no particular mechanism in mind. Consider, for example, People v. Acosta, in which the defendant led the police on a forty-eight mile high speed chase through Orange County, California. Two police helicopters pursued Acosta, and one helicopter moved in a way contrary to FAA regulations, leading to a collision between the two helicopters and the deaths of three police officers. Interestingly, the court upheld the jury’s finding of proximate causation because the result was not “highly extraordinary” but found no malice because “[a]lthough he stated he knew his conduct was dangerous ‘to the bone,’ nothing connected the statement to the aerial surveillance. His knowledge that there were helicopters involved in the pursuit does not suffice. In the absence of more evidence, no reasonable juror could find a conscious disregard for a risk which is barely objectively cognizable.”
How would Moore have us determine “fit”? What if Acosta has the fleeting thought “those helicopters might crash”? Would that be sufficient for fit? What if he just thinks “hey, this is dangerous to the bone and someone might die?” Is Acosta more culpable and more responsible for the deaths if, while speeding along the highways, he contemplates every way his flight could possibly cause deaths rather than only a few such ways?
We believe Moore’s “fit” analysis of culpability cannot be applied to recklessness or to its limiting case of knowledge. The reckless actor may realize that his reckless act risks a variety of harms in a variety of ways with varying probabilities of those ways producing those harms. It is the entire basket of risks of harms that renders the actor culpable, not any particular fine-grained harm produced by a particular fine-grained causal sequence. So when a reckless act results in any particular harm through any particular causal sequence, that harm and causal sequence were part of what made the generative act reckless. All resulting harms, however caused, “fit.” Fit, if applicable, rules out nothing. It does not work for knowledge and recklessness because there is no plan in the sense of a purpose to cause the harm. Without a plan to cause the harm in a particular way, there is nothing to govern the “fit” relationship between the culpable mental state and the actual causal mechanism that produces the resulting harm. The reckless (or knowing) act is culpable because the actor knows it is too risky, but its riskiness is the sum of all the ways the act might cause various harms.
Indeed, once Moore moves beyond purpose and contrivance, it is doubtful that his “fit” analysis can do the work he assigns to it. Notice that what Moore wishes to do is to recategorize proximate causation questions as culpability questions (100, 160). He states, “I share Hart’s and Honore’s (and the case law’s) intuition of non-responsibility, but I do not think this conclusion can be defended on causal grounds. Rather, it is a culpability discrimination doing the work of most of these cases.” (277). This move frees him from metaphysics, and puts us just in the world of moral questions—when is the fit not sufficiently tight so as to hold the defendant responsible for the harm he causes? (Or, as Moore would have it: which results reflect on his agency more directly and thus have more bearing on what he deserves?) But notice that Moore in adopting intuitions is adopting, well, intuitions, and he cannot tame the culpability beast he has created with the rigor with which he destroyed foreseeability and harm within the risk.
As Hart and Honore catalog, there seem to be a number of factors that go into proximate causation coincidence judgments (124–125; Hart and Honore 1985). The event has to be statistically unlikely, itself a cause, independent of the defendant’s act, uncontrived by the defendant, and not pre-existing when the defendant acts. And, among these factors Pierre—a case of contrivance—is actually an easy case for Moore. Contrivance crucially depends upon the actor’s subjective mental state because the actor intends what would otherwise be a bizarre conjunction of events. But the other factors by which the common man performs this matching function are not subjective. For instance, consider independence. Alex decides he wants to kill Bob by smashing his head with a bat. He hits Bob, but what happens is that Bob is knocked into a tree, and this causes the tree to land of Bob, which is what kills him instantly. Now, nowhere is it within Alex’s contemplation to kill Bob this way, but, Hart and Honore tell us, because the cause is not independent of Alex—Alex causes the tree to fall on Bob by knocking him into it, our intuitions hold firm that Alex caused Bob’s death. Moore cannot account for this. Nor can he account for the requirement that the coincidence be statistically unlikely. Assume Carl consciously disregards the risk that he will burn down the forest by throwing a cigarette down. However, the risk that he contemplates is bizarre. He imagines the cigarette’s being extinguished, but its embers being caught in the tail of a wild animal, which will then distribute them at a camp site that will catch fire. When an ordinary evening breeze comes by and causes the fire, the fire will not be caused the way that Carl contemplated it. Will he get a discount because he never contemplated this ordinary means by which cigarettes cause fires? We doubt it. No matter how outside the range of the actor’s contemplation, if the event occurs in the usual way, no one’s intuitions discount Carl’s responsibility. That is, it seems that independence and statistical likelihood are less susceptible to being reframed as culpability judgments, as opposed to causal ones.
Our concern with respect to taming proximate causation intuitions proceeds further. We concur with Moore with respect to his analysis of harm within the risk—all harms are within the risk. But Moore cannot deny (anymore than we can) the intuition that some harms just seem, well, outside that risk. The foot broken because the ten pound can of rat poison is left by the stove and falls on a child’s foot just seems to be the wrong sort of harm to be associated with the risk of having rat poison by a stove. The question for Moore is whether he can let unprincipled intuitions run rampant with respect to subjective mental states and then fully contain them when we reach negligence liability.8 That is, even if harm-within-the-risk is both conceptually and normatively problematic when applied to negligence (and we agree with this claim), and even if purpose, knowledge, and recklessness are not plagued by these same difficulties because they are anchored in the defendant’s subjective description of the harm (209–210) (another claim we do not dispute), we worry that Moore, having vindicated the layman’s causal judgments under the cloak of culpability, will find it hard to contain them.
Morality and the Nature of Deontological Constraints
Moreover, causation “is the big dog in the pack of four.” (77).
[w]hen we act (rather than omit) because we intend to achieve something bad (rather than foresee or risk it), and when our act strongly causes that bad state of affairs to come about (rather than not cause it at all, or only weakly cause it), and when that bad state of affairs strongly depended for its existence upon the act done (rather than weakly so depended, either because that state of affairs would have occurred slightly earlier or later anyway in possible worlds very close to the actual world); then we may not justify what we do by its good consequences. (76–77).
In making this claim, Moore rejects a different conception of deontological constraints, namely, the view that what we cannot do is appropriate another person, not that what we cannot do is intend to cause harm to them. Moore has since explicitly rejected this possibility, arguing that it is either “redundant” or “wrong” (Moore forthcoming). He states, “If ‘using’ means no more than, “acts strongly causing an intended harm,” then it adds nothing to the act-intent-cause factors already in play in my own analysis in chapter 3” (Moore forthcoming). He then claims it is “wrong” if it the physical consumption of the victim in the lifeboat cases is barred by the constraint.
Let us first take up why “using” does not lead to the wrong result. Killing Richard Parker, of Regina v. Dudley and Stephens fame, to use his flesh as food violates the “using” constraint, even if, contrary to Moore, he was “almost dead” (66). (The lifeboat survivors may have been “excused” for this act, but given our version of the deontological constraint, they were not justified (Alexander and Ferzan 2009)). The facts that Parker was expected to die soon and that his flesh could keep the others alive for awhile surely go a long way towards making the cannibalism excusable; however, they do not justify it any more than would his patient’s being very old justify a surgeon’s harvesting the patient’s organs to save others. (We discuss Surgeon below.)
Second, using is not redundant, as it does not take into account the actor’s intent in determining what is permissible. In our book, we argue for the following propositions: First, if a risky action (or omission) requires a justification—and most risky omissions do not—then whether the act (or omission) is justifiable turns on whether the actual good consequences of the action (or omission) outweigh the actual bad consequences of it, with the proviso that the otherwise justifying good consequences must not be produced by using a person’s body, labor, or talents without his consent. Second, whether the actor is culpable turns on the actor’s beliefs about the justifying or de-justifying facts. Third, the actor’s purposes in acting, as opposed to his beliefs about justifying and de-justifying facts, cannot by themselves render his action culpable (as opposed to criticizable)—although if he is culpable, his motivations can bear on how culpable he is (Alexander and Ferzan 2009).
Take two stock pairs of cases contrasting justified action and unjustified action—Trolley/Fat Man and Strategic Bomber/Terrorist Bomber:
Trolley: D is at the switch. He is aware that there are five trapped workers on the main track and one trapped worker on the siding. He knows that turning the trolley is a justified act even if it kills the one trapped worker. He turns the trolley onto the siding but not in order to save the five. He does so because he hates the one. We say his act is justified even if his motive is base because of the causal path by which the good effects are produced, which does not require that the one worker be or remain on the track. It does not matter for objective justification or for culpability why he turns the trolley so long as he has the relevant beliefs about its consequences—whether he does so in order to save the five, or because he just likes throwing the trolley switch, or because he hates the one.
Fat Man: There is only one trolley track with five trapped workers. There is a Fat Man on the footbridge over the track. D is at a switch that will raise the footbridge and cause Fat Man to fall onto the track. If D does so, Fat Man’s presence on the track will stop the trolley and save the five, but it will also kill Fat Man. Saving the five through a causal sequence that depends upon Fat Man’s body being on the track in the trolley’s path does not justify dropping Fat Man onto the track. D, aware of these facts, pushes the button, which drops Fat Man, whose body stops the trolley, resulting in the five being saved. But D does not push the button in order to do this! He pushes the button because he likes to push buttons, or because he likes to watch the footbridge open. He may even regret that Fat Man is on the footbridge. Nonetheless, his act is not justified. Although he does not act with any intention vis-à-vis Fat Man, what his act will do to Fat Man makes his act unjustified. The fact that it saves a net four lives does not alter this verdict because of the causal path by which this comes about. But its being unjustified is unconnected to D’s intentions, which can be myriad.
Strategic Bomber (SB): You know the set-up. The bombing of the munitions factory will kill the 100 civilians who work there, but it will deprive the enemy of munitions and cause it to surrender. It would end the war even if the 100 civilians were to live. Ending the war is a goal worth the loss of 100 enemy civilians. The bombing is justified. And it is justified even if SB’s reason for bombing the factory is not to end the war but to kill Joe, a civilian employed by that factory.
Terror Bomber (TB): The bombing of the factory will not end the war by depriving the enemy of munitions. It will do so, however, by terrorizing civilians because of the deaths of the 100 civilians who work in the factory. Bombing the factory is unjustified. However, TB bombs the factory, not to terrorize the civilians, but because the likes to watch falling bombs. His bombing is unjustified despite its ending the war. But that is not because of his intention but because of the causal route through which the war is won.
“Use of another as a means” refers to the causal path by which the good effects and bad effects of the act are related. It does not refer to why the actor acted. It is the causal structure, not the actor’s intent, that makes for unjustifiability. In Trolley and SB, do the deaths of the one worker and the civilian workers (the bad effects) cause the otherwise justifying good effects (the saving of the five, the winning of the war)? No, they do not. In Fat Man and TB, do the harm to Fat Man and the deaths of the civilian factory workers (the bad effects) cause the otherwise justifying good effects (the saving of the five, the winning of the war)? Yes, they do. The causal relations between good and bad effects, not the actor’s intentions, are what make an otherwise consequentially justified act unjustified. (Note: For us, because actual results don’t matter, it is what the actors believe about the causal mechanisms they are setting in play, not the truth of those beliefs, that makes them culpable.)
Moore believes that if our causal (“usings”) theory of how bad effects do or do not justify good effects is correct, then we would have to say that of one who does not rescue a trapped worker in the path of the runaway trolley because the trapped worker’s body will prevent the trolley from proceeding on to kill five trapped workers further down the track, that he has violated a deontological constraint by using the one trapped worker (Moore forthcoming). Similarly, he implies that our theory would condemn as a using D’s failure to rescue V so V’s organs can be used to save five dying patients.
We deny that our theory has these implications. If D has no duty to rescue the one trapped worker or V, then his omitting to do so for the stipulated reasons is not a using. He is not basing and need not base his justification for failing to rescue on these good effects—for his failure to rescue requires no justification in the first place. On the other hand, if D did have a duty to rescue the one worker or V—for example, if they were his children or his spouse—then his failure to rescue them would require a justification, and a justification that did not rest on using them.9
Thus, it is not the intent to use that violates the deontological constraint against usings. Rather, whether the deontological constraint is violated turns on the existence of facts that would justify an act (or omission) that requires a justification. Acts that create even the slightest risk of harm require some justifying reasons—facts that, given their probability, suffice to make the risk a reasonable risk to impose. Omissions generally require no justification to avoid being culpable, though they can, of course, be morally criticizable even if they are not culpable. When they do require justification, that justification cannot entail a using.
Finally, when an act is justifiable, that justification is not defeated by an intention to use. Thus, in Trolley, if diverting the trolley will save the five workers on the main track, that alone will justify the diversion. And the justification is not defeated merely because the defendant who diverted the trolley did not care about the five workers and only wanted to kill the one worker in order to harvest his organs for research. Take away the five workers who are saved, and we have an unjustified using, even if the research promises to save several lives. But given the presence of the five workers, the intent to use becomes immaterial.
In Causation and Responsibility, Moore endeavors to discipline the law’s use of causation. The book is rigorous in its analysis, breathtaking in its scope, and masterful in its effort to apply metaphysics to morality and law. Ultimately, we remain unconvinced by Moore’s definition of deontological constraints and by his quantitative analysis of causation. But we are impressed by, and have learned much from, the effort. And for us, that made grappling with the book more than worthwhile.
Antony Duff argues that moral responsibility is strict, so all that is required is causation, but the transition from moral responsibility (being called to answer for one’s act) to moral liability requires culpability. On the other hand, Duff argues that for the criminal law, criminal responsibility requires both causation and culpability (Duff 2007).
This suggests something important about doing this kind of ethics, which is that one cannot do it without simultaneously doing the philosophy of mind, the philosophy of action and events, the metaphysics of causation, and the like. Such I have always thought in seeking to use metaphysics to moral advantage in my lifetime preoccupation with moral responsibility and blameworthiness. (42).
For commentary on this scalarity question, see Rosen (forthcoming).
If I squeeze your neck blocking air from reaching your lungs, I have merely prevented air reaching your lungs. Yet what I had to cause to make the relevant counterfactual true is the state of closure of your wind pipe, a condition in which it is impossible for air to pass through. In such cases I cause a state of affairs to exist (wind pipe closure) that is very close to the type of state of affairs complained of, an absence of air passing to the lungs. Thus, I did not cause the absence; but what I did cause (to be a preventer of air) is so close to the absence of benefit that I may be as morally responsible as one who causes poised air to enter the lungs. (457).
Alex Broadbent voices a similar worry about scalarity in his review of Moore’s book (Broadbent 2011).
See, e.g., 302 for reasons as causes.
Moore admits this in his forthcoming work: “The idea is admittedly underdeveloped in the book…” (Moore forthcoming).
Citing Moore p. 320.
Moore clearly has this goal: “The place to ask and answer these questions of fit is with respect to mental states, not causation. With respect to negligence, which is not a mental state, I shall urge … that we should not ask the fit question at all, anywhere.” (102 n. 69).
If D is the parent of the one worker in Trolley, then, like a stranger, he is permitted not to switch the Trolley from the five to the one. But like the stranger, he is also permitted to switch the Trolley, despite his obligation to his child. That obligation does not cancel out the balance of evils that favors switching from the five to the one any more than it permits D to switch the Trolley from the one—his child—to the five. Those five are, after all, the children of other parents. And if D does switch the trolley from the five to his child, the fact that it will save the five trapped workers is all the justification D requires. Even if D intends to use his child’s organs to save five additional people—that is, to use his child as a means—that does not de-justify his switching the Trolley; for the switching is already justified without the use of the child as a means.
Where D’s parental duties do come into play is in situations such as Trolley in which D can switch the trolley from the track leading to his child to one on which no one is trapped. In such situations, D is obligated to switch, whereas a stranger would be permitted but not obligated to switch. And if D does not switch to the vacant track because he wishes to harvest his child’s organs to save five strangers, his failure to switch is a wrongful using of his child as a means, whereas the stranger’s failure to switch for the same reason would not be a wrongful using.