Print This Post Print This Post

Recall the big picture, which can all too easily pass us by. Kamm is exploring the part of ethics that deals with harming others. Her perspective is non-consequentialist: she is interested in permissions and constraints on harming that go beyond the good and bad effects of our actions. In layman’s terms, she is pursuing rules or principles that might block certain actions regardless of the consequences. (Connoisseurs should bear in mind the codicil about threshold-deontology, though.)

The central puzzle here is that sometimes the good ends do seem to justify the evil means, and sometimes they don’t. Diverting a trolley from a track where it will kill 5 to one where it will kill 1 seems ok to many, but chopping up an innocent person to transplant his organs into 5 who will otherwise die seems wrong.

In earlier chapters Kamm has focused on some traditional non-consequentialist explanations of trolley intuitions, particularly the thought that perhaps the distinction between intending and merely foreseeing harm to innocents in the pursuit of the greater good is what licenses such departures from the general strictures on harm. But, having criticized these approaches and pursued some variations of her own (the because/in-order-to distinction), in chapter 5 Kamm takes a different approach. Instead of explaining permissible harm in terms of mental states of the agent (what he intends, foresees, etc), she pursues a theory that focuses on causal structures. Her line of thought is a variation on “downstream” theories—theories saying that we may act to bring about evil for the sake of the greater good, but the evil must be causally downstream from the good.

Section A: Kamm begins in section A (138) with a pure downstream theory—the greater good justifying the lesser evil must itself cause the evil. She motivates it by noting the permissibility of moving the 5 to a place where their presence will cause a landslide killing the 1. A downstream theory could obviously account for this.

B: But this causal structure isn’t necessary. In the standard trolley redirection case, the evil of the 1 dying doesn’t cause the greater good of the 5 being saved. Instead, the trolley’s being on the other track where it kills the 1 is just a part of or aspect of the situation in which it doesn’t strike the 5. The redirection of the threat is the same event as the 5 being saved. Kamm calls the evil here the “non-causal flipside” to the greater good. (Sometimes the relation involves identity, sometimes constitution.) So we can employ means that cause a lesser evil when they have the greater good as their flipside. (Analogy for the perplexed: the existence of a valley isn’t caused by the creation of two nearby mountains ranges; the former is just the flipside of the latter.)

Kamm suggests the flipside notion can explain a great deal. It’s wrong, for instance, to push a fat man off a bridge to stop the trolley from killing 5, but it’s ok, she claims, to spin a lazy susan containing the 5 so that they are saved from the trolley and the 1 is moved into harm’s way. The 1 in the lazy susan moving into harm’s way is just the flipside to the good of the 5 being saved; by contrast, the fat man is a case of causing the evil so that it can in turn cause the greater good—a case of forbidden downstream causation.

C: Next, Kamm discusses cases of parallel causation: the good and the evil are caused in parallel, with neither a flipside of the other. E.g., setting off a bomb that redirects the trolley and (separately) kills the 1. Here she enters into complexities I can only darkly gesture toward, like a second-rate fortune-teller. The cases she distinguishes are extremely subtle. For instance, she considers driving to the hospital to save the 5, where (a) driving causes vibrations that cause rocks to fall and kill the 1, or, in a variation, (b) driving on top of loose rocks that fall and kill the 1. She says that (a) is ok, but that (b) is wrong, since (b) involves our means to the greater good “overlapping” with the cause of the evil.

To be fair, Kamm reaches this conclusion only after presenting several other cases and discussing overlap in more detail, but the cases at this point really do require a fine palate to distinguish. From this and other cases she argues that the significance of causal influence is in part determined by whether we introduce into a given context the cause of evil, and whether our actions involve overlap with the cause of evil. She also suggests that rather than its being crucial that evil be downstream, it’s crucial that evil not be upstream.

D-E consider further refinements and qualifications of the downstream-oriented theory Kamm presents. In particular, Kamm emphasizes the notion of sustaining the good achieved or made possible by our actions. In the loop case—where the trolley is redirected but will loop back to kill the 5 unless it strikes the 1 on the loop—after the redirection, we have only created what Kamm calls the “structural equivalent of the good,” meaning a state that would be the greater good if there were no further threats (such as the trolley looping back). If the structural equivalent cannot be sustained, then redirecting is wrong, Kamm thinks, even if doing so generates a greater good. Thus, if the redirected trolley will depress a button diverting another trolley from running over 3 innocents, we still cannot redirect if the original trolley will loop back to kill the 5.

The loop case also brings out that evil may sustain the greater good even though it seems generally wrong for it to produce the greater good. It does this by defeating a defeater—by preventing the trolley from looping back round and thus preventing the structural equivalent from turning into the greater good itself.

With these observations in place—observations that no doubt we’ve all entertained from time to time—Kamm summarizes her views up to this point (159):

“Lesser evils* to innocent, nonthreatening people who would not face comparable threats, for whom continuing life is a good; and who have not consented to the evils* should be causable given our act either (a) by what (at least) initially sufficiently justifies these evils*…or (b) by means (or effects of means) that have (at least) the structural equivalent of the greater good as their noncausal flip side or aspect. The motto here is: ‘We must have the possibility of at least initial sufficient justification all the way down for evils* that cause, or that are directly caused by what causes, the greater good.”

(Initial justification, I should add, means that an act is justified in so far as the structural equivalent offers a prima facie justification for the lesser evil, albeit one that could still be undermined by the introduction of novel threats.) This is the Doctrine of Initial Justification.

F: Kamm immediately points out that the Doctrine isn’t necessary for an act to be permitted, however. For certain evil side effects need not be initially justified. Suppose the trolley is diverted so as to save one of the 5 and (a) kill two bystanders whose being hit doesn’t produce any further good, and (b) hit someone’s leg so as to stop the trolley and save the other 4. Diverting is permissible, Kamm says, even though the side-effect of killing the 2 isn’t initially justified since that doesn’t produce any further good that would render the act right absent defeaters. Kamm comments that what seems to be at work here is a greater good in the process of working itself out which can justify the otherwise unjustifiable side-effects. At least one component of the greater good—saving the 1—justifies hitting the man’s leg, and that in turn is a component of the greater good of saving the 5 that does justify killing the 2. (I must confess to engaging in some kripkography here.)

Kamm concludes the core of the chapter with the Doctrine of Productive Purity (164). This paragraph length doctrine lays out two conditions. Very roughly (ignoring modal conditions, sustaining sub-sections, etc), they are that (1) if an evil cannot be sufficiently justified initially, it cannot be justified by the greater good it is necessary to produce, though the evil can be justified by the greater good whose components cause it; and (2) for an act to be permissible, evil side effects or evil means we employ must be the effects of a greater good in the process of working itself out.

Here Kamm pauses to reflect on the deep nature of inviolability this doctrine and the other principles she has explored embody. Kamm suggests that it is related to the difference between substitution (as discussed in earlier chapters) and subordination. When the few are sacrificed to save the many, the few don’t have a complaint, since their loss is matched by the loss to the many that is averted. This may sound consequentialist, and it may prompt the transplant case once more, but Kamm claims that what’s objectionable in the transplant case is that evil there is a causal means to the greater good, which she takes to mean that the victim is subordinated to the good of others.

She adds that when we send a threat to one person rather than another, one person occupies the very same position another would have anyway. This is substitution. On the other hand, in harming someone as a causal means to saving another, the first person occupies a different position than the second would have had the threat been realized. Here the victim is subordinated since his position makes essential reference to his usefulness in producing the other person’s good.

These are the central claims of the chapter, so I will set aside the concluding postscript that follows, and turn to questions and issues for discussion.

(1) Kamm aspires to giving a deep, searching account of the morality of harm. But the insistence on intricate case-studies seems especially odd in this chapter, even if we grant all the case-intuitions are correct. Take one of her fundamental claims—that the morality of harm (at least large tracts of it) can be developed without reference to the mental states of agents. Clearly this claim that morality is fundamentally about the causal structures of our actions and not our mental states is important, but there is no sustained discussion of it apart from specific case-studies. What we’d expect would be a discussion of such issues as unrealized threats (e.g., failed assassination attempts) or even more general topics like what the point of moral criticism of others is. What are the prospects, I wonder, of getting to the bottom of the morality of harm just by case-studies while ignoring all such general issues?

(2) Empirical evidence indicates that people’s intuitions about intention and causation are heavily influenced by their moral beliefs. (Joshua Knobe’s website details much of this work.) In many situations people are more likely to attribute intention and causal influence to people when they judge their actions to be wrong. If this is so, how good an idea is it to make moral status depend on causal influence (or agent intentions)?

(3) Consequentialists lurk in the shadows (trenchcoat collars turned up) waiting to debunk case-intuitions with evolutionary and other explanations. The more disconnected the intricate principles of a normative theory are from things we reflectively endorse as important, the greater the threat of debunking. If reflection on cases reveals that most of us judge acts to be wrong because of some subtle interplay between the ordering of causes, side-effects and overlappings, it isn’t clear to me why this would have any normative authority for us, rather than being a candidate for debunking. The theory that we should be nice to people may ultimately go back to impulses related to kin-altruism, but that doesn’t seem very threatening, since we’re inclined to endorse being nice anyhow. The intricate theory of causes and overlappings seems more worrisome since it’s often obscure why any of that should matter, since it doesn’t seem to connect to the things we care about. One answer, of course, is to proceed to connect the intricate causal structures and overlappings to deep and fundamental principles that we do recognize and endorse. The brief discussion of substitution and subordination is Kamm’s attempt at this. It would be interesting to know how persuasive others found that attempt.


  1. 1. Posted by Jeff Huggins | August 4, 2007 2:50 am

    Thanks, Dan, for the helpful summary and thought-provoking questions. Here are my responses/thoughts to the questions as you numbered them, plus a few other observations:

    (1). You ask, “What are the prospects, I wonder, of getting to the bottom of the morality of harm just by case-studies while ignoring all such general issues?” And, by “general issues”, I assume you mean some of the issues you mentioned (e.g., the role of the mental states of agents, if any; and the point of moral criticism of others, if any) as well as other, perhaps even more general, issues.

    First, I think there are different issues at play, and Kamm seems to risk conflating them. Purposefully not using the word “morality” here, for clarity’s sake, it is true that in at least one very important sense, only outcomes matter. For example, if we humans somehow manage to kill ourselves (as a species), whether we do so through a combination of bad intentions and negligence, or through good intentions and negligence, it will not make much difference after we are all gone. The very point of human “morality” will be defeated in either case, and nobody will be around to even speak the word. On the other hand, in other senses (very real ones, I would argue), intentions do matter a lot. First of all, although intentions (or motivations) do not always result in actions that are consistent with those intentions (or motivations), and there are exceptions in both directions, on average, in many types of situations, the two are tied (at least loosely) together. In other words, consider how the subject of morality might be taught in the future: If it is taught that intentions do not matter, Wow!, watch out. If that view were ever fully taught and accepted, by a large population, we might well have all sorts of people trying to intricately justify modest harms that they do (strongly) intend by simply pairing them to an arguably “greater good” of some sort. As long as the modest evil does not lead to the good, but rather accompanies or follows it, then all is well, I guess?

    Of course, the justice system (as imperfect as it is) differentiates between intended harm, gross negligence, accident, and so forth. So, to me, the answer to the question, Do intentions matter?, depends on what aspect/dimension of morality/ethics one is talking about: Teaching it, applying it (personally), modeling it, and etc. Scientists understand (to a degree) the role of intentions/motivations in terms of how we humans try to “read” each other, who we choose to trust (or not trust), and so forth. In fact, there is a great book, Robert Axelrod’s “The Evolution Of Cooperation”, and it would be very difficult for us humans to navigate the types of cooperative dynamics mentioned in that book if we didn’t try to “read”, or if we didn’t care about, each others’ intentions. In fact, it seems to me that one of the points of human communication, and the degree we value honesty, is that we can discuss plans and intentions when deciding who to trust, who to date, and so forth.

    On a broader level, I don’t think one can get to “the bottom of the morality of harm” without considering several key (and general) factors along with the case-studies. For example, the “effective” function of human social-moral dynamics. And, the origins of human social-moral dynamics. And, the (substantial) imperfections of the human being. And so forth. These create a larger (and more grounded) context within which human intuitions can be understood and explained, and with which sensible principles can be derived, in my view.

    (2). I think I already expressed some views on this question under (1), above.

    (3). You write, “The more disconnected the intricate principles of a normative theory are from things we reflectively endorse as important, the greater the threat of debunking.” I agree. You also write, “One answer, of course, is to proceed to connect the intricate causal structures and overlappings to deep and fundamental principles that we do recognize and endorse.” And then you ask, “It would be interesting to know how persuasive others found that [i.e., Kamm’s] attempt [i.e., to do that].”

    My answer is this: So far, I find the material very interesting, thought-provoking, and worthwhile. On the other hand, based on my “gut feel” and on my own understanding of morality, I feel that the more intricate principles are not accurate in mapping how humans actually feel, think, and work. The more intricate the cases, and the more intricate the thinking, and the more intricate the principles or doctrines, the farther removed they seem to be from how humans work and even from the “effective” function of morality itself, it seems to me. This is not to say that the more intricate principles won’t provide as “output” right moral decisions in some, many, or even most situations. They may well do so, or at least they shouldn’t be too far off. But, if they do so, they probably don’t do so in the same ways that most human beings do so. And, (and this is important in my view), if the intricate principles are primarily based in intuitions alone, and if they are not solidly grounded in a broader understanding of humans and of the “effective” function of morality itself, and in an understanding of human social-moral dynamics (e.g., dynamics of cooperation as discussed in Axelrod and later works), then they may lead in incorrect and unwise directions, in some cases.

    Since we are discussing intuitions, it helps to note the status of the (unscientific, but still revealing) “Kamm Poll” regarding the Munitions Grief Case. As I write this post, of 38 responses so far, 15 responded “yes”, 13 “no”, and 10 “not sure.” This shows that intuitive responses can be “all over the map”, and far from conclusive, especially in highly intricate cases. This variation reflects many aspects of human-ness, of course.

    This post has been long enough already (sorry!), so if I add more thoughts on this particular subject, I’ll add them in a future post.

  2. 2. Posted by David Wasserman | August 6, 2007 1:34 pm

    Dan, I think you’ve done a truly admirable job of distilling the essence of a chapter that’s dense and convoluted even by Kamm’s standards. I think, but I’m not sure; I’ve only been able to read about half the chapter and have may have missed dozens of key cases and distinctions. But the temporal precedence of good over evil does seem, despite the many qualifications, to play a critical role, particularly in the claim that one can do more evil to sustain than to produce the greater good. This emphasis is apparent in the early contrast of Loop and Tractor (136-38): In Tractor, the five face an overdetermined death, with a trolley coming at them from one direction, a tractor from another; a non-looping diversion will save the five from the tractor as well as the trolley, but only by placing one man in the tractor’s path. Kamm finds the diversion in Tractor impermissible. She argues that DTE can’t yield this judgment: although the two threats are more distinct in Tractor than Loop, she claims that the agent in Tractor still acts only because of the harm to the one, not in order to bring it about. But despite permission from DTE, the diversion in Tractor is not OK because (roughly) the evil of the one’s death is caused by producing the greater good of the five’s rescue. The diversion in Loop is OK because the evil of the one’s death is caused by sustaining the greater good achieved by the diversion, against the new threat posed by the looping. In contrast, the threat from the tractor pre-existed the diversion.

    Assuming that one can clearly and consistently distinguish pre-existing and newly-created threats, and producing vs. sustaining the greater good of protecting people from those threats, (an assumption I make only because I don’t have the time or energy to challenge it), the obvious question is why the distinction has any moral import. I suspect that whatever intuitive appeal there is in the producing/sustaining distinction arises from the conviction that you may do more harm to hang on to what you’ve got than to acquire something equally good. But that conviction just seems to reflect the “endowment effect” that utilitarians and their psychologist allies have sought to expose as questionable source of deontic constraints. This seems to be the just the sort of intuition one should distrust: once it’s stripped of spurious association with bona fide endowment rights and interests (e.g., from property rights and deep attachments to familiar and cherished people and objects), the producing/ sustaining distinction has no apparent link with other strongly-held moral convictions, and has a plausible psychological explanation based on features even more clearly devoid of moral content. (The latter alone would not discredit a moral intuition – that’s a genetic fallacy – it’s the absence of a good moral explanation coupled with the presence of a plausible psychological one).

  3. 3. Posted by S. Matthew Liao | August 7, 2007 5:14 pm

    Dan, echoing David, I too think that you’ve done a fantastic job in getting to the heart the matter of this difficult chapter.

    As you have noted, one of Kamm’s central claims here is that morality is fundamentally about the causal structures of our actions and not our mental states. And I fully agree with you that it would be good if Kamm considered issues of unrealized threats, the point of moral criticism, etc. In fact, I think it is precisely in these areas that one can test and put pressure on Kamm’s enterprise.

    In my view, a plausible account of nonconsequentialism (NC) holds that the goodness or the badness of the state of affairs is necessary for determining the rightness or wrongness of our conduct but it is not always sufficient, that is, state-of-mind considerations are sometimes also necessary for this task. This implies that the goodness or the badness of the state of affairs may sometimes be sufficient for such a task. And this does not imply that state-of-mind considerations are always sufficient for determining the rightness or wrongness of our conduct.

    Seen in this light, first, some of Kamm’s arguments against DDE and DDE(R) appear to miss their mark (I leave out DTE, because I’m not convinced yet that this doctrine is relevant in the moral context. Also, to keep things simple, from now on, I shall refer to DDE* when I mean DDE and DDE(R)). For example, Kamm argues that DDE* cannot explain

    The Splash Case. A doctor can save 5 patients if he drives to the hospital. But on his way driving to the hospital, he can see that there is a puddle of deadly acid spilled on the road. If he drives through the puddle to get to the hospital, he knows that this will cause the acid to splash on and kill 1 immovable bystander at the side of the road.

    According to Kamm, it is impermissible to try to save 5 people in this manner, but on DDE*, this act seems permissible, given that the lesser harm is merely foreseen and not intended.

    To start, it is not clear that this act would be permissible on DDE*. The reason is that DDE* may not only require that the lesser harm be merely foreseen, it may also require that one intends not to cause the lesser harm. This means, for example, that the doctor may fail DDE* if the doctor did not try not to cause the lesser harm, e.g. by slowing down so that the puddle would not splash.

    More importantly, even if DDE* does not rule out the Splash Case, as I have noted, a plausible account of NC does not require that DDE* is always sufficient for determining the rightness or wrongness of our conduct. If this is right, then the Splash Case does not undermine DDE*, given that DDE* does not claim to be sufficient for determining the rightness or wrongness of our conduct.

    Secondly, since the goodness or the badness of the state of affairs can sometimes be sufficient for determining the rightness or wrongness of our conduct, it is not enough for Kamm to show that in the cases she has presented, morality is fundamentally about the causal structures of our actions and not our mental states. That is, even if DPP were true in the cases she has presented, Kamm needs to show that DPP is true of all cases. Moreover, this also means that to call into question Kamm’s enterprise in this respect, one needs to provide just one counterexample against her claim. I think the following case from Kamm herself would put pressure on her claim:

    Bad Man Case II. A bad person sees the trolley headed toward the give. He has no interest in saving the five per se, but he knows that it is his enemy who will be the one person killed if he redirects the trolley. Suppose he would have redirected the trolley toward his enemy even if (counterfactually) the five had not been there (p. 135).

    As Kamm also observes, the bad person’s behavior does not satisfy DDE* even when the five are there, as he does not redirect because the five will be saved (p. 135). Yet, his action would be permissible on DPP, given that the causal structure here would be the same for a bad person as for a non-bad-person. Indeed, Kamm says regarding this case that “contrary to what [DDE*] implies . . . his act of redirecting the trolley when the five are there is still permissible” (p. 135). But I think a defender of DDE* would disagree and would argue that it is precisely in such a case that the state-of-mind is relevant to assessing the permissibility of one’s act. As far as I can see, nothing Kamm has said in this chapter shows that the defender of DDE* is wrong about this particular case.

  4. 4. Posted by Jeff Huggins | August 7, 2007 7:38 pm

    The question of whether morality is only about actions and causal structures (and outcomes), OR also about mental states and motivations, is an either/or question whose answer depends, I believe, on what aspect of morality is under consideration, and for what purpose. In some senses, the question itself can be misleading (especially if it is meant to apply to all aspects of morality, the subject) and, if accepted, can lead to people ignoring the “loosely close” (albeit imperfect) directional human relationship between motivations and actions.

    When asking the question of whether morality is only about causal structures, or also about mental states and motivations (at least in some cases), what aspect of the subject of “morality” is one addressing?: Whether the outcomes themselves (of a particular action) are better than they otherwise would have been? Whether the outcomes justify the efforts as well as any resulting harms? Whether we approve of the motivations that led (in one way or another) to the outcomes? Whether we would gently reprimand someone, or punish someone, for his actions and/or motivations, or whether we would celebrate them? Whether we would teach students and young ones that motivations and mental states don’t matter OR that they do (in certain ways and in certain instances)? I don’t think that a blanket statement about morality can be made, on this particular dimension anyhow, that is meant to cover all of these aspects of morality.

    IF for some reason we force ourselves to come up with an either/or answer to the question, “What one thing ultimately matters most?”, the answer would (I believe) have to be, outcomes, rather than mental states and motivations, IF we force this choice. This answer can be demonstrated via simple cases and also follows from the basic “function” of human social-moral dynamics. That said, much of the time when morality/ethics are being discussed, the discussion encompasses more than one aspect of morality, so (as long as we don’t artificially force an either/or answer) both outcomes and motivations are relevant in various ways.

    If someone’s view of morality argues that mental states and motivations are never relevant, and that only outcomes and causal structures matter, then that should be a sign (it seems to me) that there is a fault in the argument, unless that person has defined the question itself consistent with that view. In other words, if someone is essentially asking the question, “Are the outcomes themselves (of a particular sequence of actions) better than they otherwise would have been, and do we approve of the sequence of actions itself?”, then the answer that “motivations and mental states don’t matter” (i.e., that only actions and sequences and outcomes matter) is understandable and perhaps correct by definition alone.

    One reason (probably among several) that the either/or question of what matters (i.e., EITHER causal structures alone, OR causal-structures-and-mental-states-and-motivations) cannot apply to all aspects of morality is that morality isn’t “about” one OR the other in all its aspects. Human social-moral dynamics are, most foundationally, human survival mechanisms (in the sense of ongoing human survival from generation to generation; not in the sense of the perpetual survival of individuals). And, because human behaviors and human mental states and motivations are (loosely and imperfectly) linked (directionally and on average), questions of whether we survive, and how well, and how, and how we might do better at it, cannot divorce the two (i.e., behaviors and motivations), retain one, and reject the other.

Post a comment

Name: (required)

Email Address: (required) (will not be published)



(Spamcheck Enabled)

This work is licensed under a Attribution-NonCommercial-NoDerivs 3.0.