THE TROLLEY PROBLEM AND AGGRESSION

Abstract:

This essay considers complications introduced by the Trolley Problem to the discussion of whether and when harming some for the sake of helping others would be unjustified. It first examines Guido Pincione’s arguments for the conclusion that the permissibility of a bystander turning a runaway trolley from killing five people toward killing one other person instead may undermine one moral argument for political libertarianism and against redistributive taxation, namely that we may not harm some people in order to help others to a greater degree. It then considers both the bearing on Pincione’s argument of recent objections to the permissibility of turning the trolley, as well as the soundness of the objections. Finally, the essay considers the relevance of trolley cases for developing a theory of aggression, insofar as aggression is the unjustified use of force that is either foreseen or intended.

In his article “The Trolley Problem as a Problem for Libertarians,” ¹ Guido Pincione has suggested that the permissibility of a bystander turning a runaway trolley from killing five people toward killing one other person instead (when all other things are equal between the people) may undermine one moral argument for political libertarianism, namely that we may not harm some people in order to help others to a greater degree. In this essay, I will first consider Pincione’s discussion on its own and in the light of recent work on the Trolley Problem. In the rest of the essay, I will consider the implications of the Trolley Problem for the broader issue of whether and when someone’s action counts as aggression. For purposes of this essay, I shall understand “aggression” to be unjustified and impermissible use of force against others either when (i) force is intended as a means or end, or when (ii) force is caused as a side effect of a deliberate act with foresight to force occurring. Hence, I shall use “aggression” like “murder,” insofar as it conceptually implies wrong conduct. ² I believe it is consistent with libertarian theory to speak of impermissible force as “aggression.” I do not assume that all violations of negative rights involve force. ³

I. A Proposal for Turning the Trolley

Pincione accepts that it is permissible for a bystander — not merely the driver of the trolley — to turn it from striking five victims to striking one other victim, but rejects the view that it is permissible for the bystander to topple a fat man from a bridge so that his body will stop the trolley from hitting the five. Hence, the former would not involve aggression (though it involves force) but the latter would. Explaining why it is permissible to turn the trolley in the Bystander Case but not to topple the fat man in the Fat Man Case is what Judith Thomson called the “Trolley Problem” in one of her articles. ⁴ It is not true, however, that these are the cases to which she first applied the term “Trolley Problem,” contrary to what Pincione says. ⁵ She first introduced the term in referring to the contrast between the trolley driver permissibly turning his trolley from five people to one other person and a doctor impermissibly killing one person to acquire organs with which to save five other patients. ⁶ These were the cases that Philippa Foot had originally discussed involving the driver turning the trolley. ⁷ Let us call the latter contrast TP(1) and the one Pincione discusses TP(2). For Pincione’s purposes, it is important to deal with the Bystander Case because he wishes to analogize the government to a bystander, dealing with problematic situations that are not of its own making.

Pincione does not claim to solve TP(2) but he thinks that the solution to it will roughly have the following two-part form: It will be permissible for the bystander to turn the trolley when (a) it is ex ante in the interest of each person potentially threatened that the trolley be turned because this maximizes the chance of each to survive; and (b) the lesser harm is caused either by the greater good (of more saved) or by means that have the greater good as their noncausal flip side, in the sense that the five saved just is the trolley moving away. Pincione derives the (a) clause from one of Judith Thomson’s discussions of the Trolley Problem. ⁸ On her understanding, the (a) clause is interpreted so that the ex ante point is one in which each person who will eventually be on the tracks does not know (and has no indication) whether he will be on the track with five people or on the track with one person. Using this ex ante perspective is one way of arguing for the view that in a conflict situation when everyone cannot survive, there is a reason to do what saves the greater number of people. This conclusion might directly imply that when we must choose whether to save one person or save five people, we should save the five. But TP(2) involves a choice between killing one and saving five and this raises an additional problem, and clause (b) is meant to help justify a killing. Clause (b), which concerns causal structure, makes use of my Principle of Permissible Harm (PPH). I will discuss it in more detail below. ⁹ Clause (b) assumes that the five surviving would be a greater good relative to the one person on the other track surviving, and the death of this one would be a lesser harm relative to the deaths of the five. But it makes the permissibility of bringing about the greater good and lesser harm depend on how they are brought about given that we will be harming and not merely not saving one person.

Pincione thinks that when conditions (a) and (b) are satisfied, someone’s negative right not to be harmed (in this case by force) does not stand in the way of helping save a greater number of people. Hence, using my notion of aggression, there would be no aggression if the one person were harmed in this way. Since the bystander is typically thought to have a permission (but not a duty) to turn the trolley, if the permission of government “mirrored” this moral landscape, government would also sometimes have a permission to harm some citizens, even by force, in order to help other citizens, which seems contrary to a libertarian view of government. On the other hand, government would not be shown to have a duty to help its citizens by harming others.

II. Analogizing Government Action to Action in Trolley Cases

Even assuming the adequacy of (a) and (b) to explain the permissibility of a bystander turning the trolley, I have several concerns about Pincione’s interesting discussion. (i) First, he deals with the trolley cases that involve killing someone or letting others die, but he draws conclusions about government action such as taxing some to provide financial help for others. It is not clear that conditions like (a) and (b) (modified to apply to non-trolley cases) are required to make helpful action by ordinary bystanders permissible in cases where they do not kill, seriously harm, or use other force on someone to help others. For example, suppose five people would die near a rich man if we did not remove five dollars from his wallet lying next to him, though he does not wish to donate the money. (Call this the Rich Man Case.) I think it would be morally permissible to take the money in this case. Whether permissible or not, in neither case would this involve the use of force, so it is not the possibility of aggression that is at issue.

Presumably, taking is permissible because the rich man has a duty to help that he is not fulfilling, and given the irreversible great harm to those in need, a bystander may impose the costs on him. In these cases, condition (b) is not satisfied since the lesser harm to someone (of losing his money) is a mere causal means to the greater good of five living; the harm to him is not caused by the greater good or by means that have the greater good as a noncausal flip side. Furthermore, in this Rich Man Case, condition (a) might not be satisfied because we would permit taking the money even if we knew it was not ex ante in the interest of each person to have a policy of harming some to help others. This would be so if we knew that the rich man was invulnerable to needing help. If government taxation policy to save lives should mirror ordinary morality, it might be that the analogy to rely on is the Rich Man Case not TP(2), and thus neither condition (a) nor (b) would play a role in justifying such governmental actions. ¹⁰

Of course, there may be other differences between the Rich Man Case and government taxation policy that Pincione discusses, and they interfere with the latter mirroring the former. For example, improving some people’s economic condition is often different from saving their lives, and taxation may be causally useful only to do the former. If it would be wrong for a bystander to take even five dollars from someone without his consent to improve in a certain way five people’s economic condition, ordinary morality might not provide a basis for the government doing so. It might then be necessary to rely on conditions like (a) and (b) to justify causing the lesser harm. In addition, even if some people’s lives were at stake, much more than five dollars may be causally required to help them, so that it is impermissible for a bystander to take that amount. Then the Rich Man Case would also not serve as an analogy for government action. Indeed, it might be that when the rich man would have no duty to initiate help due to the size of the cost to him and/or the lesser importance of the good to be achieved, imposition of the cost by others should only be done in a way that satisfies conditions like (a) and (b). Call these Rich Man (2) cases. Such cases in ordinary morality that do not necessarily involve killing some to save others, but still require the use of conditions like (a) and (b) to justify action as much as permissibly killing in TP(2) does, might be the appropriate analogy for government actions that cannot be analogized to the Rich Man Case.

One remaining difference between Rich Man and Rich Man (2) Cases on the one hand and government taxation on the other is possibly the government’s permission to threaten use of force if that is necessary to acquire money or property needed to help others. Even if the government, like the helpful bystander, never actually uses force, unlike the bystander, it might be entitled to do so. (However, if conditions (a) and (b) could justify killing someone in TP(2), presumably they could justify the use of some sort of force to acquire someone’s money or property, at least if this force itself came about in a way that satisfied conditions like (a) and (b).)

(ii) My second concern about Pincione’s discussion stems from my belief that even if it is permissible to turn the trolley in TP(2) (and also TP(1)), it is still permissible for the person to whom it is directed to send it back from whence it came, even if this means it will then kill the five who were originally threatened or the person who turned the trolley. This may be in part because the five would be no worse off than they would have been had the trolley not been redirected, and because the redirector infringed the rights of the one person not to be harmed even if he did not violate those rights. ¹¹ If the one potential victim does this, he cannot be accused of aggression against the five or the redirector. (The permissibility of his doing this would have to be explained independently of (a), since a principle that allowed the one’s sending the trolley back would not maximize the ex ante chance of each person to live.)

If government and citizen permissions mirrored permissions in TP(2), citizens might sometimes legitimately resist the government’s morally justified attempts to tax them for the good of others. This is a problematic view, I think. The idea that citizens may legitimately resist permissible government actions is usually confined to nonviolent civil disobedience concerning gross injustices, not such economic policies. Furthermore, punishment of such resisters is also thought legitimate while punishment of the one resistor would not be permissible in TP(2) or TP(1), I think.

(iii) A third concern I have about Pincione’s discussion concerns identifying when clause (b) is satisfied. ¹² Clause (b) would exclude toppling the fat man from the bridge. Putting aside the possible moral significance of the difference between killing and imposing economic harm — which was part of my first concern — it would seem that taxation of some is a mere causal means to providing benefits to others, and so should be prohibited by a clause like (b). However, Pincione considers possible alternative descriptions of a case involving taxation. For example, he says that some might say that “ . . . the state can ‘redirect’ a recession from the potential unemployed to individuals whose extra tax burden will be more than offset by unemployment benefits.” ¹³ This description first makes it seem as though the recession is removed as a threat to some and redirected to others. But it isn’t really the recession that is directed; as the next clause makes clear, it is some new threat, a tax burden, that falls on others. Still if a recession being removed from some people causes a new threat to others, a condition like (b) could permit this. (Similarly, it would permit the trolley being turned from the five even if it consequently hits a heretofore stationary wagon, moving it so that this new threat kills one other person.)

However, if the actual causal route to removing the recession from some were to first impose the tax burden on others as a causal means to providing subsidies, a condition like (b) would not be satisfied. By contrast to this being the route, Pincione says that “[i]f the subsidies could be seen as the flip side of (the enactment of) redistributive measures . . . the PPH would authorize such measures. . . . The fact that a subsidy law makes it constitutionally mandatory for legislators who grant a subsidy to tax the rich mirrors here the fact that turning the trolley . . . makes it causally necessary to kill the one.” ¹⁴ This way of constructing the analogy makes the imposition of the taxes come after the subsidies have already been legislated, and it is the subsidies that help the poor. In helpful comments on the present essay, Pincione further said:

Unlike toppling the fat man, the taxes that I imagine in the recession case are causally independent of the greater good, e.g., many unemployed being saved from extreme poverty. It is the subsidies that bring about the greater good. This greater good comes before the taxes. Moreover, taxes are not necessary to bring about the greater good: other measures, such as printing money, can do that. The taxes are, as I imagine, “constitutionally mandatory” given the enactment of the subsidies, and this normative necessity as it were, may be taken to mirror the fact that in the Trolley Case the lesser harm is a causal consequence of turning the switch.

It seems to me that subsidies that are provided (even before taxes are imposed) are still mere causal means to the greater good of people being economically improved. If this is so, and providing subsidies also causes (by way of legal requirement) taxes to be raised on the rich, then a mere causal means to the greater good would have caused the lesser evil. This would make the case analogous to one where the bystander sets a bomb that causes the trolley to move away from the five and the bomb itself causes the death of one other person as a side effect. The PPH would rule out setting the bomb, even though using the bomb would not involve harming someone as a mere means to save the five (as in the Fat Man Case). This is because the PPH only permits means that have the greater good as a noncausal flip side to cause the lesser evil. Discussions of the Trolley Problem often ignore that it can be wrong to use a device (like the bomb) that causes the five to be saved and also kills someone else as a side effect. Attention is placed on the contrast between turning the trolley and toppling the fat man.

However, as quoted earlier, Pincione says: “If the subsidies could be seen as the flip side of (the enactment of) redistributive measures . . . the PPH would authorize such measures.” In this picture, I believe, the subsidies are not taken to cause the lesser harm of taxes. Rather the subsidies are seen as the flip side (or aspect) of some redistributive measure that also triggers taxes. However, it is only if the subsidies are also meant to constitute rather than cause some people’s freedom from threat of recession that the redistributive measure that causes harm would also have a noncausal relation to the greater good. This might be so if people being granted subsidies constituted their being saved from the recession. Suppose this is so. Then the redistributive measure that has the greater good as a noncausal flip side (or aspect) would also threaten others (via its tax component or taxes to which it leads). Pincione may think that introducing a measure so understood satisfies the PPH.

If this is his view, a comparable case in ordinary morality might not be one in which we redirect the trolley, but instead one in which we move the five potential victims away from a trolley that cannot be redirected by turning a swivel table on which they sit. Their moving is an aspect of its moving. The turning table, however, dislodges rocks that kill a bystander. (I called this the Lazy Susan Case. ¹⁵ ) It did indeed seem to me that turning the Lazy Susan satisfied the PPH. Would it also be permissible to introduce a Lazy Susan into the scenario in which the trolley is moving toward the five, placing the swivel table under them so that they sit on it, and then moving them away from the trolley, even though rocks dislodged by the turning table will kill someone else? This is a case that is closest to the latest picture we are considering of a redistributive measure that we introduce and that involves a subsidy and taxes. Hence, surprisingly, it may be the bystander introducing and turning the Lazy Susan that Pincione should have considered in connecting ordinary morality and justification for state action, rather than the standard Bystander Case (in which the trolley is redirected).

Despite this attempt at modeling of government action on the Lazy Susan Case, I remain skeptical that it is the right way to conceive of government action. This is primarily because it seems to me that subsidies in law are not yet the greater good. It is only once people have the actual money in hand (caused by the subsidy law) that we have a greater good. On this view, subsidies cause the greater good. They do not constitute it. So a redistributive measure would have a causal relation to both the greater good and the lesser evil (like the bomb).

However, suppose government funds available from some source other than taxation (such as donations or investment gains) were sufficient to provide subsidies for a short time, and such subsidies did not violate libertarian principles. Suppose poorer people having subsidized income increased their purchasing power and this led to the lesser harm of reduced purchasing power on the part of the rich. This way of causing lesser harm would satisfy a condition like (b) and the PPH. Further, suppose the reduced purchasing power of the rich (somehow) caused the government to be able to continue the subsidies to others that it would not otherwise have been able to do, and it was only because it was foreseen that this would happen that it made sense to provide the original subsidy. Then the PPH would endorse the permissibility of causing the lesser harm, even though the lesser harm is a causally necessary means to the continuation of benefits and only their continuation would justify providing the initial benefits. This would be an economic analogy to a version of the Trolley Case known as the Loop Case. ¹⁶

In this discussion, I have been accepting Pincione’s working assumption (that he accepts for the sake of argument) that the government is a bystander to an economic recession, not its cause by act or omission, and that it has no special duty to rescue its citizens from a recession. ¹⁷ However, suppose the government were morally responsible for a recession or had a special duty to rescue its citizens from it. Then it would not be a bystander but more like the driver of the trolley headed toward the five. In this case, actions to prevent the greater harm that are (at most) only permissible for a bystander might be obligatory for the government. Nevertheless, if the trolley cases are any indication, a government being responsible for causing, or for rescuing from, a recession need not imply that it is obligated to take actions beyond those permitted to a bystander. I shall elaborate on this point below.

III. New Issues in Killing Versus Letting Die in Trolley Cases

Since 2007 when Pincione published his article, several new questions have arisen regarding the Bystander Case (TP(2)). I will now discuss some of these and their bearing both on Pincione’s argument against libertarians and on when harming someone constitutes aggression.

A. Thomson’s revised view on the Bystander Case

In her 2008 article “Turning the Trolley,” Judith Thomson reversed herself and argued that turning the trolley in TP(2) is not permissible. ¹⁸ She accepted Philippa Foot’s argument that the negative duty not to kill one person takes precedence over offering positive assistance to the five people because the bystander would commit no injustice to the five in not aiding them but would commit an injustice in violating the negative duty not to harm the one. She sought to buttress her position by noting that neither a bystander nor the person he would hit would have a duty to turn the trolley on himself (if he could) rather than let the five die; doing so would be altruism. But, she claimed that for the bystander to turn the trolley on another person would require that person to be an altruist. ¹⁹

However, Thomson accepts (the possibly contentious view) that the trolley driver faces a choice between killing five people and killing one, and the negative duty not to unjustly kill five takes precedence over the negative duty not to kill one. This is so even though the one person who would be killed by the driver also need not volunteer to be killed to save the five.

Suppose it were impermissible to redirect in TP(2). Then the threat that Pincione sees (arising from ordinary moral permissions to harm some to save others) to a certain attempt to justify a libertarian state might not exist. (It is worth noting that Thomson herself may distinguish between what a government is permitted to do to a person and what another person may do. For example, she says, possibly the government may draft for military service, but one citizen may not draft another. ²⁰ )

My own view is that turning the trolley in TP(2) is permissible and does not constitute aggression against the person hit by the redirected trolley. If this view is correct, Pincione’s grounds for considering its implications for libertarianism would stand. I think it may sometimes be permissible for a bystander to impose losses on someone that the latter could permissibly decline to altruistically volunteer to save others. Unlike Thomson, I do not think that imposing a loss on someone is “requiring him to be an altruist,” ²¹ since we are not requiring him to turn the trolley on himself.

One possible reason for thinking that the bystander may kill one other person rather than let five die is that it would be permissible for the trolley driver to do this. That is, suppose that the driver is thrown from the trolley and lands near the switch that would allow him to divert the trolley still headed to the five. (Call this the Bystanding Driver Case.) Suppose this bystanding driver would become the killer of the five if he does not divert it, given that he started the trolley and did not stop it. Still in this case, at the time he faces the choice of diverting, his choice is between letting the five die from a threat he started or killing one other person. Yet it seems permissible for him to turn the trolley rather than let the five die. It might be suggested that this is because the bystanding driver’s letting the five die, unlike the bystander’s doing so, will result in his having treated the five unjustly, since he will have killed the five in virtue of also having started the threat to them. I disagree. First, someone who will become a killer of many if he starts a threat and later lets the threat go on, still may not have privileges to kill someone else instead in just any way in order to save the many. For example, I think it would be equally impermissible for the driver, whether on the trolley or standing at the switch, to topple the fat man in front of the trolley as it is for the bystander to do so. I believe this supports the view that it is how someone would come to kill one person that is crucial for the permissibility of killing. The PPH was an attempt to distinguish the permissible from the impermissible ways of killing someone. Further, I think that the importance of the “how” factor suggests that it is not crucial for permissibility of killing whether the alternative is killing a greater number of people or instead letting them die, or whether an agent becomes responsible for an injustice if he does not act.

Notice that even if one thought the ordinary bystander is not permitted to kill one person rather than let five die, there would still be a version of the Trolley Problem that involves explaining why the driver may kill someone in some ways but not others to avoid his killing more people. (Call this TP (3).)

B. How we fail to save

Before discussing the PPH further, I think it worth presenting a different type of challenge to the view that when the alternative to a bystander killing one person in TP(2) is his letting five die, he is permitted to kill. This is a different argument than Thomson presents, but it is also aimed at showing that a bystander may not kill one to save five in TP(2). Suppose a trolley will kill eight people. We can save them either (1) by redirecting the trolley so that it kills one other person or (2) by redirecting the trolley toward an empty road thus blocking the only route that we alone could (and would) use to save five other people from a different deadly threat. In this case (which I call the Saving-By-Letting-Die Case), it seems to me that we should save the eight by doing what will lead to our letting the five die rather than do what will kill the one. ²² So in this case, when faced with the choice between killing one and letting five die, I think we should let five die rather than kill one. Could this conclusion be consistent with the view that if a trolley were headed to the five when they would not otherwise soon die, we may turn it toward one different person?

I think that these conclusions are consistent. One of the things that the Saving-By-Letting-Die Case illustrates, I think, is that we should not save the means — here it is the unblocked route — to saving other people from death when saving it requires us to kill another person, even by redirecting a threat away from others toward him. Saving the means seems to be morally different from saving people that may be done by redirecting a threat to another person. ²³ If this is true, it raises another problem for Pincione’s view that killing the one will be permissible when the policy that permits it is ex ante in the interest of everyone [his (a)], and the causal structure of killing is consistent with (b). For in the Saving-By-Letting-Die Case, it could be ex ante in the interest of all not to have the route blocked, thus allowing us to save the greater number of people; and turning the trolley from eight to the one person seems to be consistent with (b).

IV. Principles of Permissible Harm

Now let us consider the PPH further, first discussing the version in Morality, Mortality: Volume II that Pincione uses.

A. The PPH and secondary permissibility

The PPH is about harming innocent nonthreatening people. Its limits do not necessarily apply to people who are threatening. Furthermore, it need not apply to nonthreatening people who have special immunity. For example, consider a war between two parties, A and B, where other countries are, and are morally permitted to be, neutrals. If country A directs a missile to country B that would kill a hundred of its innocent civilians, B is not permitted to use the only means of saving them that involves redirecting the missile to the neutral country, even though it will kill only twenty people there.

Furthermore, what the PPH rules out as impermissible in one context may become permissible in another context through what I call the Principle of Secondary Permissibility. ²⁴ That is, suppose the only way to save the five from the trolley is to topple the fat man from a bridge when this will paralyze his legs. This is a significant lesser harm to him to which the PPH is intended to apply, and according to the PPH it is not permissible to do even this to the fat man to save the lives of the five. This is so even though it would be permissible according to the PPH to redirect the trolley when it would kill another person were it the only way to save the five, and being killed is worse than being made paraplegic. The difference in how we would bring about the different harms is morally important, according to the PPH. Nevertheless, suppose that as we are about to turn the trolley away from the five, thereby killing one other person, it becomes possible to instead topple this same person in front of the trolley, stopping it and only paralyzing his legs. Doing this would be as a substitute for the permissible act we would otherwise do that would have a worse outcome for him. Given this, it becomes at least permissible to topple this person. This is so even though toppling this person would be wrong according to the PPH were it one’s only way to save the five. Indeed, toppling someone as the substitute for killing him may become the only permissible way to save the five in this case given that it is wrong to kill someone when it would be permissible to do the act that instead merely paralyzes him.

B. Revisions of the PPH

Subsequent to developing the version of the PPH on which Pincione relies, I considered other cases that led me to modify it. ²⁵ None of those modifications would rule out the permissibility of the greater good, or means that have the greater good as their noncausal flip side, leading to lesser harm. However, more recently I have been rethinking the history of cases that originally led me to propose the core ideas in the PPH. Let me briefly review that history. Philippa Foot ²⁶ thought it would be impermissible to use a gas in an operation to save five people when we foresee that the gas will unavoidably seep into a neighboring room killing one other person. I agreed with this conclusion. We might say that if the gas was used, the one person would be aggressed against (even though he was not an intended victim of harm). Similarly, I thought it was also wrong to explode a bomb that would stop the trolley from hitting five people when the bomb would kill one other person as a side effect. However, I revised the details of Foot’s Gas Case so that in order to save the five we needed to use a gas that would harm no one else. In this case, however, we foresee that when the five are saved they will breathe normally and this will move germs in the environment, otherwise safely closeted, so that these germs unavoidably kill one other person. (Call this the Breathe Normally Case.) I thought it was permissible to use the gas in this case because it is not the gas — a mere means to saving the five — that is causing the one’s death. Rather it is the greater good of the five being alive that leads to the death. Similarly, I thought it would be permissible to move five people away from an oncoming trolley when the trolley could not be redirected, even if this created a new threat, such as rocks previously stable becoming dislodged and killing one other person.

I further distinguished between (1) means to the greater good, such as Foot’s deadly gas or the bomb in my case, that cause the greater good and (2) means to the greater good, such as the trolley moving away from the five, that have a noncausal relation to the greater good in the sense that the five saved just is the flip side of the trolley moving away. I thought the use of the latter type of means to save the five was permissible, though it caused a side effect death, even if the use of the causal means such as the bomb was not permissible when it caused the same side effect. In addition, I suggested that there might be a moral difference between a causal means to the greater good “directly” versus “indirectly” producing a lesser harm. Under “directly” I included a means that we introduce (whether it involves one causal step or many) that itself causes the lesser harm, as the gas does in Foot’s case. Under “indirectly” I included something that is in the environment independently of what we do being affected by means we introduce so that the former causes harm. ²⁷ I suggested that using the latter causal means to produce a greater good might be permissible even when use of the former for the same purpose was impermissible.

In recently revisiting these cases, the following struck me: ²⁸ (a) In order for five breathing normally to be a case of the greater good itself causing the lesser harm of one dead, the harm should not depend, as it does in the Breathe Normally Case, on there being germs in the environment independently of what we do. Rather, it should involve, for example, the expansion of the five’s chests, when they breathe normally, unavoidably pressing into the chest of the one other person so that he cannot breathe. (We could suppose that the six are in cramped quarters, and if the five had remained sick and died, the one would not have been interfered with. Call this case Cramped Quarters.) If saving the five was still permissible when we foresaw that this would happen, it would make no problem for the PPH’s claim that greater good may cause lesser evil. But if it were not permissible, this would suggest that it is only when it is through an “intermediary” causative agent independently in the environment that the greater good may permissibly lead to lesser harm.

(b) Suppose, for argument’s sake, that the greater good may permissibly cause lesser harm only through an intermediary entity independently in the environment. Then the question would arise whether there really is a moral distinction between greater good and mere causal means to the greater good causing lesser harm. For suppose we know that the gas used to save the five in Foot’s Gas Case would not itself harm anyone, but releasing it would cause germs independently in the environment, otherwise safely closeted, to move and kill one other person. (Call this the Gas-Germ Case.) I suspect using the gas in this case would be permissible.

Hence, it is important to consider not only Foot’s Gas Case but this Gas-Germ Case, and it is important to consider not only the Breathe Normally Case but also the Cramped Quarters Case if we are to decide on the correct form of a principle of permissible harm. The results of this may or may not require alteration in the PPH (or the Principle of Productive Purity, an earlier revision of the PPH, that I also suggested). (An altered principle might be called the Independent Intermediary Principle.)

However, these conclusions would not affect the permissibility of turning the trolley in TP(2) because this case involves neither the greater good itself causing a lesser harm without an independent intermediary nor a mere means that has a causal relation to the greater good itself causing a lesser harm. The case comes closest to being one in which an intermediary that is independently in the environment, namely the trolley, causes the lesser harm when it moves on the second track, but its movement has a noncausal relation to the greater good.

Considering new cases also led me to think that the PPH should be seen as involving two separable components: (i) producing a greater good and (ii) morally different ways of bringing about goods and harms. The first component might not be necessary for the second to account for permissible harm.

To see this, suppose a trolley is originally headed toward killing one person. I do not think that to save himself he may set a bomb that stops the trolley when the bomb itself also kills five people as a side effect. Nor may he topple five people from a bridge to stop the trolley’s initial hit. However, I think he may shield himself if the shield would protect him from the trolley independently of whether the trolley moves away, even though he knows the trolley hitting the shield will cause the trolley to move away to another track where it will kill five others. I also suggest he may escape the trolley by jumping off the track onto safe land even if he knows this will cause a rockslide that will kill five people below. In addition, it seems to me that he may simply push the trolley away from himself even if he knows it will go onto a track killing five others. In all these cases, I do not think he is merely excused from committing aggression (i.e., he is not to be blamed for doing something wrong); his actions are not wrong and are not aggression against the five understood as unjustified use of force against them. However, the permissibility of doing these things cannot be due to the greater good, or to means that have the greater good as a flip side, causing lesser harm since his being saved is a lesser good relative to the greater harm of five being killed. Nor can it be due to any of the five being no worse off than they would have been had the trolley not been redirected to the one person to begin with, since in this case the trolley first approaches the one person, not the five. It might be argued that he may do these things because each person has a special permission to do whatever he can to save himself. But this is shown not to be true by what I have said he may not do to save himself when it would kill five people. How he brings about the deaths of the five is morally important, as clause (b) in Pincione’s considerations would suggest.

Further, suppose that those who care especially about a threatened person are permitted to do on his behalf what he would be permitted but is unable to do for himself. Then if the single person is permitted to act as I suggested, it would sometimes be permissible for a bystander to shield one person, even though the trolley hitting the shield would be redirected and kill five. It could also be permissible for a bystander to turn a trolley not only from five to one but from one to five. This may seem rather shocking but it need not be if we recall that individuals are permitted to sometimes act from a “partial” rather than an impartial perspective for the sake of those with whom they identify. However, a bystander who has no good reason not to be impartial should conjoin permissible means with pursuit of the greater good in the cases we have been considering in order that he not to be guilty of aggression.

Finally, in more work on the Trolley Problem, I also tried to locate some underlying idea that might justify the intuitive judgments and the distinctions drawn by the PPH and revisions of it. One suggestion was that there can be a moral distinction between substituting some people for others and subordinating some people to others. ²⁹ Permissible harmful substitution occurs when, for example, we push a threat from some people to others without interfering harmfully with a person as a means to bring this about. Subordination most clearly occurs when we harmfully interfere with someone as a means of benefiting others (as in toppling from a bridge). But how is subordination involved in setting off a bomb that will both move the trolley away from the five and also directly (in the sense described above) kill a bystander, which the PPH would prohibit? I suggested: Ordinarily if we faced a choice between setting off a bomb and not killing a person, we would favor not killing someone. However, suppose we favor setting off the bomb that directly kills one when the bomb going off would save the five by turning the trolley. This seems to involve subordinating the one person killed to the five because the five would be able to transmit their value to the bomb so that its being used outweighs the life of the person, and this seems wrong. Something similar would occur in the Saving-By-Letting-Die Case (discussed in III.B) if we protected a pathway needed to help the five and turned the threat from eight so that it killed one person instead; we would have allowed the five to transmit their value to the pathway.

V. Conclusion

We have considered complications introduced by the Trolley Problem to the discussion of when it is permissible and impermissible to harm some for the sake of helping others. Pincione considered the relevance of such discussions to the justification of such government action as taxation. I have considered Pincione’s arguments and also the relevance of trolley cases to developing a broader theory of aggression, insofar as aggression is the unjustified use of force that is either foreseen or intended.