Should we want the greatest good for the greatest number? (And, incidentally, should the “we” mean a numerical majority?) The Trolley problem in philosophy raised the issue. I was reminded of that in an interesting article by economist and philosopher Michael Munger, “Adam Smith Discovered (and Solved!) the Trolley Problem” (June 28, 2023), as well as in a follow-up Econtalk podcast.
The precise form of the Trolley problem was formulated by British philosopher Philippa Foot in a 1967 article. Imagine you see a runaway trolley speeding down a steep street and about to hit and kill five men working on the track. But you are near a switch that can divert the trolley to another track where only one man is working. None of the men see the trolley coming. You are certain that if you switch the track, only one will die instead of five. Should you switch it, as a utilitarian would ?
If you answered yes, consider an equivalent dilemma (quoting from Munger’s article):
Five people in a hospital will die tomorrow if they do not receive, respectively: (a) a heart transplant; (b) a liver transplant; (c) and (d) kidney transplants; and (e) a blood transfusion of a rare blood type. There is a sixth person in the hospital who, by astonishing coincidence, is an exact match as a donor for all five. If the head surgeon does nothing, five people will die tonight, with no hope of living until tomorrow.
Assuming there is no legal risk (the government is run by utilitarians who want the greatest good for the greatest number and are fond of cost-benefit analysis), should the head surgeon kill the providential donor to harvest his organs and save five lives? To answer this question, most people would probably change their minds and reject the crude utilitarianism they espoused in the preceding Trolley problem. Why?
Munger argues that Adam Smith formulated another instance of the Trolley problem in his 1759 book The Theory of Moral Sentiments and discovered the principle to solve it. Smith did not express it that way, but his solution points to the distinction between intentionally killing an innocent person, which is clearly immoral, and letting him die from independent causes, which is not necessarily immoral. Drowning somebody to kill him is immoral, but not saving somebody who is drowning may not be. Intentionally shooting an African child is murder; not giving to a charity the $100 that would save his life is certainly not criminal.
A more recent argument by Philippa Foot (see Chapter 5 of her book Moral Dilemmas: And Other Topics in Moral Philosophy [Oxford University Press, 2002]) explains that the underlying basic distinction is “between initiating a harmful sequence of events and not interfering to prevent it” (this concise formulation of her complete argument is from the abstract of her article). More precisely, she writes:
The question with which we are concerned has been dramatically posed by asking whether we are as much to blame for allowing people in Third World countries to starve to death as we would be for killing them by sending poisoned food?
Emphasizing moral agency, the basic principle is that
It is sometimes permissible to allow a certain harm to befall someone, although it would have been wrong to bring this harm on him by one’s own agency, by originating or sustaining the sequence which brings the harm.
In his 2021 book Knowledge, Reality, and Value, libertarian-anarchist philosopher Michael Huemer also considers the Trolley problem and comes to a similar solution, albeit more nuanced in extreme cases. His philosophical approach is “intuitionism,” as the subtitle of this book suggests: A Mostly Common Sense Guide to Philosophy. (My double Regulation review, “A Wide-Ranging Libertarian Philosopher, Reasonable and Radical,” gives the flavor of this book and of his The Problem of Political Authority: An Examination of the Right to Coerce and the Duty to Obey [2013].)
Anthony de Jasay’s condemnation of utilitarianism as a justification for government (coercive) interventions is based on the simple economic observation that there is no scientific basis for comparing utility between individuals; for example, it is meaningless to say that saving five men preserves “more utility” than killing one. Utility pronouncements, he writes, “are unfalsifiable, forever bound to remain my say-so against your say-so.” (See my Econlib review of his Against Politics.)
What is pretty sure is that utilitarianism, and certainly “act utilitarianism” (as opposed to “rule utilitarianism”), does not work, except perhaps in the most extreme and uninteresting cases—such as “stealing $20 from Elon Musk without him noticing and transferring the money to a homeless person would create net utility,” that is, Musk would lose less utility than the pauper would gain. Even if the statement seems to make sense, we cannot predict a single individual’s behavior, only general classes of events: perhaps that homeless person will use the $20 to buy cheap alcohol, get drunk, and kill a mother and her baby, who would have been a second Beethoven. He might even be a utility monster, deriving “more utility” from the harm he causes to others than what they lose. Even if the homeless person uses his $20 to purchase a used copy of John Hicks’s A Theory of Economic History, the story of his “gift” might spread and lead to a billion greedy people crying for the same transfer from Musk. Or they may agitate for the $20 billion to be directly expropriated by the state to finance subsidies for them.
****************************************
I tried hard to have DALL-E represent the simplest version of Philippa Foot’s Trolley problem. Despite my detailed descriptions, “he” just could not understand–which is not really surprising, after all. Even the idea of a fork in a trolley’s track with five workers on one side and one on the other side, he could not represent. I finally asked him to draw a runaway trolley with one track and five workers in the middle of the track. The images he produced were among his most surrealistic, as you can see from the featured image of this post. Given his poor performance, I mentally apologized to Philippa Foot (who died at 90 in 2010) and instructed DALL-E to add to the image “an old, dignified woman (the philosopher Philippa Foot) in deep thinking and looking at the trolley coming.” In this simple task, the robot did quite well.
READER COMMENTS
Craig
Jul 24 2024 at 10:05am
The trolley problem lends itself to many wrinkles. With respect to law, the problem is a thought experiment with respect to the imposition of criminal and/or civil liability. Let’s say the train is headed straight and won’t hit the 5 people tied to a side track or 1 person tied to the other side track. If you divert the train on purpose knowing its going to kill somebody you will be criminally liable for some form of homicide. Now block the straight track and the train is headed for the 5 people. If you do nothing, the law imposes no duty to assist so if you simply permit the train to continue on its course, no liability will be imposed on you*. Of course the wrinkle is when you pull the switch to divert the train from killing the 5 people to killing the single person. Does the person now face criminal/civil liability for the homicide/wrongful death of the person killed who wouldn’t have been killed except for the fact that you intervened? Then of course one can tweak the facts, but most have a problem imposing criminal/civil liability on the person making the decision to pull the switch and to allow the train to only kill one person, the Hollywood fact pattern is so hyperspecific and so contrived that most see how such a circumstance would be a sufficiently mitigating circumstance for a court not to impose criminal/civil liability on the person making that choice.
Today the more recent wrinkle on this is driverless cars where the AI might potentially face a choice between two harms and needs to make a choice between the two.
*one tweak is that the person works for the railroad or a police/fire agency with a preexisting duty and then ask if any liability would be imposed for NOT pulling the switch.
Pierre Lemieux
Jul 24 2024 at 12:06pm
Craig: What you say is very much in line with Philippa Foot. According to Mike (Munger), surveys suggest that half or slightly more than half the surveyed people answer that switching the track is the moral thing to do. I would interpret this as confirming that neither 51% nor 49% (0r 52% and 48%, …) of the population provide a moral criteria.
Your mentioning AI is interesting. Nearly by definition, I would say, a bot cannot make a moral choice. We cannot know and he cannot know either why he made the “choice.” One could say, à la Hayek (see Nobody.Really below), that humans follow rules that they often cannot themselves explain, but these rules have the benefit of social evolution.
nobody.really
Jul 24 2024 at 10:35am
I agree that this premise helps explain intuituions that arise from Trolley-type problems. But does that make those intuitions good/appropriate/optimal?
In his Law, Legislation and Liberty, Vol. 3 (1979), “Epilogue: The Three Sources of Human Values,” Hayek argues that we should expect tribalism to have the strongest emotional appeal to us, and we should reject it nevertheless:
Id. at 528-29 (emphasis in original)
David Seltzer
Jul 24 2024 at 11:32am
Pierre: Terrible conundrum where the death is the outcome for either five dying or one dying in the Trolley Problem. Doing nothing to save a drowning person when the person doing nothing can swim is another sad dilemma. Is there a duty to rescue? France and Canada have enacted “Good Samaritan” laws to protect those who offer assistance. If we are to think like rational economists, it seems the best trade-off is made with the lowest societal opportunity cost. I suspect there will be winners and losers no matter the decision. Sounds harsh but what else can one do?
Craig
Jul 24 2024 at 12:21pm
https://recreation-law.com/2014/05/28/good-samaritan-laws-by-state/
David, perhaps worth a brief perusal….
David Seltzer
Jul 24 2024 at 2:45pm
Thanks Craig. Read through the GS laws by state for all listed.
Pierre Lemieux
Jul 24 2024 at 12:47pm
Thanks, Nobody (as Ulysses wouldn’t say), for bringing Hayek’s challenges in the conversation. It seems to me that rationally reflecting on morality like an intuitionist philosopher (Foot, Huemer, probably de Jasay, and perhaps Munger) relies on our experience of the abstract liberal society more than on our instinctual emotions inherited from our tribal past. I think this would also be the answer of Buchanan, who said that we still need a rational criterion to judge the results of social evolution (quoting my review of his Why I, Too, Am Not a Conservative):
If you listen to the Econtalk conversation I link to between Russ Roberts and Mike Munger, the latter seems to say that Smith’s vision of society was, of course, influenced by the society he lived in and the individuals he was in contact with. If your circle of friends is made of people such as David Hume, your moral intuitions will be different than if your society is made of tribal progressives.
In other words, tribal emotions are not the same as philosophical intuitions. But the danger that Hayek warns us about is real.
nobody.really
Jul 24 2024 at 2:45pm
Cute…!
David Seltzer
Jul 24 2024 at 2:50pm
Nobody: Not only cute but Odysseus correctly assumed Polyphemus would be easily fooled. LOL!
Scott Sumner
Jul 24 2024 at 1:00pm
I see two types of utilitarianism:
Dumb utilitarianism, which engages in reasoning like the organ transplant example above.
Smart utilitarianism, which considers all the indirect costs of having hospital personal murder patients that have gone to the hospital to be helped.
I completely agree with you that dumb utilitarianism is a bad idea.
“Anthony de Jasay’s condemnation of utilitarianism as a justification for government (coercive) interventions is based on the simple economic observation that there is no scientific basis for comparing utility between individuals; for example, it is meaningless to say that saving five men preserves “more utility” than killing one. Utility pronouncements, he writes, “are unfalsifiable, forever bound to remain my say-so against your say-so.””
Aren’t alternative ethical frameworks also unfalsifiable?
Pierre Lemieux
Jul 24 2024 at 3:18pm
Scott: You write about “smart utilitarianism, which considers all the indirect costs…” Consider that calculating all indirect costs requires that the future consequences of particular events be foreseeable. But even if that were possible, weighing these consequences would require interpersonal comparisons of utility, which is conceptually impossible.
You are right to ask: “Aren’t alternative ethical frameworks also unfalsifiable?” The answer is of course that they are not in the scientific sense with empirical verification. They can, however, be submitted to logical tests (as philosophers such as Philippa Foot do). They can also be evaluated as more or less consistent with a free society, even if this is more challenging. Finally, as Anthony de Jasay notes, they must not require too much of our moral credulity. He argues, very cogently in my view, that “the basic presumption against coercion, a presumption that can be derived either from an axiom about the practice of choice, or from a social convention of ‘live and let live’ … when it involves no harm to others” is a value judgment that “demands far less of our moral credulity.” Hence, the first principle of intervention by political authority is “when in doubt, abstain.” (See his Against Politics, Chapter 8)
Scott Sumner
Jul 25 2024 at 1:39pm
I’m afraid I don’t accept those arguments. Coercion is wrong in some cases and correct in others. For instance, we often coerce children to learn to read and write. In my view, utilitarianism provides the best framework for deciding when coercion is wrong and when it is not wrong. I favor a carbon tax as a way to coerce people to use less fossil fuels.
As far as interpersonal comparisons of utility, I believe that reasonable estimates are possible. Thus I’d estimate that average utility for people living in Beverly Hills is higher than for inmates in Soviet prison camps. Yes, I cannot be certain, but we navigate through life making reasonable guesstimates. I see no alternative.
Jose Pablo
Jul 25 2024 at 4:43pm
I favor a carbon tax as a way to coerce people to use less fossil fuels.
And yet you have no way to show whether this increases or decreases “global utility”. Particularly so in this case.
To say that you favor “carbon tax coercion” and to say that you favor the people who prefer a reasonable expectation of lower temperatures at the expense of lower consumption now over the people who prefer the certainty of more consumption now over a reasonable expectation of lower temperatures in the future, is to say the same thing.
The latter (equivalent) wording just shows more clearly the arbitrariness of the coercion of others that you arbitrarily “favor”.
Jose Pablo
Jul 25 2024 at 1:44pm
Aren’t alternative ethical frameworks also unfalsifiable?
No they are not.
Any Pareto efficient course of action can be safely used as justification for government action (no interpersonal comparison of utility required). And a clear way of knowing when a course of action is Pareto efficient is by requiring unanimity in deciding whether or not to pursue this particular collective action.
The only “ethical framework” you need to agree on is that people will only engage in “voluntary actions” if they are better off pursuing them.
Thomas L Hutcheson
Jul 24 2024 at 1:18pm
At least give DALL-E credit for including the image of Ms. Foot contemplating the scene! 🙂
Pierre Lemieux
Jul 24 2024 at 3:32pm
Thomas: I agree, but I thought I did in my blurb on his image by saying “In this simple task, the robot did quite well.” You think I was not eulogious enough? Perhaps I should have emphasized that her pose was especially well pictured.
Thomas L Hutcheson
Jul 24 2024 at 1:29pm
I think the trolley problem is a nice metaphor for Climate Change policy. Roaring down the track of public opinion myriad of high cost “solutions” are being carried out: rooftop solar with net metering, “green” mandates for everything under the sun 🙂 ON the other track are a few deadweight losses of taxation of net CO2 emissions.
Do we advocate for throwing the switch or not?
Jose Pablo
Jul 25 2024 at 4:50pm
Give every single individual a certain amount of yearly “CO2 rights” and let them freely trade these CO2 rights.
Every CO2 emitter should (somehow) get enough CO2 rights to offset its emissions. Ideally by setting the “price in CO2 rights” they deem appropriate for any good and service they sell.
David Seltzer
Jul 25 2024 at 7:04pm
Pablo: It’s been done at the CME since 2022. To wit.
“CHICAGO, April 27, 2022 /PRNewswire/ — CME Group, the world’s leading derivatives marketplace, today announced that combined volume across CME Group’s voluntary carbon emissions offset contracts has surpassed 100,000 contracts traded, equivalent to 100 million carbon offset credits, or 100 million metric tons of CO2 equivalent.” Seems like a better market alternative to a carbon tax.
Jose Pablo
Jul 27 2024 at 1:06pm
We “know” is a system that works better, David. And we “know” that all that is needed is a clear assignation of the property of those “emission rights” (ideally citizens/individuals/consumers) and the existence of frictionless markets for those rights (pretty easy to achieve nowadays).
But for professional politicians (and their advisors), the pain of renouncing “ruling over others” is just too high. After all, market solutions will render them useless. And that’s painful.
Monte
Jul 24 2024 at 2:39pm
As I see it, the Trolley problem has only two solutions:
1. Self sacrifice – If the intervening agent is in a position to sacrifice him or herself, this would be viewed as a socially and morally acceptable (Pareto-efficient) act of heroism.
2. Don’t just do something, stand there! – This is one instance where doing nothing is greatly preferable to doing something. Some things are better left to fate. Que será, será.
Monte
Jul 25 2024 at 11:19am
It seems magnitude sensitivity (as in the example given by Ahmed below) influences our decision-making process and causes us to think more utilitarian. In what scenario sacrificing the lives of hundreds be preferable to sacrificing the lives of thousands?
Monte
Jul 25 2024 at 1:29pm
That is to say, in what scenario would sacrificing the lives of thousands be preferable to sacrificing the lives of hundreds? Maybe to preserve the U.S. president and his entourage if we were under attack from a foreign adversary?
Ahmed Fares
Jul 24 2024 at 3:23pm
I was discussing the AIDS crisis with my wife and mentioned that they stoned women to death for adultery in some North African countries. My wife had a cringe look on her face.
I then asked my wife to imagine that she was the leader of a North African country and AIDS has just arrived on the scene. She had two choices: stone a hundred women to death for adultery or watch a hundred thousand of her citizens die of AIDS.
She said she would stone to death.
steve
Jul 24 2024 at 4:27pm
But definitely dont stone the men. So the end result of stoning the women is that you get 100 female scapegoats and AIDS continues to spread.
Steve
Ahmed Fares
Jul 24 2024 at 5:06pm
The penalty applies to both men and women, but I stated it that way because that is what is usually reported in the media. In any event, it distracts from the topic at hand.
The problem is with the latter part of your comment. AIDS was a problem in Sub-Saharan Africa, which is code for Christian and animist Africa. Here’s some statistics:
HIV adult prevalence rate
HIV/AIDS in Africa
Mactoul
Jul 25 2024 at 12:22am
Governments deal with versions of the Trolley problem all the time. This is what, I believe, Jasay complains about–govts hurt some people to benefit others and this makes govts illegitimate from Jasay’s perspective.
But if trolley problems arise inevitably then they need to be solved. Even if we dislike the necessity.
Jim Glass
Jul 25 2024 at 1:38am
I’ve never much understood all the ink (and now bandwidth) spent on the Trolley problem. We all switch trolley tracks all the time on every scale. From a business unemploying everyone in a local money-losing branch for the sake of the larger firm, to killing ~200,000 humans by bombing Hiroshima and Nagasaki to save millions.
Because if upon checking into a hospital to have my hemorrhoids fixed there was any chance I might unexpectedly find my entire body carved up into pieces and distributed to the needy and deserving, I would never ever ever go to that hospital — and neither would you or anybody else. The entire health system would collapse. And we all know it. And we all want a health system. That’s why. And to the extent this principle was practiced at large, so would the entire society. (Interpersonal utility comparisons not relevant!)
Let’s very simply convert from body parts to money. A rich person walks into a banker’s office to get an account statement. Should the banker ‘bankrupt the providential donor by looting his funds to distribute to the poor?’ If he could, what would happen to the financial system? Or, say it was the tax collector’s office. Should the taxman ‘impoverish the providential citizen by confiscating all his wealth to give to the poor’?
Is not the latter what we call ‘socialism’? Not the Bernie Sanders-Sweden kind, but the Lenin-Bolshevik ‘what happened to the Kulaks?’ kind. Have we not discussed this before? How does society function then?
We human beings, being by far the most social of animals, really really want our societies to function in the real world. That philosophers don’t even consider this little fact in their hypothetical musings reminds me of why, when I lost my undergrad philosophy degree way back in the last century, I never bothered to look for it.
Jim Glass
Jul 25 2024 at 1:49am
The same survey given to Americans in 1945 would have garnered near the same mere 50-50 result, no doubt. But history shows the *big* majority of Americans then actually supported ‘switching tracks’ to nuke those Japanese cities and save the cost of invading the home islands. So much so that Truman’s decision (had he been reluctant) was politically forced — if he hadn’t cut casualties by switching tracks he’d have been damned and impeached by the same infuriated ’50-50 survey’ subjects, when they learned he’d had the option.
This psychology of the Trolley problem has always been much more interesting to me than the purported ethics of it. Ask people what they imagine of themselves and how they imagine they’d act in a totally hypothetical situation (how many have even seen a trolley?) — then put them in an actual real-life case and see how differently they really act. *That* is significant, endemic, and has true-life consequences. It’s a big issue in behavioral psychology. Do “Trolley philosophers” ever consider this? I don’t watch them so I don’t know. Somebody tell me.
Pierre Lemieux
Jul 25 2024 at 3:38pm
Jim: Your comment combines many (somewhat confused, I think) claims. Let me try to disentangle them by asking you to consider three indicative points:
(1) In his private dealings, no individual may (without committing a crime or at least a nuisance) choose to harm somebody else, except if you assume that declining to deal with somebody (refusing to eat at McDonald, for example) or encouraging risk by consenting adults (for example, by patronizing boxing competitions) amounts to causing harm. If one makes this assumption, indeed, every action is harmful (and might indeed replace the words “choice” or “action” with the word “harm”). The Trolley problem is different because it presents an individual with a moral dilemma different from an ordinary exchange.
(2) On your incentive argument, please note that I already answered it in advance by writing in my post “without him noticing,” which also applies to the hospital dilemma (in my mind anyway). I deal later in my post with the risk that even a one-shot expropriation or murder may become known. There might be other ways to foolproof the hospital conceptual model, but if we reject this sort of reasoning a priori, it will be difficult to rationally discuss moral and political philosophy.
(3) The Trolley problem would be a different (but related) problem if we assumed, as perhaps you implicitly do, that the government (“we”?) has forbidden the railway workers from working elsewhere than on this track and is itself at the switch.
Thomas L Hutcheson
Jul 25 2024 at 7:12am
Foote’s is a persuasive position on the allow harm/prevent harm/do harm issue.
I don’t see much utility is this for most practical matters where the the issues is, what DOES happen if we do nothing vs do something vs do something else.
Pierre Lemieux
Jul 25 2024 at 10:29am
Thomas: In Philippa Foot’s Trolley problem, we know (by hypothesis) what happens in each alternative. The urgency of doing nothing is even more imperative if, like in government interventions, we have no way of knowing who are all the persons to be harmed nor how they will be harmed, especially as time passes. It does not help to add more tracks and put different numbers of individuals on each, except if there is one with nobody. And, of course, if you start with the hypothesis that something (anything) must be done, you will conclude that something must be done.
nobody.really
Jul 25 2024 at 5:10pm
HEY–I resent that!
Pierre Lemieux
Jul 25 2024 at 10:51pm
Sorry, Nobody. Nothing personal. I had already forgotten your name.
Jose Pablo
Jul 25 2024 at 1:35pm
The question with which we are concerned has been dramatically posed by asking whether we are as much to blame for allowing people in Third World countries to starve to death as we would be for killing them by sending poisoned food?
The real question is: Is “preventing people from the Third World from entering a First World country in order to not starve to death”, akin to killing them on purpose or to “allowing” them to starve to death?
It is clear to me that forcefully preventing people from pursuing a harmless course of action that would save them is morally equivalent to killing them.
Monte
Jul 25 2024 at 4:39pm
Jose,
Hypothetical: What if, among the starving, there were individuals carrying a deadly plague that could disproportionately impact the population of the host country? What bearing might this have on the moral equivalence?
Jose Pablo
Jul 25 2024 at 5:59pm
Not sure, Monte. But yours is a very interesting question since this is precisely what Europeans did to Native Americans. Even worse since Europeans were not really starving when they entered America, at least not more than Native Americans were.
Monte
Jul 25 2024 at 7:00pm
Valid point. A consequentialist would probably act in the interest of the greater good. But I’m not sure I would sacrifice the few for the many in the moment. I suppose the best we can hope for is to never find ourselves in that situation.
Mactoul
Jul 26 2024 at 3:55am
Third world is populous relative to the First World. Under open borders, many small first world countries can be entirely submerged. Countries like Ireland, Iceland and Hungary.
This, the current residents of these countries, may perceive as harm. And who is to gainsay them?
Do you know more than them what their harms are?
Problem of knowledge, again.
Like it or not, the world is politically divided, and your notion of harm isn’t generally recognized.
Jose Pablo
Jul 27 2024 at 12:59pm
Under open borders, many small first world countries can be entirely submerged. Countries like Ireland, Iceland and Hungary
Any fact supporting this arbitrary statement?
There is an open border policy in place in the EU. And certainly, there are first and, at least “second” world countries within this space. To the best of my knowledge, Bulgaria and Romania have not been “emptied” and Germany or France has not been “submerged” (and Hungary even less so).
your notion of harm isn’t generally recognized
Yes, it seems so. But also there was a time when the notion that owning other human beings was wrong, wasn’t “generally recognized”.
Human ethics tend to evolve despite the efforts of the “troglodytes” who oppose its advancement.
Pierre Lemieux
Jul 25 2024 at 10:50pm
Thanks for all the comments. Let me just repeat what is the essence of the Trolley problem (but Philippa Foot probably says is better): If one (individual or, of course, group) is faced between the choice of, on the one hand, setting in motion, through his moral agency, a sequence of events that will directly lead to the death of one or more innocent person and, on the other hand, letting one or many persons die due to another sequence of events in which the choosing person had nothing to do, it would be morally wrong for him to choose the first option. Letting die is not the same as killing. Not preventing harm is not the same as intentionally harming.
I went on to argue that the Trolley problem is a way among others to understand that at least simple utilitarianism doesn’t work.
Comments are closed.