In my last post, I offhandedly disparaged the Trolley Problem as a serious thought experiment. Let me elaborate.
Any philosophical thought experiment contains stipulations about what is going on. In the trolley problem, it is stipulated that by flipping the switch, it *will* prevent five deaths, and it *will* cause another person’s death.
Question: do we believe that stipulation? We don’t exactly believe in it, it’s a fictional scenario. But you at least have to accept the stipulation to think about the problem on the level that it was intended.
In the variant of the trolley problem, it is stipulated that by pushing a fat man in front of the trolley, this *will* prevent five deaths, and it *will* cause the death of the fat man.
Do we believe this stipulation? To state the obvious, it’s ludicrous. Fat people aren’t nearly as massive as trains. The difference between a person and a train is so large, that one wonders why the thought experiment even bothers stipulating that they are fat, as if it’s relying on latent fatphobia to make it more believable. If a train were light enough to be stopped by a fat person, then maybe someone who could push a fat man off a bridge could just stop the train with their bare hands instead. And the problem doesn’t just require that a fat man will magically stop the train, but also that we have magical knowledge of this fact.
I think it’s a common issue with these thought experiments, they make these absurd stipulations and assume that people will just automatically accept them. But when you actually talk to people, you find that accepting the stipulation is often a difficult step.
There are also cases when it is correct to reject the stipulation! Most notably, people come up with thought experiments to justify torture, and their stipulation amounts to “what if torture works, wouldn’t torture be justified?” Okay, but it doesn’t work, and the thoughtless reenactment of this thought experiment across media is materially harmful. And the torture thought experiment really isn’t that far off from the trolley problem, whose stipulation amounts to “what if murder saved lives actually?”
The absurd stipulations of trolley problems are particularly problematic, because it’s not just philosophers thinking about it. Trolley problems are the subject of psychological research. They were asking ordinary people for their thoughts, and using this to come up with descriptive ethics. But the fact that trolley problems contain stipulations that vary in absurdity is a big confounding factor. I think it’s bad scholarship. I believe the initial research on trolleyology belongs in the trash, alongside the countless other psychology studies implicated by the replication crisis.
The demotion of trolleyology to unfunny memes is what it has always deserved.
I’m glad you brought up the use of thought-experiments to justify torture.
I’ve been bristling about this for a couple decades now, because the thought-experiment used to justify torture (usually the one about capturing someone who hid a nuclear/dirty bomb that is going to kill thousands to millions in the next , and is it justifiable to torture them to get the information to locate and disarm the bomb) The whole premise is absurd on multiple levels, yet it’s been used to wholly justify torture for any and all reasons
To me, the trolley problem seems like a rather clumsy and contrived way of asking the question, “Is it ethically allowed to do something that will cause harm if it will probably prevent more harm?” That seems like a perfectly legitimate and relevant ethical question to me. Maybe someone just needs to come up with a better scenario to illustrate the question.
It presumes lives are fungible.
That quantity is more important than quality.
Those are like spherical cows.
As John Morales points out, this problem assumes lives are fungible, which is sort of true at a statistical distance. I think the utilitarian argument that is the basis for the trolley problem is a nice sort of guideline to keep in mind, but a lousy doctrine on which to base individual decisions. The further away you stand, the blurrier the distinction between people, between one harm and another, between one happiness and another. But the world is far too complicated and contradictory for such abstract ideas to be enforceable on the spot. It seems similar to effective altruism in this sense, a fine idea in the abstract, but iffy in particular, unless you not only trust the integrity and efficiency of others, but ignore the finer details, as well as your own feelings and affinities.
As an actual test of anything beyond general principles, it’s worthless, because it is so purposely devoid of detail that whatever answer you give, one can reveal an omitted detail that makes it wrong.
!!! Yes, a lot of these kinds of thought experiments are like, we set up this really contrived situation such that the math works out in a way that it’s “better” to do something that people’s intuition tells them is bad and wrong. And then the conclusion is “see, people’s intuition is illogical.” When actually this intuition is based on the reality that it’s extremely difficult to actually predict with certainty what will happen, and it’s a good idea to err on the side of not pushing people onto train tracks.
I kind of look at trolley problems and similar ideas as a bad setup.
They’re really gamey and artificial. There are no good answers or even very revealing answers. This one in particular and others that pose extremely stressful and/or life and death situations are presenting scenarios where we already know people don’t act the way they think they will. We don’t know how we’ll react to being shot at unless we’ve been through that experience before.
This makes the answers largely meaningless. And frankly I tend to think asking the question a lot says more about someone than any answer ever could. Because asking out of more than idle curiosity feels like asking permission to make bad choices even if they carry a terrible cost for other people. It feels like they’re asking if you’ll forgive them in advance for being immoral.
Also I have to post this link https://www.mcsweeneys.net/articles/lesser-known-trolley-problem-variations
I don’t think I entirely agree, I think there could be some value in discussing the trolly problem in an introductory psychology class.
It certainly isn’t a philosophical problem, and as a moral problem it isn’t very interesting. At best it can show that there are consequences for inaction as well as consequences for action. By making the consequences for inaction greater than the consequences for action it accentuates the issue, but doesn’t really change things. The thought that there is a mathematical solutions of saving five is better than saving one is certainly not valid as there are too many undefined parameters, as John Morales points out, it assumes humans are fungible. Yet, we know our decision may change if we are aware who are on the various tracks. Is the problem easier if Trump is the one sacrifice, but you save Musk, Vance, Hegseth, Johnson, and Thune?
The interesting part comes when, after the student answers that the death of one person is better than the death of five, to imagine their best friend as being that one person. Would they make the same decision? Would their friend ask them to?
It allows a discussion about why a catastrophe half the world away generates less of an emotional response than something close to home. 5,000+ people dying in SE Asia from an earthquake generates less emotion in the US than a single child drowned in a pool. Or why smaller number, but more relatable, deaths generate a stronger emotional response than a higher number of deaths. A school shooting which kills 12 children has more emotional impact than a mudslide which kills 40 adults.
These are not discussions which are important in a philosophical sense. But a person giving some thought as to why they feel the way they do about themselves, their family, their friends, their acquaintances, and strangers can build empathy. So, while I don’t think the trolly problem is interesting philosophically, or even morally, it can have value if people think about how their answers may differ depending on who the victims are, and what that means about their own thoughts and beliefs.
Like a lot of thought experiments it can be useful in special circumstances. But also like a lot of thought experiments it has significant limitations.
Maybe someone just needs to come up with a better scenario to illustrate the question.
How about talking about known real-world situations (or at least simplified versions thereof), instead of silly-sounding “thought experiments” made up by twits who clearly don’t want to deal with real life? The further they get from reality, the easier it is to rig them to get an answer that aligns with the “experimenter’s” agenda, and not with observable reality.
I’ve heard so many “thought experiments” that are just insultingly ridiculous and not worth even a PHIL 101 class’s time — trolleys, wizard-and-nuke, dark forests, “quiet gentle rape” — that I just want to scream “WHO WRITES THIS STUFF?!” These experiments tend to say a lot more about whoever is making them up than anything else.
The difference between a person and a train is so large, that one wonders why the thought experiment even bothers stipulating that they are fat, as if it’s relying on latent fatphobia to make it more believable.
I hadn’t heard that one before, and God’s balls, yes it’s relying on fatphobia, and it’s not even “latent!” This is a bunch of middle-school-mouth-breathers going “huh huh, lookit that fat lady, huh huh, you could use her as a shock-absorber to stop a runaway train, huh huh huh…” I’d steer well clear of any “philosophy” class that offered that “thought experiment.”
i recall long ago a nephew was taking one of those elementary college philosophy courses, and one of the big questions was some relative of the old “Sophie’s choice” dilemma. Choose one of your two loved ones to die, or I’ll kill them both, says the terrorist The “right” answer, in this case, was apparently to choose neither, and let the responsibility be on the terrorist, a typically Kantian kind of answer in which the quality of the action trumps the result. Which is fine if you think there really is a god that will sort it all out in the end. If not, then you might as well choose one. Of course, as in all such thought experiments, part of what it requires is for you do decide, or someone to decide, whether the situation is real, the terrorist can be trusted, etc. But unlike the trolley problem, here it’s one or all, rather than one or the other. “Don’t play God” in this case, assuming the thought experiment is not rigged, means to defer to the least qualified player.
I think as Flex says, or at least implies, there is a sort of place for such thought experiments, but ultimately their usefulness is to remind us why they’re inadequate.
springa73 @ # 2: Maybe someone just needs to come up with a better scenario to illustrate the question.
How ’bout Larry Niven’s The Ringworld Engineers?
I recall reading many years ago a paper where the authors tried some 10 different scenarios where one is asked to choose between killing one person or letting 5 die, including killing one healthy person in order to use their organs to save 5 severely ill patients. The proportion of yes vs no responses varied a lot, but I no longer recall what the scenarios were (nor of course which scenarios elicited more agreement vs less with the proposal). This kind of variability could help us understand how people respond to these dilemmas, though of course that doesn’t say anything about which response is more ‘morally correct’ (within what system if morality?)
I did some searches in PubMed related to “trolley problem”. One example that came up was whether it was a good idea to replace table salt with potassium-nriched salt substitute. Replacing some sodium with potassium reduces blood pressure, thus leading to reduced risk of death from cardiovascular disease, but may cause some deaths among people with chronic kidney disease. (In this paper it is estimated that even among people with chronic kidney disease these would be a net positive effect, so perhaps there isn’t a true trolley dilemma here.)
Another scenario is in the case of vaccination – is it OK to require vaccination if a certain percent of the recipients will die as a result of side effects of the vaccine.
Replacing some sodium with potassium reduces blood pressure, thus leading to reduced risk of death from cardiovascular disease, but may cause some deaths among people with chronic kidney disease.
That can be remedied by clear warnings on the “lite salt” labels.
Another scenario is in the case of vaccination – is it OK to require vaccination if a certain percent of the recipients will die as a result of side effects of the vaccine.
In this case: a) serious side-effects are rare; b) some people are already exempt from vaccine mandates because of the possibility of side-effects; and c) side-effects can be watched for and dealt with as they happen, at least to some extent.
In neither of these two cases does one action deterministically/inevitably lead to certain death for the “disfavored”/”expendable” group.
@12: The “classic” SF story on those lines is “The Cold Equations”. That suffers the same issues as trolley problems – the scenario relies on societal failures all the way back down the line.
I’ve read that the author intended the pilot to MacGyver a solution, but J.W. Campbell insisted on changing the ending. Changing the ending without changing the rest of the story introduces flaws to the story.
My take was that the message was “the universe doesn’t owe you a living” (or “sometimes there’s only the least bad choice”), but that makes the story flawed as the scenario should never been allowed to occur in the first place. However there is a strand in SF of valorising “hard men taking hard decisions”, and seen in that light other interpretations have been made.
Arthur C. Clarke also wrote a lifeboat ethics story (“Breaking Strain”), which is must less remembered. (First rule of lifeboat ethics – first make sure that you’re in a lifeboat.)
Second rules of lifeboat ethics – do your best to avoid the need for lifeboats to be used
another stewart: OMFG “The Cold Equations” was an absolutely dreadful story. Like, why didn’t anyone think to give their interstellar freight shippers extra fuel, just to make extra sure all that expensive important stuff they’d paid for got to where it was needed?
And I’m pretty sure no “societal failure” would have prevented that simple precaution. This is rich and powerful people spending a little more to protect their investment, something no society — just or unjust — would get in the way of.
Not responding to anyone in particular, but I will say for context…
The way I’ve always heard the trolley problem discussed, is specifically looking at the two versions I mentioned in the OP. One where you pull the lever, and one where you push a fat man. It’s said that people are generally willing to pull the lever, and not willing to push the fat man. Because of this empirical result, it’s argued that people have more reservations when it requires taking an active roll.
But the argument seems faulty to me, because that isn’t the only difference between the two scenarios. Notably, the fat man trolley problem is pure physics fantasy. That’s what I was responding to with the OP.
I see from the comments, people have seen the trolley problem in many different contexts, and have a variety of reactions. Thanks for sharing, everyone!
I’ve never heard the version with the fat man. Yikes.
I think one of the arguments that hasn’t popped up here but maybe should is “It’s all well and good to save lives in the moment, but at some point we have to ask who’s tying people to the tracks”, which I’ve seen on tumblr … somewhere. Of course, the version where you kill the person doing the tying is killing one person so others will live, but it’s one that has an inherent moral judgement: the person you are killing was deliberately creating the situation where the other people would die.
I remember trying to construct a version that seemed less artificial. I came up with “There is only enough food to get three people through the winter, and four are trying to live on it.,” which is definitely a possible situation, but one which doesn’t require killing someone immediately.
I think the argument you’re making here doesn’t sit well with the one you made about Newcomb’s paradox, actually. My response to “a perfect predictor has decided whether you get a million dollars or not” is different from my response to “a fellow actor in the real economy has decided to give you a million dollars in the trust that you will give it back with interest” because a perfect predictor is a different thing from a real bank.
The version of the trolley problem that depends on believe a fat person would stop a trolley is absolutely ridiculous, though.
Yes, regarding the physics of the fat man scenario, if the man is heavy enough to stop the train, how would I have the strength to push him?
Raging Bee @ # 17: … “The Cold Equations” was an absolutely dreadful story.
So was The Ringworld Engineers, in that Niven & Pournelle didn’t really think through their original concept and Niven milked as much melodrama as he could out of patching it up (with further flaws).
Hmmm, Vonnegut’s Mother Night?
Pierce, be aware that Pournelle had nothing to do with Ringworld Engineers. That one is purely Niven.
The two did collaborate (mote, lucifer’s, inferno, etc), but that was later.
@Ciiriianan
I’m not committed to any particular response to the Newcomb’s paradox, and I think it’s arguably context dependent. I think the “real world” version of Newcomb’s paradox argues in favor of context dependence, because paying off a loan is in fact a context dependent decision. Context dependence also allows the possibility that the “real world” version is meaningfully different from the abstract version.
(This wasn’t really the thesis of my Newcomb’s article though. I was just trying to make a thought provoking analogy, and suggest that common hangups about Newcomb’s problems are not really essential aspects of the problem.)
I usually see trolley problems used to argue that our ethics are (descriptively) context dependent. That’s not wrong, but it raises the question of how anyone should respond to a context that is not remotely possible, and what conclusions we should draw from empirical research about it.
Pierce: I may have read “Ringworld Engineers” — if so, I don’t remember any of it. Judging by your comment, I’m not missing much.
Also, what does “Mother Night” have to do with any of this?
The trolley is running over thousands of people every day. There is a lever that stops it within reach.
You can pull the lever at any time, but to do so you have to admit you were wrong about voting for the trolley.
There are real-life trolley-like problems…
Take mammograms: a lot of people are at risk of developing breast cancer, these are the five people tied down to the tracks. An x-ray though has a chance of causing breast cancer, this is the one person (not necessarily fat…) tied down to the tracks. Do you decide to (pull the lever) screen everyone?
The trolley problem, though, dumbs this down to incomprehensibility… a stochastic problem becomes deterministic, and the setting is so unlikely that most people will think “I don’t think I’ll ever need to worry about this…”
FWIW: https://en.wikipedia.org/wiki/Trolley_problem#Original_dilemma
The canonical case was not framed as a probabilistic (stochastic) or deterministic problem.
It was framed as a moral dilemma.
John Morales@ # 22 – Please note I distinguished between N&P and N in all comments here regarding them.
Raging Bee @ # 24: Also, what does “Mother Night” have to do with any of this?
The protagonist/narrator does a Bad Thing (spreading Nazi propaganda) for a Good Cause (assisting Allied intelligence), and experiences/confronts some deep moral reckoning. I don’t recall whether he rides any trolleys in the process.
Pierce: I don’t see HW Campbell Jr’s choices as a trolley problem: he did broadcast Nazi propaganda to Americans, but that didn’t directly result in fatalities, while his Allied-intelligence actions saved lives (Allied lives anyway — Germans probably got killed by his actions, but that’s a war problem, not a trolley problem).
PS: I recently saw a cartoon about “Captain Trolley,” a brave superhero who can easily save any number of people from all kinds of deadly situations. The only catch is that he has to kill one of the bystanders first to activate his powers. Maybe someone at Marvel will tie him into an X-Persons story arc — I’m sure it would be as cheerful and uplifting as “Watchmen”!
With regard to the Mother Night issue, it’s true that Campbell does not directly kill people (at least as I recall – it’s been many years since I read it) but he has some reason to consider that his actions were not only in support of an evil and murderous regime, but that he did his job so well, and inhabited his character so convincingly, that he considers himself to have contributed to its overall evil. It’s not a straight trolley problem, but I think there’s a resemblance, at least, to the kind of moral dilemma which the trolley problem so bluntly and clumsily seems to point to. It does, at least, suggest that there’s a lot more involved in such judgments than arithmetic.
All that stuff about “trolley” problem is SO outdated.
We now have the “modern self-driving trolley problem”.
The solution: Shoot Elon Musk, TWICE.
(okay, maybe I’ve got it mixed up with ‘lifeboat with Hitler, Stalin, and a lawyer’, but the it’s similar)
M. Currie: Yes, Campbell does say this explicitly (paraphrasing from memory):
“I had tried to be merely ridiculous, but that is hard to do when people are so unwilling to laugh, so incapable of thought, so eager to believe and snarl and hate. So many people wanted to believe me.
“Say what you will about the sweet miracle of unquestioning faith, I consider the capacity for it terrifying and absolutely vile.”