Trolley problem, again
Hi!
I just watched the TV show Widow's Bay (highly recommended). One episode explored the classic trolley problem and got my mind rolling over its fascinating nuances. I see this problem was already discussed on tildes, but hope a 2026 revisit is worthwhile.
The episode promoted not pulling the lever. It argued that we can't stop all the horrible things that happen in the world. But we can control our own moral status:
"The runaway trolley is life. The lever is me... if I pull that lever it's a choice and I'm choosing to kill that person. I could never do that."
and this lovely message:
"The world is violent and mercurial. It will have its way with you…We live in a perpetually burning building, and what we must save from it all the time is love."
As sweet as that is, I find that interpretation lacking. To me, the physical act of pulling the lever isn't the core moral signal. Not pulling it isn't a get-out-of-responsibility-free card. There's still a violent outcome. That outcome happened because you made the choice to allow it. While you're not responsible for the scenario, you're still responsible for the outcome because it was entirely on you what the result would be.
I think my interpretation matches the "utilitarian" argument, which also hints at "the ends justify the means." I'm not, more generally, willing to get behind "the ends justify the means". But there is important nuance here. Sometimes we must cause harm for greater good and inaction in hard circumstances is its own moral trap. So I'm not sure what I think.
Also enjoyed exploring the variants at https://neal.fun/absurd-trolley-problems/. How would you handle them?
I'm actually quite frustrated with the trolley problem lately, because it's a binary. And the real world doesn't have to be like that. But currently I'm watching people use the justification of causing the "least harm" to discriminate against minority groups, in the name of protecting the whole.
I think choosing not to pull the lever and intervene if you cant meaningfully do something is valid. But I'd really rather IRL we focus on moving people off the tracks, changing the route, and solving the problems instead.
The Kobayashi Maru in effect. Don't choose from the least bad options, MacGyver a third way.
Yes kind of, but also real world situations are rarely actually a trolley problem where the options are pull or don't, vs., perhaps much more viscerally, shoot one group of people or another. It's direct, not distanced, it's an active choice to hurt someone.
Or currently, an active choice to hurt someone and violate the law to hope that maybe the federal government won't target us for not hurting those people. But you know, greater good right? Better pull that lever because the risk is just so great. Only costs your soul.
I'm particularly pessimistic towards people caving to this sort of framing atm, so this is certainly influencing my POV. If you view this lever as an option then the temptation to pull it will always be there.
There's some merit to exploring the problem because sometimes the danger could present itself in a binary way. The self driving car is perhaps the most salient example as it might be important to give the machine an answer when it's facing a "go straight and kill one, or swerve and kill more".
That said, I agree that for the most part it's an exercise in futility. @Zorind already mentioned how it often has only the utilitarian input play a role rather than additional societal, ethical, or otherwise important factors that may influence the answer.
Nevertheless, it can be interesting to see how one answers to see how they think and at least spark a discussion around it where you explore the topic and add additional parameters.
Yes it's a lovely thought experiment and I'm not mad at others engaging in it. Just sharing my thoughts and why I'm less keen on it than others
Though I'd say self driving cars also have "stop" and shouldn't be going that fast
With self driving cars, there's much more uncertainty, not to mention fast speeds. The self driving car should do what we teach humans to do in defensive driving: if something whack is happening, press the brakes as hard as possible.
What should the self driving car do if even at max brakes it will hit 10 orphans and 20 puppies? It should still apply the brakes and kill the orphans and puppies. Because in real life we don't know that a) if swerving will make things worse, fatality wise and b) that the breaks won't stop in time.
Things on the road go whack, just slow down and figure it out from there.
Somewhat of an aside: if we ever reached the point where an autonomous car is sophisticated enough to predict these outcomes and make moral choices about its actions, we will have come so far relative to the current technology that it would be nearly unrecognizable. Like the relationship between a middle ages wagons and the newest luxury sports car.
Asimov's Naked Sun, Robots of Dawn, and Robots and Empire explore the nature of the problem when the robots become this sophisticated pretty well.
Excellent insight, thank you. Far better to avoid facing the trolley problem in the first place.
I guess it's a metaphor for trying, desperate times. We always have and always will face them. But we do best when we reduce the need for them.
I'm happy you bring this up. Even though the exercise of the thought experiment is valid, it misses a bigger point. A lot of these options are guided by the current moral understanding of society (for lack of a better description). Even the options presented are probably biased.
My belief is that the process arriving to a choice is much more important, and should be the locus of the discussion. In the end, this how much of reality works: there is no clean, moral superior option (or if there is, you can keep refining the thought experiment until there is, such as the fat man digression). What matters is how, as a society, we arrive at a common understanding what is morally right at this moment in time. And in my opinion that means talking, talking, and more talking - and leaning into the conflict. There will be a lot of friction and pain in trying to find an acceptable compromise, but that's the way forward.
So maybe a cop out, but instead of deciding to pull the lever, I would sit down and discuss. No matter that the decision had to be made in a split second, the second decision will surely be better for it, and maybe come with out-of-the-box solutions such as, as you say, moving people off the tracks.
I recommend the original paper the thought experiment comes from, it's a rather more interesting read to me than the sort of "moral personality test" way it tends to be used. The initial context was the author presenting a series of similar situations where she felt most people would intuitively agree one side is the right choice, yet where the consensus would be different for each question despite them being mathematically equivalent, and trying to puzzle out what the underlying principles that make them distinct are to then examine the question of abortion in various circumstances.
Somehow I didn't know this was about abortion and the Doctrine of Double Effect - under which aborting an ectopic pregnancy is prohibited but removing the fallopian tube that is causing the risk to the woman's health and incidentally aborting the pregnancy in the process is fine.
I have feelings about this.
If I understand correctly the paper actually concludes the doctrine is mistaken and lands on an alternative standard the author believes fits what she considers moral intuition more consistently, but yes that's apparently the topic it originally comes from. (Though I was more sharing it because of the questions than because of the answers.)
Sure I just didn't know that was the origin!
I'm rather familiar with the doctrine. It's an interesting read.
The Trolley Problem kind of seems like an intentional straw-man argument with real world corollaries, especially in medicine. Could you amputate someone's limb without anesthesia to save their life? Could you perform surgery for the same reasons? All of these are considered altruistic choices by society as a whole, but they also impact our "moral status" in some way. It absolutely would change me on a fundamental mental level to be able to cut open a living being open. Choosing to bring myself to do that is much more impactful to me than maintaining my normal state of not doing that. I think that's the difference between the "choice" of pulling the lever in the Trolley Problem.
Specific to Widow's Bay, I recall there being a significant impact on the situation on the choice being presented by an outside and malevolent conscious "power". Meaning something presented the situation and was already making the decision to cause the deaths. I am curious if any studies have ever tested the Trolley problem between situations where an outside power explicitly/implicitly causes the situation, vs the situation being caused by the person making the decision. I personally can see some implications for how belief in a higher power may then impact our decision making processes (or not).
Agreed. It’s a thought experiment, and should really only ever be treated as a thought experiment. In any real world scenario, there are too many situation-specific factors that make it basically useless to refer to IMO.
I find your medical examples interesting, wouldn't you as a physician be obligated to put the person's life above your personal feelings?
Also that's exactly what thousands of doctors were doing before the advent of anesthesia, it's not a hypothetical scenario. They were expected to hack away at living screaming human beings to save their life.
It was always "fun" to me to see the different answers people would give in my undergraduate philosophy classes where the trolley problem was brought up.
We did the standard "go to this side of the classroom for pull the lever, go to the other side of the classroom for not pull the lever". And it was interesting to me how many people would swap their decision when it changed from the standard trolley problem to the "fat man" version of the trolley problem.
This specific argument, that apparently someone makes in the show you're talking about
always annoyed me. Because not doing something is still a choice.
The argument that "not pulling the lever" is not making a choice has never made sense to me.
Like, I think there is a valid argument that someone could make for not pulling the lever (e.g. direct vs indirect responsibility), but saying that it isn't making a choice doesn't hold water to me.
But also, in that class, the first time we were going over the trolley problem is the first time I ever heard the "medical transplant" version of the trolley problem. And I ended up as basically the only person that ended up going "in for a penny, in for a pound" and stayed in the "kill one to save five" side of the room. (I don't think I hold that opinion anymore, given resultant discussions around societal and medical ethical codes...but when it is presented with the same utilitarian math as the trolley problem as the only "inputs" to the question...)
I agree.
One interesting twist is that in the real world we rarely know outcomes with certainty before we make our choices.
How does doubt affect your choices? What if pulling the lever kills everyone on both tracks? What if emergency responders are en route and have a good chance of arriving on time and finding a way to save everyone? What if pulling the lever might derail the trolley and send it straight toward you? What about the passengers on the trolley? Which outcome is more dangerous for them?
All the doubt we have in the real world would make me far less confident to bias to action.
Yeah, there is no way to actually “apply” the trolley problem in a real-world scenario.
There are way too many factors in the real world that basically entirely mitigate the framing.
Voting in a two-party system seems to generate quite similar arguments. Even just the choice to vote - not going any further with volunteering or donations or anything - people will object to voting at all because they slightly dislike one choice and vehemently dislike the other. They commonly express the "pulling a (voting) lever is making me responsible for the choice, skipping voting leaves me outside the moral responsibility" language that is generally part of trolley problem discussions.
To add to the criticism of the Trolley Problem: It really annoys me that it's giving certain, deterministic, known outcomes, while real life is anything but that.
The moral intuition and decision making of humans has evolved over millennia based on real life situations, not on some ridiculously fake made up scenario that will never happen. The "paradoxes" you see are not showing inconsistency in human morals, it's just showing the thought experiment is really fucking stupid and wasn't designed well.
To give exact example: The entire gotcha of the Fat Man corollary relies on you believing that throwing a fat person in front of a trolley would somehow stop the trolley. That's really fucking stupid, it's just not how physics work. Sure, the experimenter could tell you it would stop it 100%, and you can agree to accept that as a given for the thought experiment... But deep down your subconscious will be screaming at you that there's a big chance you'll just be killing an extra person, because you evolved intuition on how physics and inertia works.
Also, deeper structures in your brain (instinct for self-preservation) knows that if you try killing someone with your hands, they might fight back and hurt you too. Again, the experimenter can assure you the fat person won't grab you and kill you by dragging you with them. But your brain didn't evolve in magical "we are 100% certain" made-up world, we evolved in actual real-life situations, where pushing people has real risks.
And I see zero implementation for any of this nonsense in real life. Everyone keeps bringing up self-driving cars for a reason, it's literally the only actual example anyone can think of. And it's completely useless in this example too - if your car is about to crash, you don't spend 816 miliseconds contacting Elon Musk's server to ask whether the life of 2 shoemakers is more valuable than 1 barista, 1 cook and a half-eaten sandwich. Instead, you do the sensible thing and apply the brakes.
The whole Trolley Problem discussion reminds me of classic philosophy. I understand (and admire) the desire to understand how something works and to try figuring it out by thinking about it... But at some point the thinking becomes so esotheric and detached from reality (and frankly up its own ass), it ends up having very little value in the real world. Hence why progress in our understanding was so slow until actual empirical science showed up.
Respectfully, I think if you understood more about what the trolley problem is and why it exists, you wouldn't have such a problem with it. It's not trying to be a real-world scenario -- it's trying to help us clarify our moral intuitions.
Here's a separate thought experiment, to try to clear away the baggage associated with the trolley problem. Peter Singer asked whether you would have a duty to save a drowning child, if you walked by the pond that the child was drowning in. Almost everyone responds yes, naturally. He then asked whether you would still have the responsibility even if it had some small material cost to you - your wallet was ruined, say. Almost everyone still responds yes. And if you saved one child, but the next day saw another drowning child? After all of this, Singer asks what ethical difference there is between ruining a wallet to save a child drowning right in front of you and a child dying of malaria on the other side of the world, who could be saved if you donated some money.
To say "but Peter Singer, I never run across drowning children, much less multiple" is just plain silly. That's not the point.
Similarly, the trolley problem isn't trying to teach us what to do when we see runaway trains. It's trying to help us clarify our moral intuitions. Most people answer very differently when the choice is to pull a lever versus to push someone onto the tracks. Why? The answer is important, and to say "a person wouldn't really stop a runaway train" is missing the point. The trolley problem came into being specifically to try to understand how doctors were making ethical decisions in the case of abortion. The simplicity and rigid nature of the thought experiment are features, not bugs, because it forces us to discard all of the "well, but..." and focus solely on the ethics.
Mostly an aside, but on the topic of the trolley problem, I liked this comic from a while back satirizing it a tad. I don't really have anything to add to the discussion, however 😅
For folks who avoid X it's posted on his Instagram
And reposted on tumblr here
Aah, thank you! I couldn't find it on Alex's bluesky account, xcancel.com doesn't appear to exist anymore, and I figured that people who dislike Twitter/X would also be opposed to instagram. Missed the tumblr repost though, which would've been ideal.
No worries, it was just the easiest option without an immediate LOGIN NOW IN THE APP TO SEE THIS pop up.
Tumblr asks eventually but not usually until you scroll
This page lists some Nitter instances which are still active, and this page has uptime info for a few of them. (Nitter being the software underlying XCancel.)
Inform any deities who may be listening and orchestrating the problem that you will always go with "do nothing" and in fact won't even be observing the problems. So whatever happens is on them.
I killed 57 people in the trolley game. It says that I solved philosophy.
If you follow Kant, he would say to not pull the lever since you are actively causing someone's death which is bad. To him, it's not about the outcome, but your goodwill. You can't control your goodwill's outcome, but you should still try to have a good will. If you had good will and it results in something bad, that's fine. (Might be butchering Kant a bit, I didn't learn about him with English words.)
Edit: And by goodwill, he doesn't mean something subjective, but objective. Killing someone is bad so you shouldn't do something that would cause someone to die.
Throw in a lot of sarcasm and make the question: Why not derail the trolley and dave all the people on tracks?
New problem: The people on the trolley won't survive.
Jokes aside - this really can't be put in the real world. I suppose instinct or paralyzation by the situation would overtake any reasoning.