Suppose you are a Consequentialist faced with a variation of the trolley car problem. Your options are to save five people with a 20% likelihood or one person with a 100% likelihood. Which option, if either, is better from a Consequentialist's perspective?

Now, what if you are uncertain about the 20% likelihood to save five people--the actual probability could be more or less. You are still confident you can save the single person 100% of the time. Should the reasoning change?

  • 2
    Consequentialism is a very wide umbrella for a variety of approaches that would give different answers or judge these situations as underdefined for any answer (we are told nothing about the "worth" of these people, their relation to the chooser, the nature of certainty/uncertainty, etc.). In rule utilitarianism these individual percentages are entirely moot, as decisions are made based on rules designed to work "on average". In act utilitarianism one would compute the expected utility of each action given the odds and pick one with the highest value, randomly if several have the same value.
    – Conifold
    Sep 12, 2020 at 7:54
  • ‘One may call these uncertainties objective, in that they are simply a consequence of the fact that we describe the experiment in terms of classical physics; they do not depend in detail on the observer. One may call them subjective, in that they reflect our incomplete knowledge of the world.’ (Heisenberg, 1958, pp. 53–54.)google.com/url?sa=t&source=web&rct=j&url=http://…
    – user47436
    Sep 12, 2020 at 7:59
  • Trolley problems help us to identify that we in fact have a range of attitudes to risks & harms (given means justifying ends or not). That applies even just among consequentialists, as regards decision making before consequences happen. There is a section on your case here: en.wikipedia.org/wiki/Trolley_problem#The_fat_man
    – CriglCragl
    Aug 28, 2022 at 8:44
  • I would suggest Bayesian reasoning, which mixes experiences & models to form priors for likelihood of events & then adjusts those priors continuously, is of widespread appeal among consequentialists plato.stanford.edu/entries/epistemology-bayesian It has known problems around rare events, or new circumstances where there is limited information to build models.
    – CriglCragl
    Aug 28, 2022 at 12:10

2 Answers 2


The point of trolley car problems is to try to remove any uncertainty from the process of making moral choices, because uncertainty makes the entire utilitarian consequence evaluation process unmanageable. Pragmatically, there is massive uncertainty in all moral choices, even in trolley problems.

Examples --

  1. there are people on the two lines. Are either group certain to be killed? Most people on rail lines know there can be rail traffic, and are listening/looking for rail hazards. And plan on getting out of the way on their own. Maybe the people on the lines are deaf, or stoned, or injured/pinned? If this is specified in the "problem", it is basically NEVER going to be knowledge one actually has in such a crisis.

  2. Are there people ON the trolley? If so, those people should probably be your greatest concern, per 1 above. Do you even know? Is there a curve in one of the lines, such that the trolley would derail and roll at speed? Are there other people who might be endangered by the derailing on either line? Is one only a siding, and is the siding a place the trolley could derail into a field, and come to a stop, or instead crash into a building, or a solid barrier? Should possible people in buildings be your even greater concern than either the passengers or the people on the tracks? Is one of the lines a long stretch of level that the trolley can come to a stop safely on? Can you even SEE what the two lines do, beyond where the people on them are?

  3. What are the properties of the switch? Is is on the slope, such that the trolley is at a lower speed at the switch? Is the switch analog, such that one can try to set it between the two lines? If so, would that derail the trolley, or just POSSIBLY derail it? And would that be preferable to letting it run down one of the lines?

  4. Is there a way to stop the trolley? A "fat man on a bridge" is in some of these problems, to tempt one to kill him by pushing him in the way of the trolley. Is the s feasible? a) CAN one push a much heavier person onto the tracks? What is the probability of success of this? b) CAN a person's body stop a runaway trolley? My strong suspicion on this is "no", people are soft and small compared to what is needed to stop a trolley, no matter how much fat is in the "fat person", but that is added to the uncertainties. c) are there other more likely options to stop the trolley, such as stacked railway ties, or a metal bar near the tracks? If so, what tare THEIR odds of stopping the trolley, and of your being able to get them on the tracks quickly enough?

  5. People are not of identical worth, as well as not being in identical risk. A labor gang of murderers chained together overseen by a sexually abusive deaf guard on one line, vs. a future Mother Teresa who is incapacitated by an ongoing mystic vision on the other, can tremendously change the calculus. IF you only knew.

In a crisis, one has to act quickly, and much of the important information and probabilities needed to arrive at a morally optimal outcome are not available to the actor. We have a chance to reflect when studying moral issues academically, and it should be the purpose of these studies, to prepare us for these crises. I consider trolley problems to be the ANTITHESIS of useful preparation for a crisis. The whole POINT of trolley problems is to try to force a decision between bad outcomes, and inure one to deliberately sacrificing others lives. They also are premised on no uncertainty for any critical issue. Anyone who did their moral training on trolley problems, will START by looking to sacrifice people, and will be befuddled by how to deal with real world uncertainties.

In a real crisis, one will have massive uncertainties. The actions one should take, AS A CONSEQIENTIALIST, should be based on trying to find a possible solution which does no harm to ANYONE. Applied to a trolley problem, and with the multiple uncertainties listed above, I would almost certainly try to derail the trolley at the switch, where it can derail in a wide railbed and probably not kill its passengers or the people on the track, or bystanders near the two rail lines it would careen down out of sight if it went down either. An analog switch in a middle position is the first thing I would look for. If that isn't possible, I would recruit the fat person to help me try to wedge bars or railroad ties into the gaps of the rail intersection, to try to make the wheels jump off the track. I would focus on the intersection, as straight track has little opportunity to wedge anything, and my object is likely to just be pushed out of the way. Then try to get far enough away from the track that we are not crushed by the trolley ourselves.

This recommendation is an example of how to apply uncertainty to consequentialist ethics. And try to salvage "trolley problems" as a moral training device.


Iirc the typical immediate solution is to divide good and duty, so that utilitarians can say that a choice can be descriptively best without being what we choose based on imperfect/incomplete application of the utility principle. Whether this defeats the point of this utilitarianism is another question, concerning for instance the strength of the relationship between imperative and deontic syntax. (Hare has a model of a quite strict imperative relation, in service to his attempt to reconcile utilitarianism and Kant, fwiw.)

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .