It appears that the trolley problem is, as they say, having a moment. Possibly due to its newfound relevance to autonomous cars and other robots – a relevance that would have been entirely science-fictional when Philippa Foot formulated the modern version of the problem in 1967 – it is now making multiple appearances in popular culture. In that respect it is a notable counterpoint to the claim I made years ago that analytic philosophy doesn’t make for good visual media.
Two years ago I noted how the problem is the focus of an excellent episode of Michael Schur’s The Good Place. The Wikipedia entry on the trolley problem lists several other appearance from the past decade. Perhaps most entertainingly of all, the writers of the webcomic Cyanide and Happiness have released a hilarious party game (in the matching style of Apples To Apples or Superfight) called Trial By Trolley.
In each round of Trial By Trolley, one player plays the role of the trolley driver who must choose to sacrifice people on one of two different tracks. Unlike in the original problem, the trolley’s current direction is not specified; the driver must simply choose which group to kill. The other players are each on a team trying to convince the driver to kill the people on the other team’s track; they can put “innocents” (a cute dog, the Dalai Lama) on their own track and “guilty” (an evil ex-boyfriend, a Nazi) on the other track, and modify any of these (such as specifying a reincarnation). The game gets silly very quickly, and it is supposed to.
Such a game, like many pop-cultural manifestations, veers far away from the intent of the original problem. The idea of a trolley problem never referred to the one single case where one’s trolley is about to kill five people but one can flip a switch to kill a different one. Because that case by itself really isn’t a problem: nearly everyone would agree that one should kill the one to save the five. The problem comes when the trolley driver case is contrasted with a different case, like the transplant or the fat man – not necessarily involving a trolley at all – where most people would agree that one should not kill one to save five. That juxtaposition is the point of the problem: getting newcomers to philosophy to recognize that moral choices are indeed difficult, there is not an obvious principle to which they can be reduced.
But being in the spirit of the original trolley problem is clearly not the point of Trial By Trolley, nor should it be. The game tends to reduce the trolley problem to absurdity, and overall I think that is a fine thing. I think the trolley problem is a wonderful and vivid pedagogical tool when it is used as a pedagogical tool, as Michael Sandel did – posing the problem as a way to shock students out of their certainty and allow them to see moral complexity. I think it’s considerably less helpful to try to spend time actually answering the problem, as many analytic philosophers do (starting with Foot in her original article). The attempt to answer it is perhaps the most characteristic move of late-20th-century analytical ethics, and I think it is based on a flawed methodology. I don’t think it really works to try and discover moral principles through a method of ceteris paribus, trying to isolate variables by posing fanciful scenarios as “thought experiments”. Such approaches, I think, offer us little help with difficult decisions in the real world – as Chidi memorably shows us on the trolley-problem episode of The Good Place. In addressing concrete ethical decisions we likely learn more from case studies of real choices more directly comparable to ones we might someday make ourselves. Those help us cultivate the disposition, the virtue, to act well.
At the heart of the “thought experiment” approach is the idea of “moral intuition”. At their best, such experiments show us a problem with our preexisting beliefs (“intuitions”): we come to inquiry thinking it’s good to kill one to save five in the case of the trolley driver but not the case of the fat man, so that can’t be our principle. But again, such an exercise is valuable primarily in moving us out of an unreflective state to deeper inquiry. It’s not what that inquiry should consist of. Against some analytic philosophers, it is a terrible idea to treat “intuitions” as if they were raw scientific data we are trying to codify. Rather, we need to go deeper, think on the reasons that underlie those “intuitions” – in ways that are not limited to the sphere of “morality” narrowly conceived, but are connected to our broader views of a good life, of human flourishing, and even of human nature. Doing that allows us to answer a question that thought-experiment ethics rarely seems to ask, but that that ethics makes little sense without: why should we be moral? Why should we bother acting according to our “moral intuitions” in the first place?
At least, such an approach makes little sense for humans, who enter philosophical reflection with a wide variety of motivations that are not necessarily connected to “morality” in the sense analytic philosophers describe. The thought-experiment method could potentially be helpful in the new situation that has probably led to the trolley problem’s newfound cultural infamy: that is, programming robots or AIs. In that case, the agent has no motivations other than those the programmers give it (unless we count the “motivations” provided by natural laws like gravity). If we told it to, a robot could act entirely according to utilitarian or Kantian principles – whereas the human partisans of such principles not only do not and cannot live up to them, but should not try. The thought experiments of analytical ethics, designed to carve out a middle space between utilitarianism and Kantianism, may yet have some purchase for a robot.
Thought experiments are far less useful for us humans in our own decision-making, and it is in that context that Trial by Trolley‘s mockery lands well. The trolley problem’s thought-experiment method leads us to pose increasingly bizarre scenarios, like a diabolical machine designed to kill one child out of two. Such an approach is well suited to the light-hearted mockery that Trial By Trolley provides. Effectively, it takes the absurdity of analytical thought experiments and pushes them a step further, making them even more absurd (a pack of alt-right velociraptors). The game is not in the spirit of what the trolley problem was intended to be, but I think that in an important respect it is very much in the spirit of what the trolley problem actually is.