On Trolley Problems (updated)
Robert Paul Wolff, at his blog The Philosopher’s Stone, objects to the use of trolley problems and other stylized thought experiments in which various complications are waved aside. “I am quite convinced that these sorts of thought experiments are nonsense,” he says. Wisely, he adds: “but it is not so easy to say why.”
I think that trolley problems and the like are excellent philosophical tools. They are often defended as means by which to isolate the relevant moral variable to help us assess how it does and should operate in our thinking, and I agree with this defense entirely. Like any tools, though, these thought experiments are built for specific tasks. Just as it would be silly to criticize the use of hammer because one couldn’t build a whole house using it alone, likewise, it is silly to criticize the trolley problems because one couldn’t build a whole moral theory out of them alone. Yet I hear variants of that criticism often.
This does not seem to be Wolff’s objection. His objection is that we often can’t isolate the relevant moral variable, and that we haven’t “earned” the assumption that we can. He writes:
The process of abstraction on which the process of formal argument depends presupposes that we can precisely and accurately distinguish the several elements of a real world action in such a fashion as will permit us to detach them from one another in constructing hypotheticals.
It… assumes that we can use terms like “intention” as though they were the names of simple identifiable particulars that can be safely abstracted from the complex context in which they are normally embedded… all those philosophers talking about trolley cars suppose that we can identify… intentions in abstraction from the knowledge and circumstances in which they occur, so that we can judge the intentions independently of the actual consequences or the knowledge that actors should have had, or — and this is quite important — independently of the institutional, historical, and bureaucratic context of the actions. And that, I suggest, is wrong.
I’m not sure this is correct. It isn’t that those of us who use such examples think “we can identify intentions in abstraction from the knowledge and circumstances in which they occur,” but rather that, by considering a simplified case, we make make it much easier to talk about what an agent knows and what the relevant circumstances are. But leaving this aside, Wolff thinks that assuming away the difficulty he identifies is kind of a cheat:
Whenever one of those factors is cited by someone objecting to the hypothetical example, the person posing the example will wave a hand and say “I will assume that consideration is not operative here.” Let me now say something peculiar, but really quite important. The dismissal of the objection with a wave of the hand is too easy. Not wrong, exactly, but too easy. The right to set aside that consideration has not been earned, by hard study, by experience, by history, or by personal sacrifice. Philosophy done that way requires no heavy lifting, so to speak. You can be as ignorant as a new born babe and yet, with a fertile imagination, you think to brush aside the hard-won wisdom of those who have actually been trolley car conductors or military field commanders or operating room nurses or heavy equipment operators by abstracting from it.
I am sympathetic with the idea that some philosophical positions have to be earned. What comes to mind is the lazy relativism that some students enter an introductory moral philosophy class with. Relativism could be true, but these students certainly don’t know enough to know that. They haven’t “earned” their relativism.
But I wonder if this criticism really applies to philosophers who use and know the limitations of the trolley cases. What prompted Wolff to write his post was a move by Sam Harris in his recent exchange with Noam Chomsky. Maybe (I have no idea) Harris is willing to go directly from “you should pull the lever in bystander” to some judgment of a real agent in the real world, without any additional steps or considerations, but if so, that seems more an indictment of Harris than of the thought experiments. Most philosophers, it seems to me, know that a variety of epistemological obstacles lie on the path from thought experiment to real-world recommendation.
As for the “hard-won wisdom of… trolley car conductors,” I will relate one story. A couple of years ago I found myself on a trolley in Memphis, Tennessee. This was a rare opportunity to get an expert opinion, readers, and I was not going to waste it. The trolley stopped and I approached the driver. I said: “Imagine the trolley car is out of control, speeding down the hill towards five people tied to the track and who will die when hit by it. The only options you have are to do nothing and let the trolley run its course, or turn the trolley onto a side track before it hits the five. Unfortunately, there is one person tied to the side track who you would kill by sending the trolley that way. What should you do?”
She answered, “That would never happen. There is an automatic fail-proof brake.”
“But suppose there wasn’t,” I said.
Eyeing me suspiciously, she said, “Why would I do that?” And with that she exited the parked trolley car.
Wise move, trolley driver. Wise move.
(image: photo from “Trolley report: MATA not following safety standards,” WREG News 3)
UPDATE (5/18/15): Professor Wolff elaborates on his critique in a new post.
You’re “…not sure this [ability to abstract intent from background knowledge] is correct”? Your objection, (an appeal to the obvious utility of toy models) doesn’t and shouldn’t provide certainty in this way. It seems there’s a much bigger problem for Wolff’s claim however, a problem often tipped off when someone uses an unjustified “the” as in “the several elements of a real world action”.
If concepts obtain meaning within a Quinian web of beliefs about presuppositions, and each presupposition’s meaning is obtained the same way, we can never (and shouldn’t) try to detach actions as he suggests.
The fact that neither Wolff nor any of us can detach any of our terms, words, and concepts we use cannot stop us from making arguments, nor should it. Wolff’s point seems to be a self-refuting idea / fallacy of the stolen concept. It is true and important that terms like “intention” cannot be abstracted with *complete* safety…but safety is relative to background information…so let’s get on with the business of life.
Similarly, Wolff’s objection to setting aside a potentially important consideration hardly seems valid. “1 + 1 = 2” ignores the fact that 2 things existing is different than taking one thing and another thing and performing an action over time and obtaining a result. These are very different things. Have mathematicians “earned, by hard study, by experience, by history, or by personal sacrifice” what Wolff would considers “the right to set aside that consideration”? (He really likes that “the” article…)Report
Wolff’s post is a mess, and Justin’s diagnosis seems right.
The problem with Sam Harris isn’t that he’s wheeling in “hypotheticals” to make a moral point—everybody does this all the time, and it’s a perfectly ordinary part of moral thought. (I might add that there’s plenty of this going on in The Republic and The Leviathan, some RPW-favorites.)
The real problem is that Harris is trying to evaluate the actions of a nation-state by looking at the particular intentions of its leaders, when in fact it doesn’t matter (to the relevant debate) how noble Bill Clinton’s aims were. What matters are the workings of the institutions he belongs to. You can’t understand or evaluate these complicated group actors—the military, the executive branch—just by looking into their members’ hearts. You have to consider history, bureaucratic structure, culture, and so on.
Great. But why should anyone expect the trolleyologists to disagree? More to the point, what does any of this have to do with the use of thought experiments in ethics? We can all agree that, in certain cases, certain thought experiments are irrelevant. We should also agree that certain thought experiments distort the phenomenon they’re trying to capture.
But this is no reason to think that imaginary cases are in themselves irresponsible, or that using them amounts to quixotic logic-chopping. We just have to be careful with them, as we have to be careful with every other useful tool for doing philosophy. (There are simplistic metaphors, like the “nanny state”—should we roll our eyes any time a philosopher uses a metaphor?)
If there’s an objection here, it’s got to be a specific one: that thought experiments *as they’re actually done* are too simplistic. And this is more or less what Wolff says, along with the claim that the simplifying is for the sake of “formal” arguments. Neither of these rings true when I read Part IV of Reasons and Persons, Thomson’s classic papers, or Ross’s imagined worlds.
I can’t really argue against Wolff, since he doesn’t give any reasons to think that thought experiments as they’re actually done are overly simple. But given the way he introduces the “trolley car hypothetical” (he cites the “Fat Man” variant instead of Foot/Thomson’s original “Trolley Driver”), I get the sense that he hasn’t given the thought experimenters a fair shake, or read their work with any seriousness. My guess is that if he did he wouldn’t be comparing them to Sam Harris.
(I should add: the “Trolley Problem” is not itself a thought experiment. It’s the problem of explaining why it’s okay for the bystander to divert the trolley from the five to the one, but wrong for the doctor to kill one patient to save five others.)Report
@Daniel: Excellent response – Thanks!Report
If only we had some set of well developed tools to help isolate the variables impacting our judgments about such cases..Report
You Panglossian! Whatever shall you dream up next? Telephones that don’t need cords? Medicines that prevent infection?Report
Imagine that philosophy went forward without thought experiments. Wait a minute. . .Report
I wonder if there is a worry somewhere in the vicinity that, since this is moral philosophy, the ‘simplification’ in question is not morally neutral. In other words, when we invite students to discuss in the usual kind of informal atmosphere of a philosophy tutorial their choosing to snuff out the life of some unknown person, there can be something rather glib about such discussions that is troublingReport
It’s like sex (isn’t everything?): Definitely there is something that makes a glib attitude troubling about significantly new experiences such as losing one’s virginity perhaps, but after a few thousand experiences…not so much. Perspectives change with wisdom, skill, & familiarity.
On the topic of troubling descriptions, I’d recommend the first couple chapters of: The Secret Language of Doctors. http://www.amazon.com/The-Secret-Language-Of-Doctors/dp/1443416010
My favorite: “Code Hollywood” which means to make a big show of attempting to resuscitate a patient, but failing. This is when the team knows very well the deceased needs to NOT be resuscitated.
It’s a disturbing bit of slang, but only at first. The author walks readers though his becoming comfortable with language that initially horrified him.Report
I don’t like the trolley problems because I think they’re faux-empirical. The trolley problems themselves generate the answers, but analyses tend to assume that the answers are functions only of moral intuitions or some similar construct.
In his follow up post, Wolff offers some comments on intuitions that I largely agree with. Intuitions are grounded, situated things. It’s not obvious that humans are psychologically capable of applying them in abstract conditions; it’s not obvious that they do apply in abstract conditions.Report
Some of you may want to check out my “The Curious Case of Pushing Buttons. What Can We Learn from Moral Judgments about Unrealistic Cases?”.
Justin, further to your asking the trolley-car driver whether she would divert the trolley away from the five to the one, in the UK train signallers are explicitly taught in training that run-away trains should be diverted to the centres of least population.Report