Guy Crain (Rose State College), “Three Shortcomings of the Trolley Method of Moral Philosophy”

Journal of Ethics and Social Philosophy, 2023

By Guy Crain

Trolley problems feature ubiquitously throughout moral philosophy. Not all trolley problems involve trolleys. Trolley problems can involve doctors who need organs and a lonely healthy drifter; ticking timebombs, terrorists, and torture; and even vampiric violinists. Philosophers use such cases to pump intuitions and draw conclusions about moral claims and theories. In other words, trolley problems are the central features of the method used by philosophers to draw conclusions about ethical theory and moral practice.

That method, though, is flawed. The fact that the method is flawed is no secret. Philosophers have pointed out a number of problems with it such as failing to pay attention to important systemic issues in ethical matters, failing to test important features in moral agents like resourcefulness, failing to be action-guiding, including agents with abilities unlike any actual human, and so on.

In my recent paper, I argue that there are, however, three important flaws yet to be identified in the method. First, trolley methods focus almost exclusively on high-stakes moral decisions: torture, deaths, bombs, etc. But the vast majority of moral life isn’t like that. Almost no one has to decide whether to divert a train to doom someone to death let alone whether to push someone off a bridge to get the job done. Most people spend most of their ethical resources wondering about decisions involving raising kids, interacting with neighbors and coworkers, eating healthily, whether a small behavior is rude, who to marry, whether to round up for charity at the card reader, bragging, gossiping, etc. But there aren’t trolley problems that deal with such things. And, arguably, there can’t be. What would a trolley problem about whether to write a thank you card even look like? That means the trolley method, by design, can’t really shed any light on most of moral life for most people. Instead, the trolley method makes it seem as though hot button issues are the real stuff of ethical analysis. This makes “ethics” seem like the kind of thing that matters to most people only very rarely or only to elite persons in positions of power.

Second, trolley problems feature either scantily-described third parties (such as philosophy’s infamous “Smith”) or us as second-person respondents (should you pull the lever?) being poised to make the relevant decision in the scenario. In the case of third parties, trolley problems operate as though there is an abstract agent whose features are operationally unimportant to the results of the thought experiment. In the case of second-person respondents, it is as though a trolley problem could be posed to anyone. In this way, the trolley method seems very inclusive and universal—the third-party agent has been anonymized or any of us could play the role of “you” in the scenario. The problem though is that, in fact, the method is not as inclusive or as universal as it seems. In the case of third-party agents, people do unwittingly supply a list of features to agents when envisioning the scenario; and those unwittingly supplied details can be problematically exclusionary (consider the infamous father-son-car crash-surgeon riddle). In the case of second-person respondents, trolley problems have at best been limited to WEIRD respondents (Wester, Educated, Industrial, Rich, Democratic) and at worst have been limited to white, mostly male academics. Thus, while the trolley method appears to be a means of doing ethics for “everyone,” it really only serves a fairly exclusive club and likes serves to reinforce inclusivity problems the club has had for quite some time.

Third, trolley problems present a fairly univocal model of what ethical decision-making is like. Trolley problems present tiny temporal snapshots. The information-gathering, awareness, deliberation, and decision all happen inside that small window. The decision is presented as though there is no significant historical influence of the person’s past or influence from environmental features. Granted, some decision-making is phenomenologically like this. But most decision-making is far more automatic and the result of a person’s psychological and environmental momentum. Consider the act of driving home from work. It would be weird to characterize the constant decisions make during that activity in the way decisions are characterized in trolley problems. Even saying something like “I decided to turn left at that light” connotes far more mental presence than there likely was. Loads of social science suggests that a lot of our ethical decisions are made in the same autopilot fashion as our driving behaviors (see my paper for a list of examples); that is, our current best evidence suggests that most of our behavior is not calculated or deliberative but conditioned and corralled. If the trolley method uses a conception of ethical decision-making unlike that used in most people’s ethical behavior, then it is a poor tool for analyzing most ethical behavior.  

I don’t conclude that trolley problems are useless. But they are at best blunt instruments that should not be expected to do the work of sharper tools.