thought experiments: good and bad

Philosophers are famous for thought experiments. You concoct a hypothetical situation in order to test a claim. Thought experiments are crucial because they force you think clearly about the limits of your claim. However, in practice, thought experiments leave people cold. They seem contrived and counter-intuitive. This post is about one way to help you evaluate whether a thought experiment is useful. I make no claims to originality, only usefulness.

Ethics motivates this discussion because ethicists are constantly cooking up thought experiments to test the limits of ethical theories. Let’s start with two examples:

  • Trolley car problems (see the wiki for a history): In this situation, there is a runaway car. It is about to run over five people. If you pull a switch, it will run over one person. People die either way. Should you pull the switch?
  • Wilt Chamberlain (from Nozick): Imagine that you live in a society with 100 people. Each person pays $1 to see Wilt Chamberlain play basketball. Inequality is increased as Wilt becomes $100 wealthier and every one becomes $1 less wealthy. Is this a just state of affairs?

Independent of my ethical priors, I have always found trolley car problems to be bizarre and Wilt Chamberlain to be intuitive. It was only recently that I was able to articulate the difference. Trolley car problems, in my view, are rare, artificially constructed situations. Wilt Chamberlain is a crisp form of a common situation.

If you buy this comparison, then it points to an important distinction among thought experiments. Some thought experiments are about creating unusual situations that place ethical intuitions in conflict. The basic trolley problem forces the reader to decide between a utilitarian theory of ethics and other theories. In contrast, the Wilt Chamberlain thought experiment is about making a common situation extremely clear. Every day, inequality increases because some people are luckier, smarter, or work harder.

Then, the question becomes – is the distinction meaningful? Are “trolley cars” better for thinking about ethics than “Wilt Chamberlains?” I submit that the answer is yes. Why? Pragmatic reasons. It is a very high bar to demand that a theory properly resolve all possible situations. I suspect that no ethical theory could be supported with such a standard. That’s probably why philosophers like “trolley cars.” It’s fun to cook up examples that are unrealistic or smuggle in unusual assumptions and make ethical claims come crashing down.

In contrast, an ethical theory can be satisfactory if it provides reasonable guidance over a broad range of every day concerns. Inequality is around us every day. Thus, an answer to the Wilt Chamberlain example will move you towards ethical questions that matter, like how much people should be taxed. In other word, you should move from axiomatic reasoning about ethics (take Rule X and see how far it goes) and instead use a statistical approach to ethics (does Rule X help me in an average case, or wide range of cases).

“Trolley car” examples are also bad from a sociological perspective because they tend to strip away all institutional context and draw your eye from bigger issues. For example, ethicists have spent a lot of time arguing about whether you need to save a drowning child. That’s a situation that is exceptionally rare. About 10 people (of all ages) drown per day compared to the roughly 6,700 who die each day, mainly from old age, cardiac disease, and cancer. And even among people who drown, the reaction most people have is to save the person, not stand around and do nothing.

Ethicists may argue that an obligation to save a drowning child, which may exist, automatically entails the obligation to save people in other situations, which was Singer’s point. That ignores the fact that when problems arise, people usually start coming up with institutions to deal with the problem that are better at fixing the problem than you diving into charity work (as Singer suggests).

For example, we can invent life guards to prevent drowning at swimming pools. People can fence off dangerous swimming areas. The Coast Guard can patrol the water ways. More broadly, there are lots of institutions that address poverty in various ways that are better than becoming a crusading philanthropist. Economic growth and property rights have pulled nations out of poverty. One can also argue about the effectiveness of state benefits, like poverty insurance, guaranteed income, income redistribution, or tax breaks. In other words, the “drowning child” example is a horrible way to think about social policy because it is an uncommon example that strips away all that we know about improving people’s live.

Instead of testing ethical theories with bizarre examples, I suggest you do the following instead. Come up with a list of common ethical issues. Should I pay my taxes? Should I cheat on this test? Should I buy goods from unethical people? Should I cheat on my spouse? Should I give to my local charity? Then, take your ethical theory and make sure it actually comes up with reasonable answers to a wide range of actual concerns that most people will encounter. You can do this by directly applying the theory or with “Wilt Chamberlain” examples that amplify common situations, not goofy situations. Only after the theory is robust enough to deal with every day life, then are you allowed to move to drowning children and trolley cars.

50+ chapters of grad skool advice goodness: Grad Skool Rulz ($2!!!!)/From Black Power/Party in the Street


Written by fabiorojas

June 13, 2016 at 12:14 am

Posted in ethics, fabio, uncategorized

8 Responses

Subscribe to comments with RSS.

  1. So, here’s a quick reaction based on a quick read. A thought experiment is useful if the variables it strips away from the situations it’s supposed to model/be analogous to are morally irrelevant ones, not useful if it the variables it strips away are morally relevant. You think the drowning child thought experiment is not very useful because there are lots of morally relevant differences between it and the situation it’s supposed to model (relations between wealthy persons in the developed world and poor people in the developing world). And you think that Wilt Chamberlain is useful because you think there aren’t that many morally relevant differences between the thought experiment and the situation it’s supposed to model (free-market exchange). As evidence for my claim, I’d suggest that socialists who disagree with you about the utility of the WC example do so mostly because they think there are too many important differences between what goes on there and what goes on in real world capitalism.
    Is that a fair reinterpretation of your basic point?


    Matt Zwolinski

    June 13, 2016 at 1:44 am

  2. 1. Yes, I think that is almost fair summary of my point. What is missing is not whether something more morally ir/relevant, but whether it is a common feature of the things that people need to make decisions about. So it is: thought experiment is good if it strips away irrelevant issues AND the thought experiment illustartes/exemplifies a common decision/situation.

    2. WC and socialists: If you use the Rojas Criteria for the value of thought experiments (TM), then I would tell the socialist the following:

    – I can provide a wide range of normal everyday situations that are extremely similar to WC
    – I’d ask the socialist to discuss the *common* and morally relevant issues that the WC assumes away.

    In other words, I would not try to go down the rabbit hole of trying to analyze bizarre examples. I would chooose classes of ethical situations and then ask how close the thought experiment is to this class. Don’t focus on the thought experiment, but it’s closeness to what we care about.

    WRT to WC and socialists, I’d start with a lot of simple examples that occur in real life with frequency: Is it ok to let a 5 year old girl set up a lemonade stand and collect a dollar from each kid in the neighborhood? If the socialist responded that issues like money and trade as assumed to be ok in the example, I’d then switch to common examples from non-capitalist societies. If people in a kibbutz give extra bread to a guy who sings well, is that unjust? Is a man with man multiple wives in a tribal society doing something unjust? In other words, small and large voluntary actions create inequality across many institutional contexts.

    I note that concluding that WC is acting justly doesn’t follow from the fact that WC satisfies my criteria. I think a reasonable response from the socialists is “inequality has bad consequences.” That is worth debating. Denying that WC extracts a common and important feature of social life is on its face highly counter intuitive.



    June 13, 2016 at 2:08 am

  3. […] via thought experiments: good and bad — […]


  4. Leaving aside how useful the trolley problem is for illustrating the sometimes conflicting pulls of deontology and utilitarianism, and how pertinent that conflict is for a range of real world ethical problems, the trolley problem interestingly may be relevant to the programming of autonomously driven cars in the future:


    Don Lange

    June 13, 2016 at 6:09 pm

  5. May I (without appearing pedantic) check my understanding of your argument. You say “Are “trolley cars” better for thinking about ethics than “Wilt Chamberlains?” I submit that the answer is yes.”

    However I believe you may mean the reverse – are Wilt Chamberlains better than trolley cars? In which case your argument is persuasive and I think I agree.


  6. “For philosophical problems arise when language goes on holiday.” – Wittgenstein


  7. I thought the most interesting uses of thought experiments for the social sciences are in counterfactual reasoning. Seen in that light, it could be useful for any causal question and set up.

    Counterfactual reasoning is written about in a few articles in political science (I believe Fearon has a piece on this) and a bit in history. I don’t know a good sociological piece on it.



    June 16, 2016 at 6:23 pm

  8. More on utilitarianism and autonomous vehicles. This article is getting a lot of press attention (e.g., WSJ, Washington Post)

    Bonnefon, J.-F., Shariff, A., & Rahwan, I. 2016. The social dilemma of automomous vehicles. Science, 352: 1573-1576.

    Article link:

    Summary by Science:
    When it becomes possible to program decision-making based on moral principles into machines, will self-interest or the public good predominate? In a series of surveys, Bonnefon et al. found that even though participants approve of autonomous vehicles that might sacrifice passengers to save others, respondents would prefer not to ride in such vehicles (see the Perspective by Greene). Respondents would also not approve regulations mandating self-sacrifice, and such regulations would make them less willing to buy an autonomous vehicle.


    Don Lange

    June 24, 2016 at 9:42 pm

Comments are closed.

%d bloggers like this: