A post by Ethan Landes

Philosophers love thought experiments, and imagining non-actual events or dilemmas is as core to philosophy as any other method philosophers have. Ask a philosopher to list philosophy’s most influential arguments, and the list might include Plato’s cave, Gettier’s 1963 refutation of JTB, Foot and Thomsons’s Trolley cases, and Cartesian skepticism. All of these are either thought experiments or include thought experiments playing essential roles in the arguments.

If you ask a thought-experiment-loving philosopher (and I count myself among them) why they use thought experiments, you are apt to get an answer like “it lets us control for variables that would be impossible to control for in real life”. When pushed on this, typical ways of unpacking this idea appeal to the epistemology of thought experiments. For example, metaphilosophers have defended the use of imaginary cases instead of facts by arguing that philosophers are interested in knowing things about possibility and necessity rather than actuality (Bealer 1998; Ludwig 2007). In a similarly epistemic vein, I’ve both witnessed others and caught myself telling undergraduates that philosophers use thought experiments because real-life cases are not “clean” enough.

There are nonetheless reasons to doubt that these are actual strengths of thought experiments. For one, gobs of experimental evidence has arisen in the last two decades showing that, while not utterly unreliable, our reactions to thought experiments (that is, our intuitions), are affected by all sorts of things we don’t think they should be affected by (see Machery 2017, chap. 2).  This includes the order in which thought experiments are presented (Schwitzgebel and Cushman 2015; Liao et al. 2012) and the personality of the people observing them (Cokely and Feltz 2009; Bartels and Pizarro 2011). Perhaps they aren’t such a good guide to possibility and necessity after all.

Moreover, is the cleanliness of thought experiments really a virtue? Thought experiments allow us to think about moral, epistemic, and metaphysical issues in a vacuum. If someone told us a Gettier case about whether or not Smith, a puppy killer, formed a racist belief about Jones, we would deride the case as including unnecessary details. Gettier cases are about knowledge, and including morally loaded terms just muddies the waters (See (Dennett 2011, 10) for an example of this sort of move).

Recently, however, philosophy has seen a rise in accounts that argue moral and epistemic matters are intimately tied, despite philosophers’ tendency to uncouple them in thought experiments. For example, proponents of moral encroachment argue that whether one knows or is justified depends on the moral stakes of the situation (Basu 2019; Fritz 2017). If moral encroachment is right, the connection between the moral and epistemic has been right under philosophers’ noses the whole time but only recently taken seriously. Perhaps the sterility of thought experiments prevented philosophers from seeing the relationship between epistemology and morality that messier real-life cases would have forced us to reckon with much sooner.

I do not want to deny that thought experiments have their place in philosophy. They do let us tease apart possibilities that we otherwise could not. But I do want to challenge the idea that philosophers use thought experiments solely due to their epistemic virtues. Instead, I will lay out a number of pragmatic virtues possessed by thought experiments and suggest those are what drive philosophers’ love of imaginary cases.

Pragmatic virtue #1: Thought experiments require fewer facts

Suppose I wanted to convince you that a certain philosophical theory has a counterexample. Why might I choose to develop a thought experiment rather than searching for actual examples? One immediate reason is that many philosophers aren’t trained to collect data to support philosophical claims in the same way that, for example, anthropologists are. We lack the expertise (and many of us lack the patience) to spend days or months finding suitable case studies then checking that we understand enough of the nuances of the case study in order to give an accurate verdict on it. Moreover, if we are going into the research with an agenda, we risk spending all that time and energy researching a case study only to find the facts do not bear out the way we wanted. While thought experiments also fall apart in this way, they do not require the same lengthy process of discovery and factual research.

Real life cases also take more work to present than thought experiments. Thought experiments do not completely remove the role of worldly knowledge from the discovery of counterexamples. Plato’s cave relies on implicit knowledge about what caves are and that they are typically quite dark, and trolley cases rely on an understanding that trolleys are quite heavy and run on rails. These are, to some extent, in the collective consciousness of the readers and can be taken to be part of the common ground. Having to educate readers about real-life cases, especially complicated and nuanced ones, then takes space away from being able to philosophize.

Pragmatic virtue #2: Thought experiments are portable

This leads us to another related set of pragmatic virtues, namely that thought experiments are portable in a way real-life cases and arguments are not. The portability of thought experiments goes beyond their just-discussed ease of presentation. Thought experiments are what we might call grab-and-go. It is very easy for readers to understand, adopt, and adapt in their own thought.

To see this in action, consider an unpublished thought experiment I developed to think about how it is that philosophers gain knowledge from the work of other philosophers:

Imagine it is 1963. A computer scientist turned performance artist wants to pull off an academic hoax. Using what was at the time extremely advanced artificial intelligence, the man created a program that takes any academic paper as input, builds a vocabulary from the paper, then does its best to rejumble the words into grammatically correct sentences. The program merely uses statistical method to identify words as nouns, verbs, adverbs, and other parts of speech, then recombines those words in a way that follow English grammatical structure (adjectives before nouns, verbs between nouns, etc.).

The prankster feeds in an analytic epistemology paper from the previous year, and programs the computer to output approximately 800 words. Skimming over the output to find any obvious typos and finding none, the prankster sends the paper off to Analysis under the name of his neighbour, Edmund Gettier.

The editor sends the paper off to a referee, who sends the following report: “Before I read this paper, I thought JTB is knowledge. This short, succinct paper, however, successfully undermines that view.”

The hoaxer comes clean to the editor, who then passes along the news to the referee. The referee responds “So what? From reading it, I now know there are counterexamples to the JTB account of knowledge—knowledge is not JTB”.

Therefore, even if a reader of a work of philosopher knows an author insincerely argued P, they can still know P from reading the author’s argument. When philosophers read the works of other philosophers, what they learn is based on their own considerations of the arguments, not any testimony or trust of the author.

Unlike cumbersome real-life cases and their pesky facts that need to be remembered and told to listeners, this thought experiment is easy to move. Many trained philosophers will immediately recognize the important features of the case, which are the following:

1)     A philosophical argument is produced insincerely.

2)     Someone reads the insincerely written argument and changes their mind.

3)     The readers’ belief constitutes knowledge, even after they find out about the insincerity.

That’s it. Despite all the fluff about computers and the use of Gettier’s 1963 paper as the archetypical convincing paper, there isn’t much to the case, nor is there much to remember. This makes it well-suited for spreading, since it can be easily learned, thought about later, then adjusted and spread to even more others. As a rhetorical device, it is portable.

This portability is in stark contrast to arguments. The structure of arguments are often not as apparent to the trained reader as the structure of thought experiments. Many of us (myself included) are not in the habit of making the argumentative structure of a work terribly explicit, and disentangling an argument from a paper’s prose can be a significant effort. Thought experiments are easier to understand and easier to repurpose than arguments or real-life cases.

Pragmatic virtue #3: Thought experiments are harder to argue against

Over and above their portability, thought experiments are more rhetorically powerful than arguments because they are less prone to refutation. Whereas thought experiments, at least well-designed ones, offer little for critics to respond to, arguments are rhetorically weak because they have multiple points of attack.

The rhetorical weakness of arguments is inherent in its structure. By arguing for a conclusion by using premises, each jointly necessary for the conclusion, denying any particular premise is sufficient to avoid the argument’s conclusion (Weinberg 2015). This means the more premises an argument has, the easier it will be for someone who does not like the argument’s conclusion to reject the argument. Because the premises of philosophy are often so arcane and so dependent on small tweaks of language, an opponent of a conclusion can reject one or more premises in an argument at very little cost to their own commitments. In other words, it is easy to refute an argument by taking issue with a premise in a way that does not damage one’s own philosophical positions (Chalmers 2015).

In contrast, thought experiments do not present opponents with a rich target full of vulnerable joints. An opponent to a thought experiment’s verdict can only a) bite the bullet on the verdict, b) deny the author correctly identified the verdict, c) deny that the author has correctly diagnosed the reason for the verdict, or d) come up with some error theory for the verdict. These are all difficult moves to pull off and many--especially denying the verdict and biting the bullet—may strike other philosophers as cheap. Therefore, if philosophers want to win a debate, they have strong pragmatic reasons to prefer thought experiments over easier to refute arguments. 

In summary, philosophers use thought experiments because compared to other options thought experiments are easy to develop, present, and defend. This list is not exhaustive—I didn’t even touch the psychological evidence that suggests thought experiments are uniquely convincing—but it is meant to be illustrative of a larger metaphilosophical point. Even though philosophers can come up with high-minded epistemic reasons for using thought experiments, those high-minded reasons may be covering up a more uncomfortable truth. We may do philosophy the way we do solely because it suits our tastes.

References

Bartels, Daniel M., and David A. Pizarro. 2011. ‘The Mismeasure of Morals: Antisocial Personality Traits Predict Utilitarian Responses to Moral Dilemmas’. Cognition 121 (1): 154–61. https://doi.org/10.1016/j.cognition.2011.05.010.

Basu, Rima. 2019. ‘Radical Moral Encroachment: The Moral Stakes of Racist Beliefs’. Philosophical Issues 29 (1): 9–23. https://doi.org/10.1111/phis.12137.

Bealer, George. 1998. ‘Intuition and the Autonomy of Philosophy’. In Rethinking Intuition: The Psychology of Intuition and Its Role in Philosophical Inquiry, edited by Michael DePaul and William Ramsey, 201–40. Lanham, Md.: Rowman & Littlefield.

Chalmers, David J. 2015. ‘Why Isn’t There More Progress in Philosophy?’ Philosophy 90 (1): 3–31.

Cokely, Edward T., and Adam Feltz. 2009. ‘Individual Differences, Judgment Biases, and Theory-of-Mind: Deconstructing the Intentional Action Side Effect Asymmetry’. Journal of Research in Personality 43 (1): 18–24. https://doi.org/10.1016/j.jrp.2008.10.007.

Dennett, Daniel C. 2011. ‘“ My Brain Made Me Do It”(When Neuroscientists Think They Can Do Philosophy)’.

Fritz, James. 2017. ‘Pragmatic Encroachment and Moral Encroachment’. Pacific Philosophical Quarterly 98 (S1): 643–61. https://doi.org/10.1111/papq.12203.

Liao, S. Matthew, Alex Wiegmann, Joshua Alexander, and Gerard Vong. 2012. ‘Putting the Trolley in Order: Experimental Philosophy and the Loop Case’. Philosophical Psychology 25 (5): 661–71. https://doi.org/10.1080/09515089.2011.627536.

Ludwig, Kirk. 2007. ‘The Epistemology of Thought Experiments: First Person versus Third Person Approaches’. Midwest Studies in Philosophy 31 (1): 128–59.

Machery, Edouard. 2017. Philosophy within Its Proper Bounds. First. Oxford, United Kingdom: Oxford University Press.

Schwitzgebel, Eric, and Fiery Cushman. 2015. ‘Philosophers’ Biased Judgements Persist Despite Training, Expertise and Reflection’. Cognition 141 (August): 127–37.

Weinberg, Jonathan M. 2015. ‘Humans as Instruments: Or, the Inevitability of Experimental Philosophy’. In Experimental Philosophy, Rationalism, and Naturalism, 179–95. Routledge.