Some Thoughts on Thought Experiments

Thought experiments are extended analogies. They appeal heavily to the hearer’s intuition. They are stories intended to show an argument as true or another as false. They are not typically formal arguments in and of themselves.

A thought experiment is most often functioning as a counterexample to a specific premise or argument. For example, pretend I claimed “all people do only that which they desire to do.” You may combat this by presenting a brief illustrative story showing that the man with a gun to his head does not want to break a window, but he does want to live.[1] Or perhaps you state “one cannot be tempted if it is impossible to fulfill that temptation,” and in response I share an anecdote of a man who is tempted by forbidden chocolate cake in the refrigerator all night, only to discover it had been eaten the afternoon before without his knowledge.[2]

As an actual, formal argument thought experiments aren’t very good. The reason is that the thought experiment is essentially question-begging. It assumes, in asserting the counterexample’s possibility, that the principle being advocated is false. However, the thought experiment holds an even greater value than formal argument, if done correctly. It appeals to the intuitive nature of the hearer. It raises the point that, if the principle in discussion is in fact true, this highly-intuitive scenario is in fact false. The idea is the counterexample is so undeniably obvious that it is far more plausibly true than the premise in question.

How to Defeat a Thought Experiment

Some may disagree and have different ideas (share them below), but these are just mine.

1.      Show the thought experiment as false.

This involves attempting to show the counterexample does not accomplish what its conclusion suggests. This can be very difficult, especially since one who takes this track is liable to commit what I call the “distinction without a difference” fallacy.[3] The key to success here is to show the conclusion (e.g., “the point is that one may be tempted even though its fulfillment is impossible”) is not suggested from the thought experiment itself.

2.      Show the thought experiment has not overcome the initial argument.

This is a tactic which grants the conclusion but not its force or applicability. In continuing with our example, the opponent of the experiment may say, “yes, but the temptation is due to epistemic limitations, which do not apply to our overall discussion.” Again, a potential problem is to point out a distinction without a difference, so make sure it truly makes a difference!

3.      Suggest the thought experiment has deleterious consequences which are counterintuitive, or at least less intuitive or plausible than the original premise.

This involves a bit of a counterfactual form. “If what you’re saying by this thought experiment is true, then Christ was not omniscient, nor was he omnipotent.” Obviously, if true, this is a high price to pay (one that orthodox Christians will not accept). Therefore, the thought experiment loses its force. The danger in this tactic is the tendency to overstate or overdramatize the case. One must be sure that the consequences really are unavoidably that certain way. If they are not (that is, if it is possible to be otherwise), then discuss the alternatives and why they are less plausible than the evidence for your position. In this way, you have honestly framed the debate in terms of the thought experiment, and shown it by no means renders your argument implausible.

Constructing a Thought Experiment

In constructing a thought experiment, you must be aware of a few things.

1.      Make sure the analogy works.

If you are attempting to show something is logically possible, it needn’t be actually occurring nor even likely to occur. It merely needs to be logically possible! Similarly, if your counterexample is intended to show that something is in fact the case, then it must be actually occurring (or at the least, plausible to occur). Be sure to state the conclusion and its applicability to the premise clearly.

2.      Don’t overstate the case.

It’s important not to say, “this proves your premise could never be right!” but rather something less aggressive, such as, “this suggests the premise in its current form is incorrect.” If you overstate the case, you’ll typically make a mistake within the application. If your discussion partner realizes this mistake, it gives the impression the thought experiment is a total failure. While incorrect, you still do not want to have to go back and build from the beginning. Better to debate what really matters to the argument in question.

3.      Anticipate objections.

This does not mean cover everything in painstaking detail. On this blog, I have used thought experiments, specifically in relation to “Greg and Compatibilism.” In this, I try to cover common objections that would be used to invalidate it at crucial points. The key is knowing what potential objections to your experiment would be, and trying to incorporate brief details which also remain plausible. A defense of those details need not be done in the thought experiment itself.

Thought experiments are awesome tools to demonstrate the intuitive appeal of your claim. Many times, people agree intuitively with your position while expressing something which is completely opposite. This is usually due to a failure to appreciate all of the nuances involved in the topic. It can, and does, happen to everyone (myself included). Hope you have fun with it and feel free to comment below!

---

[1] Please put aside whether or not you agree with the premise or the counterexample. The point is merely to illustrate what a thought experiment looks like in the course of a discussion.

[2] Thanks to William Lane Craig for this example.

[3] This fallacy involves pointing out any difference between an analogy and its referent and proclaiming the analogy false.

All posts, and the blog Possible Worlds, are the sole intellectual property of Randy Everist. One may reprint part or all of this post so long as: a) full attribution is given (Randy Everist, Possible Worlds), b) all use is non-commercial, and c) one is in compliance with the Creative Commons license at the bottom on the main page of this blog.