Bryan Caplan  

Huemer's "The Use of Hypothetical Examples"

Scenes from the Rent-Seeking S... Occupational Licensing: Rocky ...
Mike Huemer's defense of hypothetical reasoning is so excellent that I feel like I'm cheating EconLog reader when I only quote a few sentences.  So here's his complete discussion, featuring a shout out to David Hume qua economist.  From Huemer's "Why I Am Not an Objectivist":


Unfortunately, Objectivists usually object to the use of hypothetical examples to test moral principles, on the ground that the hypothetical examples do not represent reality. How, one might ask, can we draw conclusions about how the world really is from purely hypothetical premises, i.e., premises about some imagined but not actual situations?

This objection is vaguely felt by many who object to thought experiments in philosophy in general, but it is a logical error. You can validly deduce a categorical proposition from hypothetical premises. For example:

A -> (B -> C).

B -> not-C.

Therefore, not-A.
is a valid form of inference, where the "->" stands for the "if ... then" relation (i.e. "If x were true, y would be true") (N.B. not the so-called 'material conditional' of first-order formal logic). And this form of inference is relevant to the way hypotheticals are used in philosophy to test moral principles. The typical form of thought-experiment-based arguments in moral philosophy is as follows: "If moral theory T were true, then in situation S, it would be right to do A. But in situation S, it would surely be wrong to do A. Therefore, T is false." Notice that this form of argument is perfectly valid: the conclusion deductively follows from its premises (it's a variant on modus tollens). Notice also that both premises are hypothetical - i.e., both are about what would be right if so-and-so were the case. But the conclusion is categorical.

Some Objectivists refuse to even consider statements of the form "if A then B" where A is known to be false. I'm not sure whether they think that such statements are never true. If so, this would be another logical error. The proposition, "If A then B" does not assert A. To say, "If you lose your mittens, you will get no pie," is not to assert that you will lose your mittens. Likewise, to assert, "If I were a brain in a vat, I would have no knowledge of the external world," is not to assert that I am a brain in a vat; it is not even to suggest that I might be. This is obvious to anyone who understands the English word "if". In general, to say, "If A were true, then . . ." does not imply that A is true, it does not imply that A is likely to be true, and it does not even imply that A might be true (if anything, with the use of the subjunctive mood, it implies that A is false). For example, I can say, "If I lived in Alaska, I would have more clothes than I presently do." This is true. It is true in spite of the fact that I am not in Alaska, and I know I am not in Alaska. When I say that, I am not implying that I might really be in Alaska right now, and not in New Jersey as it appears.

Some will still want to know, reasonably enough, why thought experiments are useful. Even if they are in principle capable of proving conclusions about actuality, why are they necessary? Why can we not learn at least as well through the consideration of actual or at least realistic examples? Briefly, the reason is that hypothetical thought experiments provide a means for conceptual controls that often cannot be reproduced in reality. Or, in other words, they provide a way of mentally isolating a causal, explanatory, or logical factor for examination on its own which normally, in the real world, cannot be isolated, and to do so while still discussing a concrete situation.

Let me give an example to show what I mean. David Hume once came up with this thought experiment: suppose that in the middle of the night, the paper money in everyone's wallet, safe, or other stash, suddenly doubled in quantity - so there is twice as much money, but no other changes are made. Would the country then suddenly be enormously better off - would we all be twice as wealthy as we are now? No, in fact we would have exactly the same amount of wealth as we presently do, for there would be exactly the same amount of capital around, and the same availability of labor. (Everyone could then double their prices.) What this shows is that increases in the money supply do not translate to increased wealth; it can also be used to explain why increases in the money supply cause inflation.

Of course, such a scenario is impossible: all our money cannot magically double in quantity. But that is not the point. The reason the thought experiment is useful is that this way of thinking of it enables you to mentally isolate just the one factor desired for consideration: the quantity of money. We imagine just the quantity of money changed and nothing else. In the real world, one cannot do this. In the real world, it is not possible to change the money supply uniformly (i.e. increasing everyone's money, without redistribution) and it is impossible to change the money supply without affecting the economy in some other way at the same time. So I cannot cite a historical case in which nothing but the money supply was altered. This is why thought experiments are useful.

A similar thing is true of thought experiments in moral philosophy. If we want to examine the significance of one morally relevant factor for the evaluation of actions, people, or states of affairs, it is useful to be able to imagine and compare cases which differ only in respect of this one factor of interest, whereas there may be no actual cases of which this is true.

A thought experiment, in short, is not an exercise in fantasy but a tool of logical analysis that is necessitated by the need for conceptual clarity (sc. distinguishing different relevant factors from one another in your thought), together with general facts about the nature of reality (sc. that morally relevant, or otherwise explantorily relevant characteristics do not come isolated in the real world). It is a way of concretizing abstract reasoning.

Now, if someone gives an argument against your moral theory in which the premises are true, and the conclusion follows logically from the premises, you can not escape from the argument by refusing to entertain his premises, i.e., refusing to listen to the argument. I am going to give an argument against egoism which has those characteristics. The premises are true hypotheticals, i.e. true "if ... then" statements, and the conclusion that egoism is false logically follows from them. I will not regard the mere fact that my premises are hypothetical as showing that they must be irrelevant to (i.e. can not entail) their conclusion, which is categorical. I have shown above that hypothetical premises can entail a non-hypothetical conclusion. Nor will I regard the mere fact that the hypotheticals are counter-factual, i.e., the antecedents are false, and known to be so, as showing that the whole "if ... then" proposition must be false. Both of those would be gross logical errors. If, therefore, an Objectivist wishes to answer my argument, he will have to do more than point out one of the aforementioned facts, and he will have to do more than simply refuse to listen to or refuse to think about my premises.

Comments and Sharing

COMMENTS (14 to date)
Daniel Kendrick writes:

Hi, Bryan,

I really appreciate your work and agree with most of it, but I don't know why you keep quoting this polemic of Huemer's. It is very weak, in my opinion, because it misinterprets virtually every Objectivist doctrine that it is trying to attack.

I would like to refer you to Richard Lawrence's refutation of it, which is available here:

In particular, I would draw your attention to the section replying to the section you have quoted, as well as the section immediately following it. The problem is not that you cannot prove anything by hypothetical reasons. On the contrary, reason (and any reasonable Objectivist) certainly agrees with Huemer that hypothetical reasoning is valid. The danger is misusing hypotheticals.

One way of misusing hypotheticals is by inventing a fantastical scenario very different from our own and deducing a certain conclusion from that scenario, and then proceeding to apply that conclusion to the facts as they really are. Lawrence illustrates:

Why should any ethical analysis for such a way-out situation, however repulsive the conclusion may seem, affect our acceptance of one ethical theory versus another? Imagine that Huemer got Objectivists to agree that if rat poison were safe and nutritious for human beings, it would be acceptable to sell it as food. It would be very misleading for him to turn around and say, "Objectivists think it is sometimes OK to sell rat poison as food! They're evil!" To develop a hypothetical filled with assumptions that are not true in the real world, and which probably never will be true in the real world, and then denounce a moral theory for the conclusions it draws from it, is no less inappropriate than the rat poison example.

Huemer's argument against Objectivist egoism is entirely of this character.

Huemer wants to establish that ethical egoism is absurd, so he creates the following hypothetical proposition: if you were an egoist, then if a homeless person were obstructing your path on the way to work and if you had a disintegration ray, then you should disintegrate the homeless person. Now, the vaporization of innocent people is regarded by most everyone (including Objectivists) as evil. So, according to Huemer, if that hypothetical proposition is true, then egoism is evil. So far, so good.

At this point, an Objectivist egoist would say that the hypothetical proposition is false: if you were an egoist, you should not disintegrate the homeless person. There are many reasons for this: you might get caught and punished; you would feel guilty, since a well-adjusted person develops a respect for the value of human life; that person represents a potential trader and value-producer who could benefit you directly or indirectly in the future; even if not, that vaporization would be a violation of his rights, and having a society in which rights are respected is very much in your interests; etc.

But Huemer really wants to show that egoism is absurd, so he says: stipulate away all of those things. You won't get caught, that person is completely worthless, nobody will ever know it happened, etc. If there were no selfish reason not to do it (and some marginal reason to do it), then the egoist should vaporize the homeless person. Therefore, egoism is evil.

It is, of course, true that if there were no selfish reason not to do it and some selfish reason to do it, the egoist should vaporize the homeless person. In fact, that formulation is begging the question. Obviously, if it is taken as a given that it is in your interest to do something, then egoism says you should do it. The question is: what would the facts have to look like for the vaporization to really be in one's interest, and in light of those facts, would the vaporization still clearly be considered evil by "everyone"?

Let us consider just one of the stipulations: that the person must be of no potential present or future, direct or indirect, value to you whatsoever. Now, there are only two cases where I think that could be true: a) the person is a corpse or a "vegetable," or b) the person is some kind of criminal directly threatening your life and/or property. Once we keep this new context in mind, the vaporization does not look so evil anymore. In case a), that person would not have any rights, so the vaporization would not be wrong in that regard (although it would still likely violate the rights of the family which an egoist should respect, but that's another matter). In case b), the vaporization would be an act of self-defense, so the vast majority of people would not reject it.

At this point, Huemer would undoubtedly object and say: "No, imagine a parallel universe in which a completely innocent, conscious homeless person would be entirely worthless, thus making it in your interest to vaporize him." So, instead of changing the facts of the person, we will change the facts of the environment. But such a bleak and cruel world would obviously require very different socio-political systems from our own. It could, for instance, be a zero-sum world of enormously limited resources, such that every inhabitant would have to kill to survive. Assuming that one still found that world worth living in (it likely would not be; note that Objectivism does not say that one's life is unconditionally valuable), killing any potential competitor would be necessary and moral behavior.

It seems that want Huemer really wants is: "Imagine a world that is exactly the same as our own, except that it would be in your interest to vaporize the homeless person." That is the logical contradiction. It is like saying, "Imagine a word in which everything were exactly the same, except that water freezes at 50 degrees C." That isn't possible. If water froze at that temperature, lots of other things would be different, too. And many facts about our world would have to be different to make vaporizing the homeless in one's self-interest, so many that the hypothetical becomes irrelevant.

And consider Huemer's implied premise: homeless people are so worthless that, if we didn't have arbitrary laws and customs protecting them and if we acted in our own rational interests, we should vaporize them all. But because we have fundamentally groundless moral intuitions that tell us this would be wrong, we shouldn't do it. That is, I think, deeply immoral. People are valuable to each other because of what they can do for each other, not in spite of it, as Huemer seems to think.

Again, I refer you to Lawrence's article for further problems with Huemer's essay.

[comment edited by commented--Econlib Ed.]

Hugh writes:

A lot also depends on the skill with which the hypothetical is drafted.

To my way of thinking the "push-the-fat-man-onto-the-tracks-to-save-5-lives" type of hypothetical is just too far-fetched to be of much use.

A secondary problem is that far too much weight is then attached to the answers to these hypotheticals.

handle writes:

Moral hypothetical:

You're Jack Bauer, and if you don't mutilate and torture the potentially innocent detainee in your custody, the risk of a nuke going off in LA goes up 50%. Should you torture him? Maybe torture his family.

A consequentialist utilitarian probably believes yes, but the socially desirable response is, "Torture is just wrong. Always.". And trying to give a principled response is just likely to ruin ones reputation when one is smeared as " pro-torture" in general.

So, there are occasions when moral hypotheticals may be useful in private, but useless and/or abused in public (I.e. on the internet).

The point of extreme cases is supposed to demonstrate that reasonable moral principles and real world factors must balance against each other, and in extraordinary circumstances even a cherished rule that applies 99.99999% of the time in ones real world context may have to give way to other considerations.

But if every argument you make risks one being smeared as "against the cherished rule" then it's foolish to engage in such a public discussion, and inappropriate to accuse people of anti-intellectualism when they decline to take the bait.

And that's the problem we have here. If every time one speaks about immigration or borders some hypothetical is posed in which an equivalence is attempted to be drawn to slavery (abolitionism) or worse, then the intended rhetorical and social effect of honest dialogue are so obvious to all parties that it is clear the hypothetical is not being posed in good faith.

AS writes:

Hypotheticals are used all the time in formal math proofs. Math couldn't exist as we know it without them.

Hansjörg Walther writes:

I guess there are at least two reasons why hypotheticals are not as slam-dunk as you or Michael Huemer think they are:

For example, someone states "for all X (moral or factual) assertion p holds". You construct an example that is in X and where p does not hold. Hence the statement is false. The logic is impeccable, so everyone would have to agree with you.

The first point here is that many people suffer from what I would call "logical blindness" or at least "logical myopia", and cannot see it. People like you and I are almost physically driven to accept the reasoning, we cannot let illogical things stand.

But I have met lots of people who can easily shrug the most blatant contradictions off or don't even see the point. "This is just one example, but "for all" still holds mostly. What's the problem?" Probably we surround ourselves with people who have gone through some math/logic filter, and that's why we tend to over-estimate how common people like us are outside our bubble.

But that's not my main objection. Even if someone does not suffer from logical blindness, he can make a simple adjustment. And he can claim that you just misunderstood him because he did not make all of his (to him obvious) assumptions explicit.

All this someone has to do is change the statement to "for all X that we can expect to encounter (moral or factual) assertion p holds". Of course, this is a weaker claim. However, for all practical purposes it has the same meaning.

To refute this argument you would have to come up with a counterexample that is also realistic. If you don't, the other side can admit that there might be some outlandish cases, and can still insist that you have not refuted their argument.

This all is not to say that I don't understand the value of hypotheticals and they don't resonate with me. But I am not as sure as you are that they are the best way to convince people. It depends on who your audience is (Objectivists look like fair game).

And it depends on whether the "we can expect to encounter" part meshes well with an argument. If someone makes grand "apriori" claims and is in hypothetical-land himself, it doesn't. So shooting someone like Hoppe down with hypotheticals could work.

But if someone is more modest, he can easily qualify his argument and make it immune to unrealistic hypotheticals.

Daublin writes:

Bryan, you have convinced me that hypotheticals are useful, but this is not a strong defense of hypotheticals. The article not just ignores the strongest argument against using hypotheticals, but it claims that argument is not true without supplying any evidence.

The articles says we should think of a hypothetical as an implication "A -> B". That's just it, though: in classical logic, an implication is automatically true whenever its premise is false. You don't even have to look at the conclusion, but it's often the conclusion that is so very interesting in a hypothetical.

Here are some examples that are all true:

If 2+2=5, then I'm a monkey's uncle.

If it's raining, then the streets are wet.

If it's raining, then the streets are dry. (Because it's not
raining over here).

I agree that reasoning from hypotheticals is useful, but you have to dig deeper to defend it.

Greg G writes:

Wait a minute. If "Objectivists usually object to the use of hypothetical examples to test moral principles, on the ground that the hypothetical examples do not represent reality" then how do they justify the fact that fiction with comic book like characters has been the main vehicle used to sell their philosophy?

James writes:

So far fetched and unrealistic hypotheticals show that Objectivist principles lead to absurd conclusions in far fetched and unrealistic scenarios. That seems to leave three options:

Objectivists could search in earnest for more robust principles that don't lead to absurd conclusions even in far fetched scenarios.

Or Objectivists could present their philosophy as a special case, to be embraced tentatively only for as long as the world remains similar to the environment in which the principles of Objectivism were first stated.

Or Objectivists could admit that their principles don't work well in far fetched scenarios when pressed, but spend the rest of the time talking as if their principles are absolutes, always and everywhere relevant and true.

I know which approach I've most often seen Objectivists taking.

Alfred Centauri writes:

The problem, as I see it, with Huemer's "If moral theory T were true.." argument is that essentially no moral theory is safe from it.

To see this, let's generalize the argument he uses against egoism as so:

(1) If ethical theory T is true, then one ought to perform any action that is good according to T

(2) Therefore, if ethical theory T is true, then if A is good while killing 4 million innocent other people amidst gruesome agony, you should do A

(3) You should not kill 4 million people, etc. just because A is good according to T

(4) Therefore, ethical theory T is not true

Now, one might argue that A would never be good according to ethical theory T if A involved killing 4 million people etc., but according to Huemer:

The creator of the counter-example gets to stipulate what goes on in the example. So I get to stipulate, by fiat, that, in the hypothetical situation, I do not receive reprisals for my action, et cetera.

In other words, according to Huemer, objecting to (2) on the grounds that it is absurd does not allow one to escape the conclusion that ethical theory T is false.

Is it any wonder then that that such an argument is not the least bit compelling?

Daniel Kendrick writes:

Excellent point, Alfred.

James Oswald writes:

I find it hard to believe that there is a school of philosophy which refuses to engage in hypotheticals and is nonetheless taken seriously. All planning consists of hypotheticals, all comparisons of actions are hypotheticals. If you're going to say "Stalin was evil", you are engaging in a hypothetical thought experiment where Stalin didn't kill so many people.

I understand not wanting to engage far fetched thought experiments because saying:
If A than B
Not A
Therefore: Nothing. There are no conclusions to draw, so it's useless to talk about. I guess it depends on which As you find possible or plausible.

matt writes:

Alfred, your argument doesn't follow for any moral theory that condemns the gruesome murder of those four million people. That's kind of the entire point of the argument.

for example, suppose that your entire moral theory consisted of the premise "killing people is wrong". how exactly do you not kill people while killing 4 million people?

Alfred Centauri writes:


"killing people is wrong" does not constitute an "entire moral theory" and neither does a *list* of wrong actions.

It's not enough to make a bald assertion that some action is wrong; an ethical theory presents one or more moral principles that one can apply to any act to determine if it is right or wrong.

From the Wikipedia article "Normative Ethics":

Most traditional moral theories rest on principles that determine whether an action is right or wrong

Moreover, it is the form of the argument that I'm calling into question; the "while killing 4 million people etc." is just a place holder. It could very well be, e.g., "while causing 10 million babies to be born with profound birth defects".

Remember, I get to stipulate the hypothetical situation. The recipe should be clear.

Alfred Centauri writes:

(2nd attempt - 1st reply appears stuck in moderation)


"killing people is wrong" does not constitute an "entire moral theory" and neither does a list of wrong actions.

It's not enough to make a bald assertion that some action is wrong; an ethical theory presents one or more moral principles that one can apply to any act to determine if it is right or wrong.

From the Wikipedia article "Normative Ethics":

Most traditional moral theories rest on principles that determine whether an action is right or wrong

Moreover, the "while killing 4 million people etc." is just a place holder. It could very well be, e.g., "while causing 10 million babies to be born with profound birth defects".

Remember, I get to stipulate the hypothetical situation. The recipe should be clear.

Comments for this entry have been closed
Return to top