What Is Question-Begging?

Everything is relative

Be it Reviewer #2 or some random reply-guy, the neural pathway between reading an argument and saying the words “question-begging” is very short and well-trained among those learned in the ways of philosophy. As soon as there is an argument people disagree with, it seems someone will inevitably deem it question-begging.

In spite of this, figuring out what it means to beg the question is quite a hairy issue once you consider it more thoroughly. The textbook definition is something like: An argument begs the question when the premises assume the truth of the conclusion. That might sound neat, but it’s very hard to make work. To remain justified in rapid-firing on the question-begging accusations, we ought to try to find an answer here. Anything else would be question-begging.

Not a Formal Feature

As said, it’s tempting to give an account where begging the question is a matter of the premises assuming the truth of the conclusion. More specifically, it has to be the conjunction of the premises that has to assume the truth of the conclusion. After all, we can have question-begging arguments like:

1. The Bible says that God exists

2. Everything The Bible says is true

3. Therefore God exists

Neither (1) nor (2) require that (3) is true on their own. Instead the problem is that the two premises in conjunction require the truth of the conclusion. So that’s the answer: An argument begs the question just in case the truth of the premises assume the truth of the conclusion.

Whoopsie! That’s probably not a good definition, as it makes every deductively valid argument question-begging. Consider the textbook argument:

1. All men are mortal

2. Socrates is a man

3. So Socrates is mortal

The claim that all men are mortal, and that Socrates is a man certainly assume that Socrates is mortal. If Socrates were not mortal, then that would provide a counterexample to at least one of the premises, so they must assume that he is.

The problem should be pretty obvious: The standard definition of validity is that the premises jointly entail the truth of the conclusion—the premises could not be true without the conclusion also being true. But if the premises cannot be true unless the conclusion is true, then they certainly assume the truth of the conclusion!

[For a more thorough discussion of this point, and some other relevant issues, I recommend yakiimo’s excellent post on the subject. I also just recommend their blog on the whole!]

Not Inherent to Arguments

Begging the question isn’t about the truth-conditions of the premises and conclusions in the argument, or the relations between these. That is, it’s not inherent to the argument itself.

Rather, begging the question is about the epistemic features of the argument in relation to some dialectical context. In actual usage, arguments are not just arguments for a particular proposition, but arguments against alternatives (where an alternative to P is just a proposition that entails ¬P). Suppose I give the argument:

1. I turned off the stove when I left

2. If I turned off the stove when I left, then it’s still off

3. So the stove is still off

If the context here is that we are out for dinner, and you ask whether the stove is off, this seems like a perfectly appropriate argument, even if explicitly saying (2) may be a little odd.

However, if the context is instead that our arsonist neighbor has been sneaking into our house when we’re away and turning on the stove, then it suddenly seems question-begging! Nothing about the argument itself changed. The problem instead lies in the change of targets.

In the first argument, the target alternative to the conclusion is something like that I forgot to turn off the stove. By giving the argument I express that I have a clear and distinct memory of turning off the stove, and this is sufficient to disprove the alternatives we consider relevant in the “dialectical” context.

In the second case, one of the relevant alternatives is our neighbor having turned on the stove. My giving the argument is not much help here, as you obviously won’t accept (2) given the alternative in question. In the first case our neighbor turning on the stove is strictly speaking still an alternative, but it’s so far out and unlikely that it doesn’t matter whether the argument works against it.

Is the correct account then just that question-begging arguments fail to rule out relevant alternatives? It can’t quite be that. The argument above really does strictly rule out the arsonist neighbor-alternative. What makes it problematic or question-begging is that you would never accept (2) unless you already thought that the alternative in question had been ruled out.

Instead we have to zoom out one step, and look at the justifications for the premises, rather than the premises themselves. More specifically, arguments only beg the question with regards to some person (usually your interlocutor), and with regards to some alternative. They do this when the premises presuppose the falsity of the alternative in question in order to have any force.

Or put in analytic philosophese:

An argument begs the question against an alternative A for some agent S, just in case the reasons S has for believing¹ the premises presuppose the falsity of A.²

The obvious next question is then what in the world “presuppose” is supposed to mean, as that word is doing a lot of the heavy lifting. I'll give an example to explain it, before I give the definition.

You come home and see what appears to be an apple on your table. Having studied some early analytic philosophy, you reason:

1. Here is an apple

2. Therefore there is at least one apple on my table

If we’re a little charitable, we can allow that this is a valid argument. Not only that, but it seems to be of a non-question-begging kind!

However, now you remember that you actually have a lot of fake plastic apples lying around after you decided to order 1000 online because you wanted 1000 plastic apples. You wonder whether the thing on your table is an apple or a replica. To clear yourself of doubts, you reason:

1. Here is an apple

2. Therefore there is at least one apple on my table

First, it should be clear that the argument is fine in the first instance, and question-begging in the second. That’s because in the first instance, there are no relative alternatives that involve you having appearances as of seeing an apple, whereas there are in the second—and you’re trying to reason against one such alternative.

That is, in the first case, the alternative(s) is something like “there are no apples or apple-like objects on my table”. If this were true (and assuming more elaborate skeptical scenarios are not relevant alternatives either), you would not have the appearance as of seeing an apple. And since your reason for believing in premise (1) is that you have the appearance as of there being an apple on the table, everything checks out.

Your reasons for believing in premise (1) are the same in the second case. However, here the alternative is “there is a plastic apple on my table”. If this alternative were true, you’d still have the appearance as of there being an apple there, and so it doesn’t work anymore.

Another way of putting the problem in the second case is that for your reasons for believing (1) presuppose that the relevant alternative is true in order for them to justify (1); the reasons you have for believing in (1) can’t make you anymore confident than you already are in the alternative being false.

Again, allow me to translate into philosophese:

An inference from premises P to conclusion C presupposes the falsity of alternative A just in case the reasons for believing P are undercut by A’s being true.

Admittedly, the “undercut” here is still somewhat vague. But this vagueness is sort of deliberate: One way of cashing out the principle is in terms of evidential sensitivity—that the reasons do not reliably distinguish A from P. But I’m not confident that this is the only way arguments can be circular. I think the structure of this account is right, but I’ll leave it open how many ways this “undercutting” can be cashed out.

But yeah, sensitivity is one way, and it’s the one that’s been working in the cases so far. In the apple-case, your visual appearances are not sensitive to the difference between the apple being real and it being plastic; and in the stove-example, your memory can’t tell you whether the arsonist neighbor snuck into your house after you left.

[If you’re interested, I try to flesh out this idea a little more rigorously in this post, in order to show why Moore’s proof of the external world is circular.]

Conclusion

So begging the question is not a feature of arguments themselves. Rather arguments only beg the question against certain alternatives, given certain reasons for believing their premises.

So why do certain arguments feel inherently question-begging? Consider again the argument from the beginning:

1. The Bible says that God exists

2. Everything The Bible says is true

3. Therefore God exists

The reason that arguments like this seem inherently question-begging is that there is no plausible dialectical position where it would ever be convincing to anyone. No one who accepts that everything The Bible says is true would not already believe in God!

But if we imagine some weird Christian who accepted that The Bible is infallible, but were somehow still an atheist, then the argument in question would be perfectly appropriate! They could plausibly be convinced that, “oh yeah, God really does exist!”

Interestingly, the account I give doesn’t obviously capture cases where the problem seems to be purely formal—e.g. where one of the premises just is the conclusion. For instance:

1. You should subscribe

2. Therefore you should subscribe

I’m not sure this is a problem, though. It really does seem like this is a different phenomenon than the more interesting cases of begging the question. But if we want, we could just add a clause to the account saying “oh yeah, and also don’t just lump in the conclusion explicitly among the premises”.

Minor Polemical Postscript

At the end of their aforementioned post, Yakiimo claims that this result—that begging the question depends on agents and contexts and whatnot—means that the truths of logic are relativistic in a sense, and that this is very surprising. I think that claim might be a little overblown!

What we’ve shown is that what counts as a good argument is relative. But then I don’t think the truths of logic are about what counts as a good argument. Rather they’re about what inferences preserve truth (at least that’s the sense of logic that would be surprising if it were relative) and that part certainly hasn’t been shown to be relative!

You Might Also Like:

1. Maybe it should just read “the reasons S has or would give for the premises” or somesuch. After all, arguments can be question-begging even if no one believes them. It just requires that we imagine some reasons that could be given in favor of the premises.

2. This is quite similar to Sinnott-Armstrong’s account that Yakiimo discusses, and it could probably count as a “subjective epistemic account” according the the nomenclature they lay out there.