Begging the question

The same claim in other words

Lena: We should switch the team to the new scheduling app. It’s the most user-friendly option out there, because no other app is as easy to use.

“Most user-friendly” and “no other app is as easy to use” say the same thing. Lena has stated her conclusion twice, with a “because” between the two, and anyone who doubts the first half will doubt the second for exactly the same reason. This is the clear-cut case: one premise, and it’s the conclusion reworded.

A circle with a few steps

Omar: The Hillside Gazette is a reliable paper.

Jess: How do you know?

Omar: Its fact-checking column reviews the reporting every month and says it’s accurate.

Jess: Why trust the column?

Omar: It’s published in the Gazette, and the Gazette is reliable.

No single sentence repeats the conclusion, which makes this harder to see. But follow the support back and it arrives where it began: the paper’s reliability rests on the column, and the column’s reliability rests on the paper’s. Jess has been given nothing that someone unsure about the Gazette could check without already trusting it. This is circular reasoning in a chain.

A disputed word that carries the conclusion

A cycling club is arguing over whether riders on pedal-assist e-bikes may join its Saturday group rides. One member says: “Our rules say the rides are for bicycles. E-bikes are motor vehicles, not bicycles. So they can’t come.”

This isn’t circular: “e-bikes are motor vehicles” doesn’t restate “e-bikes can’t come.” But whether a pedal-assist bike counts as a bicycle under the club’s rule is exactly what the club is arguing about. Calling it a motor vehicle settles that question by choice of words, and a member who thinks e-bikes should be allowed will reject the premise for the same reason they reject the conclusion. In a club where everyone already agreed that e-bikes are motor vehicles, the same argument would be fine, which is why this form depends on the context.

In its barest form, the premise is the conclusion:

• P
• Therefore, P

That form is valid: if the premise is true, the conclusion can’t be false. Patrick Hurley and Lori Watson note that arguments which beg the question are normally valid, and Hans Hansen suggests it is better thought of as an epistemic fallacy (a failure to give grounds for knowing) than as a logical one. It’s a clear case of the difference between an argument being valid and its being any good.

Depending on the conclusion without repeating it. Hansen describes an example adapted from the philosopher John Biro: All members of the committee are old Etonians; Fortesque is a member; so Fortesque is an old Etonian. Given the second premise, Hansen explains, the first can’t be known to be true unless the conclusion is already known to be true, so the argument begs the question even though no premise repeats it. If there were an independent way to know the first premise, such as a bylaw that only old Etonians may serve, the same words would make a good argument.

• Restating the conclusion: the premise says what the conclusion says, in other words. “This diet works because it’s effective.”
• Circular reasoning in a chain: the conclusion returns as a premise after one or more steps, as in the Gazette example. Walter Sinnott-Armstrong and Robert Fogelin define circular reasoning as an argument in which a premise used “directly or indirectly” to support the conclusion is equivalent to it. The longer the chain, the harder the circle is to spot.
• A question-begging premise that isn’t circular: a premise with no support independent of the conclusion, in a context that needs one. Sinnott-Armstrong and Fogelin’s example has the form if P is false, then I’m a monkey’s uncle; I’m not a monkey’s uncle; so P. No premise repeats P, but nobody could have a reason to believe the first premise without already having the same reason to believe P.
• A loaded term: a word that builds the disputed point into the description, as with “motor vehicle” above. It is sometimes called a question-begging epithet.
• The unstated key premise: Hurley and Watson also count arguments that leave out a shaky premise the conclusion needs and cover the gap with “obviously” or “clearly”. On Sinnott-Armstrong and Fogelin’s definition (see below), a missing premise begs the question only when it has no support independent of the conclusion and the context calls for some.
• A question that presupposes the answer: a loaded question builds an unproven claim into a question. Bradley Dowden’s Internet Encyclopedia of Philosophy entry calls that a form of begging the question; Hurley and Watson list it as a separate fallacy of presumption.

Two neighbors are easy to confuse with it. Proof by repetition simply asserts a claim again and again; begging the question dresses the repetition up as an argument, with a premise and a “therefore” (Hansen calls the simplest case “a disguised instance of repetition”). And a no true Scotsman move can end up somewhere similar: once “true” members of a group are defined as the ones who fit a generalization, any argument that they all fit it assumes what it was supposed to show.

Two senses of “begs the question”. The phrase has two common uses, and they are easy to confuse:

• The logical sense (this page): an argument assumes what it’s supposed to prove. The Latin name, petitio principii, is translated by Hurley and Watson as “request for the source”, and Hansen explains that Aristotle described the fallacy as asking for the very point one is supposed to prove. Douglas Walton notes that the English term can be confusing because of its unfamiliar etymology.
• The everyday sense: “raises the question”, as in a sports announcer’s “his injury begs the question of whether he’ll be back for the playoffs” (Sinnott-Armstrong and Fogelin’s example). The American Heritage Dictionary dates this use to at least the early 1900s and reports that it has nearly displaced the logical sense in everyday speech; in its 2013 survey of the dictionary’s usage panel, slim majorities (55% and 58%) accepted example sentences in the newer senses, and 79% accepted the logical sense.

Sinnott-Armstrong and Fogelin point out that the two aren’t entirely unrelated: an argument begs the question, in the logical sense, when it leaves open an unanswered question of why anyone who denies its conclusion should accept its premises.

An argument whose premises are closely tied to its conclusion isn’t automatically question-begging.

• The premise isn’t in dispute. Sinnott-Armstrong and Fogelin’s definition has two parts: a premise lacks any reason independent of the conclusion, and the context calls for one. Reasoning from premises everyone in the conversation already accepts (the rules of a game, a club’s bylaws, the axioms of a proof) needs no further support.
• Nothing is being passed off as support. “No dogs are cats, so no cats are dogs” restates its premise, and Hurley and Watson count it as sound and not fallacious, because it creates no illusion that independent evidence is being offered. Making explicit what a claim already says is a legitimate use of logic.
• The circle is large and informative. Dowden points out that a dictionary defines every word using other words in the same dictionary. That’s circular as a whole, but it’s useful; a two-word circle (“endless loop: see loop, endless”) is not. Properly built recursive definitions also use the term being defined without fault.
• It’s an explanation of a feedback loop, not a proof. Walton notes that a circular sequence of answers can be a correct explanation. His example is a mutual admiration society: Betty likes Bruno because Bruno likes Betty, and the other way round. Pointing out that circular structure explains their behavior; it isn’t a fallacious argument.
• The arguer has an independent reason when asked. A premise you reject isn’t question-begging if the other person can support it without assuming the conclusion.

The test: could someone who doubts the conclusion accept the premises without first giving up that doubt?

A definition and a measurement

Tomás: A “century ride” means riding 100 miles in a day. My GPS logged 103 miles today, so I rode a century.

Friend: That’s circular. It only counts as a century because you defined a century that way.

The definition is doing some of the work, but nobody disputes it: it’s the standard meaning of the term. The question at issue, how far Tomás rode, is settled by the GPS log, which can be checked without assuming the conclusion. This is the premise isn’t in dispute condition above: the premise that could be doubted has independent support.

A self-reinforcing trail

Visitor: Why does everyone cut across the lawn at that corner instead of using the paved path?

Groundskeeper: Because there’s a worn dirt path there, and the path is worn because everyone cuts across there.

The groundskeeper’s answer goes in a circle, but it isn’t trying to prove that people cut across the lawn; the visitor can see that they do. It explains why the habit persists by describing a feedback loop, in which the path draws walkers and the walkers deepen the path. This is the feedback loop condition: the circle is in the process being described, not in the reasoning. (It isn’t a complete explanation, since something had to start the path, but it isn’t a fallacy.)

A premise that restates the conclusion feels convincing to anyone who already believes the conclusion, because it sounds true to them. Hurley and Watson observe that such arguments tend to reinforce what people are already inclined to believe, and that rewording creates the illusion that independent evidence is being offered, helped along by phrases like “of course” and “after all”.

It’s also often hard to see. Sinnott-Armstrong and Fogelin list several ways the dependence gets hidden: rewording the conclusion, leaving the premise that repeats it unstated, or stating a claim early and only using it as a premise for itself much later, after several tangents. And a word that carries the disputed point, like “motor vehicle” for an e-bike, sounds like plain description rather than a claim that needs defending.

Walton describes begging the question as a fault in persuasion dialogue (a conversation whose goal is to convince someone of a claim they doubt), because a circular argument is useless for that purpose. He developed this dialectical view at length in Begging the Question: Circular Reasoning as a Tactic of Argumentation (1991). Sinnott-Armstrong and Fogelin reach a similar conclusion: whether an argument begs the question depends on the context and on what the audience disputes, which is why the same argument can beg the question in one conversation and not in another.

• Ask how the premise is known. “How would you know that, if you didn’t already know the conclusion?” If the answer comes back to the conclusion, you’ve found the circle.
• Trace chains back to their start. Write the support out step by step, and look for the conclusion reappearing in other words.
• Flag loaded terms. When a description settles the disputed point (“that’s not a bicycle, it’s a motor vehicle”), ask for the argument for the description.
• Ask before accusing. Sinnott-Armstrong and Fogelin warn against charging every opponent with begging the question. A premise you reject may have independent support you haven’t heard, and asking for it may teach you something.
• Don’t swing to the other extreme. An argument that begs the question hasn’t shown its conclusion is true, but it hasn’t shown it’s false either. The conclusion may still be right; it needs a different argument.