Just asking questions

The petty cash

At the coffee machine, Lee says: “I’m not accusing anyone. But petty cash came up short in March and again in May, and Dana was closing up both of those months. Funny coincidence, isn’t it? I’m just asking.”

A colleague says that sounds like an accusation. Lee replies: “I never said Dana took anything. But can you explain the timing?”

Each sentence is a true fact or a question, but the only reason to put those facts side by side is to suggest that Dana took the money. Lee gets the effect of an accusation without offering evidence for one, then turns the burden around: the colleague is asked to explain the timing, as if a coincidence nobody has explained were evidence of theft (see Argument from ignorance and Shifting the burden of proof). This is the clear-cut case.

A question in a headline

A gadget blog runs the headline “Is the Brightline robot vacuum secretly mapping your home for advertisers?” The article explains that the vacuum builds a floor map to plan its cleaning route, notes that “some companies” sell customer data, and ends: “Brightline didn’t respond to our request for comment. You be the judge.”

Nothing in the article connects the maps to advertisers; the only link is the question in the headline. But readers expect a publication to know more than they do, so a question it chooses to ask suggests the matter is really open, and perhaps that there’s something to it. This version looks respectable because no one is accused of anything in so many words. The linguist Tom Roberts describes the tactic as letting a writer “plant the seed of controversy while evading responsibility for its flowering.”

A string of innocent questions

In a neighborhood group chat, after the bakery on Elm Street changes its bread: “Why did they switch flour suppliers right before their inspection? Why won’t they say who the new supplier is? Why did two of their bakers leave this spring? Just curious.”

Any one of these could be idle curiosity, and each may have a dull answer: a better price, a supplier that asked not to be named, bakers who moved away. Asked together, they build a story that nobody has stated and so nobody has to defend. The sign is where the questions are aimed: at the group chat, not at the bakery, which is the one place the answers could come from.

• The headline question. A journalists’ adage known as Betteridge’s law of headlines, which Roberts quotes, says “Any headline that ends in a question mark can be answered by the word no.” Roberts notes it isn’t literally true of every headline; it captures the intuition that a question headline can put forward a weakly supported claim without committing to it.
• “Isn’t it interesting that...?” A true fact followed by an invited inference. The fact can be defended; the inference is never stated.
• “Why won’t they deny it?” Treating the target’s silence, or failure to answer, as support for the suggestion. (In a Kafka trap, even a denial is counted as support.)
• “I’m just asking” after the fact. A claim that was made fairly directly is recast as a question once someone pushes back.

Not the same as a loaded question. A loaded question builds an unproven claim into the question itself (“Have you stopped cheating at cards?”), so any direct answer concedes it. A just-asking question usually presupposes nothing: “Is the vacuum mapping your home for advertisers?” can be answered “no” without conceding anything. The suggestion comes from the fact that someone chose to ask.

About the name. “Just asking questions” is a colloquial label for the tactic, borrowed from what people say when challenged; logic textbooks don’t list the error under this name. The move itself is much older and is studied under the names innuendo and insinuation. The argumentation theorist Douglas Walton, discussing personal attacks, wrote that the ad hominem “is so often such a powerful tactic precisely because it is put forward on a basis of suggestion or innuendo, instead of solid evidence. And the tactic is to try to use the innuendo to shift the burden of proof to the other side.” The philosopher Elisabeth Camp describes insinuation as communicating something “off record”, in a way that permits deniability. Roberts’s 2024 paper “Just-asking questions” analyzes the version carried out with questions, which he says gives a writer “the plausible deniability of ‘just asking questions’”.

Asking a question is not an accusation, even a pointed question, and even when the answer might be embarrassing.

• The question goes to someone who can answer it, and the answer is accepted. Asking the bookkeeper how the petty cash log works is inquiry, not insinuation.
• The suspicion is stated as a suspicion, with its evidence. “I think the timing looks odd, here’s why, and I may be wrong” puts the claim on the record, where it can be checked and answered.
• The question really is open, and it’s treated as open. Roberts points out that many question headlines and talk titles (“Could humans survive on Mars?”) introduce a sincere look at an issue nobody can yet settle.
• The question tests a claim someone else made. Asking “How do you know the meeting was canceled?” puts the burden on the person who said so, which is where it belongs.
• You’re listing possibilities in order to check them. “Could the leak be the dishwasher hose?” while troubleshooting is a hypothesis to test, not a suggestion left to do its work.

The test: if the answer came back “no”, would the asker accept it, and would they be willing to state and defend the claim their question suggests?

Asking the person who knows

Priya (to the office manager): Petty cash was short in March and May. Is there a pattern to when that happens? Could it be how the log gets filled in at closing?

Office manager: Both months, someone paid for a delivery in cash and forgot to log the receipt. I found both.

Priya: Good. Can we add a line for that on the form?

The facts are the same as in the petty cash example, and Priya also asks questions. But she asks the one person in a position to answer, her guess doesn’t point at anyone, and she accepts the answer when it comes. This is the someone who can answer condition above.

A question headline that answers itself

A cooking magazine runs the headline “Does searing meat really seal in the juices?” The article describes a test in which seared and unseared steaks were weighed before and after cooking, reports what the scale showed, and answers the question in its last paragraph.

Like the robot vacuum headline, this raises a question the reader can’t settle. The difference is that the article does the work: it collects evidence, gives an answer, and would have printed “no” if that’s what the test showed. The question sets up an inquiry instead of standing in for a claim. This is the really open, and treated as open condition.

It works because people communicate far more than they literally say. The philosopher H. P. Grice called meaning conveyed without being stated an implicature. One of his examples is a reference letter for a candidate for a philosophy job that says only: “Mr. X’s command of English is excellent, and his attendance at tutorials has been regular.” Nothing negative is said, but a reader works out that the writer thinks the candidate is no good at philosophy. Grice also noted that a conversational implicature can be canceled, for instance by adding a clause that says the speaker is opting out. “I’m just asking” works like that kind of cancellation, except that it arrives after the suggestion has already been received.

Roberts offers an account of why questions in particular carry suggestions. Asking a question puts an issue on the conversation’s agenda. When the asker is someone expected to know more than the audience, such as a newspaper, the audience infers that nobody knows the answer, so a question can make an unlikely possibility look like a live one. He adds that this goes further than ordinary implicature: asking the question commits the asker to nothing at all, which is what makes the deniability so easy.

There is experimental evidence that the question alone can do damage. In two preregistered studies with 506 participants, Letourneau and Gawronski (2024) closely replicated the first study of a 1981 paper by Daniel Wegner and colleagues: questions insinuating something negative about a person reduced favorable impressions of that person. The authors conclude that merely asking a question about a false proposition can influence opinion without any false assertion being made.

It doesn’t always take bad faith. People do wonder aloud, and a string of questions can come from real curiosity. The suggestion reaches the audience either way, which is why the conditions above matter more than the asker’s intent.

None of these has been tested experimentally as a remedy.

• Ask for the claim. “What do you think happened?” If they’ll state it, the evidence can be discussed. If they won’t, there’s no claim on the table to accept.
• Answer the question if it has an answer, and see whether the answer is accepted. A question that survives every answer wasn’t really a question.
• Don’t take on a burden that isn’t yours. “Can you explain the timing?” doesn’t have to be answered before the suggestion counts as unsupported.
• When you’re the one wondering, take the question to someone who can answer it, or say plainly what you suspect and why.