"Maria and Peter are students and meet up for a late dinner. Peter asks Maria whether Tom is at the party that they intend to go to after dinner. Maria answers that Tom is at the party. After all, Tom had told her that he would be at the party. When they arrive at the party, it turns out that Tom had changed his plans, and is not at the party. Was Maria's answer true or false?"

#truth #philosophy #cognition

(please spread for visibility, I would like this to be as wide as possible)

1/2

Maria's answer was true
32.6%
Maria's answer was false
67.4%
Poll ended at .

A new study shows that there is much, much less agreement on the answer to this question than I would have expected. Even after reading about the study, I still expect people in my bubble to have the same answer as I do. Let's see. But this probably means that the meaning of truth, in the general population, is simply different from what I would have assumed. And explains a number of public discourses.

2/2

https://reason.com/2026/05/15/the-surprising-divide-over-what-counts-as-true/

New study investigates why people disagree about what's true

A new study finds that what people think about facts, authenticity, or coherent beliefs explains why they disagree about what is true.

Reason.com
@vrandecic so some people equate someone lying to someone's statement being false? Then they should have a different word for someone being unintentionally wrong?
@janjko yeah, I have the same problem. I would say Maria never lied. But for me, that doesn't mean what she said is true.
@vrandecic @janjko as far as she knew, it was true.
@edgeofeurope @vrandecic @janjko No, she had no way of knowing it was true.

@bnlandor @edgeofeurope @vrandecic @janjko We can never be absolutely certain of anything at the level of logic.

This seems mainly to be a problem about the practice assigning logical truth values to real life language acts.

It seems we can all agree on the practical meaning, consequences, and so on, and whether the different parts of the bundle of things we might mean by a statement being true are satisfied and what they might be contingent on (the personalities of the people, their circumstance, etc)..

Where we disagree seems to be on this truth-value labelling pursuit which, a bit like village cricket, I'm pleased some people are passionate about, but I'm not sure I'm one of them.

@chiffchaff LOL, sorry for my brief reply; what I meant was that the question posed at the start has a clear logical answer for me, as Maria could have used some form of modal or evidential marker. By not doing so, even if she wasn’t lying intentionally, her statement lends itself to a fairly straightforward assessment of falsity. @bnlandor @edgeofeurope @vrandecic @janjko

@protoeuskaldun @bnlandor @edgeofeurope @vrandecic @janjko sorry, I thought you were making a claim for the foundational importance of truth values of utterances in building a model of the world.

Maria could have been clearer about her level of knowledge, to make the surface meaning match the pragmatic purpose, but that can be a tedious fool's errand if you go too far with it in real life.

@chiffchaff @protoeuskaldun @bnlandor @edgeofeurope @vrandecic @janjko
At what point could she be 100% sure anyway? Even if he had told her earlier that he was at the party he could've left by then...
If she knew him to be chained to the wall at that party? Even then he might have died by now, which opens another can of worms about the definition of *being* at the party :-p

It's obvious that she could only answer with what she knew at the time and that a "yes" implies "as far as I know"

@Doomed_Daniel @protoeuskaldun @bnlandor @edgeofeurope @vrandecic @janjko it could be an alien hologram of him, or she could be afflicted by psychosis, or whatever. I think it's okay to say false things by accident when the world is conspiring against you, but here, as long as the questioner had a fair understanding of the answerers level of knowledge, I think "is" is just a placeholder for a complex construction already understood by context by both parties. If I ask "Is there a film at 11?" at 9, I think I can say yes truthfully, without saying "as long as there isn't a major international incident in the next two hours. That "is" is clearly about scheduling.

@chiffchaff @Doomed_Daniel @protoeuskaldun @bnlandor @edgeofeurope @vrandecic @janjko

Or you can just cut out all the uncertainty and assumption and say "he said he will be at the party" or "yes, the film is scheduled at 11".

@bnlandor @edgeofeurope @vrandecic @janjko

I think she can be pretty sure that it's true Tom told her he would be at the party.

What Tom said to her wasn't true.

@HikerGeek @bnlandor @edgeofeurope @vrandecic @janjko Yes, but that wasn’t the question. If the question was if Tom *said* he would be at the party, then yes. As it is written, Maria’s statement is false (even though not an intentional lie).
@bnlandor @edgeofeurope @vrandecic @janjko Which Peter knew. So his question must be understood as being about her belief according to the information available to her.

@tessarakt

No, that logically does not follow. By normal convention all requests for information are accompanied by the implicit qualifier, "if you know".

This is precisely why LLMs never answering "I don't know" and instead hallucinating an answer is such a violation of the social compact.

@bnlandor @edgeofeurope @vrandecic @janjko

@siderea @tessarakt @edgeofeurope @vrandecic @janjko Does that affect the truth value of her statement in any way? Not any potentially implied statement, but the actual statement.

@bnlandor
Of course. Allow me to demonstrate.

Imagine that you say to Maria, "Oh, hey, I see there's a party at Jane's place. Is Tom at the party?" and Maria, knowing perfectly well that Tom is not at Jane's party but Susie's party across town, and not wanting you to know that, deliberately trades on ambiguity to deceive by saying, "Yes, Tom is at the party." Further imagine that you subsequently find out that Tom was not at Jane's party but at Susie's party, and Maria knew that, and you confront Maria, "Hey! Why did you tell me that Tom was at Jane's party when you knew he was at Susie's party?"

Imagine that Maria replies, "I did no such thing. I didn't tell you that Tom was at Jane's party. I said he was at *the* party."

This would be pretty universally agreed to be a specious argument.

This is because the truth or falsity of an answer to a question is only determinable in the context of the question it is answering.

So it turns out to matter pretty enormously to the truth or falsity of Maria's answer to Peter just which question she was answering when she uttered it.

@tessarakt @edgeofeurope @vrandecic @janjko

@siderea @bnlandor @edgeofeurope @vrandecic @janjko "This would be pretty universally agreed to be a specious argument."

However, the US Supreme Court might consider it literally true under the Bronston literal truth defense. https://en.wikipedia.org/wiki/Bronston_v._United_States

Bronston v. United States - Wikipedia

@siderea @tessarakt @bnlandor @edgeofeurope @vrandecic @janjko I don't see the comparison to an LLM. LLMs are expected to say "I don't know" and fail to when they produce answers they have no basis for in either direction. Maria did have (very strong) basis for her answer, even though it happened to be wrong.

@orman Depends on what you mean by "expected". A lot of people expect LLMs to say "I don't know" and notoriously LLMs rarely if ever say "I don't know" and this is this a violation of that expectation if you have it.

@tessarakt @bnlandor @edgeofeurope @vrandecic @janjko

@siderea @orman @tessarakt @bnlandor @edgeofeurope @vrandecic @janjko

Expecting LLMs to say "I don't know." is an unreasonable expectation. That's not what they're built to do. And LLMs don't actually "know" anything, in any strict logical "thinking" sense. They're designed to produce plausible sequences of text. The truth or falsity of their statements is irrelevant. If fed text of many humans lying, as training input, the LLM will do the same, without regret or even "thought."

@siderea @bnlandor @edgeofeurope @vrandecic @janjko that is an overly simplified view of language semantics and pragmatics.
@bnlandor @edgeofeurope @vrandecic @janjko but this is the default. I automatically parse her statement as "I believe he will be there", because the actual fact is unknowable at that point. We just use shortcuts to not use "to the best of my knowledge" in every sentence we say
@jay_peper @bnlandor @edgeofeurope @janjko yes, otherwise our daily conversations would be very difficult. I think that more precise speaking is only needed in high-stakes situations.

@vrandecic @janjko

I agree. What she should have said is that he said he would be at the party. Then it wouldn’t be false either way.

@pomegranate_stew @vrandecic @janjko I find people often express frustration when I use conditional statements, but I find it frustrating that it's apparently a social norm to express unwarranted certainty.

@vrandecic @janjko this. What she said was not true, and I don't understand* how that can be controversial because it was factually incorrect. That doesn't mean she was lying, it just means she was wrong

*I understand better after reading the article but it still boggles my mind

@raphaelmorgan If Tom was at the party at the time Maria answered the question (which she would not have been aware of either) but had left before the others arrived, would that change your view on this?
@kaspi yes, then she'd be telling the truth (No more effectively than an LLM, one could argue, because she doesn't actually know the truth and is just supplying her best guess as if she does know). I was going off of the original prompt that said "Tom had changed his plans," suggesting he never went, but if he simply left before they got there that's another story
@raphaelmorgan The issue I see with the original example is that it uses a social situation as backdrop. In which one would normally assume conversational norms and phrases. According to Grice's maxim of quality https://en.wikipedia.org/wiki/Cooperative_principle#Grice's_maxims Marie told the truth. A reader of the story could assume that the question was not asked by an external observer (asking for general facts) but by Peter within the situation's context, thus asking what Maria knows about Tom's plans. Everything else would be silly.
Cooperative principle - Wikipedia

@raphaelmorgan As someone within the system you are talking about, you can only make true statements about things you can validate/observe at the time you make your statement. Everything else would be assumptions or speculation, even if most of these often turn out correct, because the system is rather stable. We are basically talking about cats in boxes here.

Expecting only 100% true statements at any time would make normal conversations nearly impossible.

@raphaelmorgan So my personal opinion is that the researchers should have phrased their question more clearly and made it obvious, that they were looking for objective facts and not an interpretation of Maria's situational response.

My own experience tells me, that when people ask questions about stories (e.g. like book discussions in school), they want to know what the writer and characters in said story did/want to express.

Is there even a clear definition of truth?

@kaspi there isn't a clear definition of truth, at least according to the article. I thought there was prior to reading it, but evidently only about half of people in the study think truth means what I think it means, so I guess it's not that clear 😂

@kaspi @raphaelmorgan the lovely thing about the question is that it is intentionally aiming at the ambiguity between two different understandings of what "true" means. I mean, you're right, the question could be phrased so the ambiguity is resolved, but the whole point of the question was to intentionally achieve the opposite. Also, unlike here, they didn't ask the question to a general population, but as a follow up to a questionnaire that was aiming to understand the different interpretations of "true", several months later, with the goal to see if the abstract questionnaire predicted the answer to this intentionally ambiguous question.

It did.

@raphaelmorgan @janjko yes I was in the same camp. I couldn't even understand the other side! Reading the article and the discussion here showed me how differently the word "true" is understood.
@vrandecic @janjko Yes. Is this not a common interpretation?
A false statement in good faith doesn't fall into the same category as lying, for me. Maria did make her statement too definitive based on, essentially, hearsay. But not exactly a lie.
I would have replied, as far as I know, yes.

@GinevraCat @vrandecic @janjko I'd say she makes her statement on more than hearsay. Tom said he'd be at the party. Unless Tom had a track record of changing his mind or lying, she had enough information to expect him to be there.

Now, if Peter was an attorney about to start questioning a witness, she might want to be more explicit, but she was having dinner before a party...

@ignaziop1977 @GinevraCat @vrandecic @janjko

Yes. In the most strict logical interpretation, Mary has no way of knowing if anyone will be at the planned party. There might have been a gas explosion or plane crash or terrorist attack that destroyed the entire party location.

But unless Mary had reason to believe that Tom "changes his plans" about going to parties, she has reasonable reason to believe what Tom told her. So she did not *lie*.

@ignaziop1977 @GinevraCat @vrandecic @janjko

I could "split hairs" a little and say that Mary probably does not have good reason to believe that Tom *is* at the party *right now*, as he may be having dinner elsewhere right now, and will be showing up "late," as they are.

But this is a casual social conversation where "is" and "will be" and "probably" and "to the best of my knowledge" should generally be assumed.

@vrandecic @janjko

There's a valid argument that Maria lied by implication: by answering "he is there", she was also implying "I know that he is there", and that was a willful or irresponsible misrepresentation of how much knowledge she had of the situation.

So she wasn't lying about whether or not he was there. But she was, arguably, lying by implication that she was in a position to say whether or not he was there, which she wasn't.

@siderea @vrandecic @janjko

This is a really good point, and it puts into words my gut feeling about the matter.

@vrandecic @janjko I do remember reading something somewhere (well that's helpful) about conceptions of lying across cultures where in some places Maria's statements, even though she was not purposely stating a falsehood, would be considered lying
@meeoo @janjko oh, that sounds very interesting! I wonder if it's just a lexicographic issue. If you accidentally remember, I'd be curious.
@vrandecic @janjko after posting this I tried to find it just to confirm I wasn't making things up. But I have no idea where I encountered this
@vrandecic yes, it was a justified belief, but not a true one
@janjko
@Tattie @janjko since this is referencing the JTB definition of knowledge, I try to understand how the concept of knowledge fits into the original question, but I run into circles very quickly and can't lay it down.
@vrandecic @janjko yeah, I've noticed a lot of people saying "I lied" when they mean "I was wrong," and it bugs the shit out of me
@WizardOfDocs @janjko oh, even about themselves? I've seen it about others, but not so much about themselves.
@vrandecic @janjko Usually about themselves, at least on YouTube.
@WizardOfDocs @vrandecic In Croatian it's also somewhat customery to say "I lied" when something you said turnes out to be false. But it's said in a joking, self deprecated manor.
@janjko @vrandecic yeah, I think in North America it also started as a joke, but now it's so deadpan that it might as well not be
@janjko @vrandecic Her answer was false but she wasn't lying; she was simply wrong. It's only lying when you knowingly make a false statement.
@irina @janjko That's also how I understand the terms.
@vrandecic @irina @janjko if Maria had said yes or no without knowing about Tom, this would neither have been a lie, but bullshit.
What the Hell Bullshit Is: Revising Frankfurt’s Definition

Difficulty: What the hell In his work On Bullshit, Harry Frankfurt famously defined bullshit as, roughly, a product of a speaker who is indifferent to the truth. Note that bullshitting is different…

Philosophy without Bullshit
If you claim something is true when you know that you don’t know if it is true or not, then that’s a lie, even if it turns out to be true. @vrandecic @irina @janjko
@BenAveling @vrandecic @janjko Yes; you need to qualify it with "I think X" or "I'm practically sure that X"