Context: this is a follow-up to my last question Is the hallucination hypothesis always the best explanation?

Suppose A has a subjective experience (or multiple subjective experiences) that leads them to believe in the existence of X (they claim to have observed/experienced X). There are two extreme cases:

Case 1: if A is the only person on the planet who claims to have subjectively experienced X, then the hypothesis that A hallucinated X is considered highly plausible.

Case 2: if X is reported to have been experienced by all or most people on the planet, we normally say that X is a mundane experience, or in other words, that there is a high intersubjective agreement about the existence of X, and therefore the hallucination hypothesis is typically considered highly unplausible.

But what about cases situated between these two extremes?

For example, imagine that there is a very peculiar experience that only 0.1% of the world population has access to under very specific conditions/circumstances. People who have this experience typically conclude that Y exists, with very high intersubjective agreement (within this privileged group), although the remaining 99.9% of the world population would disagree, since they have never experienced Y (we could say that they are Y-blind).

I'm using 0.1% as an example, but more generally, what is the minimum threshold of intersubjective agreement (1%, 10%, 50%, 99%) that would make belief in Y justified (from the point of view of the people who (claim to have) experienced Y)? Does the amount of required intersubjective agreement depend on (the nature of) Y, and if so, how?

Additional thoughts

Is it possible to be justified in trusting one's subjective experiences even if no-one else claims to have experienced the same? I'm thinking of hypothetical situations such as a person being abducted by benevolent aliens who take them on a tour around the galaxy and then bring them back to the Earth, without producing any sort of physical evidence that this person could provide to convince others. In this case, the intersubjective agreement would be 0%, and yet would this person be justified in their belief that the things they experienced were real? We can extend this thought experiment to people who claim to have personally experienced ghosts, demons, angels, deities, God, astral projection, nonduality, psychic abilities, witchcraft, mystical experiences, miracles, paranormal stuff, etc. By the way, this question probably overlaps with Can I know something but not be able to justify it to anyone else?

EDIT: Someone in the comments objected to my 0.1%/99.9% hypothetical example with the following counterexample: Give us some credit! Only 0.008% of the world population experience tritanopia, yet there is no serious doubt among the 99.992% of us who don't experience it that it exists. Very good point. I concede the critique. The remaining 99.9% might in fact agree, disagree or withhold judgement depending on the epistemological standards of each individual, so the blanket assertion "everyone else would disagree" would not hold true in many cases. However, if the experience Y being reported by the minority group is quite extraordinary and worldview challenging, the rest of the people are more likely to react with skepticism and explain away the experience with different hypotheses (e.g. "they are lying", "they hallucinated", "they were deceived by an illusion", "they are crazy", "they are misinterpreting their experiences", "they are exaggerating", etc.)

  • 3
    It is not about agreement on "experiencing" exclusively. Even peculiar experiences can be indirectly corroborated. If the 0.1% independently and stably agree on what is experienced then Y is, at least, a shared illusion with some tangible basis to it. If, moreover, they share some biological or psychological trait plausibly linked to why only they experience Y (seeing ultraviolet, for example), instruments can detect it, etc., then existence of Y as what they take it for becomes plausible. And conversely, even 100% agreement does not guarantee that it is not a shared illusion (rainbows).
    – Conifold
    Nov 21 at 0:01
  • 1
    A low percentage of people have experience of Nonduality.
    – Scott Rowe
    Nov 21 at 11:54
  • @ScottRowe Thanks for suggesting that example. I added it to the list.
    – Mark
    Nov 21 at 13:00
  • 1
    @ScottRowe One might reasonably wonder how much of non-duality claims is from distinct experiences, and how much is from shared experiences where they simply reached a different conclusion from others. Those that reject non-duality are certainly more likely to not put such experiences into a separate category.
    – NotThatGuy
    Nov 21 at 13:08
  • @NotThatGuy yes, whatever it may be, Nonduality seems to be just a human capability. Riding a roller coaster is a bit like tripping and falling down, but the drop lasts much longer and doesn't hurt as much. Those planes that produce weightlessness for a few seconds is farther, and being in orbit farther still. But to dismiss orbital weightlessness because one has not experienced it is not sound.
    – Scott Rowe
    Nov 21 at 13:36

7 Answers 7


Intersubjective agreement isn't required at all, strictly speaking. But it does help.

  • For one particular topic, if we grant that some reasonable portion of humans are rational, it suggests that those people would all reach the "correct" conclusion, given the same data. It's possible that a person is the only rational one, but it seems highly unlikely at face value (especially given the similarities between humans in terms of biology and environment, and given our ability to share information).

    The same can't necessarily be said about irrational people, since a flaw in one's thinking could end up with infinitely many different conclusions (for all the things that aren't true).

    Although humans have similar tendencies, so they tend to have similar flaws in their thinking, and we are a social species capable of and inclined to convince others of our beliefs. So that's how you end up with many people unjustifiably believing the same falsehood.

  • If other people reached different conclusions from the same data, their analysis of the situation could help you identify flaws in your thinking. This is especially the case if they used to reach the same conclusion, but then realised the flaw in their thinking.

  • Presenting your justification for a particular conclusion to someone else could enable them to identify flaws in your thinking.

(Some) people also generally discuss epistemology, fallacies and what makes for a good justification for believing something, which should help with all of the above (especially the first point).

Of course I'm assuming that people have had similar experiences to one another. If someone were to say they're the only person to ever have experienced anything like what they experienced, that would be quite a bold claim.

Well, there has to be a first person to experience any given thing, but modern humans already have a long history to draw from of others humans experiencing things, so it's not very likely for any given person to experience something that's never been experienced. The only place to reliable find "first experiences" would be in science and technology with people making new discoveries and inventing things. A scientist that just discovered something may very well be justified in believing things related to their discovery that no-one else believes, at least until they share their discovery with others and/or others reproduce it.

  • 2
    Agreement comes with slowly expanding the realm of shared experience and assumed true things at the margins, asking people to go one step farther at a time. I think this is the purpose of Logic, to show all the steps in an agreed way. If something is too "far out", people will just dismiss it. But the shared base is not evenly shared, and people stand on different edges of it, unfortunately.
    – Scott Rowe
    Nov 21 at 11:53

I just watched a Star Trek episode in which someone says "I know this, trust me". The captain immediately swings into action, because he knows the person and trusts their judgement. In the real world, absent exigent circumstances, is this justified? A significant proportion of the population say that they know God. Corroboration among people who think similarly is useless. I do not know what they mean by knowing God nor do I believe in God. I take no action and am not obliged to. As for subjective experiences, in a real sense, I know nothing but that I am a conscious being that can perceive and think. I can reasonably infer the existence of other minds. I have no reasonable expectation of convincing anyone else of personal experiences that I cannot produce evidence for.

  • 1
    Partly it depends on the situation. If it was a case where there is an emergency and there are no other proposals for avoiding disaster, then yes, act on the info from someone you trust. It still rests on the 'captain' to give the order, regardless who suggested it. The old saying is, "There are no atheists in foxholes"
    – Scott Rowe
    Nov 21 at 15:09

To what extent is intersubjective agreement required for one to be justified in trusting one's subjective experiences?

There's no convention by which one can answer this question across all societies. However, from the point of view of the empiricist who is a fallibilist, it is highly desirable. Should one trust one's own experience? Well, the first time one should doubt, and then if one can repeat the experience, the doubt should be diminished, and by the 10th trial, one should be somewhat confident. But one then has to make the case to others.

Therefore, in science, intersubjective agreement is called peer-review, and the trick is to create a report of the first-person perspective in such a way that the experience is open and repeatable by anyone, and to encourage as many people as possible to confirm the experience which isn't always an easy thing to do.

One large impediment is that prior theory helps to determine perceived events. This was the downfall of logical positivism (the discovery of theory ladenness) and afterwards, Carnap's ideas again were attacked with the Myth of the Given which denies that sense data are strictly speaking empirical, and not shaped by selective forces of the mind. Ultimately, the scientific thinker must be mindful that the best route to trust one's subjective experience is through personal rational and empirical disconfirmation as Popper argues for in his Conjectures and Refutations. A fallibilist is always happy to concede defeat.

I would, of course, urge caution at believing one is justified or not. Justification is on the continuum with no justification on the one end to many forms of arguments for justification on the other, and thus it is not a question of whether or not something is justified, but to what extent. The prior is a hallmark of black and white thinking, and in the real world, one is often greeted by a collection of confirmatory and disconfirmatory data that one has to make sense of.

  • As for real world scenarios, many reputable people have claimed to see advanced craft operating in military airspace with impunity. Cmd. David Fravor and Lt. Cmdr Alex Friedrich, along with 2 other pilots famously claim to have witnessed and chased such a craft in advanced fighter jets. Even with various videos and CIC witnesses, there is much hostility and skepticism to their claims. en.wikipedia.org/wiki/Pentagon_UFO_videos
    – J D
    Nov 21 at 15:03
  • This answer is great, is the 'scale of justification' applicable to inductive arguments? Nov 25 at 11:47
  • @RichardBamford Doubly so. One has two axes to evaluate induction, which is by definition a probabilistic argument. First, by comparing the argument with other arguments, and by the range of evidence offered by the argument itself. Presumably, one has considered the range of evidence during the construction of the argument. Every argument by induction entails an acceptance that conclusion is only probably true, therefore there is a duty to consider why the argument itself is probably wrong. People who believe there is no need for peer review miss a fundamental point...
    – J D
    Nov 25 at 15:36
  • They are overconfident in their ability to evaluate their own evidence and argument built atop it, and are usually blind to the fact that our language is constructed to reflect our preferences without conscious awareness of what those preferences are. Consider the case of the Seekers here in Chicago who refused to grasp the implications of disconfirmatory evidence. While this is extreme, this failure to take into account others' is rather ordinary in most people's belief systems.
    – J D
    Nov 25 at 15:43
  • One often hears the term reality-expectancy discrepancy and in the negative outcome disconfirmed expectancy. Also consider en.wikipedia.org/wiki/Dunning%E2%80%93Kruger_effect. en.wikipedia.org/wiki/Ne_supra_crepidam
    – J D
    Nov 25 at 15:45

There should always be some room for skepticism, even in the face of high intersubjective agreement.

For example, there are many common optical illusions, which most people mis-interpret (even if you know that it's an example of one, it can be hard to see the reality). Thus, there is high agreement for the incorrect subjective experience.


Trust or distrust is a subjective experience. It arises both in the presence and in the absence of intersubjective agreement. Justification efforts sometimes arise alongside the trust or distrust in the context of intersubjective interactions.

Distrust does tend to arise in the context of intersubjective disagreement. I tend to trust my judgment and to distrust the judgment of others in the context of disagreement. Or I tend to trust their judgment and distrust my judgment sometimes in the context of agreement or agreement to disagree.


Strictly speaking, intersubjective agreement is in-and-of-itself a personal subjective experience, and is therefore not entirely trustworthy.

In short; intersubjective agreement isn't a foolproof way of establishing the existence of other subjects. You rely on your own personal subjective experiences of other people to make these inferences.

For example, while you're probably right to question the giant Godzilla monster tearing up the city in front of your eyes, and you're probably right to run away with all the other screaming bystanders, there's no way to prove that the other screaming bystanders are, in fact, real.

While I'm not personally a solipsist, it's hard to get around the simple fact that, somewhere in the chain of reason, we make some leap of faith to the notion that other subjects are real, and if we do so, we must accept that the information gleaned from these other subjects hinges upon that leap of faith. Kant talks about similar notions in his Critique of Pure Reason.

TL;DR - If you're going to use intersubjective agreement to justify your subjective experiences, at some point you must use some other source of justification to justify your subjective experience of intersubjective agreement.

Edit: As a tangential point, it's also worth noting that even if you concede that other subjects exist, you don't have a reliable way to ensure their testimony is trustworthy. They might be lying, mistaken, or outright insane.


This is an extremely straightforward application of Bayes' theorem.

Suppose we give prior values to:

  • The probability P(H) that person A has an hallucination (using what we know about A, or what we know about humans in general);
  • The probability P(X) of person A experiencing event X (using what we know about X in general).

Further suppose that event "A claims to have experienced X" is equivalent to the event "A either hallucinated or experienced X", and suppose that X and H are independent.

Then what you are asking for is to calculate P(X | X or H).

Bayes' theorem states:

P(X | X or H) = P(X or H | X) P(X) / P(X or H)

Tautologically P(X or H | X) = 100%, and together with our assumptions this gives:

P(X | X or H) = P(X) / P(X or H) = P(X) / (P(X) + P(H) - P(X)P(H)) = 1 / (1 + P(H)/P(X) - P(H))

Putting precise prior values on P(X) and P(H) is a bit hard, but they're both pretty small, so this is approximately the same as:

P(X | X or H) = 1 / (1 + P(H)/P(X))

As a matter of example, if we estimate that 1 out of 1000 people will hallucinate that they have seen a God, and we estimate that there is only 1 out of a billion chances that a God will show itself to a specific person, then the credibility of A's claim to have seen a God is: P(X | X or H) = 1 / (1 + 0.001 / 0.000000001) = 0.0001%.

Conversely, if we estimate that 1 out of 50 people will hallucinate that their spouse is cheating on them, but actually 1 in 10 spouses is cheating, then the credibility of A's claim to have been cheated on is: P(X | X or H) = 1 / (1 + 0.02 / 0.1) = 83.33%.

  • epic answer, thanks
    – user67675
    Nov 22 at 13:28
  • It should have been P(X) / (P(X) + P(H) - P(X)P(H)) (with subtraction instead of addition), since you have to remove the intersection, not add it. Otherwise, interesting answer, but you failed to model intersubjective agreement. What if multiple witnesses report the same observation? How do you account for that?
    – Mark
    Nov 22 at 13:55
  • I find the notion that someone can reasonably estimate "whether God has shown themselves to someone" to be a little absurd. Nov 23 at 0:04
  • @Mark then it likely increases the probabilities (multiple eyewitnesses reporting the same event make it more credible), or decreases them depending on the observation (multiple people all claiming they won the same lottery means they're probably wrong if we know that there is at most one winner). With a few more assumptions it should be possible to calculate exactly how that affects the probabilities.
    – Stef
    Nov 23 at 10:15
  • 1
    @ConnieMnemonic Indeed it is. The discussion here is not how to get a reasonable estimate of whether God has shown themselves to someone, but rather how to update your prior doubts when new information arrives.
    – Stef
    Nov 23 at 10:20

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .