I’ve often heard the phrase “I don’t know, but I know you don’t know” in my apologetics conversations. An accusation of arrogance and rebuke to walk me back to a proper agnostic posture typically follows. It’s as if one’s claim to know is an insult to another’s uncertainty. Why is that? It’s hard to say precisely, but I suspect the underlying motivations are primarily emotional since it is rarely a tenable philosophical position. So let’s take a look at why one ought to avoid such phrases.

What does it mean to know, and what kind of knowledge is in play? When discussing a controversial topic, propositional knowledge leads to the above accusation. No one has ever told me: “I don’t know, but I know you don’t know your wife or how to ride a bike.” Instead, we are interested in claims about things taking the form of person S knowing that proposition P. So what does it mean for S to know that P? Traditional views of knowledge vary around the notion of justified true belief (JTB.) For S to know that P:

S must have justification for their belief.

P must be true.

S must believe that P.

The latter two aspects of the tripartite view are relatively straightforward. However, it is the aspect of justification that is controversial. What does justification mean? Mostly it is sound reasoning for believing something. But what is sound reasoning? A minimalist response called the deontological view (DV) gives us a place to start by placing a low burden on the knower:

S is justified in believing that P if and only if S believes that P, while it is not the case that S is obliged to refrain from accepting that P.

Given this view of justification: I know my keys are hanging downstairs; if it is the case they are hanging downstairs, I believe it, and I am not obliged to refrain from accepting it. What would obligate me otherwise, you might wonder? Some other knowledge acting as a defeater would be an example. Say my wife says she sees my keys in the car, not where they usually are. If I trust her assessment more than my recollection, I am obliged not to believe the keys are where I initially thought they were.

On the other hand, I would not know it’s raining tomorrow in Tallahassee, even if I believe this proposition and it turns out to rain tomorrow if my sole justification for thinking is the prediction of a fortune cookie. I’m obliged to refrain from believing the printed prophecies in fortune cookies. Fortune cookies confer no justification in this case; if it’s my only justification, I don’t have any.

There are more rigorous approaches to justification than DV. In our current culture of scientism, some form of evidence is often a requirement. Does S have evidence for believing that P? Is it good evidence? These controversial criteria are debatable within epistemology. Giving the modern skeptic the benefit of the doubt, I’ll concede justification for the knowledge we are discussing requires more than a lack of defeaters obligating me to refrain from believing. I’ll go as far here to say it needs some positive external grounding — a sound argument based on evidence being a good example. Now that we have defined knowledge let’s move on to the problem.

So how could S’, who doesn’t know that P, know S doesn’t know that P? The short answer from the JTB perspective of knowledge is uncomplicated. Leaving out the question of the sincerity of S by assuming S believes that P: S’ would have to know that P is false or that S has no justification. Take my example of the car keys. Let’s say S’ tells S: “I don’t know if your car keys are hanging downstairs or not; I just know you don’t know that they are hanging downstairs.” How could S’ know this if she doesn’t know whether or not they are hanging downstairs? She can’t focus solely on the truth value of P because she needs to know whether P is true or false. This truth-value angle is a dead end for S’.

But let’s imagine S’ took an epistemology class and challenged the justification-claims of S. “Why do you think your keys are hanging downstairs,” she asks? Now S may have all sorts of justification for believing that P (I won’t bore you with examples.) S might be justified in believing that P. Therefore, S’ has a burden here because she is claiming to know S is unjustified in believing that P. That burden is not just challenging to satisfy with the car-keys example but in your typical real-world discussion of complex and controversial topics. Let’s look at something more representative to see what I mean.

S claims the universe has a cause for its existence (P). S’ says: “I’ve seen the evidence and arguments, and on balance, I don’t know, maybe it does, maybe it doesn’t. I merely know you don’t know that it does.” S’ freely admits she doesn’t know if P is true or false. S’ is agnostic on P. After S tries to persuade S’ with arguments Q…Qn, S’ says: “I find all of your arguments unpersuasive.” Does S’ remaining agnostic and unpersuaded mean S has no justification for believing that P? Not necessarily, and in many cases, not likely.

Say S gives the following argument Q as justification for P:

P1 – Things that begin to exist have a cause.

P2 – The universe began to exist.

P (conclusion) – Therefore, the universe has a cause.

The above deductive argument is valid; if the premises are true, the conclusion follows inescapably. Therefore, if S’ knows S has no justification for believing that P, she must know (at least): Q does not confer justification. But how could she possibly know this without knowing either P1 or P2 is false or unjustified? She can’t. She cannot logically infer P1 or P2 is false merely because she is unconvinced P1 and P2 are true. Nor can she know the belief of S in P1 or P2 is unjustified unless she knows all of the justification-claims of S for the premises.

What typically transpires while discussing a valid argument like Q is S’ says: “You have not convinced me P1 or P2 is true. So your argument is unpersuasive.” That’s fine; how persuasive Q is to S’ is partly up to S,’ but that hardly means Q is unsound, thereby not justifying S. With a deductive argument, justification is conferred as follows:

Q justifies S if Q is valid, and upon S performing their epistemic duty (considering the arguments and evidence) for P1 and P2, the conjunction of the premises is more plausible than its negation. 

Of course, this doesn’t mean P1 and P2 are true! Nor does it say the conclusion (P) is accurate. Remember, we are talking about justification, not truth values. Of course, S’ being agnostic on P1 and P2 may enter into a regress and attack the rationale for believing the premises. But this rarely happens, and when it does, the problem is pushed back to the next level. When S’ fails to ask for justification from S for P1 and P2, we know S’ doesn’t know S is unjustified.

I’ve given a somewhat technical explanation as to why the phrase “I don’t know, but I know you don’t know” is usually philosophically untenable. The person who levels this claim must determine the truth of the proposition. If they fail this, they must learn all of the possible ways you justify your belief. Otherwise, they don’t know if you know or not. A more straightforward way to address this kind of unreasonableness would be to respond: How do you know I don’t know? And as for humility: If you don’t know something is true, admit it without attacking your interlocutor. The genuinely humble attitude is: Maybe you do know; I’m just not sure. This response is more reasonable and more likely to get the other person to consider your position.