“Why is the sky blue?” asks a curious child on the drive to school in the morning. The child’s parent, being a worldly person, happens to know the answer. “Blue is scattered more than other colors because it travels as shorter, smaller waves.” responds the parent. “Well why is blue scattered more?”

It’s a classic parenting scenario along with “Are we there yet?”, where a parent answers a question like, “Why is grass green?” and a incessantly curious child craves more with the word, “Why?” In moments of great patience, the child may get three, four, perhaps even five good answers, but when the reasons why approach facts about sub-atomic particles, it becomes increasingly difficult to even answer. The practice of asserting facts and asking for reasons to believe those facts is an affair that epistemologists are interested in. If it were possible to answer all of a child’s questions, what would the series of answers look like? What possibilities are there?

This question is known as the “regress problem” in epistemology. If this conversation needed not end and the mother was all-knowing, would the conversation go on forever? The problem that this presents, and the reason it’s so hard to answer children’s unceasing questions, is that this seems to have no end. If this is the case, how can we ever raise the credibility or warrant of a claim? If reasons never reach some end, some inherent truth, how can justification ever reach our beliefs?

What possible solutions are these to this, and by corollary, how can we satisfyingly answer curious kids? Well, the logical space seems to only have a few options: 1. The reasons end, that there is a foundational reason; 2. The reasons loop back on themselves, that reasons need only be coherent; 3. The reasons go on infinitely, that there is never a “last reason”; 4. We are just forming beliefs arbitrarily.

“If we keep going, we’re going to get something foundational.”

Perhaps we build all of our justified beliefs on a bedrock of unquestionable foundations. This is plausible because there could be a set of reasons which it just does not make sense to question.

For instance, imagine again a conversation between a parent a child, this time, say a father and his daughter. The father notices that that there is a blue smear on the living room wall, and on the basis of this forms the belief that his daughter was painting today. “You were painting today? Can I see your painting?”, he asks. The daughter, having not told her father she painted, wants to know his reason for thinking she painted. He responds, “I see the blue smear on the wall over there.” The daughter, in the mood to investigate the world, asks her father “What is the reason you believe that you see a blue smear on the wall over there?”

The intuition of foundationalism, the theory which posits the end of the regress, is that questions like this, and questions about other foundational beliefs, are not valid questions. The father may be justified in responding, “What do you mean what is my reason for believing that I see a blue smear? I have no reason, I just am be appeared to as if there is a blue smear.”

There are problems for this view, however. For instance, what foundation is there for mathematical knowledge? Is the father’s reasons for believing not that “When I am being appeared to as if something, then that something?” What are the conditions for a foundational belief?

“If we keep going, my coherent reasons may repeat themselves.”

Imagine if in the process of describing to someone why the sky is blue, you at some point gave two separate reasons that both cannot be true. It would be perfectly natural for someone to question how you could hold both of them simultaneously, and you would likely try to resolve the conflict, to make your reasons cohere.

This is the intuition behind the coherentist response to the regress problem, where the structure of justification is such that you will eventually loop back around on reasons. In the genealogy of your justification for any proposition, if the cycle is sufficient large, hold coherentists, then you have knowledge.

The problem that this view faces is that it is a longer form of circular reasoning. Where it seems to be acceptable to assert a proposition, and then when asked for a reason, supply that same proposition. Furthermore, there are plenty of coherent systems which are not true.

“This is just going to go on infinitely.”

The feeling that I get when I had a conversation with a child like this is that it just never stops. There is always another reason for believing something, it seems. For instance, if you say, “It’s twelve o’clock.”, and you’re asked “Why is it twelve o’clock?” Well, it is true that a reason that it is twelve-oclock is that it is not 11:59, it’s also not 11:58, …

There’s certainly an end to my knowledge, there’s probably an end to human capacity, but that doesn’t mean there’s an end to potential reasons for believing any given thing. This is the claim and intuition of infinitism. The problem for this view is that if there’s always another reason to believe something, how can you “hook up” a proposition to the truth? Foundationalism has a bedrock, but infinitism needs to come up with an account for raising the credibility of a proposition without foundations to be make it usable.

“Eventually, I’ll have no reason for believing what I do.”

The troubling aspect of the regress problem is that none of the answers are straightforwardly right, none of them are obvious. Yet if none of them are the right view, if the question of the structure of justification is a valid one, then we necessarily cannot be justified in any of our beliefs.

And this would be especially unsatisfying for inquisitive minds.