Epistemologists talk about various epistemic features of belief. These include "justification," "warrant," "rationality," "probability," and others. Anyway, of these four terms, the term that has been most widely used and discussed, in the past twenty years or so anyway, is justification. There is a set of beliefs about what differentiates knowledge from mere belief, called the theory of justification, this theory proposes to answer the question of which is: "What is justification?" or "When is a belief justified, and when not?"
It's normal to say that more than just beliefs are justified; actions, emotions, laws, theories, a whole panoply of things can be justified or not. But in epistemology it's mainly beliefs that are said to be justified or not. So the topic is not best stated as "justification," but as "justified belief". This raises a question: "When is a belief justified, and when not?" Another way of putting this is to ask: "What does the term, justified belief, mean?
One popular way of explaining the theory of justification is to say: A justified belief is one which we are within our rights in holding. But this raises another question: what sort of rights? Not political rights, of course; not even moral rights, generally the rights in question are taken to be "intellectual" rights. The idea is that we want to be responsible about what we believe. We don't want to just go off and believe anything. In some sense, it's bad just to believe whatever happens to pop into your mind. According to the theory, you have an intellectual responsibility or obligation, to believe what is true and to avoid believing what's false. If you're being intellectually responsible, then you're within your intellectual rights in believing something; and the theory says you're justified in your belief.
Thus, justification is a normative notion. That means that it has to do with norms, rights, responsibilities, obligations, and so forth. The standard deffinition is that a concept is normative iff it is a concept regarding or depending on the norms, or obligations and permissions (very broadly construed), involved in human conduct. It is gennerally accepted that the concept of justification is normative, because just because it is deffined as a concept regarding the norms of belief.
Here's a third general feature of justified belief: If a belief is justified, there is something which justifies it. The thing which justifies a belief can be called its justifier. In this case, proponents of the theory of justification say: If a belief is justified, then it has at least one justifier. An example of a justifier would be some evidence that I accept. For example, if a woman is aware of the fact that her husband returned from a business trip smelling like perfume, and that his shirt has smudged lipstick on its collar, the perfume and the lipstick, can be evidence for her belief that her husband is having an affair. In that case, the justifiers are all the perfume and the lipstick, or more specifically her acceptance of that evidence; the belief that is justified is her belief that her husband is having an affair.
But this begs and important question: What sort of thing can be a justifier?
Not all justifiers would have to be what can properly be called "evidence"; there might be some totally different kind of justifiers out there. But to be justified, a belief has to have a justifier.
So far, I've just been reciting some general features of justification that pretty much everyone agrees on. But now we're approaching the point where people disagree. Because what people disagree about is what the justifiers are. In other words, the following question is the main bone of contention: What sorts of things are justifiers? There are, to simplify, three answers -- three different views about what sorts of things can be justifiers:
(1) Beliefs only.
(2) Beliefs together with other conscious mental states.
(3) Beliefs, conscious mental states, and other facts about us and our environment (which we may not have access to).
I'm not going to give you examples of each of these three kinds of justifiers; I'll save the examples for the theories of justification that I'm about to explain to you. Right now all I want you to see is that there are those three different views about what sorts of things that justifiers can be.
But at least sometimes, the justifier of a belief is another belief. When, to return to the earlier example, the woman believes that her husband is having an affair she bases that belief on other beliefs -- namely, the lipstick and perfume. Strictly speaking, her belief isn't based on the evidence itself -- after all, what if I didn't believe it? What if I thought that all of that evidence were just a hoax? For that matter, what if the evidence existed, but I didn't know about it? Then of course, my belief that O. J. is a murderer wouldn't be based on that evidence, just because I didn't know it was there at all; or, if I thought the evidence were a hoax, then surely my belief couldn't be based on that evidence. So here is my point: when my belief is based on evidence, actually what my belief is based on is another belief, namely, a belief or beliefs about the evidence.
For convenience and brevity, let me use letters, P's and Q's, to stand for beliefs. So the idea we're considering now is that a belief, Q, can be justified by another belief, P. So P justifies Q. Now I want you to consider a possibility -- see what you think of this. Suppose that P -- the justifier -- is itself, not justified at all. Suppose it's just totally arbitrary. I picked the belief out of a hat; I got it from a fortune cookie. Then would it be possible for P to justify Q? Surely not.
Let me give you an example. Suppose I believe that there is intelligent life on Mars, and I base this belief on a further belief, that there is a feature on the surface of Mars that looks like a face, and that this face could only have been made by intelligent life. So the justifying belief is: That face-like feature on Mars could only have been made by intelligent life. And the justified belief is: There is intelligent life on Mars. But suppose further (something that really is the case) that the justifying belief is itself totally unjustified. I am in no way in my intellectual rights if I suppose that this face-like feature on Mars could have only been made by intelligent life; that view of mine is totally irresponsible, intellectually speaking. Would my belief that there is intelligent life on Mars be justified then? It has a justifier, doesn't it? Yes. But the justifier is itself not justified. So definitely belief in life on Mars is unjustified; if it's going to be justified, it has to be justified by some belief which is itself justified. If the justifying belief isn't itself justified, then it can't do any justify anything else, no matter how much I might want it to.
So I think we can accept a general rule:
If a belief, Q, is justified by another belief, P, then P must itself be justified.
If P isn't justified, then it sure can't justify Q. The only way that P could justify Q is if P is itself first justified.