Foundationalism and Coherentism are two popular theories of epistemic justification. I'll briefly define each.

Foundationalism: some of our beliefs are foundational, that is, they are self-evidently true, and they do not require demonstration to be justified and known. According to strict foundationalism, everything we know is either a foundational belief or a conclusion derived from foundational beliefs.

Coherentism: none of our beliefs are foundational, and the truth of a belief can only be confirmed by its coherence with other beliefs. Thus, knowledge arises from a network of interdependent and mutually reinforcing beliefs. Whether a belief is justified or not depends on whether it agrees with one's other beliefs.

Now here's my thesis: there are two kinds of knowledge: 1) non-reflective or immediate knowledge, and 2) reflective, mediated, or self-conscious knowledge. The former is justified according to foundationalism, and the latter is justified according to coherentism.

Some things we know without having to reflect upon them. Like other sentient animals, we humans know what we perceive. If we feel hunger or hear a loud noise, we have immediate knowledge of these perceptions. This knowledge is foundational. I don't need to demonstrate that I hear a loud noise in order to know that I hear it. Even if the noise is a hallucination, I still know that I hear it.

But clearly, it is possible for one to doubt that one perceives what one perceives. Parmenides, for instance, seemed to believe that everything is one, from which it follows that perception is some kind of illusion, and perception does not provide any genuine knowledge, foundational or otherwise. Now Parmenides, like any other sentient animal, had immediate, unreflective knowledge of the things he perceived. But given his monistic beliefs, he did not believe this immediate knowledge of his was truly knowledge. Thus, he knew foundational things, like the noises he heard and the odors he smelled, but he did not know that he knew them. I'm only using Parmenides as a random example. If you disagree with my interpretation of him, pick another example, like a skeptic who rejects all forms of knowledge, or a mystic who believes that this world is an illusion that we must transcend.

So we can unreflectively know certain things in a foundational way, but our reflective or self-conscious knowledge that we know these foundational things is not foundational. For if the knowledge that we knew certain things were foundational, it would be self-evident. But since there is no universal agreement about what can be known, it is evidently not self-evident that we know anything. In other words, what we know is, in some cases, self-evident and foundational, but the fact that we know something is never self-evident and foundational.

So how do we know which of our beliefs are foundational? I would argue that we know this by examining our entire network of beliefs. In this process of self-reflection, we will notice that some of our beliefs are derived from others. For instance, if we believe that someone is moving toward us, we can observe that this belief is derived from sensory information. Or if we believe in some complex mathematical theorem, we can observe that this belief is derived from more basic rules and axioms. And once, through this process of epistemic self-examination, we arrive at a belief that is seemingly not derived from more basic beliefs and that seems impossible to seriously doubt, we will have reason to believe that we have discovered a foundational belief. For instance, I cannot think of a more basic principle than the principle of non-contradiction, and I can observe that, if I doubt non-contradiction, all my other beliefs become incoherent. So the principle of non-contradiction appears to be foundational.

I think this synthesis combines both what is appealing about foundationalism and coherentism. Foundationalism is appealing 1) because our knowledge can't be justified by an infinite regress or circular reasoning and 2) because coherentism seems to conflict with our intuition that some beliefs are self-evident, especially beliefs regarding perception. The synthesis accounts for 1) because, since reflective knowledge is subsequent to non-reflective knowledge, our knowledge is ultimately justified by foundational beliefs, and we need not appeal to an infinite regress or circular reasoning. It also accounts for 2) because we do have self-evident beliefs, although they are non-reflective.

Coherentism is appealing because 1) a belief is only known if it is justified, and foundational beliefs do not seem to be justified, 2) because different philosophers disagree about what can be known, making it seem as though nothing is self-evident and there are no foundational beliefs. The synthesis accounts for 1) because foundational beliefs, while indemonstrable at the non-reflective level, must be demonstrated at the reflective level. It makes sense that when engaging reality in a practical way, we wouldn't need to justify all of our beliefs, and some facts we would just figure out intuitively or non-reflectively. But when thinking philosophically and trying to determine what can and cannot be known, one cannot justify a belief simply by claiming it is foundational. Rather, one must show how, in order for beliefs to properly cohere with one another, certain beliefs must be foundational. For instance, we can discover through reflection that our beliefs will become incoherent if we deny that the principle of non-contradiction is foundational. The synthesis also accounts for 2) since people can know self-evident things without knowing that they know them. A philosopher can know something yet belief that he does not know it. This is possible because he knows it in a non-reflective way, whereas he believes that he does not know it in a reflective way.

So in conclusion, we form immediate, unreflective beliefs about reality based on foundational knowledge, and then, by reflecting on these beliefs, we work them into a coherent system that allows us to know them in a reflective and self-conscious way. So unreflective knowledge is justified according to foundationalism and reflective knowledge is justified according to coherentism.

Also, just to be clear, I think both basic perceptual knowledge, such as 'my foot hurts' and certain logical principles, like non-contradiction, are foundational.

Also, please let me know if someone else already thought of this.