Foundationalism and coherentism (October 2003):
Beliefs, it seems, are often justified by other beliefs. "Why do you know A?" "Because I know B, and A follows from B." But then one can go on to ask "Well how do you know B?" And B follows from C which follows from D which .... This is roughly the substance of a problem known as the regress problem. How can any of these beliefs be justified?
Here are three possible answers: (1) Our beliefs aren't unjustified. Any justification must appeal to another belief, and so beliefs are structured in a never-ending chain and (reductio ad infinitum) skepticism follows. (2) There is a chain of beliefs, but it is not never-ending. Some beliefs require no further belief. These are the foundational beliefs upon which the other beliefs are ultimately justified. (3) The structure of belief is not a chain, but a closed structure with no beginning and no end.
I'll focus on these last two approaches, which are often associated with foundationalism and coherentism respectively. According to foundationalism, most beliefs are based on -- and follow directly from -- other beliefs. A few beliefs, however, need no further justification; these are foundational beliefs. According to foundationalism, the structure of knowledge is like a tower or a pyramid, with a small base providing support for many beliefs.
This seems very attractive, but it is not without problems. The foundationalist must answer questions such as "What are some examples of foundational beliefs?" and "How can a foundational belief be justified if not by appeal to another belief?" In response to the first question, a foundationalist might give special status to beliefs about the contents of one's own mind. In response to the second, a foundational can sketch an account of self-justifying beliefs.
Coherentism can be seen as a reaction to foundationalism. For those who are puzzled by the notion of foundational beliefs but also want to reject the skeptical hypothesis, coherentism is appealing. But coherentism itself isn't without problems. Coherentism's critics frequently charge that coherentism fails because it is dependent on circular reasoning. A coherentist must answer to the question "How can A provide justification for C, when C provides justification for A?" One answer is that viewing the coherentist structure as a circle or otherwise closed chain is a misrepresentation. A coherentist structure is more holistic. Each belief may rely on all the other beliefs, even including itself. Coherentism is not represented by a pyramid, but by an arch, where there are no foundational pieces and yet a stable structure emerges from mutual support.
Additionally, one might argue that coherentism doesn't seem to provide justification, but rather merely consistency within a system. It seems that many alternate systems (or arches) are possible and so accepting any specific one seems arbitrary. It seems that coherentism makes it difficult to consider anything less than "the system as a whole'. I am not certain that this points to a problem with coherentism, but it does represent a very different -- holistic -- way of thinking about knowledge than the foundationalism approach provides.