Foundationalism versus coherentism

Intuitive problem

Take an ordinary belief: "the pharmacy on the corner closes at nine." Ask why you believe it. You might answer: "because the sign on the door says so." Ask why that counts. You might answer: "because a printed sign is normally an accurate report of a store's posted hours." Ask why that counts, and the chain gets harder to close off cleanly: perhaps "because signs like this are rarely wrong in my experience, and being wrong here would be low-stakes and quickly corrected." Three steps in, the chain has not stopped so much as changed shape. It could keep going, if you demanded a reason for that last claim too. It could double back, if pressed hard enough, on something resembling the first claim, that this sign in particular is reliable. Or it could halt, if at some step you simply refuse to treat "why?" as a live question anymore.

That is the regress problem stated in miniature: for any belief that is justified by another belief, the question "why does that one count?" can be asked again, and asking it enough times reveals that the chain of reasons has to take one of three shapes. It can run on without end, each reason resting on a further one that has never yet been given. It can loop, so that some belief early in the chain reappears later as its own support. Or it can terminate, at a belief that is not itself justified by anything further back. Two of these shapes have become the dominant constructive theories of justification. The third has a real, minority defender.

Definition

Foundationalism takes the termination option. It holds that some beliefs are basic: they are justified, but not by any other belief. A basic belief might be justified by direct sensory experience, by immediate rational insight, or by some other non-inferential source, and the details of which source are exactly what different foundationalist theories argue about. Everything else in the structure is a non-basic belief, justified in turn by basic beliefs or by chains of non-basic beliefs that eventually bottom out in basic ones. The image is architectural: a foundation that does not itself need support, holding up everything built on top of it.

Coherentism takes the loop option, but reframes it. Where the regress problem treats circularity as an obvious flaw, coherentism denies that a belief's justification has to run in a straight line at all. On this view, a belief is justified by how well it fits into a whole system of beliefs held together by mutual support, consistency, and explanatory connection. No belief in the system is more basic than any other; each is justified by its place in the web, not by resting on something outside the web. The regress does not terminate on this account because it was never supposed to be linear. Every belief answers to the others, and the whole is judged as a whole, so the image is a raft at sea, no single plank load-bearing on its own, the structure held afloat by how the planks are lashed together.

Infinitism takes the remaining option seriously rather than treating it as automatically disqualifying. Peter Klein's contemporary defense argues that a chain of reasons can confer justification even if it never terminates in a basic belief and never repeats, provided that at each step in the chain a further, distinct reason is genuinely available to the believer, whether or not it has actually been given out loud. On this view the demand "why do you believe that?" always has a further answer waiting, and the availability of that answer, not its actual production in conversation, is what makes the earlier belief justified. Infinitism has fewer defenders than the other two, largely because an endless, non-repeating supply of distinct reasons is a demanding thing to guarantee for any ordinary belief, but it is a live response to the regress, not a footnote to it.

Each of the three has a debt it has to pay. Foundationalism owes an account of why some beliefs get to stop the chain, given that anything short of certainty seems to invite the same "why?" that started the regress. Coherentism owes an account of why a well-connected system of beliefs is thereby true, or even likely to be true, rather than merely internally tidy; a large, detailed, self-consistent work of fiction is coherent without being justified. Infinitism owes an account of how a chain that never bottoms out can transmit justification at all, when at no finite point has anything actually been secured.

Common mistakes

The first mistake is assuming justification has to be linear in the first place, so that any theory not ending in a foundation looks like it has simply failed to finish. Coherentism denies the linear picture on purpose; judging it by whether it produces a foundation misreads what it is claiming.

The second mistake is treating coherentism as a theory on which any sufficiently large, self-consistent set of beliefs counts as justified, full stop. Coherentists build in more than bare logical consistency: they typically require explanatory connections between beliefs and some responsiveness to experience, precisely to block the objection that an elaborate, self-consistent fiction would otherwise qualify.

The third mistake is assuming a foundationalist's basic beliefs have to be certain or infallible. Most contemporary foundationalists require only that basic beliefs be justified without depending on further beliefs, which is compatible with those beliefs later turning out to be mistaken.

A fourth, closely related mistake is confusing this coherentism, a theory of what justifies a belief, with the coherence theory of truth, a theory of what makes a proposition true. A justification can cohere with a belief system built from propositions that are true by correspondence to the world; coherentism about justification takes no stand on what truth itself consists in.

Test yourself

Pick a belief you actually hold and can state in one sentence, something you would defend if a stranger pressed you on it. Write down the belief, then ask why you hold it, and write down that answer. Ask why that answer counts as a reason, and write down a second answer. Do this for two or three steps, as with the pharmacy sign above.

Now look at what you have written and diagnose it. Does the chain point toward a step you would treat as needing no further defense, something you would call obvious or directly given rather than argued for. That is a foundationalist structure. Does the chain instead keep circling back toward claims that support each other, so that no single step stands alone. That is a coherentist structure. Or does each step you write down open onto a further, distinct reason you had not yet stated, with no natural stopping point in view. That is closer to an infinitist structure, however briefly you can sustain it before running out of patience.

Success on this task has two parts, both checkable against what you wrote. First, a correct diagnosis: the shape you name, foundationalist, coherentist, or infinitist, should actually match the chain you produced, not the shape you would prefer your beliefs to have. Second, a stated objection: once you have named the shape, write the one specific debt that theory owes, an account of why the stopping point does not need defense, why the system's coherence tracks the world, or why the chain's non-termination still transmits justification, and say in a sentence whether your own chain has paid it.