Coherentism holds that a belief is justified to the degree that it coheres with the rest of one's belief system. Rather than grounding justification in a foundation of basic beliefs, coherentism understands justification holistically: beliefs support one another through logical entailment, probabilistic support, and explanatory relations. Quine's 'web of belief' metaphor captures the idea that beliefs face experience as a corporate body, with central beliefs (logic, mathematics) being more resistant to revision than peripheral ones. The main objection — the input problem — charges that a perfectly coherent system might be entirely disconnected from the world.
Compare coherentism with foundationalism using the same set of cases: perceptual beliefs, mathematical beliefs, beliefs about the past. Ask which theory better explains why revising one belief forces us to revise others.
You already know about the regress problem: if every belief requires justification from another belief, the chain either goes on forever, circles back on itself, or must eventually terminate in beliefs that justify themselves without support. Foundationalism resolves the regress by stopping it — basic beliefs serve as the bedrock that justifies everything else without themselves needing justification from above. Coherentism takes the opposite approach: it denies that any beliefs are foundationally self-justifying and instead holds that justification is always a matter of how beliefs fit together into a mutually supporting whole.
The core image is Quine's web of belief: imagine all your beliefs as nodes in a network, connected by logical entailment, probabilistic support, and explanatory relationships. No single belief is an island — each one gets its justificatory status from its connections to the rest. Beliefs near the center of the web (logic, mathematics, fundamental physical laws) are more resistant to revision because they are connected to so much else; revising them would require revising enormous portions of the web. Beliefs at the periphery (particular perceptual claims about what I'm seeing right now) are more revisable because fewer connections tie them to the core. When experience conflicts with the web, you can revise at any point — you might revise your perceptual belief, or you might revise your background theory. The web responds as a whole.
What makes a belief justified on this account? Not that it rests on a foundation, but that it coheres with the rest of the system: it is logically consistent with other beliefs, it receives probabilistic support from them, it explains other beliefs, and is explained by them. Notice that coherence is richer than mere consistency — a system can be perfectly logically consistent (no contradictions) yet totally incoherent in the relevant sense if its beliefs have no explanatory connections to each other. If I believe it is raining and I believe the streets are dry, that's inconsistent. But if I believe both that Napoleon won at Waterloo and that pi equals approximately 7, those beliefs are consistent yet entirely unrelated — the web has no structure there at all. Genuine coherence requires integration and mutual support, not just the absence of contradiction.
The most serious challenge to coherentism is the input problem (sometimes called the isolation objection). Suppose I construct an extraordinarily coherent system of beliefs — internally consistent, explanatorily integrated, rich with mutual support — but the system is entirely detached from reality. A paranoid delusion that hangs together flawlessly; a fantasy world perfectly self-consistent but never tested against experience. Is that system justified? Coherentists say justification is a matter of internal relations, so such a system would appear to qualify. But this seems wrong — surely knowledge has to be anchored to the world somehow. Coherentists have responses (they argue that sensory experiences are incorporated into the web, not left outside it), but this objection marks the real fault line between coherentism and views that insist on a more direct connection between justification and truth or reliability.
Topics in reflective domains aren't scored by quiz answers. Read, reflect, and mark when you've thought it through.