Knowledge

In plain terms

Knowing something is more than believing it, and more than believing something that happens to be true. A person who guesses correctly did not know.

So a third ingredient is needed - grounds, evidence, reliability, something that connects the belief to the fact rather than leaving the match to luck. That gives the classical formula: knowledge is true belief with the right kind of backing.

The formula survived for a long time and then broke on a simple observation. You can have good grounds, and a true belief, and still have got there by accident, because the grounds support the belief for reasons that turn out to be wrong while the belief happens to be right anyway. Repairing that has occupied epistemology for sixty years without producing an agreed answer.

As a philosopher would put it

The tripartite analysis makes knowledge justified true belief. Gettier's 1963 counterexamples show it insufficient: a justified belief can be true through luck in a way that disqualifies it, and the cases generalise easily.

The responses split. Internalists add a fourth condition - no false lemmas, or defeasibility - and keep justification as something accessible to the knower. Externalists change the third ingredient entirely: reliabilism makes knowledge belief produced by a reliable process, whether or not the knower can cite the process, and virtue epistemology makes it success attributable to an intellectual competence.

Williamson reverses the direction of analysis altogether: knowledge is not a compound of belief plus extras but a primitive mental state, the most general factive one, and belief and justification should be explained in terms of it rather than the other way round.

The Indian pramāṇa traditions ask a different question - not what knowledge is but what sources yield it - and Nyāya's account of pramā as non-erroneous cognition arising from a reliable instrument arrives at something close to reliabilism from a different starting point.

What is actually disputed

Whether knowledge can be analysed at all.

Sixty years of post-Gettier proposals have each met counterexamples, which some read as a sign that the project is misconceived. Williamson's knowledge-first programme takes that seriously and treats knowledge as unanalysable, which is either a breakthrough or a surrender depending on who is describing it.

A second split is over access. Internalists hold that what justifies a belief must be available to the believer, on the ground that otherwise a person could know without having any idea why. Externalists reply that this makes knowledge unavailable to children, animals and most ordinary perception, all of which we unhesitatingly call knowing. And contextualists argue that the standards shift with what is at stake - that a person can know where their car is parked in conversation and not know it in a courtroom, without anything about them changing.