The JTB Analysis
On this page(8)
Three notions have now been separated: belief, truth, and justification. The obvious next move is to put them together and see whether they add up to knowledge. That move produces the most-discussed proposal in modern epistemology, and the clearest example anywhere in philosophy of a theory being demolished by a three-page paper.
What an analysis is
Definition
Necessary and sufficient conditions
A condition is necessary for something when that thing cannot hold without it, and sufficient when its holding is enough. A set of conditions that are individually necessary and jointly sufficient picks out exactly the cases in question, admitting no others and excluding none.
Definition
Analysis of knowledge
The project of stating conditions that are individually necessary and jointly sufficient for someone to know that something is the case, thereby saying what knowing consists in.
The standard is demanding, and it is worth seeing why each half matters.
If a condition is not necessary, then something can be known without it, and the analysis excludes a genuine case of knowledge. If the conditions are not jointly sufficient, then something can satisfy all of them without being knowledge, and the analysis admits a case that should be excluded. An analysis fails if it errs in either direction, and the history of this particular project is a history of failing in the second direction.
Two further requirements are usually imposed. The analysis must not be circular, so "knowledge is belief that amounts to knowledge" is disqualified whatever its other merits. And it should illuminate rather than merely track, in the sense of explaining why the cases fall as they do rather than listing them.
Check your understanding
The proposal
Definition
JTB analysis
The account on which a person knows that something is the case when three conditions hold: the claim is true, the person believes it, and the person is justified in believing it. Knowledge is justified true belief.
Written out for a subject S and a claim P, the analysis says that S knows that P if and only if P is true, S believes that P, and S is justified in believing that P.
The appeal is immediate. Each condition rules out a recognisable way of falling short, and together they seem to cover the ground.
Why truth
Drop the truth condition and you get the result that people can know false things. Nobody says this. When it turns out that the accused was elsewhere, we do not say that the detective knew he was guilty and was wrong; we say he thought he knew.
The linguistic test is that "knows" is factive: from "she knows that the bridge is out" you may infer that the bridge is out, and the inference is not available for "believes", "suspects", or "is sure".
Why belief
Drop the belief condition and knowledge becomes detachable from the knower's mind, so that a person could know something they take no attitude toward at all. The truth would then be knowable by a filing cabinet.
The unconfident examinee from the first article is the standing challenge here, and the reply given there was that the case shows a failure of confidence rather than an absence of belief.
Why justification
Drop the justification condition and knowledge collapses into true belief, and the lucky bettor knows the coin will land heads. That result is unacceptable to essentially everyone, and the reason is the one Plato gave: a true belief that could as easily have been false, held for no reason, is not the kind of thing anybody wanted.
Check your understanding
What the analysis does well
Before the trouble, the account's successes should be recorded, because they are the reason it took decades to replace.
It handles the ordinary cases correctly. The person who looks at a working clock, forms the belief that it is three o'clock, and is right, satisfies all three conditions and knows. The person who guesses satisfies two and does not.
It explains the three characteristic ways of failing to know, and it explains them differently. Being wrong, not believing, and having no reason are distinct defects, and the analysis assigns each its own condition rather than treating them as degrees of one thing.
It makes knowledge a status that a belief can gain and lose without the belief changing content, which fits how the concept is actually used. The same belief you held on flimsy grounds yesterday can be knowledge today because the evidence came in.
And it locates the disputes. Once knowledge is analysed into three conditions, disagreements about knowledge become disagreements about one of the three, and the third condition is where nearly all of them turn out to live.
Where it fails
In 1963 Edmund Gettier published a paper of about three pages titled "Is Justified True Belief Knowledge?", and answered no.
His method was to construct cases satisfying all three conditions that are plainly not knowledge. The recipe is general rather than a trick about a particular example. Take a belief that is justified and false, infer from it something that happens to be true, and the inferred belief is justified, true, and believed, while its truth has nothing to do with the reasons for holding it. The believer is right by accident, and being right by accident was the thing the justification condition was added to exclude.
The response was immediate and the repairs were many, and the intermediate tier works through both the cases and the attempted patches. What matters at this stage is the structural point rather than the details. Justification and truth were introduced as separate conditions, and nothing in the analysis requires them to be connected. Gettier's cases exploit exactly that gap.
Objection
The failure is not a failure of the JTB analysis but of the project it belongs to. Sixty years of proposed conditions have each been met by a new counterexample, and that pattern is what you would expect if knowledge is simply not the sort of thing that decomposes into simpler parts. The right conclusion is that the analysis of knowledge should be abandoned rather than repaired.
Reply
Timothy Williamson has argued for something close to this, holding that knowledge is a basic mental state that resists analysis into belief plus further conditions, and that the explanatory order should run the other way, with knowledge used to explain evidence and justification rather than being built out of them.
The position is serious and it is not free. Giving up the analysis means giving up the explanation of why the three-way division of failures looks so natural, and it means saying something about why the failed analyses got as close as they did.
There is also a moderate reading available. The project can fail as a search for an exact definition while succeeding as a way of mapping the terrain, since the counterexamples are what taught the field about luck, about the connection between reasons and truth, and about the modal conditions the intermediate tier examines. On that reading sixty years of failure was productive rather than wasted, which is a claim about the history that the history supports.
Check your understanding
What has been settled
The analysis is false, and it is still the right thing to have learned first. Every subsequent theory is stated as a modification of it, and the vocabulary of the field is its vocabulary.
One question has been deferred throughout this tier and can no longer be. Justification was said to require reasons, and reasons are usually other beliefs, which will themselves need reasons. The next article asks what happens when that demand is pressed.
Further reading
- Edmund L. Gettier, "Is Justified True Belief Knowledge?", Analysis, volume 23, number 6 (1963), pages 121 to 123.
- Plato, Theaetetus, 201d to 210a, for true belief with an account and its rejection.
- Timothy Williamson, Knowledge and Its Limits, Oxford University Press (2000), for the knowledge-first alternative.
- Stanford Encyclopedia of Philosophy, "The Analysis of Knowledge".
Quiz yourself on this article
8 questions. Miss one and it comes back until you get it right.
Reached the end?
Mark it, and the sidebar keeps track of where you have been.