The Gettier Problem: Knowledge, Justified True Belief, and the Responses of Contemporary Epistemology

In 1963, Edmund L. Gettier (1927–2021) published in the journal Analysis a three-page article titled “Is Justified True Belief Knowledge?” (Analysis 23, 1963, pp. 121–123). In fewer than a thousand words, Gettier overturned a consensus of more than two millennia: the definition of knowledge as justified true belief. Rarely has so short a paper produced so lasting an impact in the history of philosophy.

1. The Classical Definition: JTB

The tripartite definition of knowledge — knowledge as Justified True Belief (JTB) — is often traced to Plato’s dialogue Theaetetus (c. 369 BC), where Socrates successively examines and refutes three definitions of knowledge. In the passage known as the “account” definition (logon didonai, 201d–210a), something close to the JTB conception appears, though Plato ultimately rejects it as well.

Whatever its historical origins, the canonical definition that Gettier targeted can be stated thus: an agent S knows that p if and only if:

1. p is true (truth condition);
2. S believes that p (belief condition);
3. S is justified in believing that p (justification condition).

This definition seemed intuitively compelling and philosophically robust. It requires that what one knows be true (knowledge is not about falsehoods), that the subject genuinely believe it (knowledge is not mere lucky guessing), and that there be good reasons for the belief (knowledge is not a hunch).

2. Gettier’s Counterexamples

Gettier presented two cases that satisfy all three conditions of the JTB definition yet in which, intuitively, the agent does not possess knowledge.

2.1 Smith and the Job

Smith has applied for a job. Smith has strong reasons (justification) for believing that Jones will get the job (the president of the company told him so, and Jones has ten coins in his pocket). On this basis, Smith forms the justified belief that “the man who will get the job has ten coins in his pocket.”

As it turns out, Smith himself will get the job — and Smith also has, unbeknownst to himself, ten coins in his own pocket. The proposition “the man who will get the job has ten coins in his pocket” is true, and Smith was justified in believing it — but the reason it is true has nothing to do with the reason Smith believed it. Intuitively, Smith does not know that the man who will get the job has ten coins in his pocket.

2.2 Jones and the Ford

Smith has strong evidence for believing that Jones owns a Ford (Jones has always owned Fords, recently offered Smith a ride in a Ford, and so on). On this basis, Smith validly infers the disjunction “either Jones owns a Ford, or Brown is in Barcelona.” Smith has no idea where Brown is.

As it happens, Jones does not own a Ford (he rented one); but, by pure coincidence, Brown really is in Barcelona. The disjunction is true. Smith was justified in believing it. Yet Smith does not know the disjunction is true — because the reason it is true (Brown being in Barcelona) is entirely independent of his justification (which was about Jones’s Ford).

2.3 The Structure of Gettier Cases

What the two cases have in common is a structure: the agent’s belief is justified by evidence that could, in principle, support the truth of the proposition, but the proposition turns out to be true for a different reason — one unconnected to the justification. There is a “causal break” or “deviance” between justification and truth. The agent has justification, belief, and truth — but only by accident. This is enough to show that JTB is not sufficient for knowledge.

3. Post-Gettier Responses

Gettier’s paper inaugurated nearly six decades of philosophical effort to “repair” the definition of knowledge. The major strategies are:

3.1 No False Lemmas: Clark

Michael Clark (1963) immediately proposed a fix: knowledge requires that the agent’s justification does not pass through any false proposition (false lemma). In the Gettier cases, Smith’s justification passes through a false belief (that Jones will get the job; that Jones owns a Ford). If we require that every link in the inferential chain be true, the cases are blocked.

Objection: Roderick Chisholm and others constructed Gettier cases that do not involve explicit false lemmas — cases in which the justified belief is true by coincidence without any identifiable false premise in the inferential chain.

3.2 Reliabilism: Alvin Goldman

Alvin Goldman (b. 1938) proposed, in “A Causal Theory of Knowing” (Journal of Philosophy, 1967), that knowledge requires an appropriate causal connection between the fact that makes the proposition true and the agent’s belief. In the Gettier cases, the belief is caused by the wrong evidence — there is a causal break.

Goldman later developed reliabilism: in Epistemology and Cognition (1986), the justification condition is replaced by the condition that the belief be produced by a reliable cognitive process — one that, under normal conditions, predominantly produces true beliefs. Knowledge is true belief produced by a reliable process.

Reliabilism faces the evil-demon problem: if a Cartesian evil demon systematically deceived a subject whose cognitive processes were objectively unreliable in the hypothetical scenario, the subject intuitively might still have justified beliefs — suggesting that justification does not depend on the objective reliability of the process alone.

3.3 Tracking Theory: Nozick

Robert Nozick (1938–2002), in Philosophical Explanations (1981, ch. 3), proposed the tracking theory: S knows that p if and only if:

1. p is true;
2. S believes that p;
3. Sensitivity: if p were false, S would not believe that p (counterfactual conditional);
4. Adherence: if p were true (in nearby worlds), S would believe that p.

The sensitivity condition blocks Gettier cases: in the job case, if it were not true that the man who would get the job had ten coins in his pocket, Smith would still believe it — because his belief is based on evidence about Jones. Hence the sensitivity condition is not satisfied, and Smith does not know.

Objection: Nozick’s theory is not closed under known logical implication — which seems counterintuitive. If I know I am human and know that humans are mortal, I should know that I am mortal; but Nozick’s theory can block this in certain cases.

3.4 Safety: Sosa and Williamson

Ernest Sosa (b. 1940) and Timothy Williamson (b. 1955) independently developed the notion of safety as a condition for knowledge: S knows that p only if, in the conditions under which S believes that p, p could not easily have been false. Unlike Nozick’s sensitivity (about possible worlds where p is false), safety concerns nearby worlds where S believes in the same way.

Safety captures the intuition that knowledge cannot be mere luck: if S could easily have believed that p even if p had been false — as in Gettier cases — then S does not know.

3.5 Virtue Epistemology: Sosa and Zagzebski

Sosa, in Knowledge in Perspective (1991) and A Virtue Epistemology (2007), proposes that knowledge is true belief that manifests cognitive competence — a stable intellectual virtue of the subject. In Gettier cases, the truth of the belief is not due to the agent’s competence but to chance — hence there is no knowledge.

Linda Zagzebski (b. 1946), in Virtues of the Mind (1996), carried virtue epistemology to an extreme and argued that any analysis maintaining the structure “true belief + X” would be vulnerable to cases analogous to Gettier’s. The structural reason is that it is always possible to separate the truth of the belief from the additional condition X through a double coincidence. Zagzebski concluded that the theory of knowledge must abandon the search for necessary and sufficient conditions and adopt an approach centered on the intellectual excellence of the agent.

4. Knowledge-First: Williamson

Timothy Williamson, in Knowledge and its Limits (2000), proposed a radical reversal: rather than analyzing knowledge in terms of simpler components (belief + truth + justification + X), we should take knowledge as a primitive mental state and analyze belief, evidence, and justification in terms of knowledge.

Williamson’s central thesis is that knowledge is the most general mental state: “E = K” — S’s evidence is precisely what S knows. Justification is defined in terms of evidence, and evidence in terms of knowledge. The attempt to analyze knowledge into simpler components is doomed to fail — as decades of counterexamples have demonstrated. The concept of knowledge is irreducible.

This position, known as knowledge-first, inverted the traditional order of epistemology and has been enormously influential in recent decades.

5. Zagzebski’s Definitional Pessimism

Before her turn to virtue epistemology, Zagzebski argued in “The Inescapability of Gettier Problems” (The Philosophical Quarterly, 1994) that the Gettier problem is insoluble within the JTB+X analytical framework. The demonstration is elegant: any condition X added to JTB can be “separated” from truth by a double coincidence — first, one ensures that the belief is justified (in the sense of X); second, one introduces a coincidence that makes the belief true for a completely different reason. The Gettier problem is structural, not accidental.

Zagzebski’s conclusion is not sceptical — it is not that knowledge is impossible, but that knowledge cannot be defined by adding conditions to true belief. Epistemology needs a different architecture.

6. Philosophical Significance of the Problem

Why does Gettier’s three-page article have philosophical importance so disproportionate to its size? Because it showed that philosophical analyses can be systematically falsified by counterexamples, even when they seem intuitively correct. This method — seeking counterexamples to philosophical definitions — became the central procedure of analytic philosophy in the following decades.

Moreover, the Gettier problem revealed that the notion of “justification” in the JTB definition is underdetermined: for some, justification is having good reasons; for others, it is being produced by a reliable process; for still others, it is manifesting an intellectual virtue. The answer one gives to the Gettier problem depends on what one takes justification to be.

Finally, Williamson’s proposal to treat knowledge as primitive raises a deeper question: analytic philosophy, in seeking to define knowledge, may have chosen the wrong direction. Perhaps knowledge is not a compound but a foundation.

Essential Reading

• Edmund Gettier, “Is Justified True Belief Knowledge?”, Analysis 23, 1963, pp. 121–123.
• Alvin Goldman, “A Causal Theory of Knowing”, Journal of Philosophy, 1967.
• Alvin Goldman, Epistemology and Cognition (1986).
• Robert Nozick, Philosophical Explanations (1981), ch. 3.
• Ernest Sosa, Knowledge in Perspective (1991); A Virtue Epistemology (2007).
• Linda Zagzebski, Virtues of the Mind (1996).
• Linda Zagzebski, “The Inescapability of Gettier Problems”, The Philosophical Quarterly, 1994.
• Timothy Williamson, Knowledge and its Limits (2000).

← Articles