[Essay] The Gettier Problem

Guillaume Hansali
Guillaume Hansali
[Essay] The Gettier Problem
Is there really a sheep in the field?

Essay question

Describe and explain why Gettier-style cases demonstrate that the tripartite account of knowledge is unsustainable. How should one go about offering a theory of knowledge that is immune to Gettier-style cases, do you think? Can one offer a theory of knowledge that is immune to Gettier-style cases?

Essay written for the Theory of Knowledge online course at Oxford:
Constraint: 1500 words.


Knowledge is both intuitively trivial and elusively hard to define.
It has been epistemologists' quest to determine whether we can convincingly analyze knowledge in a way that agrees with our intuition.
A successful analysis would be a set of clear conditions that are necessary and jointly sufficient to qualify instances of knowledge.
The traditional contender has been the tripartite justified true belief or JTB for short, but while the epistemic community widely accepts JTB as necessary conditions to knowledge, JTB's sufficiency is still very controversial.

In this essay, I will introduce Gettier-style cases and explain why they undermine JTB's sufficiency. I will then discuss different approaches to modifying JTB to counteract such cases and illustrate a few past attempts and why they arguably failed.

JTB's insufficiency

JTB (justified true belief), also known as the tripartite analysis of knowledge, states that;
S knows that p iff (if and only if)
(i) S believes that p
(ii) p is true
(iii) S is justified in believing that p

where S is a subject, and p a proposition. (Ichikawa 2018)

JTB as an analysis of knowledge would be instrumentally valuable if it could be reliably used to determine whether an agent has knowledge in a given situation. In other words, if justified+true+belief were jointly sufficient.

Now, sufficiency is a strong claim. To refute it, one only needs to suggest a situation in which an agent has acquired justified true belief, but we nonetheless intuitively wouldn't confer her knowledge.

Edmund Gettier offers two such cases in his short but devastating essay Is Justified True Belief Knowledge?.

Case 1: Smith and Jones apply for the same job. Smith believes from testimony that Jones will have the job, and from observation that Jones has ten coins in his pocket, and therefore infers that (p) the man who will get the job has ten coins in his pocket. However, unbeknownst to him, Smith ends up getting the job and, too, happens to have ten coins in his pocket (Gettier 1963, p.122).

Case 2 (shorter version): Smith has good reasons to believe that Jones owns a Ford, but is clueless about the whereabouts of Brown. Convinced about the car ownership, he infers that (p) either Jones owns a Ford, or Brown is in Barcelona. However, as it happens, Jones doesn't own a Ford, but Brown is in Barcelona (Gettier 1963, p.122).

In both cases, the proposition (p) is true, and Smith has reasonably valid justification for his belief. However, we intuitively wouldn't grant him knowledge; his belief is inferred from a false proposition and happens to be true by chance.

Let me introduce two more famous Gettier-style cases that will become handy later on.
Chrisholm's sheep in the field: subject X sees something that resembles a sheep in the field and infers that (p) there is a sheep in the field. What he sees is, in fact, a dog disguised as a sheep. However, there happens to be a real sheep in the field behind the hill (Ichikawa 2018).

Goldman's fake barns: subject X is driving on a road where most of the barns on the side of the road are fake but look pretty real. At one point, he looks at one of the barns and forms the belief that (p) the barn he is seeing is real, assuming that all the barns are. As it happens, the barn he is looking at right now is one of the few real barns (Ichikawa 2018).

The Gettier problem

Modifying the tripartite to make it immune to Gettier-style cases is commonly known as the "Gettier Problem" (Hetherington).

The four cases mentioned earlier all share the following pattern: in each case, the agent logically infers an accidentally true belief from a justified false belief.
For example, in case 1, the false belief is the proposition that John would get the job. In the sheep in the field case, it is mistaking the dog for a sheep.

Two characteristics seem to be at play:
Fallibility of the justification: the justification doesn’t always ensure the trueness of the proposition, i.e., the justification could have been valid and the proposition false (Hetherington)

Luck: the proposition nonetheless ends up being true by luck (Hetherington). In the fake barn case, for example, looking a few seconds earlier would have shown a fake barn, and the belief would have been false.

General approaches to solving Gettier Problem

An intuitive approach to solving Gettier Problem would be to eliminate fallibility and luck from our analysis of knowledge. All four cases would be ruled out as non-instances of knowledge because of the fallibility of their justifications and the luck involved in getting to a true belief.

However, this comes with a few challenges.
Eliminating fallibility rules out any form of justification that could be fallible in principle. But that would lead to radical skepticism because we are not infallible creatures, and therefore we cannot be expected to come up with infallible justifications. Consequently, we cannot know anything (Hetherington).
Similarly, eliminating luck means that the justification of a belief would always entail its truth (no luck involved), rendering it inevitable and thus infallible; we are back to radical skepticism (Hetherington).

So both fallibility and luck cannot be removed entirely, which leaves us with the conundrum of determining how much fallibility and luck are acceptable in our analysis.

Failed attempts at solving the problem

Since the publication of Gettier's essay in 1963, many modification proposals have been offered. I will introduce a few well-known examples that preserve JTB and add a fourth condition that allegedly makes it immune from being "gettiered."

The no false lemma condition

(iv) S's belief that p is not inferred from any falsehood  (Ichikawa 2018)

This condition clearly rules out the two original Gettier cases in which Smith's beliefs are inferred from false propositions.
However, it is possible to conceive Gettier-style cases where a belief is not inferred from a false proposition.
In the sheep in the field case, suppose that the disguised dog looks exactly like a real sheep, to the point that anybody seeing it wouldn't be able to make the distinction. The agent's sight not being faulty, her belief is not inferred from falsehood, but we still wouldn't confer her knowledge.

The defeasibility condition

(iv) there must be no other accessible evidence against p strong enough to undermine S's belief that p (Zagzebsk 2020)

This condition is much stronger than the no false lemma condition and rules out all four cases. For example, in the fake barn case, the agent could have stopped the car and had a closer look at all the barns and realized that most of them were just facades. But there is a matter of subjective interpretation in regards to what is enough to undermine a belief.
Zagzebsk argues that a strong version of the condition is untenable because it would undermine the independence between truth and justification conditions; the absolute absence of possibly undermining evidence is a synonym of infallibility, which leads to skepticism. She also adds that truth should be contingent on justification; otherwise, it becomes redundant in our analysis, and knowledge becomes a form of warranted belief (Zagzebsk 2020).

The sensitivity condition

(iv) if p were false, S would not believe that p  (Ichikawa 2018)

While the first two conditions were internalist by nature, this one is externalist.
It rules out the four cases as described earlier, but we could modify the sheep case as follows to defeat it;
X sees a sheep in the field, but that is, in fact, a large screen displaying a very accurate image of a sheep. Suppose that unbeknownst to X, there is a real sheep hiding behind the hill and that the screen has been programmed to display a sheep only when it detects a real one in its vicinity.
If p were false (there is no real sheep), the screen wouldn't have displayed a sheep, and X wouldn't have believed that p. The sensitivity condition is met, but X doesn't have knowledge.

Is it even unsolvable?

There are many more proposals, but so far, none has seemed to unequivocally convince the epistemic community.

As a way to settle the debate, Zagzebsk argues that any analysis of knowledge based on true belief + x can be gettiered as long as there is some degree of independence between x (for example, justification) and getting to the truth. Because justification doesn't entail truth (contingency requirement), the connection between the two could always be severed by creating an adequate Gettier situation and reattached afterward by introducing luck (Zagzebsk 2020).


As long as the Gettier Problem remains unsolved, any account based on the tripartite will be deemed insufficient to qualify knowledge.

Some epistemologists have given up trying to solve the problem and have been looking at ways to dissolve it instead by proposing accounts of knowledge that include more luck, or dismissing Gettier cases altogether out of a lack of practical relevancy (Hetherington).
Some even claim that knowledge may not be analyzable after all; we will always find cases that go against our intuition (Ichikawa 2018).

On the bright side, while largely unfruitful, attempts at solving the Gettier problem have sharpened our intuitive conception of knowledge, so at least, we are not empty-handed.


  • Gettier 1963: Edmund Gettier, Is Knowledge Justified True Belief? (Analysis 23, 1963)
  • Hetherington: Stephen Hetherington Gettier Problems (Internet Encyclopedia of Philosophy) URL= https://iep.utm.edu/gettier/#H15
  • Ichikawa 2018: Ichikawa, Jonathan Jenkins and Matthias Steup, "The Analysis of Knowledge", The Stanford Encyclopedia of Philosophy (Summer 2018 Edition), Edward N. Zalta (ed.), URL = https://plato.stanford.edu/archives/sum2018/entries/knowledge-analysis/.
  • Zagzebsk 2020:  Linda Trinkaus Zagzebsk, The Inescapability of Gettier Problems (Oxford University Press, 2020) ISBN: 978-0-19-752917-1

Great! Next, complete checkout for full access to Guillaume Hansali
Welcome back! You've successfully signed in
You've successfully subscribed to Guillaume Hansali
Success! Your account is fully activated, you now have access to all content
Success! Your billing info has been updated
Your billing was not updated