Skip to content

In Defense of the Justified True Belief

by: Ammar Ahmad (University of Michigan) | December 5, 2023


The Gettier Problems have posed seemingly inescapable problems for fallibilism and epistemology overall.1 Modern epistemologists have had to tread the line between falling back into the skeptic consequences of infallibilism and creating unreliable theories of knowledge.2 David Lewis has somewhat evaded the Gettier Problems by relying on a context-sensitive form of infallibilism, in which an individual knows that something if and only if that individual’s evidence eliminates every possibility in which ‘not-that-something’ (except for those possibilities that he is “properly ignoring”).3 Other philosophers, such as Dana Lenkin, for example, have argued that knowledge must bear a causal or explanatory connection to the belief we hold.4 This externalist causal condition creates immediate trouble, which Baron Reed detailed, since knowledge is then fixed to be definable only in retrospect, derailing the fallibilist project.5 All this is to say, there has not been a sufficient response to the Gettier Problems: Either they must be answered or they are flawed. I will argue that they can be answered and if that is not sufficient, then they are flawed. Justified true belief (JTB) places the individual in the best possible position for gaining knowledge, which is the most productive theory of knowledge.6 The Gettier Problems do not pose a serious threat to fallibilism; instead, fallibilist skepticism is a natural symptom of our conceptual understanding of fallibilism. In other words, I argue that JTB is sufficient for our knowledge because it avoids the epistemic link between belief and truth, a detriment to any fallibilist knowledge theory.



JTB should be accepted to mean the formal epistemic concept of ‘knowledge’ on two accounts. First, knowledge can be accidental because there is no way to rule out all possibilities of luck in a given situation. In other words, in coming to know anything, there is no way to ensure that we have eliminated the possibility of accidental knowledge. Eliminating such possibilities would require infallabilist knowledge to truly know anything, ultimately leading to a separate problem
of skepticism.


Second, in looking at any case where it has been found out that knowledge was gained through luck, we come to find that the individual’s action would not have changed in any way even if there was no accidentality.7 Of course, in finding out that the knowledge was gained accidentally, we came to know more about the case in question, but the individual’s behavior is not altered upon realizing accidentality. And so, practically speaking, JTB is a satisfactory condition in our theory of knowledge, even in the face of Gettier examples.


Gettier Example

The Gettier Problems present epistemic cases in which an individual gains accidental knowledge of something. One famous Gettier Problem is the following: An individual, Mary, walks by a clock tower that says 4:00 PM, leading her to justifiably believe that it is 4:00 PM. Yet, unbeknownst to her, the clock tower is broken, but it just so happens that it is actually 4:00 PM. Thus, Mary had a justified belief that it is 4:00 PM and it just so happened to be true.


Under the JTB view, Mary had knowledge that “it was 4:00 PM”. Mary’s belief satisfied both conditions of JTB — truth and justification. Mary’s belief is justified because clock towers tend to convey the time accurately and her belief is true because it was actually 4:00 PM. 


The Impossibility of Eliminating Accidentality

However, the accidentality of the situation causes some problems for JTB. The case posed earlier is clearly different than if Mary had walked by a functioning clock tower. In the working-clock tower example, she has justified belief and it is true, but it is different from the broken-clock tower example since her knowledge is not accidental. Accidentality seems ‘off’ in a theory of knowledge because the justification should have failed, even if the belief turned out to be true.Can we alter the justification portion to eliminate cases of accidentality in order to allow Mary to know things in a non-accidental way? And if so, what are the implications of such a demand? 


One way to eliminate cases of accidentality is by raising the threshold for justification. Perhaps Mary should have waited to believe that it was 4:00 PM until she confirmed with various sources, such as her watch, phone, and the clock at home. But what if all these time-tellers happened to be malfunctioning, and yet, by luck, each time she checked the time, they happened to convey it accurately just then? In other words, what if, even under intense scrutiny, accidentality persisted nonetheless? 


This hints at a bigger problem: trying to raise the threshold for justification to rule out accidentality in Gettier cases may show that we cannot prove that our other knowledge is not accidental without resorting to infallibilism. Said differently, the implications of having to have formed the belief based on more thorough justification would be quite daunting, even for non-Gettier instances. If Mary walked by the functioning clock tower and saw that it was 4:00 PM, she can never prove that it was not an accidental instance. Any proof can be accidental. If she went to the United States Naval Observatory, it could still be the case that they ‘accidentally’ miscounted and yet it just so happened to be the right time.9 Perhaps the atomic time is actually off, and yet it just so happened to be the right time. Of course, in the Gettier example, we know that the clock tower is broken and Mary’s knowledge is accidental. That is the fundamental difference between the two. But perhaps some knowledge that we have is, unknown to us, accidental. We can never holistically rule out all the alternatives of accidentality. 


Skepticism threatens our ability to gain fallibilist knowledge. If we want to avoid this rabbit hole, we must allow for justification to remain intact and refuse to raise the threshold for justification. Thus, for Mary to ever have fallabilist knowledge, we should allow her to avoid the exhausting task of ruling out alternatives of accidentality. Even in the Gettier cases, simple knowledge that there is accidentality should not mean that we redefine our epistemological theory to have to rule out accidentality, especially if we cannot prove that our present knowledge is not itself accidental. Knowledge of accidentality can refine our justification when we come to find that it was obtained through luck, but it should not prevent us from holding onto our current knowledge.


Practicality of the JTB

Moreover, consider the consequences of Mary’s JTB that it is 4:00 PM. If Mary walks to class at 4:30 PM and is told that the clock tower is broken, the clock nonetheless provides her with the temporal knowledge that allows her to get to her class on time. In other words, she could perhaps never come to find out that the clock was broken, and nonetheless, it functionally served its purpose, even if it was on faulty grounds. 


To be more precise, Mary’s belief that “it is 4:00 PM” was not negated by the fact that the clock tower was broken yet it happened to convey the time accurately. That piece of information adds nuance to the matter, in that when Mary formed that belief, she was not aware that the clock tower was broken and the timing was pure luck. Despite this contextualization, Mary’s core belief that “it is 4:00 PM” was an accurate piece of information, and it allowed her to practically go about her day — since the truth of the matter was not altered, Mary would go about her day as normal. Thus, if cases of accidental knowledge do not influence the individual’s behavior, we should take them to be somewhat irrelevant in practical terms. And so, even in cases of accidental knowledge, JTB is a practical theory of knowledge.


An Epistemic Link

Perhaps the problem with the JTB can be countered, however, on the grounds that in the cases of accidentality, the justification portion of the JTB gets called into question. In other words, the valuation of the premise “it is 4:00 PM” is not problematic at all to JTB challengers, but the justification is problematic (that is the implicit belief that the clock tower is accurate and has led Mary to her belief). Perhaps we do not need to raise the threshold for justification, as mentioned earlier, but it is nonetheless on shaky grounds since it could have easily led to a falsity. Thus, there must be, not simply more justifications, simply a right justification that links the belief and truth.10 


Had Mary walked by the clock tower just a minute later, she would have been led to believe something false, namely “it is 4:00 PM” when in reality it was 4:01 PM. So, the justification that “the clock tower is a good indicator of time” would have led her to a false belief. Said simpler, the JTB seems to be grounds for some future errors. From this, we would conclude that the JTB is not an appropriate theory of knowledge because the justification portion is problematic in epistemic or even practical terms!11 If the truth of a belief has nothing to do with its justification, then we should reject JTB.


An Epistemic Link and Infallibilism

This counter-argument, however, misrepresents the link between justification and truth because it requires justification to somehow entail truth in all scenarios when that is not really what justification is supposed to do in a fallibilist framework. In other words, the counter to the JTB requires that an individual must have the ‘right’ sort of justification for her accidentally true belief. But we must not link the two because then we fall into a problem of infallibilism.


To push this counter-argument to its direst consequences, we find that to require the absolute right justification leads us to a serious predicament — how do we ever know that we have the correct justification for any of our knowledge?12 Similar to raising the justification bar, the epistemic link between truth and justification leads us to infallibilism and, further, skepticism. And so we are left with the following question: How do we avoid this dangerous link and yet allow justification to be satisfactory for accidental knowledge? 


‘Available Justifications’

To allow for JTB to survive the problems that arise out of accidentality, namely that the justification could have led the individual astray if it were not luck, we should look towards the justifications available to Mary. When Mary glanced at the clock tower and saw that it said 4:00 PM, she could not help but believe that “it is 4:00 PM.” If she did not have her phone, watch, etc. then this is especially true, since there is no alternative justification available to her, she should rely on this one. 


By relying on the only justification available, we form a belief. The truth of that belief is not implicated by this justification.13 And so, when Mary formed her justified belief that it was 4:00 PM, she relied on the only thing available to her at that time strictly — the clock tower (assuming that her phone was dead and she did not have a watch and so on).


Revisiting the scenario stated earlier under the anti-JTB viewpoint, we find that if the clock tower was not broken, Mary’s justification would have been sufficient. If the clock tower was broken and yet it was 4:00 PM, Mary’s justification would no longer be sufficient.14 Regardless of this, Mary had only one reliable justification available. This justification could be correct or incorrect, but in this scenario, it was correct, but by accident. Even if she later realized that her justified belief was accidental, that does not dismiss the true belief nor the fact that she relied on the best justification she had at the time. It would be absurd to assume that she must forfeit her knowledge that “it is 4:00 PM” on the grounds that she should have found some other, hidden justification for her belief this one time. And since merely looking at the clock tower prompted her to form a belief about the time (since the clock tower is taken to be the only justificatory source for time-knowledge at this time), it is clear that the justification is somewhat separate from the real-world truth value of the belief. In other words, if an individual’s belief is true, then we should judge their knowledge strictly on their available justification.15


Best Hypothesis and JTB

Some philosophers have argued that there is a fundamental difference between the best available hypothesis and justified true belief.16 Jeffery Olen has countered a similar point to mine and argued that it is better to characterize the aforementioned scenario as a “good guess” that is “coherent” with Mary’s other beliefs. But is it really a “good guess” that Mary knew that it was 4:00 PM? Revisit the impossibility of ruling out cases of accidentality in everyday knowledge. How do we properly delineate between a “good guess” and knowledge when the question of fallibilist skepticism looms? Without knowing it, Olen has fallen into an infallibilist territory.17 Unless there is a clear way to distinguish between the coherence used in a “good guess” and in cases of knowledge, Olen’s view seems ad hoc


The Gettier Problems should not be interpreted to be some immovable mountain in the face of a coherent fallibilist theory of knowledge. The Gettier Problems simply represent cases in which accidentality is not realized by a subject but realized by us. And so given that JTB places the individual in the best possible position, epistemic accidentality should be a salient part of the theory unless we want to rule out all possible alternatives. That we have ‘caught’ an instance of accidentality should not be surprising given that justification does not in any way entail truth. 


Moreover, what characterizes the feeling that Mary gets when she looks at the clock? Is she supposed to have induced herself to not believe that it was 4:00 PM, opting for some “best guess” feeling instead? The implications of Olen’s paradigm have consequences of skepticism. Perhaps Olen would counter by stating that Mary’s belief should be characterized as a “best guess,” but again, I would reiterate that there must be some explanation as to why Mary’s belief is knowledge in one instance but a “best guess” in another. The fact that there is not a clear answer to this should not lead us to modify JTB.18 The Gettier Problems cannot be ‘solved’ because they are borne from our definition of fallibilism. Instead, we should understand the Gettier Problems properly: to accept that justification does not entail truth is to accept that Gettier cases can occur, but there is nothing all too special about them except that we know they are accidental. To rule out that accidentality is to spiral into Descartes’ dogmatic problem.19 Thus, we should pivot to recontextualizing the Gettier Problems as non-problematic instead of monumentalizing them as some “inescapable” problem for fallibilism. 



The JTB puts an individual in the best position to obtain knowledge even in cases of accidentality. First, there is no way to rule out accidentality even in regular cases of knowledge. So unless we want to bleed into infallibilism, the JTB stands as a cohesive theory that explains our current knowledge and sets us up for more in the future. Second, the JTB is a practical approach to knowledge since, even in cases of accidentality, the behavior of the individual is not altered in any way upon realizing the luck of their knowledge. Lastly, since justification and truth are separate, we must judge an individual’s true belief strictly on their available justification. All in all, this allows the JTB to avoid infallibility problems and justify accidental knowledge.

End Notes

1. See Edmund L. Gettier’s Is Justified True Belief Knowledge? (1963). It details various cases of accidental knowledge. This leaves fallibilism in an awkward position which either requires another condition to be added to JTB or a modification of our definition of JTB.

2. See Linda Zagzebski’s The Inescapability of Gettier Problems (1994). She concludes her paper by arguing that “the best position is assumed to be imperfect, for such is life. Properly functioning faculties need not be working perfectly, but only well enough.” I consider this a major reason for maintaining JTB as the fundamental requirement for any working theory of knowledge.

3. See David Lewis’s Elusive Knowledge (1996). Lewis argues that “S knows that P iff S’s evidence eliminates every possibility in which not-P – Psst! – except for those possibilities that we are properly ignoring.” A skepticism still persists: “S knows that P iff S’s evidence eliminates every possibility in which not-P- Psst! – except for those possibilities that we are properly ignoring. That ‘psst’ marks an attempt to do the impossible – to mention that which remains unmentioned.”

4. See Dana Nelkin’s The Lottery Paradox, Knowledge, and Rationality (2000), p. 390.

5. See Campus Reed’s Fallibilism (2012). Camp expands on this point by arguing that by “requiring a causal or explanatory connection between the belief and the fact that makes it true may mean that it is impossible for anyone to ever know anything about the future, about general facts, or about abstract objects.” 

6. See Brian Weatherson’s What Are Good Counterexamples? (2003). He argues that JTB is not just productive, but sensical: “What’s distinctive about the faulty intuitions, I argue, is that respecting them would mean abandoning a simple, systematic and largely successful theory in favour of a complicated, disjunctive and idiosyncratic theory.”

7. The ‘Smith owns a Ford or Brown is in Barcelona’ example might pose some problems considering that the disjunctive statement may have induced the individual to act differently. For instance, if the individual utters ‘Smith owns a Ford or Brown is in Barcelona,’ the actionable implication of either element are not necessarily acted on due to the utterer’s ignorance of which part of the disjunction is true. But I will not be exploring this Gettier example because I do not find its valuation to implicate action in any clear way. This convenient logic can prove anything, including impossibilities. Suppose I have reason to believe that ‘Brown is in Barcelona or there are elephants on Mars.’ If it turns out Brown is not in Barcelona, then the implications are a bit unreasonable. Thus, if anything, this form of reasoning should not even be considered “justification” at all. 

8. In that the “justified” part of the “justified true belief” condition is “justified” only in that it usually yields knowledge, but in accidental cases of knowledge, the “justified” seems to be the problematic part of the condition since it should have failed to yield knowledge. In cases of accidentality, it is luck that allows for alignment between the belief and the truth, not justification. That is why I focus on amending justification in this section.

9. The United States Naval Observatory (USNO) maintains the U.S. Department of Defense reference for time and time interval. USNO has an ensemble of atomic clocks, which is used to derive a time scale called UTC.

10. This problem with justification is not that it is not investigative enough. Rather, this segment of the paper focuses on how justification seems to be unrelated to the truth of the belief. 

11. Epistemic terms since this same justification could have led to a false belief; practical terms since if the JTB could have led her to be late to class (had it been actually 5:00 PM), then the counter-argument is that the JTB is actually not practical. 

12. As was earlier stated, this externalist theory relies on a temporal ‘fixing’ of knowledge that does not allow for immediate knowledge of anything. 

13. But we certainly think that the belief based on the justification is true to some extent.

14. Or we would need another thing besides justification. Regardless, my focus is on how Mary can rely on justification and truth strictly to come to knowledge.

15. See Keith Lehrer and Thomas Paxson Jr.’s Knowledge: Undefeated Justified True Belief (1969), which essentially argues that “any satisfactory theory of justification, some knowledge must be undefeated completely justified true belief, and the rest is basic.” I diverge from their outlook because I argue that some knowledge formation is inevitable. Thus, unless you want to be an unhealthy skeptic like Peter Unger, relying on a theory of knowledge that puts you in the best position is much more productive and characteristic of sensible human behavior.

16. In Jeffrey Olen’s Is Undefeated Justified True Belief Knowledge? (1976) he argues that “accounts such as Harman’s and Lehrer’s fail to make an important distinction, the distinction between an undefeated justified true belief that constitutes knowledge and an undefeated justified true belief that constitutes the most reasonable hypothesis.” He further states that “what is needed, apparently, is an additional condition to undefeated justified true belief. This condition must require that S’s justification be sufficient to yield knowledge in the given situation.” The merits of such a case echo a similar concern that arises from the causal link of truth and belief, and the problems of such a theory have already been detailed and discussed.

17. In his proposed cases, individuals “do not yield knowledge because, even though [they are] justified in believing them, [their] justification is not sufficient for the given situation.” 

18.  See Lukasz Lozanski’s The Gettier Problem No Longer a Problem (2007). Lozanski argues that the Gettier cases are simply “reference muddling” and our problem with them arises out of a problem with language and ill-defined concepts. I will not explore his idea here as I think it produces major issues for fallibilism in general which would not be a productive course for this paper.

19. See Descartes’ first meditation in his Meditations on First Philosophy (1641).