Epistemology/Printable version


The current, editable version of this book is available in Wikibooks, the open-content textbooks collection, at

Permission is granted to copy, distribute, and/or modify this document under the terms of the Creative Commons Attribution-ShareAlike 3.0 License.

What is epistemology

What is epistemology and why should you learn it? Well, epistemology is the branch of philosophy that deals with knowledge and surrounding concepts such as justification, belief and credence, and is tightly interconnected with many other areas of philosophy such as metaphysics, ethics and philosophy of science, just to name a few. Epistemology aims to explain how we know the things we do, whether that's mathematics, science, history, or even just the things that we see or hear right in front of us. Another important project of epistemology is in characterising when and why we should believe some things and not others. Therefore, epistemology is not only a central area in—and useful introduction to—philosophy as a whole, it also helps us to know the limits and nature of our knowledge, helps guide our intellectual practices by showing us when we are truly justified in our beliefs, and it helps us to develop key philosophical skills such as critical thinking and intellectual humility.

Epistemology is a broad and exciting area of philosophy as represented by the range of issues covered in this textbook. One of the central questions of epistemology has been "what is knowledge?" and as a result, the first section is dedicated to analysing the concept of knowledge. It is important to note with regards to this section that there are actually many kinds of knowledge. For example, I could say "I know Bob, I've met him many times before" or I could say "I know blue, I've been lucky enough to be acquainted with that colour in all its shades!" In both these examples, I am expressing a type of knowledge called acquaintance knowledge. Likewise, I could say "I know how to ride a bike, that's easy." In this example, I am expressing a type of knowledge called ability knowledge. Neither of these two types of knowledge are the focus of the analysis of knowledge in section I. The analysis of knowledge has historically focused on the analysis of propositional knowledge. This is the type of knowledge we have when we know of the truth or falsity about a proposition. Examples of propositional knowledge are "I know that 2 + 2 = 4" and "I know that Bob is a philosopher". Analysis of this kind of knowledge has a long history but remains a difficult and contentious question in epistemology and philosophy more broadly.

Another argument important throughout the history of philosophy is whether experience or reason is the true source of knowledge. In section II, we will explore the historical debate between Rationalists and Empiricists in the early modern era about how we gain knowledge of the world by examining perception and reason as sources of knowledge. Philosophers today not only consider perception and reason as sources of knowledge but have begun to also consider other sources of knowledge including memory, introspection and testimony. Testimony is different to the other sources of knowledge we have talked about so far in that testimony has an inherently social component, it relies on whether or not and under what circumstances we should trust others. Building on this, we will consider the social practice of science as a source of knowledge. This consideration of social epistemology will be echoed and developed further in section IV.

I think that one of the first thoughts many students of epistemology have is about scepticism and the possibility that we are living in the Matrix or that we are a "brain in a vat" as most philosophers use as such a sceptical example. Indeed, Descartes' methodological doubt and corresponding evil demon thought experiment led to perhaps the most famous line of philosophy ever written or spoken: "I think, therefore I am." Section III will consider the problem that radical scepticism poses for epistemology by exploring solutions to two defining problems in epistemology. The first is Agrippa's trilemma which threatens the idea that any of our beliefs could possibly be justified and the second is sceptical scenarios we have talked about above and "the closure principle".

In section IV, we will explore some exciting problems that are posed by the social context of knowledge. How can we understand widespread disagreement in many areas even amongst equally educated people and how should we respond to it? Do we have any responsibilities to others in our intellectual lives and can we even commit epistemic injustices against others? Could such epistemic injustices even rise to the systemic level? Can the social structure of inquiry bias the knowledge we acquire and how can we alleviate this problem? Should we trust experts, how can we know who the experts even are, and what problems do fake news and the internet play for these questions? These are all questions asked in the areas of social and applied epistemology.

Hopefully this introduction has given a taste of the issues in epistemology covered in this textbook and created an appetite for the content of the rest of the book, as well as epistemology more generally. Even more so, hopefully this book is useful for those interested in philosophy and epistemology and provides an accessible introduction to the subject.

Printable version What is analysis? → 

What is analysis?

In our everyday speech, we often use very complex or lofty concepts such as "justice" or "knowledge" but when asked "what is justice?" or "what is knowledge?", we might find it hard to give an exact answer. Despite the fact that we presumably know what these words mean (we can use them perfectly fine when we need to!), we seem to only have a fuzzy idea of what they are or how they work. One of the methods that philosophers use to make our understandings of the words we use less fuzzy is called conceptual analysis. Conceptual analysis rests on the idea that we really do already know what these words mean so that we can reliably test a definition or "analysis" against examples to see if it is right. First, we propose an analysis of the concept we're interested in by placing conditions on what counts as falling under that concept. Then, we test this analysis to see if there are any counterexamples which show that it is not quite right. Counterexamples can come in two forms: (1) the analysis may miss a genuine case of the concept which it should have captured or (2) the analysis may include a false case of the concept which it should not have. If we find either of these counterexamples, we need to alter the analysis appropriately to fix the problem.

Individual necessityEdit

In scenario (1), our analysis misses a genuine example that it should have captured. For example, we may try to analyse the concept of a chair by saying that a chair is "an object that is (i) wooden, (ii) has four legs, and (iii) is something that people sit on". When considering this analysis, we might realise that it misses chairs which are made of plastic and chairs which have less than four legs (e.g. see the picture to the right). Here we can see that when our analysis misses a genuine example, it is because it is too strict and contains too many conditions. Some of the conditions in our analysis aren't individually necessary and have to be removed. In the chair example, we see that not all chairs are wooden and they don't all have four legs so these conditions can be removed. Doing this we are left with an analysis saying "a chair is an object that people sit on".

Joint sufficiencyEdit

In scenario (2), our analysis includes something which it shouldn't by mistake. For example, consider the analysis of chair that we have left from the last section: "a chair is an object that people sit on". There are many things that people sit on that are not chairs. Horse riders sit on horses while they are racing, Christians sit on church pews on Sundays, and park-goers often sit on benches or even the grass. Here we can see that when our analysis captures something it shouldn't, it is too broad and we need some extra conditions to rule out the false cases. Overall, our conditions are not jointly sufficient to guarantee that a case is a genuine example rather than a false example. We could add the condition that chairs must not be animals but this wouldn't rule out all the false cases we have thought of. Furthermore, simply adding "not [counterexample]" conditions to an analysis seems ad hoc, it rules out the things it should but it also doesn't give us much insight into why those counterexamples don't count or into what truly makes a chair a chair.

Having seen how difficult it is to analyse even the concept of a chair into its necessary and sufficient conditions, it might become clear here why the analysis of knowledge has been a historically debated topic worthy of an entire section of this textbook. Throughout this section, we will explore various analyses of knowledge and test them to see if they have any counterexamples that mean their conditions are not all individually necessary or jointly sufficient to properly define knowledge.

 ← What is epistemology Printable version Justified true belief → 

Justified true belief

In this chapter, we will introduce and discuss the tripartite (or "JTB") analysis of knowledge. The tripartite analysis of knowledge has been held by many philosophers throughout history and can trace its origins all the way back to Plato's discussion of a tripartite analysis of knowledge in the Theaetetus. The tripartite analysis of knowledge is often called JTB knowledge after its conditions of justification, truth and belief. Each of these conditions is claimed to be individually necessary for knowledge and together they are said to form the jointly sufficient conditions for knowledge.

The tripartite analysis of knowledge: A subject S knows that p if and only if
  1. S is justified in believing that p,
  2. p is true, and
  3. S believes that p

where p is a proposition.

Attacks of the JTB analysis have come in two main forms: (a) in this chapter, we will discuss objections to the individual necessity of each of the conditions and (b) in the next chapter, we will discuss a famous attempt to show that these conditions cannot always guarantee us knowledge and so are not jointly sufficient.


The justification condition is placed on knowledge to make sure that we don't know things by pure luck. It is reasonable to say that without any reasons one way or the other to believe a proposition is true or false, we would not have no rational way to judge whether it is true or not and so we can only be right about that proposition by luck alone. Justification can be thought of as anything that gives us those good reasons that we need to legitimately believe that a proposition is true. To understand why philosophers see "lucky knowledge" as a problematic concept, let's look at the following example:

  • You and a friend are playing a game where you try to guess the outcome of a coin flip
  • Your friend shouts out that they know that the coin is going to land heads up
  • By pure chance the coin does land on heads
  • But the coin could just have easily landed on tails making them completely wrong!

This example aims to show that there is something faulty in the idea that we can know things without having the right kind of justification because if we don't have justification for our belief, all that matters is whether our beliefs happen to come true or not. There is no difference between an expert undertaking complicated research in an area to come to a conclusion and a person that makes a random guess on the subject so long as they both happen to be right in the end. All it takes to change the guesser to a knower is the toss of a coin (until we add a justification condition to our analysis, of course!).

Attempts to argue against the necessity of justification to knowledge have attempted to create an equivalence between the JTB analysis of knowledge and true belief so that true belief alone is really all that is needed. For example, consider this passage by Plato from the Meno:

Soc. If a man knew the way to Larisa, or anywhere else, and went to the place and led others thither, would he not be a right and good guide?

Men. Certainly.

Soc. And a person who had a right opinion about the way, but had never been and did not know, might be a good guide also, might he not?

Men. Certainly.

Soc. And while he has true opinion about that which the other knows, he will be just as good a guide if he thinks the truth, as he who knows the truth?

Men. Exactly.

Soc. Then true opinion is as good a guide to correct action as knowledge

Opponents of the tripartite account could use this argument to make the claim that there is no meaningful difference between justified true belief and true belief and so justification isn't needed. However, even though true belief and justified true belief are both good guides to action, this doesn't mean that there is no meaningful difference between them. As we saw from the coin flip/expert research example, the difference between a person who merely has right opinion and somebody who has justified true belief is that the person who has justified true belief has a certain sense of epistemic legitimacy. If we want the status of knowledge not to be granted to people who happen to believe true things but are motivated by ignorance, intolerance or epistemic laziness, then our analysis of knowledge must make sure that it guarantees a certain bare minimum of epistemic legitimacy.


Most philosophers believe that truth is the most secure and uncontroversial of the conditions on knowledge. To understand why, let's reconsider our previous example:

  • You and a friend are playing a game where you try to guess the outcome of a coin flip
  • Your friend shouts out that they know that the coin is going to land heads up
  • Unlucky for them the coin lands heads down

If your friend were to continue to claim that they knew that the coin would land heads up after it lands the opposite way, you would probably think they'd gone a little bit insane.

The basic point is, you can't know something is it isn't even true in the first place – knowledge simply aims at the truth.


Augustine defined belief as "to think with assent". In other words, belief can be defined simply as thinking that something is true. If you were to hear somebody say that they believe that ghosts exist but they don't really think it's true that they exist, you might think that they don't really believe in ghosts at all. Now let's consider what reaction we would have if we were to hear somebody say "I know that ghosts exist but I don't really believe that they do." It seems like the rational reaction to this statement would be similarly to think that this person doesn't really know that ghosts are real, for how could they possibly know something that they don't even believe.

However, a counterexample by Colin Radford has attempted to show that not all of our knowledge needs to be believed, some of it may be known implicitly or subconsciously.

  • Albert learned that Queen Elizabeth died in 1603 at school years ago.
  • When asked when Queen Elizabeth died, Albert doesn’t believe she died in 1603 because he can’t remember being taught about her at all. He doesn’t have any belief about which year she died because he is uncertain.
  • Randomly guessing, he answers 1603 and he can also answer many more similar history questions correctly with accuracy far better than chance.
  • Radford argues that Albert must know subconsciously and that this is an example of knowledge without belief.

Somebody who believed in the JTB analysis of knowledge could deny that this counts as an example of knowledge, but they would have to provide an explanation for how Albert could possibly get so many questions right without knowledge (for it seems very plausible that this example really could happen in real life).

 ← What is analysis? Printable version Gettier cases → 

Gettier cases

In 1963, Edmund Gettier released one of the most influential philosophy papers of the 20th century entitled Is Justified True Belief Knowledge?. Maybe deciding he should quit while ahead, he has never released another paper on epistemology since. In the previous chapter, each of the conditions of the tripartite analysis of knowledge were scrutinised to see if they are really necessary for knowledge. In this chapter, we will study the critique that Gettier made of justified true belief. In his paper, Gettier presents two counterexamples to the JTB analysis of knowledge which purported to show cases in which a person could have a justified true belief that fails to be knowledge. These examples and other similar examples which attack the sufficiency of the JTB conditions are now referred to as "Gettier cases".

Gettier's counterexamplesEdit

Smith doesn't know how many coins are in his pocket
Jones owns a very reliable clock

In his first counterexample, Gettier gives us the case of two people applying for a job:

  • Smith and Jones apply for a job
  • Smith hears the interviewer say “I’m going to give Jones the job”
  • Smith also sees Jones count out 10 coins from his pocket
  • Smith forms the belief “the person who will get the job will have 10 coins in their pocket”
    • This is justified: Smith has good reason to think Jones will get the job and that Jones has 10 coins in his pocket
  • But actually Smith gets the job! Unknown to him, he has 10 coins in his pocket!
  • So his justified belief has turned out to be true by random chance and satisfies the JTB conditions for knowledge

In the last chapter, we discussed why most philosophers see epistemic luck as a deadly blow for any analysis of knowledge and this is exactly what Gettier has shown can happen for the tripartite analysis of knowledge. In his next case, Gettier utilises the idea of disjunction to arrive at the same result. Disjunction is when you combine two propositions (1) and (2) together to make the proposition "(1) or (2)" which is made true if either one (or both) is true.

In his second counterexample, Gettier gives us another case involving Smith and Jones:

  • Smith sees Jones driving a Ford
  • Smith thinks to himself that not only does he have justification for believing Jones owns a Ford but also the belief “Either Jones owns a Ford or Brown is in Barcelona”
    • He is justified in believing this is true because he is justified in believing that Jones owns a Ford (and this alone would make his belief true)
  • But Jones doesn’t own a Ford, He was driving a friend’s car!
  • By complete chance Brown is in Barcelona

Once again, Gettier has provided an example where Smith has justified true belief by a complete coincidence – can this possibly be considered true knowledge?

Russell's counterexampleEdit

Previously to Gettier's paper in 1963, Bertrand Russell provided this example in his 1948 book Human Knowledge: Its Scope and Limits:

  • Jones owns a very precise, reliable clock
  • Jones comes downstairs in the morning to check the time from his trusty clock
  • It shows the time to be 10 o'clock
  • Jones forms the belief “It’s 10 in the morning”
    • Justified: his clock has always been reliable and precise in the past
  • However, last night at 10pm, the clock stopped and hadn’t moved since
  • By chance, it is 10 in the morning! So Jones has JTB
  • If Jones had gotten down the stairs even a minute later, he would have been wrong

This example not only shows that there are possible cases of justified true belief which fail to be knowledge, but also that such cases can even be uncomplicated everyday cases of belief!

How to respond to the Gettier casesEdit

Most contemporary philosophers agree that justification, belief and truth are not sufficient for knowledge precisely because of the type of examples that Gettier presents. This leaves epistemology in big trouble, if this common sense analysis of knowledge isn't enough, what can we replace it with? We will briefly consider one obvious response.

Simply adding a condition ruling out epistemic luck solves the problem of Gettier cases. However, it is not a solution favoured by philosophers. This is because this solution is ad hoc. Something is called ad hoc when it is created to solve a problem and for which we have no other independent reason for creating. This is a problem because it means that this solution is trivial and uninteresting; it tells us nothing new about how knowledge works and why it is not lucky. A good analysis of knowledge would rule out epistemic luck without needing to add a "no epistemic luck" rule because it would tell us something substantive about how and why our knowledge is not lucky.

The "no epistemic luck" analysis of knowledge: A subject S knows that p if and only if
  1. S is justified in believing that p,
  2. p is true,
  3. S believes that p, and
  4. S’s belief that p was not formed as the result of epistemic luck

where p is a proposition.

In the following chapters of this section, we will explore the various ways in which philosophers have responded to Gettier cases. We will discuss the prospects of adding an extra condition to the JTB analysis (sometimes called JTB+X), then the views of those philosophers who have argued that this kind of analysis of knowledge has been doomed from the start, ending finally on the consideration that maybe all attempts to analyse knowledge are doomed, maybe knowledge is simply unanalysable.

 ← Justified true belief Printable version Adding an extra condition → 

Adding an extra condition

In the previous chapter, we saw that Gettier cases provide motivation for the idea that justified, true belief is an insufficient analysis for knowledge. In this chapter, we will consider the possibility that the tripartite analysis does not provide the sufficient conditions for knowledge because it is missing a necessary condition. In other words, we will consider whether adding an extra condition on top of justification, truth and belief will rule out Gettier cases as knowledge.

No false lemmasEdit

A natural first thought for a new condition to add to the tripartite analysis might be to rule out beliefs inferred from false premises. A proponent of this view could then argue that Gettier's examples do not count as knowledge because the justified, true beliefs in these examples were inferred from falsehoods and so were not properly formed to count as knowledge.

Summary of the "no false lemmas" analysis
The "no false lemmas" analysis of knowledge: A subject S knows that p if and only if
  1. S is justified in believing that p,
  2. p is true,
  3. S believes that p, and
  4. S’s belief that p is not inferred from any false premises

where p is a proposition.

Gettier case False premise
The first Gettier case:
  • Smith and Jones apply for a job
  • Smith hears the interviewer say “I’m going to give Jones the job”
  • Smith also sees Jones count out 10 coins from his pocket
  • Smith forms the belief “the person who will get the job will have 10 coins in their pocket”
    • This is justified: Smith has good reason to think Jones will get the job and that Jones has 10 coins in his pocket
  • But actually Smith gets the job! Unknown to him, he has 10 coins in his pocket!
Smith infers his justified true belief that "the person who will get the job will have 10 coins in their pocket" from the premise that Jones will get the job. However, this premise was false because Smith got the job.
The second Gettier case:
  • Smith sees Jones driving a Ford
  • Smith thinks to himself that not only does he have justification for believing Jones owns a Ford but also the belief “Either Jones owns a Ford or Brown is in Barcelona”
    • He is justified in believing this is true because only one of the parts has to be true to make the whole thing true (and he thinks the first part is true)
  • But Jones doesn’t own a Ford, He was driving a friend’s car!
  • By complete chance Brown is in Barcelona
Smith infers his justified true belief that “either Jones owns a Ford or Brown is in Barcelona” from the premise that Jones owns a Ford. However, this is false because the Ford he was driving was a friend's and not his own.

As can be seen from the table above, the "no false lemmas" analysis rules out the original Gettier cases provided in Gettier's famous paper. However, there have been Gettier cases that have been provided for the "no false lemmas" analysis. For example, consider the following "fake barn county" example:

  • Jones is driving through fake barn county
  • In fake barn county, it is tradition for the locals set up many cardboard cut-outs of barns
  • Jones sees a lot of fake barns thinking that they’re real barns
  • Then, Jones happens to see a real barn!
  • Jones has justified true belief that there is a barn

In this example, Jones has a justified true belief that there is a barn. But not only does he have a justified true belief that there is a barn, he has a justified true belief that is not inferred from any falsehoods. This is because Jones' belief that there is a barn is inferred directly from his visual perception of a barn and so is not inferred from any premises at all. However, Jones could very easily have been looking at a fake barn with the exact same visual justification and his belief would have been false. Therefore, even with the "no false lemmas" condition, there can still be epistemic luck.

Sensitivity conditionsEdit

Another way we could rule out the Gettier cases as counting as knowledge is by considering their relationship to the truth. The key insight behind adding a sensitivity condition is that knowledge should not only be true, but it should also track the truth. To fix the tripartite analysis, then, we just need to add a sensitivity condition that ensures knowledge must be sensitive to the truth (as shown in the table below). A proponent of this view could argue that we can rule out Gettier's examples as cases of knowledge because the justified, true beliefs in these examples are insensitive to the truth.

Summary of the truth-sensitive analysis
The truth-sensitive analysis of knowledge: A subject S knows that p if and only if
  1. S is justified in believing that p,
  2. p is true,
  3. S believes that p, and
  4. if p were false, S would not believe that p

where p is a proposition.

Gettier case Insensitivity to truth
The first Gettier case:
  • Smith and Jones apply for a job
  • Smith hears the interviewer say “I’m going to give Jones the job”
  • Smith also sees Jones count out 10 coins from his pocket
  • Smith forms the belief “the person who will get the job will have 10 coins in their pocket”
    • This is justified: Smith has good reason to think Jones will get the job and that Jones has 10 coins in his pocket
  • But actually Smith gets the job! Unknown to him, he has 10 coins in his pocket!
If “the person who will get the job will have 10 coins in their pocket” were false (i.e. if Smith didn't have 10 coins in his pocket), Smith would still have believed it anyway because he believed that Jones was going to get the job and saw him count out 10 coins in his pocket. Therefore, this belief is insensitive to the truth.
The second Gettier case:
  • Smith sees Jones driving a Ford
  • Smith thinks to himself that not only does he have justification for believing Jones owns a Ford but also the belief “Either Jones owns a Ford or Brown is in Barcelona”
    • He is justified in believing this is true because only one of the parts has to be true to make the whole thing true (and he thinks the first part is true)
  • But Jones doesn’t own a Ford, He was driving a friend’s car!
  • By complete chance Brown is in Barcelona
If “either Jones owns a Ford or Brown is in Barcelona” were false (i.e. if Brown was not in Barcelona), Smith would still have believed it anyway because he believed that Jones owned the Ford that he was driving (which he doesn't). Therefore, this belief is insensitive to the truth.

Adding a sensitivity condition, like a "no false lemmas" condition, rules out Gettier's original examples. However, can it also rule out the fake barn county counterexample? Initially we might think that it can. After all, if the barn was not there for Jones to see, he wouldn't believe that there was a barn there. However, what if inside the real barn had been a cardboard cut-out of a barn set up. Then, if there was no barn there for Jones to see, he would have still believed there was a barn there because he would have seen a fake barn and mistaken it for a real barn. Saul Kripke takes this example one step further to show how counterintuitive a sensitivity condition could be. Consider the following case:

  • Jones is driving through fake barn county
  • In fake barn county, it is tradition for the locals set up many cardboard cut-outs of barns
  • Cardboard cut-out barns are always green in fake barn county and real barns are always red
  • Jones happens to see a real barn (with a cardboard cut-out barn inside)
  • Jones forms the justified true belief that there is a barn
  • Jones also forms the justified true belief that there is a red barn
  • If there had been no real barn there, Jones would still believe there is a barn but he would not have believed there was a red barn
  • Therefore, in this case, Jones' belief that there is a red barn is sensitive to the truth but his belief that there is a barn is not
  • Therefore, in this case, Jones knows that there is a red barn but he doesn't know that there is a barn!

A general problem with JTB+X solutionsEdit

Linda Zagzebski has argued that all analyses of knowledge which simply add an extra condition X onto the tripartite JTB analysis to acquire a JTB+X analysis will always be vulnerable to Gettier cases. This is because the fallibility of justification (as well as any added condition) will always be able to be able to be subverted to create Gettier cases. In fact, Zagzebski even provides an easy recipe with which to create a Gettier case for any JTB+X analysis of knowledge:

  1. Find a case of justified false belief which satisfies condition X
  2. Make the belief true by complete luck
  3. You have a Gettier case!

We will discuss Zagzebski's solution to this problem in a later chapter. However, in the next chapter, we will look to a completely new way of analysing knowledge called "reliabilism".

 ← Gettier cases Printable version Reliabilism → 

Virtue reliabilism

Throughout this section of the book, we have seen how troublesome Gettier cases have been for the analysis of knowledge. In the early 1980s, Ernest Sosa presented a view which he believed solved various disputes in epistemology, including the foundationalist/coherentist debate (covered in Section 3). More importantly for this section of the book, Sosa also believed that his theory could resolve the Gettier cases with a new condition on knowledge called aptness. This theory, now called virtue epistemology (more specifically virtue reliabilism), was the first attempt at a virtue approach to epistemology and soon led to many other philosophers presenting their own versions of virtue epistemology, such as Zagzebski's form of virtue responsibilism which will be presented in the next chapter.

The virtue reliabilist analysis of knowledge: A subject S knows that p if and only if
  1. S believes that p, and
  2. S's belief that p is apt

where p is a proposition.

What is the "virtue" in virtue epistemology?Edit

In ethics, non-virtue approaches often characterise morally good actions according to things such as consequences or obligations. In these approaches, understanding the morality of actions is taken to be the foundation of ethics which once achieved, can be used to understand whether or not a person is moral. For example, it may be discovered according to a certain moral theory that stealing is wrong. Once this has been done, the moral character of a certain person can be understood in terms of the amount of thefts they commit or in another similar way. Virtue ethics, on the other hand, aims to solve various philosophical problems in ethics by instead starting with the question "what makes a subject a good moral agent?" and then turning to determine the morality of actions in terms of the types of virtues (or vices) that motivate those actions. For example, it might be answered that empathy makes a person a better moral agent and that stealing is wrong because it is not the kind of thing an empathetic person would do. Similarly, virtue epistemology aims to solve problems in epistemology by starting with the characteristics of a subject that makes them a good intellectual agent and then determining which beliefs are good beliefs to hold by the types of virtues (or vices) that motivate the holding of those beliefs.

To understand why virtue reliabilism is called virtue reliabilism, we will have to explain the difference between two types of virtues. Sosa's virtue reliabilism is based on a type of virtues called faculty virtues. Faculty virtues are virtues that make a thing good at achieving its tasks or goals. For example, a faculty virtue of a knife could be as simple as the knife being sharp, because this will make it better at chopping things, and this is exactly the thing that we use knives for. Similarly, epistemic faculty virtues are simply things about a knowing subject that makes them better at knowing things and can be things as simple as having reliable eyesight, hearing or memory (reliable processes as found in the previous chapter). Faculty virtues, therefore, are quite different to most ethical virtues. Most ethical virtues don't only make you good at achieving your goals, they are good in themselves and have a degree of moral praiseworthiness that is not present in faculty virtues. This type of virtues which are praiseworthy and good in themselves are called trait virtues. In the next chapter, we will explore a position called virtue-responsibilism that argues that epistemic virtues should not just be faculty virtues, but should be epistemically praiseworthy, and should therefore consist of trait virtues.


To explain his idea of aptness, Sosa often uses the example of an archer. When an archer takes a virtuous shot and hits their target, their shot satisfies what Sosa calls the "three As" of virtues: the shot is accurate, it is adroit, and it is apt. The shot is accurate because it hits its target, it is adroit because it is skillful and will reliably go wherever the archer wants it to go, and it is apt because the shot is accurate because it is skillful and reliable. The addition of the aptness condition is the important step that distinguishes the shot as not just reliable, but virtuous (and therefore distinguishes reliabilism from virtue-reliabilism). Imagine an amazing archer who shoots an arrow into the air in a skillful way. The arrow is blown astray by a rogue gust of wind but in a sudden flash of luck, it bounces off the side of a barn door and hits the target anyway. In this case, the arrow shot was adroit (skillful/reliable) and also accurate, but it wasn’t virtuous. It wasn’t virtuous because the shot wasn’t accurate because it was skillful, it was accurate because the archer got very lucky. From this example, we can note two things: (1) if a shot is apt, then this implies that it is both accurate and adroit, and so a virtuous shot can simply be described as an apt shot (rather than an accurate, adroit, apt shot), and (2) aptness not only implies accuracy and adroitness but it ties them together so that they are no longer independent conditions from one another. The change from reliabilism (which can be thought of as describing knowledge as accurate, adroit belief) to virtue-reliabilism is not adding a new independent condition but instead it is telling us how the accuracy and adroitness of a belief must be related to one another.

Aptness can be translated from virtuous archery to epistemic virtues by considering what it means for beliefs to be accurate, adroit and apt. Beliefs are accurate when they are actually true. They are adroit when they are formed by a reliable process. They are apt when they are true because they were formed by a reliable process. Therefore, for Sosa, a virtuous intellectual act is an act that produces a true belief and it produces that belief due to the utilisation of a reliable process.

Gettier cases and other problemsEdit

The aptness condition is motivated by the type of problem underlying Zagzebski's general critique of JTB+X theories. As justification is fallible and, therefore, vulnerable to Gettier cases, so too are reliable processes and so the simple addition of an extra independent condition to reliabilism is not enough to solve the Gettier problem. Instead, Sosa introduces a condition that is supposed to rule out the possibility of epistemic luck by tying truth and reliability of process together in such a way that excludes luck from playing a role in knowledge. Because knowledge is apt belief and because apt belief is true because it is formed by a reliable process, knowledge cannot be true by luck because luck is not a reliable process. However, some philosophers question the legitimacy of this line of thought. Consider the fake barn county example again. In that example, a person is driving through fake barn county where it is custom to erect realistic cardboard cut-outs of barns. This person looks at a real barn whilst driving through fake barn county and forms the belief that they are seeing a barn. Here, a belief is formed from the reliable process of eyesight and that belief is true because the eyesight was reliable and acting as it should. But the person had already formed the belief that they were seeing a barn many times in fake barn county in the exact same way and had been completely wrong! Some philosophers use this example to argue that virtue reliabilism is still vulnerable to Gettier cases even with the added condition of aptness. Others argue that this isn't really a case of apt belief because the belief is true more because of luck than because of the reliability of eyesight and so the belief can't be apt. Sosa himself argues that this actually is a case of knowledge. The reason that this seems so counterintuitive, according to Sosa, is because when we usually think of knowledge, we are thinking of a special type of knowledge that Sosa calls reflective knowledge.

The difference between what Sosa calls "animal knowledge" and what he calls "reflective knowledge" is a difference between our evaluation of our beliefs. For Sosa, animal knowledge is just apt belief. When we see things in front of us or hear something besides us, we have an apt belief formed from a reliable process but we haven't necessarily thought to consider whether we should really trust that reliable process. Therefore, animal knowledge describes cases where it seems that we do have knowledge (such as knowing that a table is in front of you when you see one) without a requirement that we have any consciously thought-out justification for whether or not we should hold that belief. Another advantage of the concept of animal knowledge is given by its name: Sosa wants to be able to describe the knowledge that animals have and apt belief sets the bar for knowledge low enough that animals can have knowledge formed via their senses. On the other hand, Sosa describes reflective knowledge as "apt belief aptly noted". This is the type of knowledge where we don't only have apt belief about the world but we have also evaluated the reliability of this belief itself. Upon this evaluation, which would be some reliable method for evaluating beliefs, we have come to the apt belief that our initial belief about the world is reliable. This type of knowledge is valuable because it makes us reflect upon our beliefs and come not only to know things about the world, but also to come to a greater understanding of the world and the ideas in which we believe.

 ← Reliabilism Printable version Virtue responsibilism → 

Virtue responsibilism

In response to Sosa's formulation of a virtue epistemology that rested on the virtuous use of reliable faculties, other philosophers began to formulate their own versions. Some thought that although Sosa had been right to bring attention to intellectual virtues, that he was wrong in not conceptualising intellectual virtues as praiseworthy character traits more akin to the virtues of virtue ethics. The contributions of this group of philosophers led to the birth of virtue responsibilism. Although virtue epistemologists had many different motivations for developing this theory, such as to better understand our intellectual responsibilities to others or as a way of guiding intellectual practice, in this chapter we will be focusing on the work of Linda Zagzebski whose project was focused on the use of virtue responsibilism to provide the proper analysis of knowledge. Zagzebski defines knowledge simply: knowledge is "belief arising out of acts of intellectual virtue". However, this simplicity hides a complex theory of intellectual virtues that at once attempts to solve the Gettier problem in a way similar to Sosa, whilst at the same time explaining the acquisition of high-level knowledge such as that acquired by a brilliant scientist or genius detective.

The virtue responsibilist analysis of knowledge: S knows that p if and only if
  1. S believes that p, and
  2. S's belief that p arises from an act of intellectual virtue

where p is a proposition.

An act of intellectual virtue A is an act which:

  1. arises from the motivational component of A,
  2. is the kind of act characteristic of what a person with virtue A would do, and
  3. is successful in reaching the truth because of these features of the act.

Acts of intellectual virtueEdit

Acts of intellectual virtue have three components.

  1. When people have moral virtues such as generosity, compassion, kindness and so on, they are motivated to make the world a certain way. For example, a compassionate person seeing somebody in pain would want to relieve there suffering and so would be motivated to do something to make the person feel better. In direct analogy, the first feature of an intellectually virtuous act is that it must arise from the motivational component of an intellectual virtue. In the case of compassion, the motivational component is a desire to relieve the suffering in the world. In the case of curiosity (as an example of an intellectual virtue), the motivational component is the desire to discover new things and to come to a deeper understandings of the things we already know. Likewise, intellectual humility may motivate us to hear out other perspectives before we form conclusions, and intellectual courage may motivate us to propose creative solutions to problems even if they seem outlandish to others. These are all examples of intellectually virtuous motivations.
  2. Virtuous motivations are part of what makes an act praiseworthy but it isn't the whole story. Consider the example of a person that acts kindly to somebody else but only because they do not want to seem mean to their friends (not because they are genuinely motivated by kindness). It seems that in this case, the act of kindness itself is morally valuable even if it is not motivated by virtuous reasons. In general, people who are not virtuously motivated can still do the "right" thing and this is just as true for intellectual action as it is for moral action. To put it in other words, part of the value of a virtuous act is that it is the kind of thing that a virtuous person would do under the circumstances and, therefore, another component of an intellectually virtuous act is that it is the kind of thing that an intellectually virtuous person would do.
  3. Finally, acts of intellectual virtue must also be successful in meeting the ends of the virtuous motivation because of these previously mentioned features. In the case of the compassionate person described previously, the motivational component of compassion is the desire to relieve the suffering in the world and the end of this motivational component in the given scenario is to make the hurting person feel better. For an act under these circumstances to be truly virtuous, it must succeed in making the hurting person feel better, otherwise it would be a failed virtuous act, and it must achieve this success because it was virtuously motivated and because it was the kind of thing a virtuous person would do, otherwise it would just be lucky. For acts of intellectual virtues, truth is the end that must be successfully achieved and it must be achieved due to the first two components. Similarly to Sosa's aptness condition, this condition requires that the truth of the belief be tied up with the virtuousness of the act and there is a similar argument that because of this it can avoid Gettier cases.

Some virtue epistemologists have critiqued Zagzebski's theory of intellectually virtuous acts. For example, many virtue epistemologists see the analysis of knowledge to be a less important task of epistemology than the task of telling us what we should believe and how we should conduct our intellectual inquiry. These virtue epistemologists, called "virtue anti-theorists", argue that the central task of virtue epistemology is to study particular intellectual virtues to fully understand their unique value and that any attempt to make systematic connections between all virtues and knowledge is doomed to failure. In the next chapter, we will study similar thoughts behind an approach to epistemology called "knowledge first epistemology" which holds that knowledge is unanalysable. A more particular argument against Zagzebski's theory is that the conditions for a virtuous act are too strict. For example, consider a person who donates to charity out of kindness but the money is lost in transaction before it got to the charity. In this case, it seems that even though the act of giving to charity failed to achieve its virtuous end, it is nonetheless a virtuous act. We might also worry that the condition that we have intellectually virtuous motivations for a belief to count as knowledge may be too strict. This will be discussed more in the next section of this chapter and the distinction between high and low grade knowledge.

High and low grade knowledgeEdit

Gettier strikes againEdit

A Gettier counterexample to Zagzebski's virtue responsibilism provided by Heather Battally is shown below:

  • Brenda is a detective investigating the murder of an accountant for a big corporation
  • There are 2 suspects: the CEO of the corporation and the accountant’s husband – there is evidence against both of them and they both have a motive
  • Brenda considers the evidence and comes to the conclusion that, although it’s mixed, it points more towards the CEO than the husband
  • Out of intellectual humility, Brenda is motivated to hear out the opinions of her investigative team so that she can feel more certain that she has come to the truth (virtuous motivation)
  • The CEO of the company knows the evidence against him is strong, so he uses his massive amount of money to hire a hypnotist to brainwash the investigative team into believing that the accountant’s husband committed the murder
  • After listening to her investigative team (the kind of act an intellectually humble person would do), Brenda is convinced and forms the belief that the husband is the murderer
  • It turns out that the murderer actually was the accountant’s husband
  • If Brenda hadn't been virtuously motivated and acted as an intellectually humble person would do, she would not have believed that the husband committed the murder
  • Therefore, Brenda is successful in reaching the truth due to the appropriate features of her actions and so according to virtue responsibilism has knowledge
  • BUT we would never really consider this knowledge because the investigative team was brainwashed - Brenda only has knowledge through luck!
 ← Virtue reliabilism Printable version Maybe knowledge can't be analysed → 


One argument that rationalists have used to argue that not all knowledge is derived from experience is that some of our knowledge is innate. Innateness is a hard concept to pin down but innate knowledge can be thought of as knowledge that is unlearned. One way of understanding the debate between rationalism and empiricism is that rationalists have argued that experience cannot adequately explain all of our knowledge and so we need to appeal to innate knowledge whereas empiricists have claimed on the contrary that all of our knowledge can be derived from experience and so we don't need innateness as a source of knowledge. This is not a perfect way of dividing rationalists and empiricists but, historically, the denial of innate ideas has been an important part of empiricism and has even been proposed as a defining part of empiricism. Whether or not empiricism can be defined so simply, innateness is an interesting source of knowledge that has remained an important issue for debate throughout the history of philosophy up to the present day.


An animation of the geometry problem that Socrates walks the slave through.

In the Meno, Plato argues for innate knowledge via a dialogue between Socrates and a slave who has had no training in mathematics. Through a series of questions, Aristotle prompts the slave to go through the following reasoning:

  • The square at the bottom left is 1x1 so it has an area of 1
  • The question is: what square will have double the area (i.e. area = 2)?
  • It isn’t a 2x2 square because we see the area of the big 2x2 square is made up of four little squares (so the area is quadrupled)
  • But if we cut these little squares in half then the areas of each of them is halved.
  • Doing this forms a rotated square; adding up the four halves we find that its area is ½ + ½ + ½ + ½ = 2
  • So this is the square with double the area!

Plato argues that because the slave has not been educated in mathematics and because he is led to discovering the answer for himself rather than being given the answer, they must be remembering knowledge that is given to them innately. Opponents of this argument note that the questions Aristotle asks in the dialogue can be read as leading questions and it seems that he may simply be providing the slave with the answer through cleverly framed questions.

In the Phaedo, Plato uses the example of equality to argue that our abstract ideas must be innate. Once again Socrates is portrayed in dialogue, this time questioning a man who believes that we can gain our idea of equality through experience of things such as equal pieces of sticks. Socrates argues that nothing we ever experience is perfectly equal and so falls short of our idea of perfect equality. But if our idea of perfect equality cannot be found in our experience, then it must come from somewhere else, specifically it must be innate. This applies to all of our ideas which are too general and abstract to be found anywhere in our experience and so seems to indicate that we must have many innate ideas. Whilst Plato here is arguing for innate ideas rather than innate knowledge, it is still important in a discussion of innateness as a source of knowledge because (as Locke argued) it is plausible that for us to know a proposition we must have all the ideas contained in the proposition. For example, to have innate knowledge that "there is a God" requires that we have an innate idea of God that gives the proposition meaning.


The first book of Locke's An Essay Concerning Human Understanding is dedicated to arguing that there are no innate ideas or knowledge. Locke's broad argument throughout book I is that if there was any knowledge that was innate, then everybody should agree to it but according to Locke there are no propositions that all people agree to because infants, "idiots", and those that haven't considered the propositions do not agree to them. This means that none of the knowledge we have could possibly be innate. Furthermore, innate knowledge requires that we have innate ideas of the things we know about but we do not have any innate ideas because infants do not have these ideas (e.g. ideas of God, equality, impossibility not in infants). Locke considers many possible counterarguments that could be used to counter his point.

  • We have innate knowledge but it is uncovered by reason. Locke argues that if the knowledge is uncovered by reason then it cannot possibly be innate because reasoning is the process of inferring what is unknown from what is known. Furthermore, if innate knowledge is uncovered by reason then would this not mean that almost all of our knowledge including knowledge that does not seem innate at all but rather seems learned including all mathematical theorems would be innate. This seems to make the innatists claim trivial, if their definition of innateness is this broad then all of our knowledge (including learned and derived knowledge) counts as innate. Locke continues to argue that it is not even true that children agree to the knowledge that is thought of as innate when they come to reason because the knowledge that is thought of as innate is far too abstract for children to consider at the age they come to reason and even many adults may not think of it. This is because knowledge that is often thought of as innate is abstract and logical truths such as "it is impossible for the same thing to be and to not be".
  • We have innate knowledge but it is only agreed to when proposed and understood. Once again Locke argues that this makes innateness too broad because even quite trivial propositions such as "bitter is not sweet" are agreed to when proposed and understood. He also questions whether knowledge that is only agreed to when proposed to someone could possibly be innate by asking the rhetorical question "doth the proposing them print them clearer in the mind than nature did?"
  • We have innate knowledge but it is only known implicitly. Locke argues that if knowledge is innately imprinted onto the mind that it cannot be implicitly known because according to Locke "No proposition can be said to be in the mind which it never yet knew, which it was never yet conscious of". Locke goes on to argue that implicitly known propositions are incoherent saying "that a truth should be innate and yet not assented to, is to me as unintelligible as for a man to know a truth and be ignorant of it at the same time". These arguments rest on an assumption that all areas of the mind are open to conscious access. As we will see, this is a point that Leibniz utilises in his defense of innate ideas.

Locke hopes to strengthen his case against innate knowledge by showing that all of our knowledge can be derived from experience with the use of his empiricist theory. According to Locke, we are born a "tabula rasa" or blank slate without any knowledge or ideas. Through perception we are furnished with all of our ideas and our knowledge. These ideas can be broken down into simple ideas which are the basic building blocks of all of our ideas and which themselves cannot be broken down. These ideas can be combined in many different ways to create all kinds of complex ideas. For example, our idea of a horse may be built from ideas such as the horse having a certain shape, colour, movement and sound (e.g. HORSE = HORSE-SHAPED + BROWN + GALLOPING + NEIGHING). The ideas that we have can be compared with one another to see how similar or dissimilar they are from one another. We can come to our abstract ideas such as equality by a process of abstraction in which we compare our ideas and remove all the differences between them until we get to an idea that only has the similarities between all of our ideas that have equality. Likewise, we can come to a general abstract idea of "human" by comparing all our ideas of particular people and taking away all the differences between people until we get an abstract idea that applies to everyone.


In his New Essays Concerning Human Understanding, Leibniz puts forward a similar argument to Plato's argument about equality. Leibniz argues that experience cannot provide us with knowledge of necessary truths because we only ever experience particular instances but we cannot derive any necessary truths from particular instances because it does not follow that just because something has happened that it must keep on happening. Leibniz then says that the only way that we could possibly know these necessary truths (if not by experience) is that they are just innate. As Leibniz puts it:

Although the senses are necessary for all our actual knowledge, they are not sufficient to provide it all, since they never give us anything but instances, that is particular or singular truths. But however many instances confirm a general truth, they do not suffice to establish its universal necessity; for it does not follow that what has happened will always happen in the same way. Necessary truths, such as we find in pure mathematics and particularly in arithmetic and geometry, must have principles whose proof does not depend on instances nor, consequently, on the testimony of the senses [...] and so the proof of them can only come from inner principles, which are described as innate.

Leibniz also responds to Locke's arguments against innate knowledge by utilising a conception of the unconscious as well as arguing that some of our innate knowledge comes in the form of tendencies in our way to think and reason rather than in propositions. For example, in response to Locke's suggestion that our mind begins a blank slate without any ideas or knowledge imprinted onto it, Leibniz writes the following passage:

I have also used the analogy of a veined block of marble, as opposed to an entirely homogenous block of marble, or to a blank slate – what the philosophers call a tabula rasa. For if the soul were like such a blank tablet then truths would be in us as the shape of Hercules is in a piece of marble when the marble is entirely neutral as to whether it assumes this shape or some other. However, if there were veins in the block which marked out the shape of Hercules rather than other shapes, then that block would be more determined to that shape and Hercules would be innate in it, in a way, even though labour would be required to expose the veins and to polish them into clarity, removing everything that prevents their being seen. This is how ideas and truths are innate in us – as inclinations, dispositions, tendencies, or natural potentialities

Contemporary debates about innatenessEdit

[Chomsky, maybe Fodor, moral innatism, innatism in other areas]