What is knowledge? An Overview of contemporary theories of knowledge

December 28, 2024 | 5,213 words | 25min read

Knowledge is one of the most important concepts we humans possess. It has not only allowed us to advance society tremendously through scientific achievements, such as visiting the moon or sequencing the human genome, but it also helps us on a smaller, personal scale in daily life. Whether it’s knowing when a supermarket opens or closes, or understanding how to prepare food, knowledge is essential.

With the ever-increasing connectivity of our modern world through the internet and social media—platforms that have democratized free speech—nearly everyone can share their opinions and knowledge publicly, spreading their ideas far and wide. In this context, understanding what knowledge truly is becomes even more critical. This is especially true in a time when many people believe we live in a post-truth era, where the truth of a fact matters less than the emotions it evokes. How can we avoid falling into this trap if we don’t even know what knowledge is in the first place?

The central question this article seeks to answer is: What is knowledge? Specifically, what is its essence, and how can we define it in a way that is consistent with both everyday usage and with challenging examples that test our common sense and assumptions?

The theory of knowledge, or epistemology, encompasses far more than this single question. It also investigates inquiries such as: How does knowledge relate to truth? Do our senses provide the same kind of knowledge as abstract reasoning? Must a claim be justified to be considered knowledge? These related questions will be addressed when relevant to the primary question. However, this article is not an exhaustive account of all positions on knowledge. Rather, it offers an overview of common perspectives and arguments.

Basics of Knowledge

The first thing to note is that knowledge is not the same as a fact. There are countless facts in the world, but just because a fact exists doesn’t mean we know it. Nor is it enough to simply record it. For example, imagine a coin inside a shuffled shoebox. It’s a fact whether the coin landed on heads or tails. We might even write on one piece of paper that it landed on tails and on another that it landed on heads. Yet, this doesn’t mean we know whether the coin landed on heads or tails. To claim knowledge, we need some form of access to the fact.

Knowledge, unlike tangible goods such as gold or bread, always belongs to someone. However, it doesn’t have to belong to a single individual. For instance, we can say that a band collectively knows how to play a symphony, even though each musician only knows how to play their specific part.

This leads us to deeper questions: Is there knowledge that extends beyond a group of people? Can non-humans possess knowledge? And what distinguishes belief from knowledge?

Epistemology of Knowledge

There are three primary methodologies for investigating what knowledge means: Skepticism, Empiricism, and Rationalism.

Skepticism

How do you know that you are currently wearing socks? You might say that you can see them on your feet or feel them against your skin. But isn’t it possible that you are dreaming right now? While you might think this is unlikely, it remains a possibility.

It doesn’t even have to be an extreme case like dreaming. Our senses can deceive us in ordinary situations. Optical illusions, for example, work precisely because they exploit the unreliability of our sensory data, making us perceive things that aren’t happening in reality.

This raises a fundamental question: Do we have any reliable way to truly know something?

Academic Skepticism

Academic skeptics argue that it is impossible to know anything with certainty because there are always scenarios—like dreaming or hallucinating—that we cannot rule out. Historically, this school of thought developed as a response to Stoicism.

The Stoics believed that we can distinguish between impressions and judgments. Impressions are the raw sensory experiences we encounter, while judgments are the acceptance or rejection of those impressions. According to Stoicism, knowledge arises when a judgment is so clear and rational that it cannot be mistaken.

Academic skeptics, however, challenged this view. While they agreed with the distinction between impressions and judgments, they argued that no impression could ever be so clear that it’s immune to doubt. A person could be hallucinating, intoxicated, misremembering, or dreaming, making it impossible to be certain about any impression.

However, the argument of academic skepticism is self-defeating. If it’s impossible to know anything, how can we claim to know with certainty that knowledge itself is impossible?

Pyrrhonian Skepticism

To address this paradox, Pyrrhonian skeptics proposed a different approach: they argued that we should suspend all judgments—not only about knowledge itself but also about the process of attaining it.

But this leads to another question: If we suspend judgment on everything, how do we decide basic matters, such as when to eat or drink? Pyrrhonian skeptics would say that such behavior can be guided by instinct rather than knowledge.

Arguments Against and For Skepticism

Moorean Shift(Dogmatist)
One well-known counterargument to skepticism, put forward by G.E. Moore, is known as the Moorean Shift. Moore argues for the existence of external objects using a simple yet effective argument:

1. Here is one hand.
2. And here is another.
3. Therefore, there are at least two external objects in the world.
   C: An external world exists.

A skeptic might claim that one cannot truly know they have hands, but Moore essentially reverses the burden of proof onto the skeptic. If common sense tells us that our hands exist, then it is the skeptic—not Moore—who must provide evidence to challenge this. In other words, the burden of proof lies with the party disputing the common sense or the “null hypothesis.”

Inference to the Best Explanation(Abductive)
This argument, championed by Bertrand Russell, acknowledges the skeptic’s point that sense data might be misleading or distinct from reality. However, Russell argues that just because something cannot be proven with absolute certainty does not mean we cannot rationally infer its existence.

Russell posits that the simplest hypothesis is often the most reasonable to believe. The hypothesis that external objects exist is simpler than the idea that we are dreaming or are brains in vats.

Skeptics, however, counter this by noting that simplicity does not guarantee truth. For example, just because a detective has many clues pointing to one suspect does not mean the suspect is the actual culprit. Similarly, skeptics might concede that external objects are likely to exist but maintain that we cannot know this with certainty.

Additionally, skeptics argue that external objects existing might not necessarily be the simpler hypothesis. For instance, Descartes’ evil demon hypothesis, where a demon manipulates our sense data to create a false reality, provides a coherent and equally comprehensive explanation for our experiences.

Semantic Approach
The semantic approach to countering skepticism relies on how language functions. According to this view, meaning is not strictly tied to internal definitions within our minds (Semantic Internalism), but rather to causal associations between words and the world (Semantic Externalism).

For example, people once thought water was a fundamental element. Now we know it is the compound H₂O. Despite this shift, the meaning of “water” remains consistent across both understandings because it refers to the same physical substance through causal links.

Hilary Putnam builds on this to argue against skepticism. He suggests that a skeptic’s claim—such as “I am a brain in a vat”—cannot be true for someone who understands the sentence. This is because, in a virtual reality, the word “vat” would have a different causal association than it does in the real world. A person in such a scenario would only ever interact with simulated vats, so their word “vat” would not mean the same as in our reality.

A skeptic could respond, however, by introducing a scenario like this: “Until yesterday, you lived in the real world, but today you became a brain in a vat.” In this case, since the person once had real-world experiences with vats, the meaning of the word “vat” in their sentence aligns with its real-world meaning, preserving the skeptical hypothesis.

Rationalism of René Descartes

In his work Meditations, René Descartes begins by questioning all the assumptions he, and we, have about what we think we know. He does this by pointing out the fallibility of our senses, eventually introducing his famous concept of the evil demon, an omnipotent being that manipulates our minds by sending us all the sensations we currently experience. Against such an adversary, how could we ever know that something is true? The rest of Descartes’ Meditations is dedicated to reconstructing our worldview from the ground up.

Descartes argues that the first undeniable truth is that one exists. He famously states, Cogito, ergo sum (“I think, therefore I am”).

However, this is not enough to construct an entire worldview. Descartes then asks: What is the principle that allows us to say “I exist”? He answers that the principle is a clear and distinct perception of what I am asserting.

Descartes then categorizes ideas into two groups: internal ideas (e.g., a flying eagle-headed horse) and external ideas (e.g., a fireplace). As long as the evil demon deceives him, he cannot be certain these ideas are his own.

But Descartes argues that some ideas are greater than others, with the greatest idea being God—an infinite and perfect being. Since the source of perfection cannot come from something less than perfect, God must be the source of all perfection (a controversial causal principle).

He continues by claiming that human errors arise because God gave us free will, including the freedom to accept imperfect ideas. He also argues that our senses can deceive us because they are not intended to convey ideas but are instead focused on bodily sensations that ensure our survival. Descartes concludes that we can be certain we are not dreaming because a benevolent God would not deceive us into being trapped in a lifelong dream.

Empiricism of John Locke

John Locke worked on his theory for 20 years, aiming to challenge Descartes and address the issue of finding the boundaries between opinion and knowledge. His approach was based on the historical plain method.

Rationalists claim that humans possess innate ideas, which we can use to reason our way to a worldview. Locke disagreed with this. He first raised the objection that children often lack the ideas that rationalists claim all humans are born with. One might counter that children will eventually discover these truths as they mature, but this would make the idea of innate knowledge redundant.

Locke argued that the mind is a blank slate (tabula rasa), and that it is shaped by sensation and experience. Through these, humans learn and gain knowledge. This approach, which focuses on sensation and experience, is called empiricism.

Empiricists do not believe that all ideas come from experience, only the raw material for them. Reason can then be used to construct knowledge and ideas.

Locke defines knowledge as the perception of the connection and agreement, or disagreement, of any of our ideas. He differentiates between three types of knowledge:

• Intuitive knowledge: Ideas we grasp immediately.
• Demonstrative knowledge: Ideas we come to agree with, but only through a chain of reasoning.
• Sensitive knowledge: Knowledge of particulars based on sensory experience.

For Locke, knowledge is a tool for the pursuit of happiness. Therefore, if something appears to be a reliable source of pleasure or pain, it is worth considering. Locke is not concerned with the question of the skeptics, as he believes that knowledge is primarily practical, helping us navigate the world effectively.

Theories of Knowledge

Cynical Theory of Knowledge

A subject S knows that a proposition P is true if and only if S believes P to be true.

In this theory, knowledge and opinion are treated as the same thing. It argues, “What is knowledge if not the opinion of the elitist few?” According to this view, knowledge is nothing more than the opinion of those whom society deems to have high social standing.

However, this theory is disproven by the fact that knowledge is not reserved for experts or leaders but is one of the most common terms used in everyday life. Furthermore, even Nobel Prize winners can be shown to be wrong.

Protagoras’ Theory of Knowledge

Not everyone believes that truth is objective, meaning that a statement applies to everyone in the same way. Protagoras, for example, believed that two people standing outside, one feeling it as warm and the other feeling it as cold, would both be justified in saying they know it is currently cold or warm and would both be correct in their assertions.

A subject S knows that a proposition P is true if and only if: S perceives P to be true according to their personal experience or perspective.

This is similar to the Cynical Theory of Knowledge, but while cynics emphasize power dynamics, Protagoras argues more broadly for relativism, independent of power structures.

One problem with this conception of knowledge is: What if you were incorrect in the past? If Protagoras is correct, then our past selves, who were mistaken, should also be correct at the same time. A counter to this would be that the “you” of the present is not the same person as the “you” who made the mistake. In other words, things are only true for the “you” of right now, but not for the “you” of a moment ago.

Another problem arises if I believe the theory to be false: According to Protagoras’ view, if I believe something is false, then it must also be false for me, since knowledge is what you believe.

Justified True Belief (JTB)

What is the difference between ‘know’ and ’think’? For example, consider the sentences James thinks he turned off the oven vs. Brian knows he turned off the oven.

One distinction is that in the first case, James might only think that he turned off the oven, but in reality, he forgot to do it. In the second case, this cannot be true. As such, one requirement for knowledge seems to be factuality—in other words, we can only know true propositions.

It would make sense to say James thinks he turned the oven off, but he didn’t, but the sentence Brian knows he turned the oven off, but he didn’t seems nonsensical. Knowledge seems to require some level of confidence, or in other words, the knower needs to believe their knowledge to be true.

But this is not enough for knowledge. We could have someone who is very confident in their claim, but for the wrong reasons. For example, a parent might feel very confident that their child is innocent when accused of a crime, but that doesn’t mean they know it. One possibility to amend our requirements is to add that knowledge requires justified belief—it is not enough to simply think one is correct; one must also be justified in thinking so for it to count as knowledge.

More formally, we can define the theory of Justified True Belief (JTB) as:

A subject S knows that a proposition P is true if and only if:
(i) P is true;
(ii) S believes that P;
(iii) S is justified in believing that P.

The Gettier Problem

There is a problem with this definition, known as the Gettier problem. Let’s imagine a person wandering in a deserted train station, having lost track of time. They want to know the time, so they check the station clock, which reads 1:17—this is the correct time. But here’s the twist: the clock is actually broken, and the person happened to glance at the exact moment when the clock stopped at 1:17.

With how we’ve defined JTB, we would say the person knows it’s 1:17—since (i) it is true, (ii) they believe it’s 1:17, and (iii) they are justified in believing it is this time, having checked the clock. But most people would argue that the person didn’t really know it was 1:17.

This suggests that knowledge is more than the sum of justification and true belief.

Modified Justified True Belief

The reason we think the person in the train station doesn’t know the time is because they have a false belief. We could modify our definition to exclude this scenario.

More formally:

A subject S knows that a proposition P is true if and only if:
(i) P is true;
(ii) S believes that P;
(iii) S is justified in believing that P;
(iv) The justification is not based on a false belief.

But there’s still a problem with this definition. Sometimes, even though a person has a false belief, we can still say they know something. For example, if a detective gathers physical evidence and interviews one hundred people about a murder, and all of them point to the same person as the murderer, but one of the witnesses is lying and didn’t see the murder, we would still say the detective knows who the murderer is. A single false belief doesn’t invalidate all the other evidence leading the detective to their knowledge.

We could try to modify the definition of knowledge again to allow for some false beliefs, while the majority of the evidence remains true. However, this is very hard to formalize.

Causal Theory of Knowledge

According to the causal theory of knowledge, experience-based knowledge requires the knower to be appropriately causally connected to a fact.

For example, when one looks at a burning barn, the visual evidence of the burning barn causes us to have the belief that the barn is in fact burning. As such, we have a causal chain linking us back to the fact of the matter.

More formally, we can write it as:

A subject S knows that a proposition P is true if and only if:
(i) P is true.
(ii) S believes that P is true.
(iii) S’s belief that P is true is appropriately causally connected to the fact that makes P true.

Let’s think back to the clock example, where the person was accidentally correct in telling the time using the broken clock. In this scenario, there is no causal relationship between the broken clock and the fact that it was 1:17. Therefore, we wouldn’t consider it knowledge.

This definition also works for less obvious knowledge. For instance, every educated adult knows that Julius Caesar was assassinated, but most couldn’t tell you where they heard about it. However, there is still a causal chain that connects the fact of Caesar’s assassination to eyewitnesses, to ancient historians, to history books, and finally to history teachers or pop culture, which pass the knowledge on to us.

However, the causal theory has a major flaw. Imagine the following scenario: Henry and his son are driving through the countryside. While driving, Henry points to different objects such as barns and tells his son what they are. Henry is causally linked to the actual barn through his visual perception. But here’s the twist: they are driving through “Fake Barn Country,” where residents put up fake barns by only placing their facades, and through sheer luck, Henry is currently looking at the only real barn.

Does Henry know he is looking at a barn? Many people would say no, since there are thousands of fake barns, and he was just lucky enough to be looking at the only real one. But according to the causal theory, Henry knows it’s a barn because he is causally linked to the barn through his vision.

In other words, the causal theory of knowledge does not account for environmental factors.

Reliabilism

Reliabilism argues that what makes knowledge knowledge is the fact that it is arrived at through a reliable mechanism that creates true beliefs.

More formally, we can write it as:

A subject S knows that a proposition P is true if and only if:
(i) P is true.
(ii) S believes that P is true.
(iii) One has arrived at the belief that P is true through some reliable process.

The problem, however, is how reliable the process needs to be. Does the process need to create a true belief with an 80% likelihood, or does it need to be 99% reliable?

Let’s imagine a scenario: You have a lottery ticket with a chance of 1:1000 to win. The winner has not been announced yet, but with a likelihood of over 99.9%, you have lost. However, most people would say you don’t know you have lost before the announcement is made.

Knowledge First Approach

The knowledge first approach argues that knowledge is fundamental—something so basic that it cannot be analyzed further, similar to an axiom in mathematics. Instead of knowledge being made up of belief and additional conditions, this theory holds that belief is made up of knowledge.

To illustrate this, consider the question: what is more fundamental, a perfect circle or rounded corners? A perfect circle can be used to define rounded corners, but the opposite is not true. As such, the circle is the fundamental starting block.

Nozick’s Tracking Theory

In knowledge theory, there are two camps: the internalists, who believe that without supporting evidence, you do not know something (think back to the example of Julius Caesar’s assassination—we know it, but we don’t know from where we learned it), and the externalists, who believe that knowledge is a relationship between a person and a fact, and that one does not need to always know where one’s belief comes from for it to count as knowledge.

Nozick’s tracking theory is an externalist theory, which argues that knowledge is the tendency to believe something when it’s true and not to believe it when it’s false.

For example, if a doctor always prescribes the same medicine to every patient they meet, they will likely be wrong more times than right, and thus they wouldn’t be said to “know” what the correct treatment is.

More formally, we can write it as:

A subject S knows that a proposition P is true if and only if:
(i) P is true.
(ii) S believes that P is true.
(iii) If P were not true, S would not believe that P.
(iv) If P were true, S would believe P.

If someone has a constant delusion of reading, they can’t know that they are reading because (iv) is satisfied but not (iii). In other words, (iii) and (iv) together suggest that knowledge is belief that tracks some fact.

However, this theory also faces challenges. Imagine a nephew is visiting his grandmother in the hospital. The grandmother forms the belief that her grandson is healthy enough to visit her because he’s there. However, the family could be so concerned about her health that the nephew would visit even if he were sick, and the grandmother wouldn’t know. This scenario violates (iii), meaning she can’t know if her grandson is healthy or not. Yet, it feels like a family plan shouldn’t influence whether the grandmother can know it or not.

The theory can be amended by adding that the method of belief formation should be reliable or good. However, this raises the question of how reliable or good the method needs to be.

Relevant Alternatives Theory (RAT)

The Relevant Alternatives Theory (RAT) of knowledge suggests that knowing involves excluding all possible relevant alternatives.

More formally:

A subject S knows that a proposition p is true if and only if:
(i) p is true.
(ii) S believes that p is true.
(iii) S is able to rule out all relevant alternatives to p.

Consider the following scenario: You are at the zoo, and in the enclosure in front of you, you see a black-and-white striped animal. Do you know that the animal is a zebra?

Intuitively, we would say yes, and the Relevant Alternatives Theory would agree, considering all other animals that could possibly be in a zoo. However, a skeptic might pose the challenge: How do you know it isn’t a donkey painted to look like a zebra?

If we want to know it’s a zebra, we need to rule out this scenario, which can be difficult. How relevant is it to know whether the zoo owner is a prankster? Do we need to consider and exclude this possibility?

The challenge is determining which alternatives are relevant and which are not. But there’s an even bigger problem for this theory.

Relevant Alternatives Theory tells us that if we are standing 20 meters away from the zebra, we know the animal is a zebra, and we know that if it’s a zebra, it’s not a cleverly disguised donkey. But at the same time, we do not know if it’s a painted donkey. This violates the principle of closure, which is fundamental to logic.

Contextualism

Contextualism emerged as a way of keeping the parts of the Relevant Alternatives Theory without violating closure.

Words are slippery. What does tall or small mean? Is a banana small? Compared to a skyscraper, definitely. Compared to a crumb of dust, probably not. Another example is the word tomorrow, which refers to a different day depending on the context.

Even though the meaning of words can change depending on the context, we don’t need a completely different dictionary.

Contextualism is the belief that words like know are context-sensitive. In other words, once the context of the word is established, we know what it means. The standard needed for belief to count as knowledge can be much lower in some contexts (e.g., a street encounter) than in others (e.g., a formal academic debate).

In a context where we are concerned that the zebra in the zoo might be a donkey painted to look like a zebra, it would be incorrect to say we know it is a zebra. However, if we are not concerned about that possibility, it would be correct to say we know it is a zebra.

For example, if someone says they know the sandwich in front of them is a chicken sandwich, normal people might wonder if it’s really chicken or tuna, while a skeptic might question if the sandwich is a hologram. Both are valid concerns but require different standards of evidence to count as knowledge.

The idea of contextualism is that when we use the word know, we express something different depending on the situation. This is not necessarily a theory of knowledge, but more about the language of knowing, which can be used in various other theories.

The biggest objection to contextualism is that it may be seen as a form of skepticism in disguise. Skeptics don’t have a problem with the lower standard of knowledge in certain contexts, but with the higher standard in others.

The counterargument is that the higher standard is not more valuable than the lower standard; they simply have different use cases. A lower standard is faster to reach with less expenditure but is less accurate, while the higher standard is slower and more accurate. For contextualists, asking which is better is like asking which weekday tomorrow ultimately is.

Internal Relative Invariantism

Imagine you’re heading to work and suddenly ask yourself, “Did I turn off the oven?” You then remember and say, “I know I turned off the oven.” Now, picture a detective from the police finds you and asks, “Did you turn off the oven?” You might respond, “I think I did turn off the oven,” replacing the word know with the weaker version think.

Another example is the famous idiom “treat every gun like it is loaded.” In a normal situation, one might act as though they know the gun is loaded. However, if a robber breaks into your home, you will need to check twice whether the gun is loaded or not.

In both situations, you know whether the gun is loaded or not, but when the situation becomes more serious and your life might be at stake, you need more evidence before making a judgment.

While contextualism is only a theory about the language of the word know, internal relative invariantism is a theory about how knowledge works. In this view, a judgment is not true just based on the situation, but true full stop. A skeptic would be wrong in saying that you lack knowledge in the first scenario, and that they are merely using a different standard of evidence.

A problem with this theory is how easily intuitions can shift. In the story with the oven, how can the meaning of the word know change so easily based on what a skeptic might say?

Testimony

How does testimony play a role in knowledge? Can we gain knowledge by listening to others?

Just like with everything in philosophy, there are many approaches to these questions. I will briefly mention some of them but not go into depth:

• John Locke argues that testimony never provides you with knowledge. At best, it provides a high probability, which can later be disproven. For Locke, you can never know if you were born, because your beliefs about this come from family members or your birth certificate. You can only say it is highly probable that you were born.

• Reductionism claims that we can gain knowledge through testimony and that it should not be treated differently from any other sensory information. We use past experiences to collaborate and verify the testimony.

• Default View holds that testimony is a special kind of information gathering that is independent of other factors. This view argues that testimony is distinct, just like reason is distinct from perception.

Imagine a Wikipedia page, created by many volunteers. A Lockean would say that the information we get from Wikipedia is a highly probable opinion rather than knowledge. A reductionist would say that any knowledge we gain depends on the character we infer the creators of the Wikipedia pages to have. Finally, a default view theorist would say that Wikipedia provides knowledge of the facts it reports.

Conclusion

With this article on knowledge, I have now explored three major philosophical topics: meta-ethics, normative ethics, and epistemology. While I found certain problems in the popular positions of the first two areas, I feel a greater sense of dissatisfaction with the theories of knowledge. The issues in epistemology seem harder to reconcile, and I find myself unhappy with all the positions explored.

I find some merit in contextualism and believe it makes sense, but it doesn’t provide a clear definition of what knowledge is. For now, I must be content with viewing knowledge as a fundamental building block and accepting that the standard of evidence differs depending on the situation.

References:

• Knowledge: A Very Short Introduction
• Gettier Problem
• Moorean Shift