by James Beilby and David K. Clark

How can we know the world around us? How can we know God? How can we know anything at all? These are some of the questions of epistemology, the study of theories of knowledge.

Epistemology has two main goals. First, we want to find as much truth as possible. And second, we want to avoid as much falsehood as possible. These two goals stand in tension with each other. I can easily acquire very large amounts of truth. If I were totally gullible, I’d believe just everything I hear. That would give me the largest number of true beliefs possible. But the problem is that along with all the true beliefs I’d acquire, I’d also obtain many false beliefs. So I’d have some needles of truth hidden in a very large haystack of error. That wouldn’t help me much.

Similarly, I could easily avoid as much error as possible. If I were completely skeptical, I’d disbelieve everything. That would safeguard me against every falsehood. But the problem is that I’d miss out on all truth whatsoever—and some truth might be very important. So that wouldn’t help me much either.

No one urges us to believe absolutely everything. But some very important and influential thinkers do advise us to believe nothing (or very little)—or at least they recommend that we believe only when an idea is incredibly well supported. This is skepticism. Skepticism puts most of its energies into avoiding error, and very little effort into finding truth. So how can we develop an understanding of epistemology that goes beyond skepticism? How can we balance our desire for truth with our need to avoid error?

Truth and Knowledge

It’s critical to distinguish truth and knowledge. Too many people equate these two concepts, with chaotic results. But truth and knowledge are different concepts. Put simply, true affirmations are those that correspond to reality. So truth is a characteristic of statements that properly describe aspects of the real world. This is called the correspondence view of truth.

The correspondence view of truth isn’t a method for testing truth claims or discovering knowledge. It’s a definition of what we mean when we say that a statement “is true.” According to the correspondence view, what makes a statement true is reality itself. A statement like, “This car is red,” is true, simply if the car in question actually is red. Truth doesn’t depend on anyone knowing the truth. So, for instance, even if no one’s around to discover that it’s 115° at 2:00 p.m. on August 15, 1977, in the middle of Death Valley, it’s still true that it’s 115° out in that desert. The statement, “It’s 115° at 2:00 p.m. on August 15, 1977, in the middle of Death Valley,” is true even if no one thinks about it. Truth is independent of human minds.

The word knowledge denotes a person’s proper understanding of the true nature of reality. This proper grasping of reality can be knowledge by acquaintance. In this sense, we know what the color blue looks like. An accurate perception of reality can also take the form of knowledge of true statements that describe that reality. Both of these are important. Knowing a friend is more akin to knowing by acquaintance, and it’s more important than just knowing about a friend. But knowing true statements is also important. In fact, the two kinds of knowing are related, because knowing by acquaintance entails the truth of descriptive statements. If I know a friend named Greg, then I know many true propositions, including “Greg exists” and “I count Greg as a friend.”

For a belief to count as knowledge for a person, it must meet three conditions. First, knowledge must be true. We don’t just mean that someone thinks the idea is true. We mean that the idea is true. Members of the Flat Earth Society (believe it or not, there is such a thing!) think that the earth is flat. Do we count their belief as knowledge? Of course not! They believe the earth is flat, but their belief is false and hence can’t count as knowledge. Genuine knowledge is true.

Second, knowledge must be believed. We must believe a claim (that is, we have to hold a belief as true) in order to know it. Of course, believing something isn’t enough to make it true, and not believing it doesn’t make it false. But without believing, a true idea isn’t knowledge for us. Suppose it’s true that one of my great-great-grandfathers was a Confederate Army lieutenant whose troops played a key role at the Battle of Fredericksburg. Now suppose I don’t know this fact and don’t have any particular beliefs about the lieutenant. In this case, it’s obviously true that my great-great-grandfather was this lieutenant, but it would be very odd to say that I know this about my great-great-grandfather. In fact, I probably have very few beliefs about my great-great-grandfathers. I can know generic things: eight persons who lived sometime in the last 250 years are my great-great-grandfathers. They were males; they fathered my great-grandparents; and none of them ever watched TV or received an e-mail. But since I don’t believe anything individually about any of them, I can’t be said to know anything distinctive about them as individuals. We must believe something to know it.

Third, knowledge requires some other fact that legitimates the knower’s holding that belief. The belief must arise out of this legitimating fact; it must be grounded in this “something else.” Now we’re being vague because the exact nature of this legitimating fact is very hotly debated. But the importance of this legitimating fact is that it separates genuine knowledge from true beliefs that are held purely by chance. Obviously, we shouldn’t consider a true belief as knowledge if that belief was the result of a wild guess. Say I win the lottery by guessing the winning numbers. Sure, I hoped that the winning numbers would be the first five digits of my Social Security number, but it’s wrong-headed to say that I knew that they would be the winning numbers! In sum, by the word knowledge, we mean a true belief held by a person for an appropriate reason—that is, grounded in a legitimating “something else.”

Forming and Testing Beliefs

If knowledge is true belief plus some legitimating fact, then how should we set the standards for assessing these legitimating facts? The 17th century philosopher Rene Descartes concentrated on this very problem. His philosophy set the stage for modern discussions of knowledge. Descartes’ approach posited very high—too high—standards for that “something else,” that legitimating fact that distinguishes merely true belief from genuine knowledge.

Methodism Vs. Particularism

In order to weed out false beliefs and gain genuine knowledge, Descartes required that all candidates for genuine knowledge must arise from a method. Correct method (for Descartes, the geometric method) is the key to finding true knowledge. This approach is called methodism. Methodism, in this discussion, isn’t the religious denomination. Rather, it’s an epistemic theory that stipulates this: we know any particular true belief if and only if we arrive at or produce that knowledge by following a correct method.

Here’s a specific example. Suppose someone asks me whether I know the statement, “My coffee cup is blue.” (Let’s call this statement p.) Methodism requires that before I can truly know p, I must follow a proper method by which I know p. So to know any particular truth, methodism says I must follow a proper epistemic method.

Although Descartes’ methodism may seem like a promising way to ground knowledge, it’s fundamentally flawed. Methodism requires that before I can know anything, I must have prior knowledge of the method by which to know that thing. But then how do I know that method itself? My coming to know what method to use would itself require following a prior method. This quickly leads to what’s called an infinite regress. Every time I try to answer the problem, the problem keeps appearing. I start moving back a chain of questions. But every time I move back to a prior link in the chain, the problem repeatedly emerges. It’s like asking, “What explains Michael’s existence?” If I say, “His parents,” I just raise again the very question I hoped to answer: “What explains his parents’ existence?” “Their parents?” Ultimately, given the methodist approach, there’s no way to end this infinite series of questions. In the end, if methodism were true, I’d have to know something (the right method) before I could know anything. There’s no way out of this double bind.

But there’s another approach to finding the legitimating fact that separates true belief from knowledge. It’s called particularism. Particularism starts by assuming that it’s right to know particular things directly (that is, without following a method) since we find that we already know many particular things. In certain conditions, we directly and properly form true beliefs. And we form these beliefs through a variety of means. We see a tree or hear a train. We compute things. We infer conclusions from things we see or hear. We learn from experts. We read the Bible. Each of these processes generally leads to true beliefs. And so it’s legitimate, particularism says, to count particular beliefs like these as knowledge. We shouldn’t be required to step back and first prove that, say, our vision is perfect, before we rightly know something we see. That would lead us back to the methodist trap (since we’d have to prove the method that we use to prove our vision is perfect). So it’s better just to assume that our properly formed beliefs are innocent until proven guilty. With these particular beliefs in hand as examples, we can begin to understand what knowledge is—and gradually to increase the number of things we know.

Testing Individual Beliefs

But difficulties arise when we run into contrary evidence. Let’s say that, just by looking at it, I form the belief that a particular stick is straight. I have no reason to doubt this because my eyesight’s generally very good. Then I put the stick in water, and suddenly I form the belief that it’s bent. Again, my eyesight’s pretty good. But my mind tells me that the stick can’t be both straight and bent. So which of my two beliefs is true? Or let’s say my wife helps me pick out a tie that looks gray to me. I protest: “It’s too drab.” But she assures me that the tie is a nice shade of rose. Should I trust her judgment?

It’s when this sort of thing happens that testing procedures become important. This is where we follow methods. We have procedures to help us figure out which of the conflicting things that our normally reliable belief-forming processes are telling us is actually correct. The conflict between beliefs produced by these normally reliable indicators leads us to question whether what we think we saw could really be so. I remember something in my high school physics class about light refracting when it passes through water, and this accounts for the bent appearance. Or I remember that I’m color blind in reds and greens, and this explains why the rose-colored tie looks gray to me. So what do we do about conflicting facts? We go to procedures to help us sort them out. (This is the correct insight that methodism takes too far.) Should we just give up and concede skepticism? Hardly.

What are the procedures or strategies for evaluating competing beliefs? First, our beliefs should be rational. At a minimum, this means that our beliefs shouldn’t contradict one another. This is coherence, a negative test. Say I believe both that “I’m the world’s leading microbiologist” and that “I don’t know much about microbiology.” These beliefs are obviously incompatible, and so holding both beliefs at the same time is irrational. One of the two (or maybe both) must go. Coherence is necessary. But it doesn’t guarantee truth. Incoherence is a significant red flag. It guarantees that some beliefs are false. We should pursue strategies in order to avoid holding incoherent beliefs.

Second, our belief should fit with the evidence. If a belief doesn’t fit with data we know to be true, we should give up that belief. Take the claim, “I’m the sixteenth president of the U.S.” This belief conflicts with many well-established facts: “The sixteenth U.S. president’s name was Abraham Lincoln”; “My name is David Clark”; “Abraham Lincoln is dead”; “I’m alive”; and so on. So I’m really not the sixteenth U.S. president.

Generally, we look for beliefs that fit the evidence. But notice something very important. We don’t stipulate a rule: “Every belief must be proved by evidence before it counts as knowledge.” Among other problems, that rule would land us back in methodism. The problem with making this rule into an absolute requirement for knowledge is that the rule itself can’t be proved by evidence. No evidence could ever prove that “Every belief must be proved by evidence before it counts as knowledge.” So we do look for evidence to help us, but only when it’s appropriate.

Testing Large-Scale Models

So far we’ve been talking about particular beliefs. But we also seek knowledge about large networks of truth claims. A large-scale scientific theory, for example, is a complex set of interlocking claims, all connected in a large network. Large-scale models include many different kinds of things, including scientific, historical, and even religious convictions.

Large-scale models compete with each other to see which one does the best job of explaining all (or most of) the known facts. Thus, for instance, the heliocentric (sun-centered) model of our solar system competed with the geocentric (earth-centered) model. Though this isn’t well known, both the heliocentric and geocentric models explained the available physical facts equally well for centuries. Physical observations didn’t finally confirm the heliocentric model until more than 200 years after Galileo’s controversies. Thus, the heliocentric model didn’t compete with the facts. Rather, it competed with and finally defeated the geocentric model of the solar system by doing a better job of explaining the most facts. This is one way that large-scale models gain support—by outdoing their rivals at explaining the data.

Here’s another example. When National Transportation Safety Board (NTSB) investigators are trying to explain a plane crash, they look for evidence. They know what to look for because they’ve explained other crashes by finding telltale facts that guided them to large-scale explanations. The telltale facts are clues that unlock patterns of interpretation and lead to strongly supported explanations. The NTSB puts all the data together and concludes, say, that the plane crashed because a turbine blade in one of the engines shattered. The power of this explanation to incorporate all the relevant data—like the loud explosion passengers heard and the sudden loss of airspeed reported on the cockpit data recorder—is a major reason we hold that the large-scale theory is a properly supported, interlocking set of true beliefs. The individual facts are themselves grounded in experience (such as the sound of the explosion, the report of the plane’s reduced airspeed and the shattered pieces of the blade). The large-scale theory incorporates and explains these and many other facts.

Complex explanatory models can form ongoing programs of research and investigation. They not only explain what we know already. They can also guide us to what we don’t yet know. Take, for example, the discovery of Neptune. Uranus didn’t orbit the sun as the large-scale models suggested it should. But when scientists imagined that another planet was exerting gravitational force on Uranus, then its orbit suddenly made sense. So scientists began looking for this other planet, and sure enough, they found Neptune. This is similar to “superstring theory” which developed when theorists used mathematics to explain their observations. The mathematical calculations worked out beautifully when scientists assumed the existence of things they called superstrings. The calculations are powerful in that they explain a number of related issues. So researchers posit that superstrings exist even though they can’t observe them. Research programs that guide researchers to new discoveries are progressive. This helps confirm their connection with the real world.

But testing large-scale constellations of beliefs isn’t simple. In fact, it’s sometimes impossible. Theories about particular events, like why a particular ship went down in a perfectly calm sea, may never in fact be understood. The problem might be that certain key pieces of evidence are stuck too far down on the sea floor. This means we could explain the event in principle, but can’t in fact. That is, there’s no logical reason why we can’t explain this event, but there’s a practical barrier to our understanding. So in this case, we should remain agnostic rather than claim to know what we really can’t know—at least until we develop a new submersible vessel that can get down to the wreck and find the key evidence. The truth about some complex processes might just remain hidden.

Testing models is even more complex because it requires making judgments of several different kinds. What are the facts to be explained? (Sometimes the two models will explain different ranges of data, and there’s no way to step outside the two models to know which range of apparent facts is really most relevant.) What are the criteria by which we decide which explanation is best? (Sometimes the two explanations will excel at two different criteria—one model might be simpler while the other is more helpful in guiding us to new discoveries.) So our procedures aren’t straightforward and linear. But reasonable judgments are still possible. When the NTSB investigators find a cracked turbine blade, we know we shouldn’t blame the pilots for the crash (and maybe we should blame the jet engine manufacturer). Gathering knowledge isn’t always easy, but it’s amazing how much we can learn through carefully using all our strategies in a coordinated way.

Knowledge and the Intellectual Virtues

Thus far, we’ve been discussing some of the key elements of a proper understanding of knowledge, including belief formation and testing. We’ve argued that knowledge requires true belief plus some account of that belief—something that legitimates the belief. But thus far we’ve been quite coy about what this account is. It’s time—indeed, past time—to repair that deficiency.

What is this feature that, when added to true belief, constitutes knowledge? Here scholars disagree—in fact, there are few things about which epistemologists disagree more! Thankfully, it’s not our purpose to address all the academic squabbles. Rather, we’ll offer an account of knowledge that we find persuasive. t focuses on the relationship between knowledge and the intellectual virtues.

What are intellectual virtues? Virtues are qualities of excellence possessed by a person. Intellectual virtues share some characteristics with moral virtues. In fact, many acts that are virtuous in a moral context are also virtuous in an intellectual context. Examples of intellectual virtues are honesty and courage. Being intellectually honest means making a fair appraisal of the evidence at hand, dedicating effort to reach valid conclusions, admitting personal biases that affect beliefs, and seeking to override or reduce those biases. In an intellectual context, courage involves, among other things, being willing to take a minority position when the evidence points in that direction. It also means investigating personally held beliefs with rigor.

An intellectual virtue, therefore, is a characteristic of a person who acts in a praiseworthy manner in the process of forming beliefs. But an epistemic virtue isn’t simply an instance of intellectual skill. For example, think about the ability to see sharply. This is a skill that some lucky people have from birth. This ability isn’t developed over time. (In fact, eyesight falters over time.) So it’s not particularly virtuous. Virtue relates more to what a person does with abilities or skills like incredibly sharp vision.

Further, the intellectual virtues don’t just happen naturally. Rather, they arise from habits. Like good habits (such as exercising and eating healthfully) and bad habits (like biting fingernails and gossiping), the intellectual virtues are the sorts of things that become more and more a part of us the more we practice them. Similarly, the more we practice their opposites, like intellectual dishonesty, the more difficult it becomes to respond to any given situation in an intellectually virtuous way.

Intellectual virtues influence, and are influenced by, the motivations of the one employing them. A person must come to believe something out of proper intentions. Say that a student named John hears a teacher talking about a classmate whom John dislikes. “He is nice,” the teacher says. Because of his ill will toward the student, John hears, “He has lice,” and he jumps on this bit of negative information. Even if it’s true that the student has lice, does John’s belief count as knowledge? No. Even if he believes it, it’s true, and it’s grounded in a normally reliable belief-forming process (John has good hearing), from a virtue perspective, John’s belief doesn’t count as knowledge since this belief arose in an intellectually non-virtuous way. John’s belief was shaped by his malicious attitude toward the fellow student. Given all these points, we define knowledge in this way: Knowledge is true belief that is reached or acquired through an act of virtue.

The key insight of virtue epistemology is that knowledge isn’t just an issue of whether evidence exists for specific belief at a particular time, but an issue of how a person goes about gathering evidence. So whether or not a particular belief is properly grounded for me has to do with how I formed the belief. Did I form this belief in accord with the intellectual virtues, reflecting praiseworthy habits of belief formation and testing acquired over time? Or did I form this belief in a manner that reflected slipshod handling of the evidence or haphazard reasoning processes?


We began by talking about our need to gain truth and avoid error. Skeptics are fixated on avoiding error. Their concern is the adequacy of a person’s evidence. To avoid falsehood, skeptics place a very high standard for admissible evidence. For some skeptics, the standard is so high that every belief becomes doubtful.

We agree that avoiding falsehood is vital. And given our virtue-oriented epistemology, the notion of evidence is important. But more important is whether we rightly handle the evidence we have! An unscrupulous person can twist evidence to support the position he holds. But if we’re intellectually virtuous, we’ll operate differently. We’ll treat evidence honestly, overcome our biases toward our own culture’s preconceived notions, and refuse to misuse evidence to gain power or to pretend that our own pet beliefs are superior.

So is knowledge possible? Even though people have many false beliefs, Yes! The existence of junk car yards doesn’t count as evidence against the existence of new cars. Similarly, the existence of intellectually non-virtuous people doesn’t show that intellectually virtuous people fail completely in their quest for genuine knowledge. In sum, due to human limits, some things are beyond knowing. But if we exercise the intellectual virtues, we can achieve genuine knowledge about important things. Skepticism wins some skirmishes along the way, but it doesn’t win the war!

James Beilby is an adjunct professor at Bethel College and Theological Seminary, St. Paul, MN.
David K. Clark is professor of theology and Christian thought at Bethel Theological Seminary, St. Paul, MN.

Most of the content of this article was previously published in James Beilby and David K. Clark, Why Bother with Truth? Arriving at Knowledge in a Skeptical Society (Norcross, GA: RZIM, 2000).