The Concept of Knowledge and Skepticism PDF

The Concept of Knowledge and Some Varieties of Skepticism Lyle Crawford I know the number of the bus that takes me to school. I know the name of the main character in the movie. I know the atomic number of helium. I say “know” in each case and these seem...

The Concept of Knowledge and Some Varieties of Skepticism Lyle Crawford I know the number of the bus that takes me to school. I know the name of the main character in the movie. I know the atomic number of helium. I say “know” in each case and these seem to be three examples of one thing, knowledge. But what is that? This question was the origin, in Ancient Greece, of epistemology (from epistēmē, knowledge). It’s the question we’ll try to answer in Part I. In Parts II and III we’ll use our (incomplete) understanding of knowledge to ask some skeptical questions, questions that doubt or deny the possibility of knowledge: Can I know anything about a world beyond my own mind? Is moral knowledge possible? -I. THE CONCEPT OF KNOWLEDGE- 1) CONCEPTUAL ANALYSIS a) Dictionary Definitions “What is knowledge?” Couldn’t we just look up knowledge in a dictionary? The problem with this is that a dictionary only tells us how English speakers use the word. This might be the beginning, but cannot be the end, of our philosophical work. We’re interested in the concept, not the word. Maybe English speakers have an incorrect or incomplete understanding of the concept. Imagine looking up star in a 15th century dictionary. It would say something like “an eternal point of light fixed to the inside of the giant sphere that rotates around Earth”. This definition is completely wrong. We’ve improved our understanding of stars since the 15th century. Even a modern dictionary definition might leave out important information, for example that stars shine by nuclear fusion. Maybe we can improve our understanding of knowledge in similar ways. b) Conceptual Analysis and Counterexamples A traditional method for investigating a concept is to try to formulate a conceptual analysis and test that analysis with possible counterexamples (exceptions). An analysis of concept X is a set of conditions that are said to be 1) each necessary (required) and 2) together sufficient (enough) for being X. This is the same as: 1) All Xs have these features, and 2) All things with these features are Xs. necessary necessary | | For example, here is an analysis of sport: competitive team activity sufficient Lyle Crawford © 2025 The analysis says: - All sports are competitive team activities. - All competitive team activities are sports. But the analysis fails in two ways: - It is too narrow: it leaves out some sports. o Counterexample: an Olympic marathon, a sport that is not a competitive team activity. People run marathons as individuals. The “team activity” condition in this analysis is not necessary. Not all sports are competitive team activities. - It is too broad: it includes things that are not sports. o Counterexample: teams in a science class competing in a design challenge. This is a competitive team activity that is not a sport. The conditions of this analysis are not sufficient. Not all competitive team activities are sports. This method depends on our ability to notice counterexamples. The idea is that by doing this, and formulating an analysis that avoids counterexamples, we can learn about the concept. 2) THE TRADITIONAL ANALYSIS OF KNOWLEDGE a) Three Types of Knowledge Consider three questions: Do you know how to swim? Do you know Bob, my friend from school? Do you know what time it is? It seems these different types of knowledge altogether. - If I know how to swim, or how to speak German, or how to play the violin, I have an ability. There is something I can do. This is ability knowledge. - If I know Bob, or the city of Vancouver, or the taste of chocolate, I am familiar with something. I’ve experienced it and can recognize it. This is acquaintance knowledge. - If I know that it’s noon right now, or that the Sun is a star, I have knowledge of a fact. This is propositional knowledge or factual knowledge. In many cases of ability or acquaintance knowledge, someone may have propositional knowledge, as well. If I know how to swim, I may know that I must exhale with my face underwater and inhale with it above water. If I know the taste of chocolate, I may know that dark chocolate is mildly bitter. Propositional knowledge is certainly not the only kind of knowledge worth having, but this is the concept that philosophers have traditionally been most interested to analyze. From now on, we’ll use knowledge to mean propositional knowledge. b) Belief People sometimes speak of knowledge as though it were a kind of stuff, something that may exist outside of anyone’s mind, e.g. “Wikipedia contains a vast amount of knowledge”. But we want an analysis of knowledge that also applies to the verb know. It would be weird to say that Wikipedia 2 itself literally knows something. People know things and record their knowledge in Wikipedia. The reason it’s weird to say that Wikipedia knows something seems to be that it’s weird to say that Wikipedia believes something. This points to the first condition in the traditional analysis: belief. Try saying this: “Bob knows that the Sun is a star, but he doesn’t believe that it is.” Does this make any sense? It seems not. Belief is necessary for knowledge. All knowledge is belief. This immediately raises new questions. For example, what kinds of minds are capable of believing? Is language required for belief? To fully understand knowledge, we would need to better understand belief. But our primary task here is the analysis of knowledge, so we’ll settle on a basic definition of what it means to believe a proposition: to think that the proposition is true. Is belief sufficient for knowledge? Is all belief knowledge? It seems not. There are counterexamples. Suppose I believe that my girlfriend is planning a surprise birthday party for me. But she isn’t. Actually she completely forgot that it’s my birthday. If belief is sufficient for knowledge then I know that she’s planning a surprise party even though she isn’t. Does that make any sense? Surely not. The counterexample (a false belief) shows that this analysis is too broad. Belief is not sufficient for knowledge. Not all belief is knowledge. Here's an idea that will guide our analysis: to say that someone knows something is to credit them with a cognitive achievement. Belief is not sufficient for knowledge because believing is just too easy. Anyone might believe anything! Part of the point of the concept knowledge is to refer to “successful” beliefs, in other words, true beliefs. c) Truth Truth is necessary for knowledge. Again, this immediately raises new questions about truth, in particular: What is truth? To stay focused on the analysis of knowledge, we’ll just use the most common definition of truth: a proposition is true if it “matches” or “corresponds to” the world (the facts). Someone’s belief is true if the proposition they believe is true. Now on with the analysis. Is true belief sufficient for knowledge? Is all true belief knowledge? Once again, it seems not. There are counterexamples. Walking down the street, I see a fortune teller sitting in her booth. Just then I wonder if my girlfriend is planning a surprise birthday party for me. I pay the fortune teller $30 and ask her. She gazes into her crystal ball and says “Yes, she’s planning a surprise party for you.” Because of this, I now believe that my girlfriend is planning a surprise party for me. And she is! I have a true belief. But do I know that my girlfriend is planning a party? It seems not. Something is missing. I get the facts right, but only by luck (chance). This doesn’t look like much of a cognitive achievement. A lucky true belief is like an archery competition in which an archer blindfolds themselves, spins around until they are dizzy, then wildly shoots the arrow… and hits the 3 target perfectly. It’s pointless to give them an archery prize. The prize is to honour archery skill. Similarly, if my belief is true only by luck, it’s pointless to say that I have knowledge. The counterexample of a lucky true belief shows that even this improved analysis is too broad. What’s missing in the fortune teller story? The fortune teller is just putting on a performance for money. Her crystal ball has no magical powers. When I believe that my girlfriend is planning a surprise party for me because of what the fortune teller says, my belief is not based on good evidence. My belief is not justified. d) Justification This brings us to the traditional, three-condition analysis of knowledge: justified true belief (JTB). All knowledge is JTB; all JTB is knowledge. Again, this immediately raises new questions. Perhaps the most obvious is: How good does my evidence have to be in order to justify me in believing something? We’ll come back to this below. Here we’ll briefly visit a question about the general nature of justification: Is justification an internal or external condition of knowledge? Internal and external mean internal and external to my mind and thoughts. It seems fairly clear that belief is an internal condition because my beliefs are a part of my mind. And with the way we defined truth above, truth is an external condition because the truth of my belief depends on the world “out there” beyond my mind and thoughts. What about justification? Internalism claims that justification is an internal condition. My belief is justified if I formed it by actively thinking about my good evidence and how it supports my belief. I can explain to someone else (or to myself) why my belief is reasonable and how it’s based on good evidence. Externalism claims that justification is an external condition. What matters for justification is that my belief is formed by a process that reliably forms true beliefs. I don’t need to be aware of this process or able to explain it led me to form my belief. Externalism may be more plausible if we think that, for example, infants and animals can have justified beliefs. Infants and animals don’t base their beliefs on evidence in any way that they could explain, not least because they don’t have language. But they do have mechanisms that reliably form true beliefs. Is justified true belief sufficient for knowledge? Is all JTB knowledge? From Ancient Greece to the 20th century, philosophers mostly thought “yes”. That’s why JTB is the traditional analysis of knowledge. Then in 1963, Edmund Gettier realized that this traditional analysis has a problem. Even the JTB analysis is too broad. JTB is still not sufficient for knowledge. This is called the Gettier Problem. 4 3) THE GETTIER PROBLEM a) Gettier Cases A Gettier case is a lucky JTB. It’s a counterexample to the traditional analysis (JTB) of knowledge. These counterexamples went unnoticed for so long because many are weird scenarios that we are less likely to encounter in regular life than the counterexamples we saw to the simpler analyses. We can write the formula for creating a Gettier case in three steps. Here it is along with an example. 1) Someone forms a justified belief, based on strong supporting evidence. 2) Unknown to the person, their evidence is actually faulty or incomplete. Some additional facts “defeat” their evidence. If the person knew these, they would not form the belief that they do. 3) Just by chance, their belief is true. 1) Justified Belief: I believe it’s 25˚C outside. I form my belief after checking a thermometer that I’ve used regularly for many months, which has always been accurate. It displays 25˚C. 2) Faulty evidence: Unknown to me, last night the sensor in the thermometer broke and the display is now stuck at 25˚C. If I knew this, I would not believe that it’s 25˚C. 3) Truth: Just by chance, it really is 25˚C outside. I have a justified true belief: it’s 25˚C. Do I know that it’s 25˚C outside? To most people, it seems that this belief is not knowledge. It’s just luck that it’s true! This seems to be a counterexample to the traditional analysis: JTB is not sufficient for knowledge. Not all JTB is knowledge. b) Attempting to Fix the Analysis After Gettier’s discovery, many philosophers tried to find a better analysis that avoids Gettier case counterexamples. Alvin Goldman (1967) tried to improve the analysis by first figuring out how a Gettier case actually happens. Normally when I have a JTB, there is some fact that 1) makes my belief true, and 2) causes me to form that belief. Normally, if I have a JTB that it’s 25˚C, there is a causal chain like this: Air temp. (25˚C) Thermometer display (25˚C) I see display. I believe it’s 25˚C. Goldman noticed that in a Gettier case, this causal chain is broken. The truth of my belief is not causally connected to the fact (air temperature) that makes my belief true. I would believe that it’s 25˚C even if the actual temperature were 24˚C, 26˚C, etc. Air temp. (24˚C) Thermometer display (25˚C) I see display. I believe it’s 25˚C. Air temp. (26˚C) Thermometer display (25˚C) I see display. I believe it’s 25˚C. 5 So Goldman proposed that knowledge is true belief that is causally connected to the facts that make it true. This analysis still needs narrowing. Is any kind of causal connection sufficient? I’m mountain climbing and a rare blue mountain goat above me kicks lose a rock, which falls on my head and causes a brain injury that causes me to believe: “This mountain has a rare blue mountain goat”. My belief is true and caused by the rare blue mountain goat, but it doesn’t seem to be knowledge. This causal connection is not of the right kind. What is the right kind of causal connection? Annoyingly, the obvious answer is “a justifying kind”, and now we still need to know what justification is. But Goldman later realized (1976) that the whole causal theory was completely wrong and couldn’t be fixed. His counterexample is a famously strange case. Fake Barn Land I’m driving through some farmland, looking out my window at what appear to be barns every few kilometres. But in this strange land, people do not build barns. They build fake barns, giant flat barn shapes that could be used as props on cheap movie sets. These are what I see as I drive down the road. Now I look out the window and think: “There’s a barn over there.” Normally when someone believes this while driving through Fake Barn Land, their belief is false. But in all of Fake Barn Land, there is a single real barn among the hundreds of fake barns. And it just happens that this one real barn is what I’m looking at when I form my belief! My belief is true because there really is a barn over there. And my belief is causally connected to the fact that makes it true: the real barn causes sunlight to reflect into my eyes in a barn-shaped pattern, which causes activity in the visual cortex of my brain, which causes me to believe there’s a barn over there. Yet it was obviously just luck that I formed my belief while looking at the one real barn in Fake Barn Land! I would have formed the same belief while looking at any of the fake barns up and down the road, and I would have been wrong. Once again, the truth of my belief just seems too lucky to be knowledge. So a causal connection doesn’t fix the analysis. It’s still too broad. Whether a belief is knowledge can depend on the context in which it is formed. There is still no analysis of knowledge that philosophers agree on. Some philosophers think that the whole attempt to find a conceptual analysis of knowledge is philosophically misguided. But most at least agree that knowledge requires some sort of justification. If that’s right, it’s enough to create the problem to which we’ll turn in Part II. This problem is not an analytical question about the concept knowledge, but rather a skeptical question about whether knowledge is possible. 6 4) DEGREES OF BELIEF AND JUSTIFICATION We’ve treated belief, truth, and justification as binary (yes/no): someone either believes or not; a proposition is either true or not; a belief is either justified or not. This is reasonable in the case of truth. A proposition is either true or false. But belief and justification seem to come in degrees. Someone can have better and worse evidence, and they can make use of it more or less carefully. There can be degrees of justification. This is why we set thresholds when we need to make a decision based on evidence. In law, we use thresholds such as “beyond a reasonable doubt” and “on the balance of probabilities” (more likely than not). In science, we use probabilities such as “less than a 5% chance the result is a statistical fluke” for “statistical significance”. It's normal to believe something more or less strongly, with more or less confidence. A degree of belief is called a credence and can be expressed with a number from 0 to 1, just like a probability. With a credence of 50%, for example, I believe equally strongly that some proposition is true and that it is false. Some philosophers urge people to think and communicate with credences rather than all-or-nothing beliefs. It allows us to recognize that we very rarely are (or should be) absolutely certain (100% or 0%), and to apply mathematical rules of probability to our own beliefs. What would it mean for knowledge if we acknowledged degrees of justification or degrees of belief? Here are a few possibilities, none of which are very attractive. 1) Knowledge comes in degrees. I can “strongly know” or “weakly know” a proposition, according to my degree of belief and my degree of justification. The notion of someone strongly or weakly knowing something is odd. This is not how we normally speak. But that doesn’t mean it’s wrong! Maybe we should just get used to it. 2) There are thresholds. In order to count as knowing something, I must have a credence at least ____ high, and I must have justification at least ____ strong. This suggestion immediately raises many difficult questions: How would we ever determine what thresholds are sufficient? Do we need to be able to precisely measure belief and justification, and what if there’s no good way to do that? Do the thresholds change in different contexts? 3) The concept of knowledge doesn’t apply or isn’t useful with degrees of belief and justification. Some philosophers might be ready to abandon the concept of knowledge, but this would be a major revision in our ways or thinking and speaking, even more drastic than admitting degrees of knowledge. We use the word know constantly in everyday and scientific/philosophical contexts. Philosophers’ advice to get rid of know and speak entirely in terms of “credence” and “evidence strength” is unlikely to be well received. 7  -II. EXTERNAL WORLD SKEPTICISM- 1) EXTERNAL WORLD SKEPTICISM We said above that external means outside or beyond my mind (my thoughts and consciousness). So the external world is all of reality that is external to my own mind. It includes physical space and all the objects in it. All this is external because my mental “access” to it is indirect, through my senses. My senses make contact with the external world and create the conscious perceptions, and these are what I experience directly. On this view, the external world even includes my own body, which I sense with vision, touch, etc. as well as interoception, proprioception, etc. Can I have any knowledge of the external world? Do I even know that there is an external world? External world skepticism answers: No, none of my regular beliefs about the external world such as “There is a chair” and “Here is my hand” are knowledge. It’s a surprising claim, since these sorts of beliefs seem like some of the clearest cases of knowledge. External world skeptics assume that the JTB conditions are necessary for knowledge (even if they’re not sufficient), and they deny that any beliefs about the external world satisfy the justification condition. So none of these are knowledge. 2) SKEPTICAL HYPOTHESES My sensory beliefs seem to be very well justified. In the next section, we’ll look at an argument that they’re not. The argument makes use of a skeptical hypothesis. A skeptical hypothesis has two key features: 1) If the hypothesis is true, all of my beliefs about the external world are false. 2) I cannot get good evidence that it’s false even if it is false. René Descartes (1641) famously notes reasons to distrust his senses. They are not perfectly reliable. They are subject to all sorts of distortions and illusions. Temporary disruptions (e.g. from psychedelic or anaesthetic drugs) can produce hallucinations and impairments. Since Descartes, we’ve also come to realize that neurological damage can produce profoundly weird false beliefs, such as blind people who sincerely believe they can see (Anton’s syndrome), people who believe their own limbs are not part of them (asomatognosia), etc. And then there’s the ultimate hallucination: dreaming. “I’m dreaming right now” is close to being a skeptical hypothesis. If I’m dreaming then the surroundings and the body that I seem to perceive are not real. So all of my beliefs about my current sensory perceptions would be false. But of course I have other beliefs, such as “Planet Earth exists”, “I have hands”, etc. that would still be true even if I’m dreaming. Moreover perhaps there are ways to determine that the dreaming hypothesis is false. There’s an old idea that I can pinch myself to be sure that I’m not dreaming. Maybe that’s not a good test because it's too brief and mild. But if I break my leg and I’m in agony for hours, and I don’t wake up, this might count as good evidence that I’m not dreaming. Even if “I’m dreaming right now” is not quiet a full skeptical hypothesis, Descartes’ final hypothesis is. He imagines a powerful demon who directly manipulates his mind to create illusory perceptions. 8 The demon could give me any experience at all (breaking my leg and worse), so there is no possible experience I could have that would be strong evidence against the demon hypothesis. A modern, technology-inspired version of the demon hypothesis is the brain-in-a-vat (BIV) hypothesis. Last night a crazy neuroscientist snuck into my house while I was sleeping, anaesthetized me, and removed my brain. They put my brain into a vat, a special container that keeps the brain alive. Tiny wires connect the sensory (input) nerves and the motor (output) nerves to a powerful computer that runs a highly realistic simulation of the external world. This works like a video game. When I got out of bed this morning, my brain’s motor signals went to the computer. It sent back signals from the simulation. My brain processed these as it would real sensory signals from a body (eyes, ears, hands, etc.). This created the completely convincing feeling of standing up, the visual experience of seeing the room around me, etc. If I’m a BIV then most of my beliefs about the external world are false, although some general beliefs such as “Earth exists” are still true. And I could never get strong evidence against the BIV hypothesis even if I’m not a BIV, since the computer could create any experience just as Descartes’ demon could. I could have all the same experiences as a BIV that I could have as a normal human. But although I cannot acquire new evidence against the BIV hypothesis, I do have one important kind of evidence against it: my memories. I remember a world in which a BIV is technologically impossible, and in which no one has any reason to deceive me in this crazy way. But this memory evidence is neutralized by one small modification of the hypothesis, the lifetime brain-in-a-vat (LBIV) hypothesis: I have always been a BIV. I never had a body, there never was a planet Earth, etc. My whole lifetime of experiences was fabricated by the computer (perhaps just five minutes ago). My memories are not evidence against the LBIV hypothesis because I would have all these same memories whether or not I’m an LBIV. So I’m not justified in believing that I’m not an LBIV. This feature of the LBIV hypothesis (or of any skeptical hypothesis) is key for the argument for external world skepticism. 3) THE CLOSURE ARGUMENT a) The Closure Argument The Closure Argument relies on the Closure Principle. This very plausible principle says that I can expand my justified beliefs by noticing that something is a logical consequence of things that I justifiably believe. 9 For example, suppose I’m justified in believing that 1) it’s snowing today, and 2) school is cancelled when it snows. The Closure Principle says that I can form a new justified belief that 3) school is cancelled today, since this is a logical consequence of my other justified beliefs: If I’m justified in believing (1) and (2), then I’m justified in believing (3). Now take any regular belief about the external world such as “Here is a hand”. If that belief is true then I’m not an LBIV. After all, the LBIV hypothesis says I have no real hands. So the Closure Principle gives us this statement: If I’m justified in believing “Here is my hand” then I’m justified in believing “I’m not an LBIV”. But LBIV is a skeptical hypothesis. I can never have evidence that I’m not an LBIV. So I’m not justified in believing “I’m not an LBIV.” From this we get the Closure Argument for external world skepticism. Premise 1: If I’m justified in believing “Here is a hand”, then I’m justified in believing “I’m not an LBIV”. Premise 2: I’m not justified in believing “I’m not an LBIV”. Conclusion: I’m not justified in believing “Here is my hand”. The Closure Argument is valid (logically perfect). And both its premises seem to be true. It seems to prove its conclusion. And the argument may be repeated for any belief about the external world. Assuming that justification is necessary for knowledge, the argument seems to prove external world skepticism. b) Responding to the Closure Argument Most philosophers do not want to be external world skeptics, and many have worked hard to resist the Closure Argument. There are four general types of responses. 1) Reject Premise 1 – Since Premise 1 comes directly from the Closure Principle, rejecting it requires rejecting the principle. But isn’t the Closure Principle obviously true? 2) Reject Premise 2 – To deny this premise, someone argues that they really do have evidence that skeptical hypotheses are false – despite the fact that the LBIV hypothesis and Descartes’ demon hypothesis are designed such that evidence against them is impossible. 10  The simplest example of this response is a famous argument by G. E. Moore (1925). Moore turns the Closure Argument around. Premise 1 is the same as the Closure Argument, but Premise 2 and the Conclusion are different. Premise 1: If I’m justified in believing “Here is a hand”, then I’m justified in believing “I’m not an LBIV”. Premise 2: I’m justified in believing “Here is a hand”! Conclusion: I’m justified in believing “I’m not an LBIV”! Moore’s argument is also valid. And he insists that he is absolutely certain of Premise 2 – more certain than he is of Premise 2 of the Closure Argument. So his argument wins! If Moore is right, I can discover that I’m not an LBIV just by looking at my hand. It’s that easy. This has exasperated philosophers who expect either some amazing new evidence or a clever new theory of justification. 3) Contextualism – Contextualism is a theory of justification that has been applied to the problem of external world skepticism. The contextualist response only partly rejects skepticism. It claims that Premise 2 is sometimes true, sometimes false, depending on the context. That’s because the standard of justification varies. In regular life (outside of a philosophy classroom), I’m not worried about the possibility that I’m an LBIV. In this regular-life context, Premise 2 is false: I am justified in believing “I’m not an LBIV”. That’s because in regular life, it’s just crazy to worry about being in LBIV. But when I step back into the philosophy classroom and think about skepticism, the standard of justification changes. In this philosophical context, Premise 2 is true and the argument is successful. But does it make sense that what I know could change like this depending on what I’m doing? 4) Admit Defeat – If we cannot find a compelling response to the Closure Argument, we may have to admit defeat and be external world skeptics. This could be quite an annoying, even upsetting, conclusion. But philosophers should be ready to follow their reasoning wherever it leads. 11