1 The Threat of Skepticism
1.1. What Is a Theory of Knowledge and Why Do We Need One?
This is a book about the theory of knowledge, also known as epistemology, but we need to begin by explaining what a theory of knowledge is and why we need one.
Other disciplines don’t face the same problem. There’s no problem at all in explaining what a theory of motion is and why we need one. Physical objects move through space, and it would be nice to know if there are some regularities to be found about how they move, something more informative than just the familiar principle that what goes up must come down. As it turns out, there are such regularities. A theory of motion tells us what the laws of motion are which govern moving objects. Why do we need such a theory? One reason might be simple intellectual curiosity. The world is a complicated place, and we might be interested in learning more about how it works. This is a perfectly good reason to want a theory of motion. But there are obvious practical reasons as well. If a stone is thrown into the air, one might have very practical reasons for wanting to know where it will land. If a large meteor is headed toward the earth, there are straightforward practical reasons for wanting to know whether it will collide with the earth or harmlessly pass it by. And so on. As a result, when physicists talk about a theory of motion, no one is terribly puzzled about what such a theory might be or why anyone needs such a theory.
The same seems to be true about scientific theories generally. No one is terribly puzzled about what a theory of disease might be or why we need one. Or a theory of electricity. Or a theory of inheritance. And so on.
But philosophical theories are different, and there seem to be no such obvious answers to the questions about what a theory of
Scientific Epistemology. Hilary Kornblith, Oxford University Press. © Oxford University Press 2021. DOI: 10.1093/oso/9780197609552.003.0001
knowledge might be or why we might need one. One straightforward way to address these issues is to ask what has made philosophers think they needed a theory of knowledge, and what they have thought such a theory might look like. And from the very beginnings of Western philosophy, in ancient Greece, the motivation for developing a theory of knowledge has come from the threat of skepticism.
Skepticism is the view that we know nothing at all. On its face, the view seems to have little to recommend it. We seem to know a great many things, from the utterly trivial, such as who is buried in Grant’s tomb, to the highly consequential, such as whether it is safe to travel by airplane. The reason for taking skepticism seriously is not, therefore, that it seems to be true; it doesn’t seem to be true. From the earliest days of philosophical thought, however, there have been various arguments which seemed to show that knowledge is, in fact, impossible. These arguments present us with a puzzle: they start from premises which seem undeniable, and move, by way of a series of inferences which themselves seem rationally forced upon us, to a conclusion which seems obviously false, the conclusion that we know nothing at all. These arguments present us with a problem: if we really do have lots of knowledge, as common sense would have it, and skepticism is thus false, then there is something wrong with these arguments, and we ought to be able to say what that is. Alternatively, if we cannot find any fault in these arguments, if they really do start with premises which are undeniable and move by way of a series of inferences which are themselves rationally forced upon us, then perhaps we need to take skepticism more seriously than common sense would suggest. And this has turned out to be a difficult problem because it has proven exceptionally difficult to say just where the error lies in various arguments for skepticism.
Skeptical arguments thus, historically, have provided a motivation for theorizing about knowledge.1 We want a theory of knowledge because we want to be able to explain, in the face of skeptical arguments, how knowledge is actually possible. A theory of knowledge would provide us with an account of what knowledge is, and this, in turn, would allow us to respond to the threat of skepticism.
This isn’t, of course, a practical problem. There aren’t skeptics standing around, lurking on every corner, demanding that we explain
how knowledge is possible and threatening us with some concrete harm if we cannot meet the demand. But it is an intellectual problem, and no less legitimate for all that. We all believe that we have a good deal of knowledge, and skeptical arguments seem to show that we are wrong about that. We ought to be able to respond to the challenge which these arguments present. We ought to be able to explain how knowledge is possible in the face of arguments which seem to show that it isn’t. And the first thing we will need, it seems, if we are to successfully respond to any such argument is an account of what knowledge is. That would be a theory of knowledge.
If we are to make any progress here, we will thus need to look at some of these skeptical arguments and see why they present such difficult problems. So let us turn to one of the most powerful and troublesome skeptical arguments, the argument from illusion.
1.2. The Argument from Illusion
Right now I am sitting in a chair in my office. There are some books in front of me, a desk, a telephone, a computer, a coffee cup. All of these things are in plain view; I can see them clearly. I have no doubt whatsoever that I am in my office, rather than, say, at home. I have no doubt that there are books in front of me, and the various other objects which form the tools of my trade: my desk, my telephone, my computer, my coffee cup. In a word, I know that I am in my office; that there are books, and a desk, and a telephone, and a computer, and a coffee cup all within arm’s reach. Nothing could be plainer than this. I can just see these things. To put the point somewhat pedantically, I am currently having certain sensory experiences on the basis of which I believe a host of things about where I am and what is in front of me. If I didn’t have these sensory experiences, if I didn’t, in particular, have the visual experiences I’m currently having, I wouldn’t have these beliefs about my environment. It is my visual experience which provides my basis for believing what I do about my immediate surroundings. More than this, my visual experience provides me with exceptionally good reason to believe as I do. This is not a matter of wishful thinking or hasty generalization or some failure to consider relevant evidence. My visual
experience provides all the evidence I need for believing as I do. It is on the basis of my visual experience that I may be credited with knowledge of the various features of my office. This isn’t anything momentous, but it is knowledge nonetheless.
I could, if I wanted to, get additional evidence for these beliefs. I could, for example, reach out and touch these various objects to further confirm what my visual evidence tells me. In fact, I do, quite regularly, reach out and pick up my coffee cup. But even before I touch my coffee cup, and even though I haven’t touched my telephone or various books, the evidence which my visual experience provides is evidence enough. I have excellent reason to believe as I do about the objects in my office even though I could have additional reasons for such beliefs. And it is because I have such good evidence and, of course, because my beliefs are, in fact, true, that I count as having knowledge about the objects in my office.
So what is the problem? Why does the skeptic think that I don’t know these things? Indeed, why is it that the skeptic claims not only that, in fact, I don’t know these things, but that knowledge of these things, or any other things at all, is impossible? The skeptical argument—in this case, the argument from illusion—is distressingly simple.
Consider my knowledge of my beloved coffee cup. I’ve had this coffee cup for several decades. It’s black, with penguins on it, some of whom are wearing bow ties. I’m looking right at it now and that’s how I know that my coffee cup is in front of me. The skeptic will point out, however, that I could have had the very visual experience I’m having right now—the experience on the basis of which I’ve come to believe that my coffee cup is in front of me—even if my coffee cup were not in front of me. People sometimes have hallucinations which are extraordinarily lifelike. If someone slipped some hallucinogenic drug into my breakfast cereal, I might have a visual hallucination of a coffee cup, a coffee cup just like mine, with well-dressed penguins on it, even when there is no coffee cup in front of me. And we can imagine a situation in which I have such an hallucination, without realizing that I’m hallucinating. Everything might seem perfectly normal to me. It might seem to me that I’m looking at my coffee cup in my office, just the way I do each morning, even though, in fact, I’m not looking at my coffee cup, but hallucinating.
Alternatively, the skeptic might point out, as the seventeenthcentury French philosopher René Descartes did, that all of us have dreams. There is a very common anxiety dream which students often have. I myself had such dreams on more than one occasion when I was a student. I would dream that I was walking across campus, without a care in the world, when it suddenly occurred to me that I was due at a final exam. In one version of the dream, I also realized that I had no idea where the final was being held. In another, I realized that I had not attended even one meeting of the course all semester, nor had I so much as looked at any of the assigned readings. In another version— not that I ever had this version of the dream—one realizes as one is walking across campus that one is not wearing any clothes at all. In each case, panic sets in. And then one wakes up in a sweat.
Why do we experience such anxiety in these dreams? We do so because, when we’re dreaming, we form a variety of beliefs: that we’re walking across campus; that we should be taking a final exam; that we have no idea where the exam is being held; and so on. And we have these beliefs as a result of various experiences we have while dreaming. I might believe, for example, that I’m on campus walking past the library because I have certain visual experiences while dreaming, the very experiences I would have if I were on campus walking past the library. Our dreams can be a source of various emotions which we vividly continue to experience when we first awaken precisely because those experiences are so lifelike, that is, because, in dreams, we have experiences, like my visual experience as of walking past the library, that are exactly like experiences we have when awake.
The recognition that our sensory experience in dreams can be exactly like the sensory experience we have when awake led Descartes to ask himself what reason he has to believe, at any particular time, that he is in fact awake. So let us go back to my coffee cup. I’m in my office, as I said, looking right at the cup. And I believe, of course, that I’m looking at my cup because I can see it; that is, I believe the cup is right in front of me on the basis of the visual experience I currently have. But we now recognize that I could have experiences exactly like this in a dream. I could dream that I’m sitting in my office looking at my coffee cup. And if I did have such a dream, my visual experience would be exactly like the experience I’m having right now. And this raises the
question: What reason do I have, given this fact, to believe that I am in fact awake now, in my office, looking at my coffee cup rather than asleep, at home, in my bed, with my eyes closed, and no coffee cup anywhere near me?
We’ve already said that I believe that my coffee cup is in front of me on the basis of my visual experience. But I could have an experience exactly like the one I’m having now even if I weren’t looking at the cup and even if there were no cup present. I could have this very experience if I were asleep and dreaming. So, it seems, I have no more reason, at least on the basis of my experience, to believe that I’m awake rather than dreaming, or to believe that there’s a coffee cup here rather than that there isn’t. Nothing in my experience could possibly settle this issue, because for any experience one might pick, and for any feature of my experience that might serve as a sign that I am genuinely awake, I could have an experience exactly like that while dreaming.
Thus, consider the famous pinch test. How do I know I’m not dreaming now? I pinch myself, and I don’t awaken; everything remains as it was before, and it does so because I was wide awake the whole time. Now this is, of course, a ridiculous test. It doesn’t help at all. I could dream that I pinched myself and dream that I passed the test. But what applies to the pinch test applies to any would-be test to distinguish dreaming from waking life: I could dream that I applied the test and dream that I passed it. And this seems to show, as Descartes suggests, that I can’t know right now that I’m actually awake rather than asleep and dreaming. And if I can’t know that I’m awake rather than asleep and dreaming, then I can’t know that I’m currently looking at a coffee cup, and so I can’t know that there’s a coffee cup in front of me. And if I can’t even know that there’s a coffee cup in front of me when I’m looking right at it, then I can’t know anything at all about the world around me. There’s nothing special about my knowledge of the coffee cup. Indeed, we began with this example because it seems so unproblematic. If it turns out that I can’t even know this, then there is nothing at all I can know. Skepticism is forced upon us.
Descartes raises another skeptical worry, one even more fanciful. He asks us to imagine that there is an evil demon who is all-powerful and who wishes to deceive me about absolutely everything. The demon gives me visual experiences as of looking at my coffee cup when, in fact,
there is no coffee cup in front of me. He gives me visual experiences as of being in my office when, in fact, I am not in my office; in fact, I have no office. He gives me apparent memories as of growing up in New York when, in fact, I didn’t grow up in New York; in fact, there is no New York. And so on. And Descartes then asks what reason we could possibly have for thinking that there is no such demon. Just as in the dream case, nothing in our experience could possibly rule this out. For any experience we might choose to distinguish veridical experience—experience which accurately reflects the world around us—from illusory experience, the demon, it seems, could give us experience like that in order to mislead us. And that surely shows, Descartes suggested, that we are not in a position to know that our current experience is veridical and that we are not being deceived by a demon at this very moment, and that, in turn, undermines the basis for everything we believe about the world around us. We are thus led, inevitably, to total skepticism.
A more contemporary version of Descartes’s evil demon involves a mad neurophysiologist. The neurophysiologist has you kidnapped and brought to his laboratory. Your brain is removed from your skull and put in a vat of nutrients to keep it functioning, and it is then provided with electrical stimulation exactly like the stimulation that would be supplied, for example, to the optic nerve, and to other sources of sensory input, were you, say, on campus, walking past the library. A whole simulated life is then provided for you. (Fans of The Matrix will have no problem imagining this sort of thing.) Once again, we see that there is nothing we can point to in our experience that could possibly show that we are currently embodied and in touch with the world around us rather than envatted in the mad neurophysiologist’s laboratory. And, once again, this undermines the basis for everything we believe about the world around us. And, once again, we are inevitably led to total skepticism.
These nightmare scenarios, as I will call them, all present situations in which even our most confidently held beliefs are false. I’m looking right at my coffee cup now. I know that there is a coffee cup in front of me as surely as I know anything. But in each of these nightmare scenarios—the hallucination case, the dreaming case, the evil demon, the dastardly neurophysiologist—I have the very same visual
experience I’m having right now, the same experience on the basis of which I’ve come to believe that there is a coffee cup in front of me, and yet my belief is false. What this suggests is that my current visual experience does not provide an adequate basis for believing that there is a coffee cup in front of me. Does that mean that I should hold off on believing that there is a cup in front of me until I can get additional evidence? Not at all. In each of these nightmare scenarios, there is simply no possibility of getting the kind of evidence that would allow me to determine whether I’m actually looking at the coffee cup right now or, instead, hallucinating, or dreaming, or being deceived by a demon. For all my experience shows, I could be a disembodied brain floating in a vat of nutrients, subjected to a steady stream of electrical stimulation which presents a thoroughly inaccurate picture of the world outside my vat. So there’s no further evidence I can get which would allow me to determine whether there really is a cup in front of me. And what that seems to show is not only that I don’t currently know whether there is a coffee cup in front of me, but that I couldn’t possibly know such a thing. And this is not, of course, something peculiar to my belief about the coffee cup. Rather, these nightmare scenarios seem to demonstrate that I have no knowledge at all, and that it is absolutely impossible for me to gain any knowledge at all. That is the skeptical conclusion.
One last point about these arguments. You might think that perhaps the skeptical conclusion isn’t so bad. If these arguments show what they seem to show, that we have no knowledge at all, and that knowledge is impossible to achieve, I should stop claiming to know things, and I should stop believing that I know things. That would certainly require me to change what I say and what I think. But maybe this wouldn’t be as big a change as it seems. I should stop talking about knowledge, but life could go on as before. In particular, I still have just as much reason to behave as I do right now. I just have to get used to giving up all claims to knowledge.
But the skeptical arguments, if they work at all, show far more than this. Consider the following case. I’m standing on the corner of University Avenue intending to cross the street. Just as I’m about to set foot in the street, I see a large eighteen-wheel truck barreling straight toward me. Before thinking about the skeptical arguments, I would have said that I know there is a large truck headed straight toward me,
and I know that it would be dangerous, even foolhardy, to cross the street. But I’m wiser now, and I realize that I shouldn’t claim to know such things. I lack all knowledge of such things; knowledge is impossible. But I still, one might think, have reason not to cross the street. Even if I can’t know that there is a truck headed toward me, surely I have good reason to believe that the truck will run me down if I walk into the street. And that’s all I need to stay on the sidewalk. Life without knowledge goes on because my actions can be governed by reasonable belief even if knowledge is impossible. And reasonable belief is all it takes to keep me out from under the wheels of the oncoming truck.
But this attempt to endorse the skeptical argument while limiting the scope of its consequences cannot possibly work. What the skeptical argument shows, if it succeeds at all, is not just that knowledge is impossible. It shows that we have no more reason to believe that there is a truck headed down the street than that we are hallucinating that there is such a truck. We have no more reason to believe that there is a truck barreling down the street than that we are merely dreaming that there is such a truck. We have no more reason to believe that there is a truck about to crush us than that we are deceived by an evil demon, or a mad neurophysiologist, into thinking that there is such a truck when, in fact, there really isn’t. And what all of this would mean, of course, is that we have no more reason to stay on the sidewalk than to walk boldly out into the street.
The skeptical conclusion is thus no small matter. It cannot be accommodated by some little change in our lives in which we stop talking about knowledge or thinking that we know things, and yet go on just as we did before. If the skeptical conclusion is true, not only do we know nothing at all, not only is knowledge impossible, but we have no reason whatsoever to do any of the things we do rather than anything else. We have no more reason to stay out of the way of oncoming trucks than we do to walk right in front of them. And that, if it were true, would matter a great deal.
I’m not suggesting for a moment that skepticism is true. I don’t believe it. I certainly hope you don’t believe it. Even Descartes didn’t believe it. He thought he had a solution to the skeptical problem, an adequate reply to the skeptical argument. But I hope it is now clear why there is a problem about skepticism. We have an argument before
us which seems to show that knowledge is impossible. And it isn’t remotely clear what is wrong with this argument. It would be nice to be able to say just where the argument goes wrong. What that would take is a theory of knowledge, an account of what knowledge is which would show us how knowledge is genuinely possible. And that is one reason why philosophers engage in epistemological theorizing.
1.3. Convincing the Skeptic
What would it take to have an adequate reply to the skeptical argument? One very natural thought here is this. We imagine being confronted by someone who is convinced by the skeptical argument. Let us call this person the Skeptic. The Skeptic lays out the various nightmare scenarios for us and concludes that he has no knowledge whatsoever, and neither does anyone else. Knowledge, he tells us, is impossible. And now we try to devise an argument to convince the Skeptic that he is wrong: not only is knowledge possible, but we all do know many, many things. More than this, we cannot convince the Skeptic to change his mind by way of threats, or financial incentives, or any other inducements. We need to offer the Skeptic reasons to believe that skepticism is false. We succeed in responding to skepticism when we have devised a good argument which convinces the Skeptic that knowledge is possible and we do know many things. This is what an adequate response to skepticism requires, or so it might seem.
In order to see how such an argument might go, it will be useful to begin by considering how we actually convince others with far less radical views than the Skeptic. After all, we regularly encounter people who disagree with us about matters large and small, and such disagreement often prompts a kind of rational engagement: we each try to convince the other that we are right and they are wrong. Sometimes we succeed, and they are convinced by us; sometimes they succeed, and we are convinced by them; and sometimes there is a standoff: neither party to the dispute is convinced to change their mind.
Let us take an example of such a dispute. Suppose I meet someone who is a climate change skeptic: this person does not believe that there is human-caused global warming, while I do. I think this is an
important issue, and I attempt to change this person’s mind. I don’t offer threats or incentives to believe as I do; I offer rational arguments.
Perhaps I begin by summarizing the scientific consensus: the overwhelming majority of climate scientists, I point out, believes that there is global warming and that it is caused by human activity. These experts know far more about this matter than I do or, as I suggest, than my opponent does, and in the face of such expertise, we should believe what the experts believe.
Now it might be nice—indeed, surely it would be nice—to explain what the reasons are on the basis of which the experts have come to this opinion. Nevertheless, it does seem that the argument I offer is a perfectly good reason to believe that global warming is indeed a real phenomenon. Understanding why global warming is occurring requires more than this, but this is sufficient reason to believe that global warming is actually occurring.
Nevertheless, this might not convince my opponent, and for perfectly good reason. I am happy to defer to the scientific consensus because I believe that climate scientists are better informed about the matter than I am, or, for that matter, than my opponent is. But my opponent might disagree with me about this. My opponent may know perfectly well what the scientific consensus is, but believe that climate scientists are just mistaken. Indeed, my opponent may have an explanation for why so many scientists might have all gone wrong: they are part of a liberal elite, my opponent claims, and they are driven by ideology rather than by the facts.
Now I may believe—in fact, I do believe—that this view of climate scientists is completely mistaken. But if I’m trying to convince someone who believes in such a vast liberal conspiracy that global warming is genuine, it will do me no good to take as one of my premises something which my opponent doesn’t believe. If I were entitled to do that, then I might as well have simply stated my conclusion, that global warming really is occurring, and let it go at that. But this would obviously give my opponent no reason at all to believe as I do.
We can draw a general moral from this: if my argument is to be dialectically effective against an opponent, that is, if my argument is to give my opponent good reason for changing their view, then I cannot simply assume something which my opponent rejects. And that means
that I cannot take as a premise in my argument a claim, such as the claim that we should defer to the scientific consensus, which my opponent does not accept. What we need to do, if we are to rationally convince an opponent, is to start with things that person already believes, and show them that these beliefs which we share rationally force them to change their mind about the issue which divides us. Our argument must therefore begin by getting a foothold within their body of beliefs. Without that, we offer our opponent no reason at all to change their mind.
But it takes more than premises to make an argument. Suppose I point out that the New York Times has reported something about the global warming issue, and I immediately conclude from this that matters are just as the Times has reported. Perhaps this is just an obvious conclusion to draw, as I see it: the Times says it; that settles it. Now my opponent may completely agree with me about what the Times has reported; I am then entitled to take that as a premise in my argument. But I have directly inferred something from this—that the facts are as the Times reports—without the benefit of further premises. I have treated this as a legitimate inference. Here too, my argument will not be dialectically effective—it will not rationally convince my opponent—if my opponent does not already see this as a legitimate inference. If my opponent rejects this inference, that does not mean that I cannot convince them. I may produce additional premises, premises which my opponent accepts, and make use of transitions which my opponent sees as legitimate, to show that the Times should be deferred to. But absent such additional argument, if my opponent rejects my inference, I have not produced the kind of argument which is needed to produce a rational change in view.
The kind of argument I need, then, to rationally change someone’s mind must have premises which my opponent already accepts, and move by a series of inferences which my opponent takes to be legitimate. It is only in this way that we can show someone, given what they are already committed to, that they need to change their mind about some issue. This is what a dialectically effective argument requires.
Let us now leave the global warming skeptic behind and return to the Skeptic, that is, the person who claims that knowledge is impossible. In order to rationally convince this person that knowledge is,
in fact, possible, and that we do know many things, we must provide an argument which starts with premises which the Skeptic already accepts. But what premises are these? The Skeptic insists that he knows nothing at all. So he doesn’t accept any premises. Any attempt to provide premises from which our argument might proceed will automatically be dialectically ineffective against the Skeptic because we cannot get a foothold within the Skeptic’s body of beliefs. And there is a very simple reason for this: the Skeptic does not have a body of beliefs; the Skeptic doesn’t believe anything at all.2
Not only does the Skeptic have no beliefs; the Skeptic also does not accept any inferences as legitimate. To accept that an inference is legitimate would be to acknowledge that a certain transition in thought is sufficient to transmit knowledge. But the Skeptic, being a skeptic, doesn’t accept that there are any such legitimate transitions.
So, to summarize: in order to rationally convince the Skeptic that knowledge is, indeed, possible, and that we do have a great deal of knowledge, we need to construct an argument which leads to that conclusion starting from premises the Skeptic already accepts and proceed by way of a series of inferential transitions which the Skeptic acknowledges as legitimate. But the Skeptic has no beliefs, and so there are no premises which the Skeptic already accepts, and the Skeptic does not accept any inferential transitions as legitimate. So we need to provide an argument for the conclusion that knowledge is possible which has no premises and involves no inferences. And, of course, there are no such arguments. So it is clearly impossible to come up with a dialectically effective argument to change the Skeptic’s mind. We cannot rationally convince the Skeptic.
Does this mean that the Skeptic is right? Certainly not. It does mean, however, that we should stop trying to come up with arguments that will rationally convince the Skeptic to give up his skepticism. That exercise is a losing game. It has been rigged—by denying us all of the tools of rational argumentation—in such a way that we cannot possibly win . The fact that we are deprived of all means of rationally convincing the Skeptic shows only that we should not play this game. It doesn’t show anything at all about the possibility of knowledge. If someone asked us to show how it is possible to earn money without either working or investing, we might well be at a loss. Those
