We have looked at Descartes' skeptical questions, through which he tried to establish a new form of science, and then at Hume's skeptical questions, through which he tried to reform our understanding of science. These two thinkers are representatives of the two great schools of early modern philosophy: rationalism and empiricism. These two groups of thinkers held extremely different views on the nature of knowledge, as well as on its extent and limitations.
As the period progressed, over the course of roughly two centuries, the two schools diverged further and further apart from each other, until they appeared to be fundamentally incompatible, with no common ground from which a dialogue could even begin.
Kant came at the end of the 18th century and brought the period of early modern philosophy to an end. He is universally considered one of the greatest philosophers in the history of the discipline. He and Aristotle are probably the two greatest. He touched on every branch of philosophy, and in every one of them he brought about fundamental revolutions.
He is most famous for his first major book, and major is the right word, as it is an 800-page tome: the Critique of Pure Reason, published in 1781. It was not widely read at the time, and it was frequently misunderstood, due in large part to its style. Kant was a terrible writer. His prose has been described as dry, obscure, opposed to all ordinary notions, and long-winded, and that description was written by Immanuel Kant himself, about his own book.
In response, Kant wrote the Prolegomena to Any Future Metaphysics as a shorter, more accessible presentation of his philosophy. Accessible is a relative term: it is still not easy reading, but it is considerably shorter, and it cuts directly to the essential ideas. The Critique of Pure Reason is enormous, filled with intricate details, nooks and crannies, and back alleys of argument. The Prolegomena, by contrast, goes straight to the main point, and that is what we will be focusing on in this lecture.
We are not going to chase down all the details, because there are simply too many of them. We are trying to see the world that Kant saw, to get the insight that drove his philosophy, rather than all the minutiae of it.
As with Hume, there is one big argument here, one big idea, and if you get that, you have got the book. The first part of the Critique is preparation, leading up to it; then comes the big idea itself; then Kant applies it in three different ways. So if you grasp that central idea, you are in excellent shape. As with Hume, however, it is a difficult, complicated, and counterintuitive idea, though it has proven enormously influential.
Kant was trained as a rationalist, as someone who relied on a priori reason to understand the structure of reality and to pursue metaphysics. He was committed to that program until he read David Hume. He describes this as the cataclysmic event of his intellectual life, one of the signal moments in the history of philosophy, which, in his famous phrase from the introduction to this book, woke him from his dogmatic slumbers.
By dogmatic, Kant means that he had been trusting his reason to accomplish these feats without question. He had been assuming that he could simply sit in a chair, the way Descartes did in that snowy cabin in November of 1619, turn inward, reflect on ideas, and through the process of examining their internal logic, arrive at facts about reality. The underlying premise is that the way the inside of the mind works is parallel to the way the outside world works, and that by studying the one, you can learn about the other. That is precisely how a priori metaphysics works: you do not use your senses to look around, because you assume that what you find on the inside reflects what is happening on the outside.
The key word there is assume. Kant came to believe that this entire project is dogmatic for exactly that reason: metaphysicians were assuming that the logic of the mind works in precisely the same way as the metaphysical structure of the world, and that studying the one therefore teaches you about the other. But no one was proving it. No one was establishing it, because doing so is extraordinarily difficult.
It was Hume who shocked Kant into asking this question: How do I know that the way my ideas fit together is the same way that events and objects fit together in the world? This assumption is necessary for a priori metaphysics to make any sense. Without it, a priori thinking tells us nothing about the world, and that is exactly what Hume argued.
Hume maintained that a priori thinking yields only what he called relations of ideas, that is, the way ideas relate to one another, and that this tells us nothing about the world outside of us. We can always be, as it were, the turkey who does not know that Thanksgiving is coming. The turkey assumes its trust in the farmer is fully justified, unaware that something else is going on in reality beyond its understanding. For Hume, that may well be our situation, and therefore we cannot simply rely on a priori thinking and trust that it will reveal the world to us.
For Hume, the most important relation is causality, the connection of certain events as causes with other events as effects. His central claim was that we cannot establish this relation a priori. No internal a priori logic can tell us what will make what happen in the actual world. As far as our ideas alone are concerned, we can imagine anything causing anything: we can imagine snapping one's fingers and the sun going out, and that scenario is no more or less logical than a billiard ball striking a second one and setting it in motion.
The only reason we accept the billiard ball case as real and dismiss the other is that we have seen the former happen repeatedly. Even then, Hume argues, the connection we forge between cause and effect, between the first ball striking the second and the second moving, is not made by reason at all. It is made by instinct, by what he calls custom or habit, a reflex with no rational basis whatsoever. It is simply our ingrained tendency to react in a certain way to certain stimuli. Kant found this argument brilliant, and it shook his entire way of thinking.
Kant argued that the question Hume raised did not go far enough. Hume focused on one connection, the connection between cause and effect. But what about all the other connections we make? For Kant, the fundamental nature of knowledge is the making of connections. When you know something, what you are doing is making a judgment, and a judgment is essentially the connection of a subject and a predicate.
In Hume's case, causality involves connecting a cause and an effect, connecting a person's foot to a footprint, or a ball striking a second ball to that second ball moving. This is the forging of a connection between events. But to know anything is to connect some predicate, some quality, to a subject. If I say "My dog Ned is brown," I take the subject, Ned, and the predicate, brownness, and connect them. Ned now possesses the quality, or, in classical terminology, the accident, of brownness, and that constitutes a piece of knowledge.
Kant's question, then, is how we make any of these connections, not just causality, but the full range of connections that form the entirety of knowledge. Mathematics and science are simply vast systems of such connections, whole networks of predicates joined to subjects. Kant takes Hume's question and generalizes it as far as possible: how do any of these connections take place at all?
Some of the connections we make are physical, and some are metaphysical. We saw Descartes take God as a subject and then attach a series of predicates: omniscience, omnipotence, omnibenevolence, and existence. God has omniscience, God has omnipotence, God has omnibenevolence, God has existence, and that is how he proved the existence of God. These were claims to knowledge that shared the same logical structure, but what made them metaphysical for both Kant and Hume is that they transcended all possible experience.
Consider my dog Ned. When I see Ned, I perceive the subject and the predicate of existence combined right there before me. I see brownness combined with my dog; I see existence combined with my dog. The subject and predicate are together in the world, and I perceive their togetherness. That is how I perceive a fact. But I do not perceive the existing God. I do not perceive the omnipotent, omnibenevolent, omniscient God. The connection Descartes made was therefore entirely a priori, which raises the question: how could he possibly know it?
The basis Descartes used was simply the ideas in his own head. The idea of God is the idea of a being with all possible perfections; existence is a perfection; therefore these ideas stand in this relationship to one another. But that reasoning makes the dogmatic assumption that the way ideas relate to each other inside my head must reflect the way things outside my head actually operate. If my ideas fit together in a certain way, then actual reality fits together in that same way.
That is the fundamental dogmatic assumption from which Hume woke Kant. Hume demanded that we prove it, and he concluded that we cannot. You can find relations of ideas, but that tells you only about the ideas themselves. You can perceive matters of fact, and then you know those matters of fact are real, but only because you have perceived them, and that is all. Relations of ideas and perceived matters of fact: those are the entirety of knowledge, and nothing else.
Kant thinks that Hume's question is the right question: how can we know that our knowledge, held in our minds, reflects the way the world actually is? But he believes Hume's answer was wrong. Hume gave up too much, leaving us in too difficult a position. The conclusion Hume was content to rest in, that we simply react the way we do because that is how our brains are built, was not good enough for Kant.
Kant was a figure of the Enlightenment, and for him, we must have a reason to believe something, or we do not have the right to believe it. Without a rational justification for a belief, we lack the epistemological right to hold it. That is what it means to be a rational being: to believe only what you can argue for, only what you can justify.
To simply react to the perceived constant conjunction of events, the billiard ball strikes, the second one moves, again and again, and to react as automatically as a leg jerking when struck by a doctor's hammer, is to behave as an animal. And that is precisely what Hume concluded: that we are animals, nothing more. Kant refuses this. Even amid all our philosophy, he insists, we must be human beings, not animals. We must have reasons for our beliefs.
So Kant takes up Hume's challenge and Hume's question, but he believes that in answering it, Hume simply surrendered. Kant wants an actual solution, one that gives us a rational justification for believing that our ideas and thoughts genuinely work in concert with the world around us. He will not retreat into the dogmatism of the rationalists, but neither will he abandon the possibility of knowledge. He will have an argument.
To understand what is being given up, we have to move into preparation for the argument. It is technical and somewhat tedious, but we need to work through it in order to understand the way that Kant phrases the question. In preparing for the big argument, Kant examines our judgments. All knowledge consists of judgments, and all judgments are connections, and he wants to understand the different types. To understand the nature of knowledge, we have to understand the different kinds of connections that are possible.
He lays out two pairs of contrasting terms. The first is one we are quite familiar with: a priori and empirical. Being Kant, he cannot simply say "empirical" — he uses the fancier term a posteriori. You can see the logic: "posterior" means "after," as in after we have consulted our senses, as opposed to "prior," meaning before we have consulted our senses. It is not strictly a chronological distinction, but that is roughly the analogy being used.
A posteriori is simply a synonym for empirical, just another way of saying the same thing. We will generally use the word empirical, but Kant frequently uses a posteriori, so it is worth knowing both. This first contrast is one we are quite familiar with from Descartes and from Hume. Kant, however, adds a new contrasting pair.
Kant introduces the distinction between analytic and synthetic judgments. To explain it, consider two questions. The first: how many bachelors are there in New York City? You cannot answer that a priori. You cannot simply think about the nature of bachelors or the meaning of the term. The only way to answer it is empirically, you would have to conduct surveys, consult sociological studies, and examine marriage records. There is no logical inference that can get you there.
The second question is: how many married bachelors are there in New York City? That sounds like the same kind of question, but it is an entirely different kind. You can answer it right now: none. There cannot be any, because the very definition of a bachelor is an unmarried man. A married bachelor is a contradiction in terms and therefore cannot exist. You can know this a priori, without consulting any empirical information whatsoever.
This is not a matter of married bachelors being extremely rare or difficult to find. They are impossible, because the concept itself is self-contradictory, like a four-sided triangle. If someone claimed to be a married bachelor, you would not marvel at the discovery; you would conclude that the person had misunderstood the words. The impossibility follows entirely from the meaning of the terms involved.
Certain statements, then, are answerable a priori precisely because of the meanings of their constituent words. Consider the statement "all bachelors are unmarried." Here, "bachelors" is the subject and "unmarried" is the predicate. But notice that we are not really connecting two independent ideas: "unmarried" is already contained within the subject. If we spelled it out fully, the statement would read: "all unmarried men are unmarried." All we are doing is extracting what was already there and making it explicit. This is what Kant calls an analytic judgment.
When we take a term like "bachelor," we are analyzing it, pulling it apart to see what is already contained within it and simply spelling that out. You can know this completely a priori, as long as you know the meaning of the words, because it does not add any new information. My son, when he was younger, once came up to me and asked, "Daddy, what color is Clifford the Big Red Dog?" I thought to myself that we were going to be paying for that kid's housing for quite some time, because the answer is right there in the name. Assuming he is correctly titled, he is big, he is red, and he is a dog. You know all of that simply from his name, so those terms are already contained within it; you learn nothing new. It only clarifies.
By contrast, consider my dog Ned. If I tell you Ned is brown, brownness is not part of the definition of Ned. Dogs can come in all kinds of colors. Bachelors, however, cannot have all kinds of different marital statuses. The definition of bachelor is to be unmarried, so a bachelor can only be one way, and we know this with certainty because any other answer produces a contradiction. If we said "some bachelors are married," we would really be saying "some unmarried men are married," which is a contradiction: we would be asserting that they are both unmarried and married simultaneously. We know this simply by the laws of logic, without ever consulting the world.
You cannot know what color my dog Ned is without looking at him or being told, because he could be brown, black, white, spotted, or any number of other colors; there are many possible options. With bachelor, there is no other option at all.
This is what Kant calls synthetic judgment, because you are taking something that is not contained in the subject and synthesizing it, combining it, bringing the two together. Ned is synthesized with brownness, and now this subject has a predicate that does not belong to him intrinsically. Analytic judgments, by contrast, are a priori: all you need to know is the meaning of the words. Synthetic judgments are empirical; you have to check the world.
Analytic judgments are also necessary, absolutely necessary, because no other option is possible; any alternative is a contradiction. Synthetic judgments are contingent: things could be many different ways, which is precisely why you have to go out and look. With these two pairs established, Kant sets up a table.
These four qualities, a priori, empirical, analytic, and synthetic, form two mutually exclusive and exhaustive pairs. That gives us four possible types of judgment depending on how we combine them.
Are there analytic a priori judgments? Yes, of course. The classic example: how many bachelors are married? Zero. It is both analytic and a priori. We have already seen this kind of judgment: think back to Hume. He called these relations of ideas. They are a priori and uninformative, telling us only how ideas relate to one another rather than anything about the world.
Can you have analytic empirical judgments? Not really. Analytic judgments do not require checking the world or using the senses; they depend entirely on the meanings of words. There are no judgments of this type.
Synthetic empirical judgments, on the other hand, are everywhere. "My dog is brown," "this chalk is red," "I am standing here" — the vast majority of our beliefs fall into this category. These are essentially what Hume called matters of fact, and they were his primary concern. Indeed, Hume argued that all legitimate forms of knowledge are either matters of fact, known through experience, or relations of ideas, known by analyzing concepts. Nothing else, he insisted, qualifies as genuine knowledge.
Kant identifies three types of judgment: a priori analytic, synthetic empirical, and synthetic a priori. That last category should be surprising. Hume built his entire philosophy on denying that such judgments exist. This is precisely what metaphysics claims to do: make substantial claims, such as that God exists, without any empirical information, without seeing, hearing, smelling, touching, or tasting the thing in question. That is the nature of metaphysics, and it is what Hume tried to show was impossible.
Here is where Kant's disagreement with Hume begins. Kant argues that there are, in fact, some synthetic a priori judgments. Metaphysics does claim to operate in this mode, and Kant's central question in the Critique is whether that claim is legitimate. But he also believes we already possess certain legitimate synthetic a priori judgments. The question, therefore, is not whether they are possible; it is how they are possible.
At this point in the argument, one should be a Humean. Hume's view is commonsensical: some beliefs are formed by analyzing concepts and seeing how they fit together, while others are formed by going out and having experiences. That is it. There is no way to know how the world is without looking at it, a clear, obvious, and intuitive position. But Kant says it is wrong.
Kant argues that there are ways to know substantial, informative facts about the world that do not rely on experience. Rather than dogmatically assuming that the way we think a priori simply mirrors the way the world works, he intends to prove it. He will show why it is legitimate to believe that in certain limited cases, we can examine the workings of our own mind and thereby know that the world must conform to those same workings.
This point is crucial. Hume showed us that everything empirical is contingent. No matter how much data we have, no matter how long the string of experiences, no matter how stable it has been that a certain event has happened over and over again, we can never be sure it will continue. Anything based on empirical experience is uncertain. Anything based on looking around us is dependent on what the world shows us, and we don't know what it will show us next.
The authority for empirical claims is the world, and we don't control that. So if the world shows us something surprising, we have to go with it. If the farmer comes out and grabs us by the neck, that's the new reality, that's the new normal. For Kant, if we cannot know whether eating glass is harmful or eating bread is nutritious, then we are lost. This is not merely a failure to be metaphysical, as Hume suggested; this is epistemological nihilism. It means there is no point in asking questions at all, no point in thinking. It is not a limitation; it means we have nothing.
Tom Waits, a great musician and lyricist, has one of his finest albums titled Rain Dogs. I once read an interview in which he explained what the title means. He said that dogs spend so much time marking their territory because those scent markers allow them to know their way around. It is a way of making the world familiar, making the world make sense: knowing where things are, being oriented, having a home where you belong. Then, every once in a while, it rains, and it is all gone. The world goes blank, as if every street sign has been taken down, and the dog is left in a featureless world that smells of nothing.
That, Kant believes, is precisely our situation if Hume is right. We are simply rain dogs, with no way to know anything. And if that is the case, why are we doing any of this? Why are we thinking? Why are we conducting experiments? It becomes entirely pointless. Either it is possible for us to know something, Kant insists, or we should give it all up.
The central question of the book is: how are a priori synthetic judgments possible? We know we have them — we have two entire bodies of them in math and science. Kant holds that math and science consist of a priori synthetic judgments because they are not merely contingent. We know that tomorrow the same causal links will continue to hold, that nature will be stable and reliable. That is the very nature of science.
Science does not simply say, "So far, things have been this way." That is not a scientific law — that is what Hume thinks is all we can ever say, but if that is all science amounts to, then it is not science at all. It is merely a collection of rules of thumb, a kind of history: things have happened this way up to now. For something to count as science, it must give us laws.
Laws have two crucial properties: they are necessary and universal. Science is never about one particular thing or two particular things. The lessons of Pavlov's dogs were not about that specific group of canines — they were about animal minds, about dogs in general. Scientific laws are universal: they apply to all instances of something, forward in time, backward throughout all time, and across all places. Without that, we do not have science.
And for Kant, losing science is not simply a matter of losing people in white coats mixing beakers together — it means losing everything. We would not know how to build a building, how to cook a meal, how to speak, or even whether the people around us are human. Everything rides on a priori synthetic judgments. It sounds technical; it sounds dull. It is not.
Hume has left us in a position where a priori synthetic judgments seem fundamentally impossible. But that is actually a good thing. The fact that they seem so impossible makes them a fantastic clue for critical philosophy. Kant coined the term critical philosophy to describe his project of understanding the nature and structure of the mind — how we think, how we reason — in order to understand its limitations. Drawing that boundary line tells us both what falls inside it, what we can legitimately think, and what falls outside it, what we cannot.
It is very hard to figure out how our minds work, but if we have this one thing that is bizarre, that seems ridiculous and weird, then whatever the mind is like, it must be capable of doing this thing. Any model of the mind we construct carries this requirement: it must be able to produce a priori synthetic judgments. In that sense, the very strangeness of these judgments is enormously helpful. On one hand, they seem impossible; on the other, they give us precise guidance, because we know exactly what we need to explain.
We have to arrive at an account of how the mind works such that it can produce a priori synthetic judgments — and that is precisely what Kant is going to give us. In order to explain it, however, I first have to tell you a riddle.