We can now begin the big argument — the core and center of the book. It is a difficult argument, but extraordinary in its conclusion. It has come to be called the problem of induction. Hume doesn't use that term himself, but you may hear it referred to as such.
The question he is asking is: how do we reason about matters of fact? Reasoning about relations of ideas is simple — there is nothing else it could be. Only one answer is possible, anything else is a contradiction, which is why such reasoning is both necessary and uninformative. But matters of fact are far more important and far more numerous. How do we reason about those?
There are some matters of fact we have an immediate impression of — things we can directly perceive. How do I know my shirt is blue? Simply because I see it. That is uninteresting and obvious. But that accounts for only a tiny fraction of the beliefs I hold. I have an enormous number of beliefs about matters of fact that I have no immediate perception of — things outside this room, outside the empirical data available to me at the present moment. How do I know those things? How do I reason toward them? How do I hold justified beliefs about them? That is the question Hume is now asking.
He will work through a series of steps to arrive at his answer, and along the way he will make two remarkable claims — one surprising, and one genuinely shocking. We will need to examine how those claims work, whether we can accept them, and whether any objection can be raised against them.
Hume's first question is: how do we know matters of fact that we do not have immediate, direct impressions to tell us about? His answer is through cause and effect — a causal relationship. The example he gives is drawn from Robinson Crusoe, a novel very popular at the time, about a man stranded on a deserted island who believes he is alone. Then one day he realizes there is someone else on the island — not because he sees another person, which would be a direct impression, but because he sees a footprint on the beach in a place he had never been.
Hume asks: what is the reasoning that takes place here? Crusoe has no direct perception of another person. The situation is contingent — there could be another person, or there might not be — so this is a matter of fact. The footprint is an effect that must have been caused by a human foot. Crusoe takes that effect, which is an immediate impression he has direct experience of, and uses the causal relationship to move beyond his immediate experience.
He has no need to reason that there is a footprint — he simply sees it. What he does need to reason toward is what he does not see: the other person. The way he moves from what he does see to what he does not see is through a causal relationship. Something must have caused this footprint as its effect, and the only thing that could account for it is a human foot — therefore, another person exists on the island.
Cause and effect is thus the means by which we build up our knowledge of matters of fact, extending our understanding of the world beyond what we can immediately hear, smell, touch, or taste. This, Hume argues, is precisely what science is. Science is the study of what causes what; its laws are simply laws of regular causal relationships. Science is fundamentally about bringing phenomena into causal relationships, and so if you want to understand science — as Descartes did — you must first understand causality. And for Hume, of course, causality is contingent and apparent.
We now have an answer, but it is not a complete answer, because we can ask the same question again. How do we know this? How do we reason about cause and effect? How do we link causes and effects together in an intelligible, rational, justified way? That is Hume's new question. He has his first question and his first answer, and now he has a second question about that answer — and here we arrive at the first surprising claim.
Imagine you are in a pool hall and you see one billiard ball rolling toward another and striking it. What do you expect the second ball to do? You expect it to move. The first one hits it, and the second one travels off in some direction. Hume asks, "How did you know that? Where did that knowledge come from?"
What it feels like, of course, is obvious — what else could it do? If the first ball hits the second ball, the second ball is going to take off. This seems logical, something we can simply infer. It feels as though we can logically deduce what will happen when one physical object strikes another. And Hume says we cannot.
The thought experiment he runs to illustrate this involves bringing Adam — from Adam and Eve — into the pool hall. We bring Adam in, the first billiard ball rolls toward the second and is about to strike it, and we freeze time. We freeze time just before the moment of impact, turn to Adam, and ask, "What is going to happen?" Adam has just been created. He possesses complete faculties — he is intelligent, rational, fully capable of reasoning — but he has no experience whatsoever. He has seen nothing of the world.
This is what makes it such a powerful thought experiment. We ask Adam what will happen to the second billiard ball when the first one hits it, and Hume's answer is that Adam will have no idea.
Consider someone who has never seen physical objects collide before. To us, the outcome is obvious — we have witnessed objects striking one another countless times, so of course we expect the second ball to move. It seems so familiar, so rational, so inevitable, that it feels as though we could simply think it through and predict the result. But if you had never seen it before, anything could happen. The second ball could remain perfectly still, fly straight up, explode, or do something else entirely. Without prior experience, you have no way of knowing.
This is Hume's central point: it only feels as though we can reason out causal relationships because they are so familiar to us. Hume argues that reason plays no role in learning what causes what — only experience does. If you have never witnessed a cause and its effect, you cannot know that relationship. As Hume puts it, Adam could not have known that water would drown him, or that a cloud would not.
Consider quartz. The word crystal comes from the ancient Greek for "icy cold," because quartz looks like ice. It makes up twelve percent of Earth's land surface and twenty percent of its crust. And yet it is created in lava — extreme heat produces crystal, not extreme cold. No one would ever have reasoned that out from appearances alone. Only because people have actually observed the process do we know it.
These causal relationships only feel self-evident because we have seen them repeated so many times. In reality, the only way we come to know them is through experience. Imagine bringing Adam to a pool room for the first time, freezing the moment just before the cue ball strikes, and asking him what will happen next. He would have no idea. That is the first surprising claim Hume makes. We have now answered two questions in sequence: how do we know matters of fact that are not immediately apparent? Through cause and effect. And how do we know cause and effect? Only through experience.
The next question is: how do we use experience to understand causal relationships? This is where Hume makes his shocking claim. We take Adam into the pool hall, and the first time we ask him what is going to happen, he has no idea. We leave him there and come back two hours later. We do the same experiment: the first billiard ball is about to strike the second one, and we freeze time. Adam has spent two hours watching pool games, watching billiard balls strike other billiard balls, and we ask him, "Now what's going to happen?" He says, "The second ball is going to move." He knows it easily, without hesitation. Now he can make the inference from the cause — the first billiard ball striking the second — to the effect, the second one moving. Why? Because he has had experience.
Hume's question is: how does that work? What is this arrow? How does experience underwrite it? His answer — his shocking answer — is that it is not a reasoning inference. In fact, there is no rational basis for Adam to expect the second billiard ball to move after two hours of observation. There is no more reason to expect it then than there was at the very beginning. If you believe the second ball will move, you are not doing so on a rational basis. That is a shocking claim.
The first claim was that the first time you witness something, you cannot reason it out without experience. The second claim is even more radical: even after experience — after years, after decades of experience — you still cannot reason it out. You still cannot infer an effect from a cause, no matter how many times you have seen it. This is a genuinely shocking claim.
Here is Hume's argument. I am basing my expectation of what effect will arise from a cause in this moment on all of my experience — on all the times I have seen the second billiard ball move, over and over again. Now, Hume says, let's slow way down. What exactly does that experience tell you about? What is it actually about?
What that experience tells you about is all of those past moments. In those moments, the second ball moved. No denying it. All those times one ball struck another, the second one moved. But that is all that the experience tells you. You are saying something that goes beyond that. You are not saying the ball has moved many times — Hume will grant you that. You are saying that now, in this moment, this ball will move. And Hume says there is a gap there, an explanatory space that you have to fill.
There is a space between "thousands, millions of times the effect followed from the cause" and "now it will happen again." It is not a relation of ideas. The idea that a causal relation which has held for a long, long time will continue to hold is not logically necessary — it is contingent. At any moment, the laws of physics could change. That is perfectly intelligible. There is no logical contradiction in that possibility.
Hume does not think it is going to happen. He is not saying, "Watch out — go live today because gravity will be reversed tomorrow." He is saying it is logically possible, and therefore it is not absolutely necessary that the same laws that made the ball move all those times in the past are going to be in effect now. It is not a relation of ideas. It is a matter of fact. It is contingent.
The claim that the laws which governed all these events in the past will continue the same — and that you can therefore expect the same effect from the same cause — is a contingent claim. And so we have to ask: how do you know? If it could be false, then you need a reason to believe it is true. And Hume says you do not have one. There is no reason to believe that the laws of nature, which have held for billions of years, will continue to hold in the next five seconds.
He is not saying he doubts that they will. He is not saying, "Prepare yourself." He is saying there is no reason for that assumption. He believes it. You believe it. I believe it. But we cannot prove it.
We cannot even have a reason for it, and here is why — though this is a very tricky argument, so do not be discouraged if it does not click immediately. What Hume is saying is that in order for past experience to be predictive of the future, you have to prove something first. You have to prove that the future will resemble the past — that the basic laws of nature governing these events will remain stable and continue operating the same way they always have.
This is the crucial idea: the future will resemble the past. If the future might not resemble the past, then all of our accumulated past experience becomes useless for prediction. It tells us nothing about what is about to happen, because it only tells us what will happen if the framework — the laws of physics, the laws of nature — remains basically the same. So you have to prove this, and you cannot.
It is not a relation of ideas, so it is contingent. And if it is a matter of fact, how do we determine it? We cannot have an impression of it, because it lies in the future, and we cannot have impressions of the future. So it must be established the same way we establish other matters of fact — through past experience. Why do you believe the laws of nature will remain stable? Because they always have. There has never been a moment where the law of gravity suddenly reversed, or where the billiard balls got up and started dancing. They have never changed before, so why would they now?
That reasoning sounds compelling, but notice what it is doing: it is using past experience to prove that the future will resemble the past, while the only way past experience can serve as proof is if you already assume that the future will resemble the past. You are assuming the very thing you are trying to prove — and that is a logical fallacy. That, incidentally, is what begging the question actually means. It does not mean raising or introducing a question; that would be raising the question. Begging the question is a logical fallacy in which you assume as a premise the very conclusion you are supposed to be establishing.
We would naturally appeal to the past stability of nature to predict its future stability: it has always been stable, it has never changed, therefore it will continue to be stable. But if there is a genuine possibility of it changing — and there is, because it is contingent — then all that past experience of stability cannot predict future stability, for the exact same reason we have just identified. In order to use past experience of nature's stability as evidence, you must already know that nature will be stable. But that is precisely what we do not know. That is what we are trying to prove. And so we cannot prove it.
There's a turkey who lived on a farm. The farmer was a wonderful man — every day he came out and fed the turkey splendid food. Every single day of the turkey's life, this farmer was kind and generous. Then Thanksgiving arrived. The farmer came out and killed the turkey.
Something had changed that the turkey was never privy to. The turkey didn't know the holiday season was upon him; he had no awareness of the calendar. What Hume is saying is that, for all we know, we are the turkey, and there are forces beyond us that could change without our ever becoming aware of it.
The turkey assumed the future would resemble the past. Every day the farmer had been wonderful — but then the future stopped resembling the past, altered by forces entirely outside the turkey's knowledge. Hume is saying we cannot rule out the possibility that we are in exactly that position, with some cosmic Thanksgiving waiting for us.
He raises this not because he believes catastrophe is imminent, nor as a practical warning to watch out. Rather, he is trying to show us something fundamental about human understanding. He wants to demonstrate that matters of fact — those that extend beyond our present experience — are the most important things we need to know in order to live, and yet there is nothing that ultimately supports them.
There is no reason — there cannot be — to expect that eating bread tomorrow will be more nutritious for you than eating glass will be harmful. In the past, eating glass has always been bad and eating bread has always been good. But will that hold tomorrow? You cannot know. To know that, you would have to know that bread will maintain the same properties, that glass will maintain the same properties, and that your stomach will maintain the same properties. You cannot know any of that, because the only way to prove it is by pointing to how long those properties have held — and that past experience is irrelevant unless you already know that they will continue, which is precisely what you are trying to prove.
So you cannot know whether you should eat glass or bread tomorrow. You cannot know whether you should get out of the way of a car five minutes from now. You cannot know anything. Hume seems to be leaving us in epistemological nihilism. But he will not stop there.
I have taught Hume for many years, and because of the way the schedule works, I usually end a class by telling students they cannot know whether they should eat glass or bread. I have never once received an emergency call, and I have never had a student who actually ate glass. I have never worried about it — and that is precisely because of Hume's next point.
Hume argues that this argument is airtight. You are welcome to find objections to it, and he encourages you to try, but he believes there is no response to it — that it proves you have no rational basis for expecting the causal relations you know and rely on to continue. And yet you will continue to expect them. That is because you cannot be reasoned out of a position you were not reasoned into.
If you accept this argument, as Hume thinks you should, you will conclude that you have no rational foundation for expecting everything you know about the world to persist — and yet you will still expect it to persist, because that expectation was never grounded in reason to begin with. Hume argues that the reason Adam, after two hours in the billiard hall, believes the second ball will move is not the product of a thought process. It is not because he has said to himself, "This has happened so many times, it must continue." It is because of a psychological feature of human nature. Hume calls this custom or habit.
This is a fact about the way our minds are put together, and in exploring it Hume is doing proto-psychology. Psychology did not become its own discipline until the end of the nineteenth century, with figures like Breuer, Charcot, and Freud. Before that, if you wanted to do psychology, you did philosophy — it was simply part of the discipline. In probing this territory, Hume was exploring a branch of inquiry that would eventually develop into what became known as behaviorism.
B.F. Skinner is probably its most famous practitioner, but behaviorism was created by Pavlov — a Russian psychologist very famous for his experiments with dogs. What he would do is ring a bell and then feed them. The first time he did this, it meant nothing to the dogs. But he did it over and over again, and eventually when he rang the bell, the dogs anticipated supper, and they began salivating at the sound of the bell alone.
Now, there is no dinner bell in nature. There is no natural, organic reason for dogs to associate that sound with anything having to do with eating. What Pavlov did was forge a connection — he created a cause-and-effect relationship that was entirely artificial, an artifact he manufactured simply by pairing one thing with another, first one then the other, over and over again. His conclusion was that if you have what Hume called constant conjunction — meaning that whenever one thing happens, another consistently follows in the same order — then after a while, when you expose the animal to the first event, it will expect the second. It doesn't matter how strange or how artificial or how meaningless the conjunction is; the animal will expect it.
B.F. Skinner had a very famous demonstration of this principle. He would bring a pigeon on stage during his lectures — a pigeon he had never worked with before — and the pigeon would simply turn left and right, moving around in a random pattern. But every time the pigeon turned left, Skinner would reward it with a pellet of food. Turn left, turn right, turn left — food. Turn right, turn left — food. Within a minute of rewarding only the left turns and ignoring the right turns, he could train the pigeon to turn all the way around in a full 360-degree circle. You can see it on YouTube; it's really fascinating. He forged a link through the consistent conjunction of turning left and receiving food. And this, Hume argues, is precisely how all causal reasoning works.
Descartes says we are thinking things, operating on logic. Hume says we are hairless apes responding to stimuli. We are animals, continuous with the rest of nature. We are not exceptions to it, not limited angels — we are simply clever apes, and the same things that happen to other animals happen to us.
Yes, we are very smart, and we can become incredibly good at identifying constant conjunctions with great precision — but that is all we can do. All that science is, at its core, is finding those constant conjunctions: first this, then this. It is the project of simplifying and unifying them, of finding the most universal causes.
But why is gravity 9.8 meters per second squared rather than 9.7, or 10, or 50? It simply is. Why do atoms have the particles they have, and not one more or one less? This is not merely a philosopher's complaint — scientists say it themselves all the time: we do not know why the fundamental forces are the way they are; we only know that they are. Science studies empirically how the universe works, but the ultimate reasons why it works as it does remain beyond our reach, because all we can do is trace how things occur — find these constant conjunctions.
Hume's argument, then, is that all of our reasoning rests on a foundation of arational responding.
This is very important: the belief is arational. That is not the same as irrational. Irrational means bad reasoning — it means you should not think that way. Arational means it has nothing to do with reason at all. Your digestion is arational. When you go to the doctor and the doctor hits your knee with a little hammer and your leg jumps, that is arational. You do not do that because you have thought, "I believe the right thing to do here would be to move my leg, so I will do so as soon as he hits me." No — you simply do it. Your leg does it.
That is exactly what your belief that the second billiard ball will move amounts to. You expect it to move the same way your foot jumps when the hammer strikes your knee. It is a mental reflex — not an argument, not an inference, nothing rational. According to Hume, there cannot be a reason for it. You cannot prove it, you cannot know it. There is no reason whatsoever to believe that the way things have been going so far will continue. But we believe it anyway, just as the stomach digests food and the foot jumps when struck by the hammer, because that is the way the brain is constructed.
And this is what is so fascinating about Hume. Descartes, like many philosophers, says we should only believe what we can rationally prove. Hume says that if you actually did that, you would be dead tomorrow — eating glass, failing to get out of the way of cars — because you cannot rationally prove that those are the wrong things to do. If you succeeded in being perfectly logical, the human species would end immediately.
It is our lack of reason that is our saving grace. It is our arationality that enables us to survive in this world, not our rationality. Hume is not saying rationality is bad — of course it is very good and very helpful. But it only operates on the basis of a foundation that is itself arational, and there, in an ironic echo of Descartes, is where Hume plants his flag.
Descartes said the foundation of knowledge has to be maximally rational. Hume says the foundation has to be non-rational, or else we would not know how to get by. The first implication of this is that what Hume calls his skeptical argument — that we have no rational grounds to expect causal events to continue as they have — will not actually change our behavior. Nature is too strong for reasoning, and that is a very good thing, or else we would perish immediately.
But the fact that we cannot even prove that we should not eat glass tomorrow ought to give us a little humility about proving the existence of God. It ought to give us some restraint in our metaphysical yearnings for the ultimate truths of the universe. Hume's counsel is to pull back, to reconcile ourselves with our animal nature, and to understand the world the way humans actually understand it. We need to limit ourselves to the kinds of knowledge we are genuinely capable of having.
What are those kinds of knowledge? Relations of ideas — yes, we have those. And matters of fact grounded in experience — and that is it. Those are the only legitimate forms of knowledge available to a human being, and metaphysics does not belong to either category. When Descartes proved the existence of God a priori, he was making an argument entirely within the confines of his own mind and yet claiming to establish something about what exists in the universe. That is a contingent fact, and there is no way to know what exists in the universe without empirical verification.
The idea that one can simply think one's way into the basic fabric of reality is, according to Hume, hubris — overweening pride, and a waste of a mind that could have been put to far more useful purposes. Now that we understand what is in the mind and how the mind actually reasons, we must restrain ourselves to those forms of knowledge the mind is genuinely equipped to produce. In doing so, we will achieve a far more productive creation of knowledge and a far better science.