Every time you drop a stone, it falls. Every sunrise follows a night. Every fire burns. We act on these regularities without hesitation, and science is built on the assumption that they will continue. But David Hume, writing in the 1730s, posed a question so simple and so devastating that it has never been fully answered: what justifies our belief that the future will resemble the past?
This is the problem of induction, and it is one of the sharpest knives in the history of philosophy.
What Induction Is
Before we can appreciate Hume’s challenge, we need to be clear about what inductive reasoning is and how it differs from deductive reasoning.
A deductive argument is one in which the conclusion follows necessarily from the premises. If all men are mortal and Socrates is a man, then Socrates is mortal — there is no possible world in which the premises are true and the conclusion false. The conclusion is already contained, implicitly, in the premises.
An inductive argument is different in kind. From the observation that the sun has risen every day for as long as anyone can remember, we conclude that it will rise tomorrow. From the observation that copper conducts electricity in every test we have run, we conclude that copper always conducts electricity. The conclusion goes beyond what the premises strictly contain. No matter how many cases we observe, the next case is not logically guaranteed.
This gap — between what we have observed and what we conclude — is exactly where Hume plants his flag.
Hume’s Argument
Hume’s treatment of the problem appears most carefully in A Treatise of Human Nature (1739) and is restated more accessibly in An Enquiry Concerning Human Understanding (1748). His argument can be reconstructed in a few clean steps.
Step 1: All reasonings about matters of fact rely on cause and effect.
When I conclude that fire will burn me, or that bread will nourish me, I am reasoning from past experience to a future effect. All such inferences — about what will happen, what is happening beyond my current perception, or what happened before I was born — depend on causal connections.
Step 2: Our belief in causal connections is itself derived from experience.
We do not perceive causation directly. We observe that event A has been regularly followed by event B, and we come to expect B whenever we encounter A. The “necessary connection” we feel between cause and effect is not something we read off the world; it is a habit of the mind, formed by repeated experience.
Step 3: The inference from past experience to future cases cannot be justified by reason.
Here is the crucial move. What justifies the step from “A has always been followed by B in the past” to “A will be followed by B in the future”? There are only two possibilities:
- Demonstrative reasoning — the kind used in mathematics and logic, where the conclusion follows necessarily from the premises.
- Probable reasoning — reasoning from experience.
The first option fails immediately. There is no logical contradiction in supposing that the future will differ from the past. The sun’s failing to rise tomorrow is not like a square circle — it is a coherent thought. So demonstrative reasoning cannot bridge the gap.
The second option leads to a vicious circle. To justify induction by appealing to experience — by saying “induction has worked well in the past, so it will work in the future” — is itself an inductive argument. We would be using the very principle under scrutiny to justify that principle. As Hume puts it, we would be taking for granted exactly what is in question.
Conclusion: The practice of induction has no rational foundation.
We cannot justify our inductive inferences by reason. We make them anyway — we cannot help it — but our justification is not logical. It is, Hume says, custom and habit. The mind is so constituted that repeated experience produces an expectation, and that expectation is what we call belief. But belief, however vivid and irresistible, is not the same as knowledge.
The Significance of the Circularity
It is worth pausing on the circularity point, because it is the most precise part of the argument and the hardest to escape.
Suppose someone says: “Induction is reliable because it has been tested extensively and has always worked.” This sounds like a reasonable response. But notice what it is doing. It is inferring from the past reliability of induction to its future reliability. That inference is itself inductive. The argument has the form: “Induction worked yesterday, and the day before, and the day before that — therefore induction will work tomorrow.” But this is precisely the kind of reasoning whose validity is in question.
You cannot validate a method of reasoning by using that very method. If someone doubts whether a particular measuring tape is accurate, you cannot reassure them by measuring the tape with itself. An independent check is required. And in the case of induction, no independent check — no non-circular justification — has ever been produced.
Hume’s Own Resolution
Hume is not a sceptic in the paralysing sense. He does not conclude that we should stop making inductive inferences. He concludes that we cannot stop — and that this tells us something important about human nature.
The practice of induction is not grounded in reason; it is grounded in instinct. Nature has built us to form habits, to project past regularities into the future, to feel the pull of expectation after repeated experience. This is what Hume calls “custom” or “habit,” and it operates below the level of rational justification. When I pull my hand from a flame, I do not first construct a syllogism about the reliability of induction. The expectation is immediate, animal, irresistible.
This resolution is psychologically astute but philosophically uncomfortable. It tells us why we reason inductively without telling us whether we are right to do so. The gap between the psychological explanation and the epistemic justification remains wide open.
Responses and Attempted Solutions
Philosophers have not been idle in the face of Hume’s challenge. Three responses deserve attention.
The Pragmatic Vindication (Reichenbach)
Hans Reichenbach argued in the twentieth century that even if we cannot prove induction will work, it is the best bet available. If there are any regularities in nature to be discovered, induction will find them. If there are no regularities at all, then no method of prediction will succeed — so we lose nothing by using induction. Therefore, we are rationally justified in using it, not because we know it works, but because it is the optimal strategy under uncertainty.
The problem is that this argument is itself inductive in spirit: it assumes that the structure of the future is the kind of thing that either does or does not contain regularities, and that our past success with induction gives us some reason to think regularities exist. Hume’s ghost lingers.
Popper’s Falsificationism
Karl Popper tried to sidestep the problem entirely by arguing that science does not actually use induction. Science, he claimed, proceeds by bold conjectures and attempted refutations. We never confirm a theory by accumulating positive instances; we only ever fail to refute it. A single black swan refutes “all swans are white”; no number of white swans confirms it.
This is elegant, but it has a cost. If we never confirm theories inductively, what justifies our preference for well-tested theories over untested ones? Popper’s answer — that well-tested theories have survived more attempts at refutation — seems to smuggle in the very assumption he wants to avoid: that past survival is evidence of future reliability.
Dissolving the Problem (Strawson)
P. F. Strawson argued that Hume’s problem rests on a mistake. To ask for a justification of induction is like asking for a justification of deduction. Deductive reasoning is what we mean by “valid reasoning”; inductive reasoning is what we mean by “reasonable belief based on evidence.” The demand for a further justification misunderstands the role these practices play in our conceptual scheme. Induction does not need an external foundation any more than the rules of chess need to be validated by some higher game.
This response has force, but many philosophers find it too quick. Even if “reasonable” just means “based on evidence,” the question of why evidence-based reasoning should track truth remains. Dissolving the question does not answer it.
Why It Matters
The problem of induction is not an academic puzzle confined to philosophy seminars. It cuts to the heart of what science does and what knowledge is.
Every scientific law is an inductive generalisation. “Force equals mass times acceleration” is a claim about every instance of force, mass, and acceleration — including all future instances — derived from a finite number of experiments. The entire edifice of physics, chemistry, and biology rests on the assumption that the regularities we have observed will continue. Hume’s point is that this assumption, however natural and irresistible, cannot be given a non-circular rational justification.
This does not mean science is worthless or that we should abandon it. It means that the foundations of empirical knowledge are more precarious — and more interesting — than they first appear. We are not, as we sometimes like to think, in possession of a method that guarantees truth. We have a method that has worked extraordinarily well, and a deep philosophical puzzle about why.
The Clean Statement of the Problem
It is worth ending with the argument in its starkest form, so that its force is not lost in the discussion of responses.
- Every inference from observed cases to unobserved cases is an inductive inference.
- An inductive inference can be justified either by demonstrative reasoning or by probable reasoning.
- Demonstrative reasoning cannot justify induction, because the denial of any inductive conclusion is logically possible.
- Probable reasoning cannot justify induction, because any such justification would itself be inductive — and therefore circular.
- Therefore, inductive inferences cannot be rationally justified.
The premises are clear. The logic is valid. The conclusion is one that almost no one is willing to live by — and that is precisely what makes it so philosophically alive. Hume did not solve the problem of induction. He created it, permanently, and left it for us to carry.
Leave a Reply