Hume’s Empirical Framework: Matters of Fact vs. Relations of Ideas

What does it actually mean to know something? David Hume, the 18th-century Scottish philosopher, tackled this question with a precision that still unsettles philosophers today. In his landmark work An Enquiry Concerning Human Understanding (1748), he proposed that every meaningful claim a human mind can make falls into one of exactly two categories: relations of ideas or matters of fact. This division – widely known today as Hume’s Fork – is not just a neat philosophical classification. It draws a hard boundary between what we can know with certainty and what we can only believe based on experience. Understanding it changes how you think about science, logic, mathematics, and even everyday reasoning.

The two categories of human knowledge 🔗

Hume begins with a deceptively simple claim: all objects of human reason or inquiry divide into two exclusive and exhaustive kinds. There is no third category. A statement is either a relation of ideas or a matter of fact – it cannot be both, and nothing meaningful falls outside these two.

This binary is not arbitrary. It maps directly onto two deep philosophical contrasts: the difference between a priori and a posteriori knowledge, and the difference between necessary and contingent truth. Relations of ideas are knowable independent of experience and are necessarily true. Matters of fact depend entirely on experience and are only contingently true – they could, in principle, have been otherwise.

Relations of ideas: certainty without the world 🔗

Relations of ideas cover what we today broadly call formal or abstract knowledge – mathematics, geometry, algebra, arithmetic, and definitional truths. The classic examples Hume gives are propositions like “the square of the hypotenuse equals the sum of the squares of the two sides” or “three times five is equal to half of thirty.” These are knowable through pure reasoning alone, without any need to consult the physical world.

Three features define this category. First, such propositions are intuitively or demonstrably certain – they can be proved by the mere operation of thought. Second, denying them produces a contradiction. You cannot coherently assert that a triangle has four sides; the denial is not just false, it is logically impossible. Third – and this is the part Hume stresses – they tell us nothing about the world. As the Wikipedia entry on Hume’s Fork explains, relations of ideas can only be used to prove other relations of ideas; they mean nothing outside the context of how concepts relate to each other.

Take the statement “all bachelors are unmarried.” It is necessarily true, certain, and cannot be denied without contradiction. But it does not inform you whether there are any bachelors in the world, how many there are, or where they live. Its truth rests entirely on the definitions of the words involved – not on any fact about reality. This is what Hume means when he says relations of ideas contain no formal reality despite being always true.

Mathematics as the clearest example 🔗

Mathematics is the purest domain of relations of ideas. The statement “2 + 2 = 4” requires no experiment, no observation, and no sensory data. According to the Internet Encyclopedia of Philosophy, such knowledge is “discoverable by the mere operation of thought, without dependence on what is anywhere existent in the universe.” A mathematician working in complete isolation from the physical world still has full access to mathematical truth – because that truth lives in the relationships between abstract ideas, not in the behavior of external objects.

Matters of fact: knowledge grounded in experience 🔗

The second category is everything that has to do with the actual world – physical events, causal relationships, historical occurrences, scientific generalizations. For Hume, a matter of fact is any object or circumstance that has physical existence, such as “the sun will rise tomorrow” or “water boils at 100°C at sea level.”

What defines this category is precisely the opposite of relations of ideas. Matters of fact are learned a posteriori – through sensory experience, observation, and memory. And crucially, their denial never implies a contradiction. Hume puts it directly in the Enquiry: the contrary of every matter of fact is still possible, because it can never imply a contradiction, and is conceived by the mind with the same facility as if it conformed to reality. “That the sun will not rise tomorrow,” he writes, “is no less intelligible a proposition, and implies no more contradiction than the affirmation that it will rise.”

This is a striking point. The fact that you cannot disprove “the sun will not rise tomorrow” using logic alone – you have to look out the window, so to speak – reveals something important: our grip on factual truths is always empirical, never purely rational. We know them through experience, not through deduction.

Cause and effect as the engine of factual knowledge 🔗

For Hume, our reasoning about matters of fact is almost entirely driven by the relation of cause and effect. When you know that your friend is in France even though you haven’t seen them, you’re relying on a causal chain of evidence – a message, a ticket, a phone call. According to the Internet Encyclopedia of Philosophy, causation is the only relation that allows us to go beyond what is immediately present to the senses, and along with perception and memory, it is responsible for virtually all our knowledge of the world.

But here is the problem: Hume argues that we never actually perceive causal necessity. Consider the famous billiard ball example from the Enquiry. When one billiard ball strikes another, we see the first ball move, then the second ball move. We do not see the necessity connecting them – we only see the sequence. Motion in the second ball is a distinct event from motion in the first, and there is nothing in the first that logically compels the second. The connection we believe in is something our minds project onto experience, not something directly observed in the world.

Why this distinction matters: the limits of empirical knowledge 🔗

Hume’s Fork does more than classify knowledge – it exposes its limits. Since relations of ideas are certain but tell us nothing about the world, and matters of fact tell us about the world but can never be certain, the result is a fundamental gap: nothing can be both certain and about the world. You cannot cross the fork.

This has radical consequences. By Hume’s fork, it is impossible to prove something about the world with logical certainty. Arguments for the existence of God as a matter of empirical fact, for instance, face this problem directly – if God is not a physical entity with observable effects that can be consistently verified, no matter-of-fact argument can establish divine existence with the certainty of a mathematical proof.

The problem of induction 🔗

The sharpest consequence of Hume’s framework is the problem of induction. Science depends on inductive reasoning – observing patterns in the past and projecting them into the future. We have seen the sun rise every day for millions of years, so we conclude it will rise tomorrow. We have seen water boil at 100°C thousands of times, so we treat it as a law of nature.

Hume’s response is uncompromising. No amount of past experience can logically guarantee future outcomes. There is no contradiction in imagining that the sun might not rise tomorrow – the Earth could stop spinning, the laws of nature could change. Since inductive conclusions go beyond what we have observed, they cannot be established by relations of ideas (which are purely formal), and as matters of fact, they remain permanently open to revision. Our expectations are grounded in habit and custom, not in logical proof, which means all empirical knowledge is at best probable, never certain.

This was not a minor academic puzzle. Bertrand Russell later remarked that if Hume’s problem of induction cannot be solved, “there is no intellectual difference between sanity and insanity.”

Hume’s fork as a tool for philosophical critique 🔗

Hume used his two-category framework as a radical diagnostic tool. In a famous passage at the end of the Enquiry, he proposed a test for any book claiming to contain knowledge: Does it contain abstract reasoning concerning quantity or number (relations of ideas)? Does it contain experimental reasoning concerning matters of fact and existence? If the answer to both is no, then – in Hume’s words – it should be committed to the flames, as it can contain nothing but sophistry and illusion. This sweeping criterion would disqualify much of metaphysics, theology, and speculative philosophy as meaningless.

The philosophical legacy: from Kant to modern science 🔗

Hume’s distinction did not go unanswered. Immanuel Kant famously responded by arguing that Hume missed a third possibility: synthetic a priori knowledge – claims that are both necessarily true and genuinely informative about the world. In his Critique of Pure Reason (1781), Kant argued that the mind itself imposes structures – like space, time, and causality – onto experience, giving us certain knowledge of the world’s general framework without deriving it from observation alone. This was Kant’s direct attempt to close the gap Hume had opened.

In the 20th century, the logical positivists of the Vienna Circle took Hume’s fork as their starting point, using it to argue that only verifiable empirical claims and logical/mathematical truths are meaningful. Later, philosopher W. V. O. Quine challenged the very analytic-synthetic distinction that underlies it, arguing in his landmark essay Two Dogmas of Empiricism that the boundary between the two categories is less sharp than Hume assumed. And Karl Popper, responding to Hume’s problem of induction, developed the principle of falsifiability – the idea that scientific theories can never be conclusively proven, only potentially refuted by evidence.

None of these responses has made Hume’s framework obsolete. Each takes it seriously enough to require a substantial philosophical counter-move, which is itself a measure of how foundational the distinction remains.

Perception and the structure of human understanding 🔗

Underlying Hume’s two-category framework is a deeper empiricist commitment: all the raw materials of knowledge come from experience. His Copy Principle holds that all the contents of our minds – all our ideas – are ultimately products of sensory impressions. This means that any concept we can think with must ultimately trace back to some experience. Abstract ideas in mathematics relate purely to each other. Factual ideas about the world trace back to sensory input.

This makes perception not just important but foundational. As the Stanford Encyclopedia of Philosophy notes, Hume translates the traditional distinction between knowledge and belief into his own terms, dividing all objects of human reason into his two categories as a way of showing where certainty ends and where probability begins. Our senses give us the data; our habits of mind organize it into causal expectations. But neither gives us the logical necessity that relations of ideas possess. The result is a picture of human understanding as genuinely powerful – but permanently bounded.

What do you think? If all our factual knowledge rests ultimately on habit and experience rather than logical necessity, does that mean science can never give us genuine certainty about the world – or is that an acceptable and even productive limitation? And if relations of ideas are perfectly certain but say nothing about reality, what role should mathematics and logic play in our understanding of the physical universe?

References

1. https://en.wikipedia.org/wiki/Hume%27s_fork
2. https://plato.stanford.edu/entries/hume/
3. https://philolibrary.crc.nd.edu/article/the-causes-of-things/
4. https://iep.utm.edu/hume/
5. https://iep.utm.edu/hume-causation/
6. https://www.sparknotes.com/philosophy/understanding/section4/
7. https://plato.stanford.edu/entries/induction-problem/
8. https://philosophyalevel.com/posts/hume-causation-problem-of-induction/
9. https://en.wikipedia.org/wiki/Problem_of_induction