Why Does Hume Think Causation Is Not Directly Given in Experience?

This is the tenth article in the series Understanding Philosophy: Figures, Problems, and Ideas. The fifth article introduced Hume’s empiricism and scepticism. This article concentrates on their central problem: when we say that one event causes another, what is actually given in experience, and what does the mind contribute?

Causation appears everywhere. Fire heats water, a collision moves a ball, a virus produces illness, and a policy brings about social consequences. Scientific explanation, ordinary prediction and judgements of responsibility all depend upon causal language. We commonly feel that in observing events we also observe the causal connection between them. Hume asks us to distinguish with unusual strictness between what appears in experience and the expectations that thought naturally forms.

Consider two billiard balls. One moving ball strikes another; the first slows and the second begins to move. Vision presents the colour, shape and speed of the balls, the moment of contact and the temporal order. We see the second ball move after the first arrives. But if we look for a visible property called causal necessity, no line of force appears binding the two events together. Experience gives contact and succession. It does not separately present the claim that the result had to occur.

Imagine a rational observer encountering a billiard collision for the first time. By analysing the motion of the first ball alone, could this observer infer exactly how the second must move? Hume thinks not. The character of the cause and the detailed form of the effect can be separated in thought. We can conceive the ball stopping, rebounding or behaving in a way entirely unlike familiar physical patterns without producing a logical contradiction. Our actual expectation comes from previous encounters with similar events.

One observation supplies one sequence. Repeated observations reveal that events of one kind regularly follow events of another kind. Hume calls this constant conjunction. Flame is repeatedly accompanied by heat, a particular collision by motion, and ingestion of a substance by a bodily response. Constant conjunction gives us grounds for grouping events into regular patterns. Yet however often the pattern has occurred, observation still records past repetitions rather than a logical guarantee that the future cannot differ.

Two distinct questions emerge. The first asks why we believe that the future will resemble the past. The second asks where the idea of necessary connection originates. The first produces the problem of induction. If the continued success of past regularities is used to prove that nature will remain uniform, the argument already presupposes what it seeks to establish. Uniformity cannot be proved by logic, because a law failing tomorrow involves no contradiction. Nor can induction prove it without relying circularly on induction.

The second question invokes Hume’s principle that clear ideas should be traceable to impressions. If necessity is not sensed in external objects, where might the relevant impression be found? Hume turns towards the activity of the mind. After repeated experience, the appearance of a cause carries thought involuntarily towards an anticipated effect. The felt determination of thought as it passes from one idea to the other supplies an internal impression from which the idea of necessary connection arises.

The experience of necessity is therefore not a perception of a bond inside the objects. It is an awareness that custom directs the mind towards an expectation. A person seeing flame for the first time does not know it will produce heat. After repeated experience, the appearance of flame immediately calls up the idea of heat. Regularity in events and habit in the mind jointly form causal belief. Without repeated events, no habit could develop; without the mind’s transition, repetition would not automatically become the feeling that an effect must follow.

Does this make causation a merely subjective illusion? The answer is not so simple. Hume does not say that every association counts as causal, nor does he deny stable differences among patterns of events. His definitions of cause refer both to constant conjunction and temporal order among objects, and to the way the appearance of one object carries thought towards another. The concept has a basis in worldly regularity while also including a knower’s expectation of that regularity.

What cannot be derived from this basis is a necessary power in the traditional metaphysical sense. If someone claims that a cause contains a hidden force directly accessible to reason, Hume asks for the impression from which that idea was acquired. We can provide deeper mechanisms—for example, explaining heat through molecular motion—but the mechanism itself consists of a more detailed set of regular relations. Explanation can become more precise and unified without experience ever displaying an absolute necessity beyond every law.

Hume’s analysis does not invalidate science. Scientific inquiry does not normally infer causation from one sequence alone. It uses controls, repetition, measurement, mechanistic models and counterfactual tests to distinguish causes from accidental correlations. Hume’s lesson is that the strength of a causal conclusion comes from the structure of its evidence, not from a researcher directly seeing metaphysical necessity. Probabilities can become very high and models exceptionally reliable while empirical conclusions remain open to revision.

The modern warning that correlation is not causation becomes clearer against this background. Two variables may change together because one causes the other, because both depend upon a common cause or because the pattern is accidental within the sample. Constant conjunction is central to Hume, but it does not replace every method of causal inquiry. Later approaches accept the limits of experience while developing more rigorous accounts of intervention, mechanism and counterfactual dependence.

Hume also transforms the relation between causal judgement and reason. Human beings do not first prove the uniformity of nature and only then begin acting on the past. Children and non-human animals form expectations from repeated experience without understanding a philosophical argument. Custom operates earlier and more deeply than theoretical reflection. Reason can compare evidence and correct hasty generalisation, but it cannot supply induction with an ultimate proof wholly independent of inductive practice.

The result contains both scepticism and trust. The sceptical element is that causal necessity cannot be demonstrated like a mathematical relation. The element of trust is that human cognition is naturally capable of living through regularities, while science can discipline that capacity. Hume does not recommend standing in the path of a vehicle while wondering whether it will really strike us. Stable experience provides sufficient confidence for action. Philosophy only prevents practical reliability from being exaggerated into logical impossibility of error.

Kant later argues that Hume correctly showed causality could not be extracted from experience but wrongly attributed it primarily to custom. If experience is to become an objective world of events ordered in time, Kant thinks that the category of causality must already participate in its constitution. A later article in this series examines Kant’s response in detail. What matters here is that Kant’s problem becomes possible because Hume draws the limits of experiential content so strictly.

Hume’s analysis finally forces us to distinguish three levels: the actual sequence of events, the stable patterns found through experience, and the necessary expectation formed by the mind. Ordinary language compresses them into the word “because”; philosophy reveals the distance among them. Causation is not an object placed directly before us in experience. It is a structure of judgement supported by empirical regularity and cognitive habit. We can still reason reliably, but that reliability arises from experience that can accumulate, be tested and be revised—not from a guarantee received in advance from the future itself.

From my own epistemic standpoint, the most productive way to continue Hume is not to reduce causation to a merely subjective habit. His deeper disruption is that a judgement does not become warranted simply because necessity feels compelling from within. What matters is whether a regularity continues to survive new experience, whether counterevidence can enter, whether error can be detected, and whether prediction and action withstand testing.

This is no longer Hume’s own conclusion. Hume still locates habit and expectation within the psychology of the human mind. Statistical models and artificial systems now show that pattern recognition, prediction, testing and revision can operate without human-style belief or understanding. The causal problem can therefore move from ‘Why does a subject acquire an expectation of necessity?’ to ‘What relations remain reliable under continuing tests?’ That is my extension of Hume into an epistemic condition after the knowing subject.