Hume, David (1711–76)

DOI: 10.4324/9780415249126-DB040-2
Version: v2,  Published online: 2005
Retrieved January 20, 2019, from

3. Causal reasoning

Seventeenth- and eighteenth-century philosophers offered many different views of the nature and extent of causal power. Descartes, for example, held that causes must contain at least as much ‘perfection’ as their effects and that the laws of nature can be deduced from God’s immutability, while Malebranche followed the suggestion of Descartes’ doctrine that God’s conservation of the universe is equivalent to continuous re-creation and the implications of the proposition that causes must be connected by necessity with their effects to propose that only God has genuine causal power. Spinoza held that the laws of nature could not have been otherwise and can be discerned by the intellect, while Leibniz denied the possibility of causal interaction between substances. Locke found some but not all kinds of substantial interaction to be unintelligible, while Berkeley held that the only causal power lies in the volition of minds, either divine or finite, to produce ideas.

Of the two kinds of reasoning that he distinguishes, it is probable reasoning, Hume holds, that predominates in human life, yet he finds that it has received relatively little investigation; and it is because it has been so little investigated, he thinks, that there is so much philosophical confusion concerning the nature and extent of ‘the efficacy of causes’. All probable reasoning, he argues, depends on the relation of cause and effect: whenever we infer the existence of some matter of fact that goes beyond the content of present perceptions or memories, it is always on the basis of an implicitly or explicitly supposed causal relation between what is represented by a present perception or memory and the conclusion we draw from it (THN 1.3.2; EHU 4). Yet although all probable reasoning is thus also causal reasoning, the causal relation between distinct things or events seems itself difficult to understand: causes precede their effects in time and are spatially contiguous to them (at least when they have spatial locations at all), but we also suppose that causes and effects have a ‘necessary connexion’ of some kind. In order to understand fully what the causal relation is, Hume holds, we must first understand the nature of the probable inferences that ‘discover’ it.

Contrary to those who hold that causal relations can in principle be discerned through pure thought alone, by means of the intellect, Hume argues that the attribution of causal relations always depends essentially on experience – in particular, experience that an event of one kind is regularly followed by an event of a second kind, which he calls their ‘constant conjunction’. Any event may be conceived to follow any other event (that is, the imagination can form an idea of it), and prior to experience there is no basis to suppose anything about what will actually follow a given event. Although it may well seem that some causal relations – such as the communication of motion by impact central to the ‘mechanistic’ natural science of the seventeenth and eighteenth centuries – are so ‘natural’ that we could anticipate them prior to any experience of them, this is only because early and constant experience has rendered them so very familiar. A probable inference occurs when, following experience of a constant conjunction of events of one type (such as the striking of a match) with events of another type (flame), an impression or memory of a particular event of the one type leads the mind to form a belief in the existence of an event of the other type. Hume’s investigation of the way in which this mental transition occurs is the occasion for his famous discussion (originally presented in the Treatise (1.3.6) but repeated with slight variations in the Abstract and the first Enquiry (4)) of what we now call induction – that is, the projection that what has held true of observed cases will also hold true of as-yet-unobserved cases.

As a ‘scientist of man’, Hume asks how this transition actually occurs. He argues first for a negative conclusion: that the transition is not produced – or ‘determin’d’, as he puts it in the Treatise – by reason. His argument for the negative conclusion is as follows. The transition in question spans a gap between what the mind has experienced – namely, past constant conjunction between two types of events plus a present impression or memory of an event of one of the two types – and what the mind concludes, which is the occurrence of another event of the other type, in conformity to the previously observed regularity. This transition he calls, variously, ‘making the presumption’ that ‘the course of nature continues always uniformly the same’, ‘supposing that the future will resemble the past’ and ‘putting trust in past experience’. If reasoning were to cause this transition, Hume argues, it would do so through an inference to a belief that nature is uniform, for it is only a conclusion of this kind that could span the gap and so produce the inference. Yet what kind of reasoning could this be? The reasoning could not be demonstrative, for the denial of the uniformity of nature is in every case perfectly conceivable and involves no contradiction. Nor could the reasoning be probable, for, if Hume’s previous account of that species of reasoning is correct, all probable reasoning can proceed only if it already makes the presumption of the uniformity of nature – which is the very presumption whose causal origin is to be explained. A probable inference thus cannot be the original cause of this presumption, for the making of the presumption is a precondition for all probable inference; and, as Hume puts it in the Treatise, ‘the same principle cannot be both the cause and effect of another’. Since all reasoning is either demonstrative or probable, and neither demonstrative nor probable reasoning can cause its key transition, probable inference is not ‘determin’d by reason’.

In interpreting this conclusion, it is important to recognize that Hume is not questioning whether probable inferences constitute a species of reasoning – as of course they do, by his own classification. Rather, he is questioning whether the key transition in such inferences is itself mediated by a component piece of reasoning – in something like the way that Locke regards some demonstrative inferences as mediated by other component demonstrative inferences concerning the relations between their parts – or whether the transition is instead made by some other process. It is also important to recognize that Hume’s negative conclusion is one about the causal origin, rather than the epistemic warrant or justification, of probable inferences. Hume’s argument does, however, have important consequences for questions of justification. For if the presumption of the uniformity of nature cannot originally arise from reasoning at all, then it cannot be justified by the way in which it originally arises from reasoning. Likewise, if the claim that induction will continue to be reliable is a claim of the kind that can only be supported by reasoning that presupposes that induction will continue to be reliable, then any argument intended to justify the claim that induction will continue to be reliable must beg the question, by presupposing what it seeks to establish. Thus, the much-discussed philosophical problem of how induction can be justified is rightly traced to Hume’s discussion of probable inference.

Although the mind does not make the presumption of the uniformity of nature by reasoning to a belief about that uniformity, it does indeed make the presumption in another way, Hume argues – namely, through the mechanism of ‘custom or habit’, which is the general tendency of the mind to ‘renew’ a past operation or action without any ‘new reasoning or reflection’. In the case of probable inference, the mind’s experience of past constant conjunctions between two types of events is renewed when the mind, upon the impression or memory of an instance of one type, proceeds immediately, without further thought or reflection, to form an idea of an instance of the other. The force and vivacity of the present impression or memory provides a measure of force and vivacity to the idea as well – and this force and vivacity, or liveliness, constitutes the belief that the mind reposes in the existence of the object of the idea and serves to explain its ability to affect the will. This habit-based process – which is a feature of the imagination in Hume’s narrower sense – thus produces a belief in the conclusion of a probable inference without any intermediate belief or reasoning about the uniformity of nature at all. Hume’s description of this process constitutes his positive answer, complementing his negative answer, to the original question of the nature of the transition.

Hume follows Lockean terminology in characterizing as ‘probability’ every kind or degree of assurance other than the ‘knowledge’ that is based entirely on perceiving relations of ideas, but he recognizes that the assurance derived from inferences from experience can be firm and unhesitating. Accordingly, he goes on to distinguish, within the range of probability in the broad Lockean sense, between proof, which is the high degree of belief or assurance that results from a full and exceptionless experience of the constant conjunction of two types of events, and mere probability in a second and narrower sense. There are three philosophically-approved species of probability (THN 1.3.11–12) in this narrower sense: the probability of causes, in which two types of events have been commonly experienced to be conjoined but not exceptionlessly so or only in a small number of cases; the probability of chances, in which there is uniform experience that one of a set of alternatives will definitely occur (such as the landing of a die on one of its faces), but nothing determines the mind to expect one alternative rather than another on a given occasion, so that each alternative acquires only a limited share of assent; and analogy, in which belief concerns events that are somewhat similar to, but not exactly resembling, those of which one has experienced a constant conjunction. In addition to these reflectively approved species of probability, Hume also distinguishes several species of unphilosophical probability (THN 1.3.13) – that is, ways in which features of the imagination affect the mind’s degree of belief or assurance that, upon reflection, we do not approve. In addition, he specifies a set of ‘rules by which to judge of causes and effects’ (THN 1.3.15), rules that result from reflection on the mind’s own operations in probable reasoning and on the successes and failures of past probable inferences. He thus offers a thorough, and provocative, account of non-demonstrative reasoning as always causal and always based on inductive projection from past experience.

Citing this article:
Garrett, Don. Causal reasoning. Hume, David (1711–76), 2005, doi:10.4324/9780415249126-DB040-2. Routledge Encyclopedia of Philosophy, Taylor and Francis,
Copyright © 1998-2019 Routledge.

Related Articles