Devin Ens
“Analysis of Hume’s Critique of Causation”
written for PHIL 281:
Theory of Knowledge
Sometimes it is hard to be sure what conclusion to draw from a Humean analysis, and he is easy to misrepresent. This is partly because one argument he is engaged in may raise a number of related issues that he has dealt with elsewhere, and some of his points seem contradictory. My wish is to consider some of the possible readings of David Hume’s critique of causation, as it appears in Section VII of the Enquiry Concerning Human Understanding, “On Necessary Connexion”, and their relation to the propositions of Section II, “Of the Origin of Ideas”, and Section X, “On Miracles”. I will offer criticisms and alternatives to Hume’s account(s) and conclude by picking which interpretation of Section VII best works for Hume, given certain arguments elsewhere in the Enquiry.
The following is a summary of the aspects of the problem of induction as presented in the Enquiry which concern my discussion.
Our assurance that certain sets of conditions are sufficient to produce certain effects is based on past experience that like has been conjoined with like. The belief in necessary connection entails (Hume will conclude that it amounts to) a belief that events similar to those experienced in the past will be accompanied by similar conjuncts. Such a belief may only be arrived at inductively, and induction does not discover necessity.[1]
This argument is against the supposed necessity of connection. “Necessity” here may refer to logical necessity, or it may not distinguish between this and physical necessity. To be physically necessary is to be sufficiently caused, but contingent upon the conditions of the event and the properties of all objects involved. Physical necessity is logically contingent, and is thus compatible with Hume’s view that causation can not be an analytic concept, but it is not explicitly dealt with. This may be due to wariness about making strong commitments to the existence of a physical external realm, at least in a discussion of induction which is supposing that what are directly given in experience are sense impressions.
Hume concludes at the end of Section VII of the Enquiry that the concept of causation may be preserved if we understand its definition as this: to say that a causes b is to say that the appearance of a is always succeeded by the appearance of b. [2]
This definition will not do. If it were true that causation is
equivalent to constant conjunction, then any cases of constant conjunction may
be, as legitimately as any other, treated as a case of causation. Is, then, the
shortening of days in
Furthermore, conjunction is a reciprocal relationship, causation is not. If a is conjoined with b, then b is conjoined with a. It is not the case if a causes b, that b causes a. We could suppose that cause and effect respectively refer to temporally anterior and posterior conjuncts, but if we are limited to sense impressions as our material for consideration, it must be admitted that some of the things which are considered causes are preceded by their effects (in sense).Would we say that vibrations in the ground are the cause of a train’s appearing from around the bend seconds later?
Does the flash of lightning (i.e. a brilliant, fleeting appearance of light) cause the rumble of thunder? In this case, both are effects of the same event: a bolt of electricity.
Forms of conjunction which are not cases of causation include: events having common causes (such as in the case of the flash of light and the rumbling thunder), and coincidences (like the Christmas season coinciding with the winter season in the Northern hemisphere).
It may be that the concept of events causally related may be subsumed under that of conjoined events, but it would be a peculiar species of conjunction, one such that:
-the relation between the conjuncts is a physically necessary one,
-the relation is non-reciprocal,
-the distinction between a cause and an effect is partly a temporal one: effects may not be antecedent to causes (although they may be simultaneous, as in the case of stationary causal relationships, e.g. my rear causing the impression in my typing chair). This much is contained in the concept of an effect. Sense-impressions of effects may precede impressions of or inferences to causes. This goes to show that there is a difference in the way in which we think of impressions and the way in which we think of objects. Whatever it may be that we are immediately acquainted with, we do certainly speak of more than sensual data. If talk of objects were reducible to talk of impressions, we would have no way of distinguishing between the order of occurrence of events, and the order in which related data are collected.
The following are three ways I see of considering the issue as Hume has put it, each with a different emphasis:
i) Ideas, such as causal necessity, of which we have no impressions, are mere assumptions, arrived at by habit. Our belief in causation (and maybe other metaphysical assertions) is essentially groundless, unless we define causation in terms of impressions and expectation instead of powers and necessity. Let this interpretation be the story of knowability.
ii) The way in which things come to be known, where knowledge entails a certain mental state, is through experience. The belief that certain things cause others happens to be formed by constantly observing one in league with the other. It is an automatic move that all rational creatures make. This is the psychological story.
iii) The way in which knowledge only can be and only ought to be arrived at is through experience. Causation is known precisely because, if reduced to a statement of impressions always being found together, it is empirically known. This is the normative story.
Case i requires as a premise that what we are acquainted with in experience are sense impressions. If we accept this premise, then we have two inductive leaps to make: one from the existence of sense data to the existence of external objects, and another from those objects to the metaphysical relations between them.
Let us suppose instead that what we are directly acquainted with in experience are physical objects. The concept of a physical object contains facts about causality in the form of physical laws. Therefore, the relation between the event of bodies interacting and the idea of causality is an analytic derivation from the synthetic concept of a physical object. Thus, the supposition of direct acquaintance with physical reality saves some inductive moves.
Case ii is pretty safe, but may not be consistent with his statement that “all events seem entirely loose and separate.”[3] This must mean, “seeming upon reflection,” for if psychological habit affects the way things seem in experience, and if the expectation of uniformity is a psychological habit, then things do indeed “seem” (in experience) uniform and regular.
One thing that is a central idea of both i and ii is that the idea of causation can be formed only a posteriori.
Here’s an a priori attempt.
1) That which exists cannot fail to have properties. That is, if there is nothing predicable of it, there is no thing to speak of. While knowledge of specific properties comes from experience, the concept of properties is contained in the concept of existence, which itself is presupposed by the having of concepts at all.[4]
2) The having of a property entails behaving in certain ways under certain conditions. This is true even in an idealist universe. Whether one takes properties as dispositions to appear certain ways, or as qualities subsisting in matter, any property attributed to an object will, in its conception, contain conditional statements concerning possible scenarios.
3) The conditions influencing any given entity consist of other entities in a certain spatio-temporal relation to each other. Having properties, i.e. dispositions to behave in certain ways under certain conditions, entities cannot fail to incite each other to particular sorts of behaviour, since they are each other’s conditions. This is causal interaction.
Thus the concept of causality may be derived from the very concept of an existing thing. What may only be learned from experience is which events will produce which effects, or to put it another way, what properties the things in the world possess.[5]
It is my opinion that both ii and iii are faithful representations of Hume’s intentions. The authority of the psychological story is undermined, however, by some of his own assertions elsewhere, namely by the above-mentioned one that things “seem” unconnected, and by his concern about belief in miracles.
Section X of the Enquiry, “On Miracles”, tells a normative story. He is clearly stating that one ought not to believe in miracles. The fact that people do believe in miracles shows that it cannot be the case that people do always form their beliefs based on experience, assuming, as Hume does, that belief in miracles stems from reports of miracles. While the psychological story is damaged by the existence of faith-based beliefs, the claim that actual knowledge is based only in experience is underscored in this section.
While miracles are imaginable, being compiled of ideas known from impression (that is, biblical miracles are visualisable, but defy the laws suggested by daily experience), they ought not to be believed because they are contrary (though no two empirical claims may be contradictory, i.e. they are not logically related) to experience.
If the claim were about knowability, then Hume would not be able to make this normative move. If he thought there were no reason to suppose that like will always in the future be conjoined with like, then there would be no reason for incredulity at the report of a miracle. So it must be that his claim is not one about reasons, for surely we have reasons to expect that the future will resemble the past, namely that it is precisely through such inferences that we come to know anything (beginning with the inference from internal to external), but one about reason, that it is not this, but habit, which forms the basis of our beliefs. While it may be the case that denying an empirical fact may not result in a contradiction, Hume seems to be suggesting that it would still be irrational to do so. That abstracting from events to laws is a rational, though inductive, act seems hard to deny. Thus, at best, Hume can only show that it is experience which first provides the matter for reason.
Hume, David. An Enquiry Concerning Human Understanding. (Indianapolis: Hackett, 1977)