HUME Treatise I,iii,14: Hume offers an account of all five causes: matter, form, efficient, exemplary, and final cause. Beauchamp / Rosenberg, Hume and the Problem of Causation, start with: David Hume s theory of causation is an analysis of the causal relation;... (p. 3). There can be no empirical knowledge without causal knowledge and inference: Treatise I,iii,2. / Enquiry SB 76. CONTEXT: A THEORY OF BELIEF Hume discusses causation in the context of a theory of impressions, ideas, and beliefs. Context is a Lockean theory of sensation: impressions caused by the primary qualities of their objects. Hume insists that we cannot know any such relation between our impressions and an object that causes it; only a relation between impressions and ideas we have. As far as we can know anything about causal relations, they must obtain between data, not between things in themselves. See Locke, Essay II,viii,1 (?) Treatise I,i,1: ideas are caused by impressions. Hume offers two arguments for this claim: (a) there is a constant conjunction between impressions and ideas, and impressions re first. (b) when we want to cause an idea, we do so by causing an impression. Treatise I,iii: Beliefs are complex ideas that are confirmed by being causally related to present impressions. This requires a discussion of causality, but the terms of the relation here are impressions and ideas, never their objects. (Hume discusses causation on a more general level, including causation between objects of impressions.) DISTINCTNESS IMPLIES SEPARABILITY
Important in this context: all two distinct ideas / impressions are independent. (This is why Hume must claim that lines consist of a finite number of points; points as limits of segments would not be separable.) Treatise I,i,3: there are not any two impressions which are perfectly inseparable as a general principle (as invoked I,ii,3): every two distinct impressions can be separated. Treatise I,iii,6: There is no object, which implies the existence of any other if we consider these objects in themselves, and never look beyond the ideas which we form of them.... all distinct ideas are separable... Of course. If we do not look beyond X, we do not see anything but X. Treatise I,iii,12: there is nothing in any object, consider d in itself, which an afford us a reason for drawing a conclusion beyond it. Yes, but in order to conclude that we cannot infer effects from causes, or causes from effects, we must first assume that each of them is indeed beyond the other. At any rate, it follows that for Hume, all relations between distinct items may be decomposed into sets of independent relata without in any way changing the nature of these relata. Enquiry IV,i 25: For the effect is totally different from the cause, and consequently can never be discovered in it ; In a word, then, every effect is a distinct event from its cause. It could not, therefore, be discovered in the cause,... Enquiry VII,ii 58: All events seem entirely loose and separate. One event follows another; but we never can observe any tie between them. They seem conjoined, but never connected. This is a problem for Hume s analysis of causation. Hume tries explananda of a different logical type: succession, contiguity, constant conjunction. These are genuine relations; and they can only be external relations. Causation, however, is not a genuine relation at all, but a process. NOTE this is fine as an analysis of the causal relation. Anscombe, Causality and
Determination Sosa / Tooley p. 89: there is no contradiction in me putting a kettle on the stove and the water not getting hot. There would be a contradiction between the fire heating the water and the water not getting hot. One problem with causality as a relation is: we cannot understand how even A causes event B by interpolating a further event C that is caused by A and causes B. Salmon, Propagation and Production, Conclusion: we need to understand that what mediates here is not an event but a process. SEE Salmon, At-At theory. SEE John Venn, The Logic of Chance, London 1866, p. 320: causal rope instead of causal chain RELATIONS All relations are philosophical relations. Robinson, Hume s Two Definitions, about natural relations: Naturalness is then simply the property of any relation R between a thing or event A and a thing or event B (not between the idea of A and the idea of B) whereby the observation of A and B standing to each other in the relation R is enough to induce an association between the idea of A and the idea of B. (p. 166) WHY A RELATION? Argument for the assumption that causation is a relation: the idea of causation is not found in any quality of an object; The idea, then, of causation, must be deriv d from some relation among objects (Treatise I,iii,2). This argument presupposes that if X is not a thing or a quality, then it must be a relation. What about processes? Are they things? (And the premises are also questionable: must there not be some quality in a cause by virtue of which it is a cause? Do causes not have causal properties? If so, the idea of a cause may well be derived from the idea of some property. To be sure, the next question
would still be how we can have the idea of a causal property. The point here is not that this question is easier to answer, but only that nothing forces us to formulate the not-soeasy question as one about a relation.) When Hume realizes that the idea of causation is also not found in an observable relation, he does not draw the same conclusion: that instead of a relation, it is something else. But he should. QUESTION: what is the status of processes in Hume s philosophy? He does not seem to acknowledge the possibility of perceiving something that extends into the future; cf. Treatise I,iii,6. CAUSE AND EFFECT CANNOT BE SIMULTANEOUS Hume argument is (Treatise I,iii,2, Selby-Bigge p. 76): - something that is present at a time t1 without producing E is not the cause of E. (Beauchamp / Rosenberg p. 193: proper or sole causes act as soon as possible.) - Therefore, if an effect E occurs at time t2, it is not the effect only of something that was present before t2. Something else must happen at t2 in order to produce E. - If it would be possible for a cause to be perfectly contemporaneous with its effect, it would follow from this that all causes are contemporaneous with their effects. (Hume indicates that the argument may not be perfectly convincing.) Hume s argument for the claim that cause and effect cannot overlap in time is far from convincing. It runs as follows: (P1) Nothing can be a sufficient cause of X if it is already fully present before X occurs. Therefore, it seems that if (P2) causes could be contemporaneous with their effects, then (C1) all causes would have to be contemporaneous with all their effects. But then
(C2) everything would have to happen at once. We have to reject one of the assumptions that lead to this absurdity. Hume chooses to reject (P2) and deny that causes can be contemporaneous with their effects (Treatise I,iii,2, p. 76). The problem is that it is unclear what to do with the first premise (P1): that no sufficient cause be fully present before its effect occurs. This is the contrary opposite of what Hume wants to argue for. If Hume would actually show that this premise leads to an absurdity, given further reasonable assumptions, he would give us a reason for rejecting it. But this is not what he does. He uses P1 in order to show that P2 leads to C2, which is absurd. That is, if he rejects P1, no absurdity follows from P2, and then he has no reason for rejecting P2. But P1 is a premise that Hume cannot accept in any case. In order to avoid the conclusion that everything happens at once, it would therefore have been enough to deny only P1, but not P2, and maintain that some causes can precede their effects. Beauchamp and Rosenberg state the first premise in a way that Hume might have found acceptable: Sole or proper causes act as soon as possible (1981, p. 193). Taking this as the first premise, Hume s argument leads to the conclusion that if all causes could be simultaneous with their effects, then all causes would indeed be simultaneous with their effects. But there is no reason why as soon as possible should be the same for all causes. Some sole and proper causes might act instantaneously, others may not be able to do so. WHAT S WRONG WITH CAUSATION AS A RELATION The main argument will be: if causation = causality = a mere relation, then so is simultaneity and inherence a mere relation. But then we cannot have any experience. Treatise I,iii,2: it is not that relations in general cannot be perceived. Spatial relations, for instance, can be perceived. Causal relations cannot be perceived because they involve a counterfactual element. The problem is not that causation is an internal relation (not an external one, as Hume supposes). Internal relations depend the intrinsic qualities on their relata, but here, cause
and effect cannot without loss be considered apart from the causation of which they are a part depend on the relation. The difference between causation and a relation, whether internal or external, is this: the parts of a causal process cannot be considered in isolation without loss. If relations cannot be prior to their relata, then causation cannot be a relation. The line is: Hume considers causal relations; relations (other than identity) are between distinguishable things; distinguishable things may be isolated from another; therefore, there can be no proof of a necessary connection between cause and effect. But all this depends on the first: that causation is a relation. Consider Hume s Example from Treatise I,iii,8: The idea of sinking is so closely connected with that of water, and the idea of suffocating with that of sinking, that the mind makes the transition without the assistance of the memory. Hume cannot mean that there is a causal sequence there is water -> I will sink -> I will suffocate - because that there is water does not cause me to sink. Hume must mean: there being water around me -> I sink. Once I isolate the presence of water, it cannot be seen what it causes; but when we consider it as part of the process of drowning, the connection is clear and does not need to be established by an act of the mind. Consider: there being water around X -> X s sinking -> X s drowning. Here, the three items can be seen to be 3 descriptions of the same process, provided the sinking item X needs to breathe and can t swim. (Objection: in Enquiry IV,i 23, Hume says that Adam would not have been able to infer from the fluidity of water that it would suffocate him. Hence to sink is not to suffocate.)
Likewise, ball A moving -> ball B moving is not accurate. It should be: ball A hitting ball B -> ball B being hit by ball A - and these are clearly two ways of describing one and the same process. Instead of asking whether the idea of causality may be derived from the impression of the balls, Hume should ask what the impression that one touches the other comes from. In Treatise I,iii,14 Prometheus p. 164 he speaks of an impulse or shock; in the Abstract he says that one ball touches another. What is it in the idea of one ball that tells us whether it hits the other one? NO IMPRESSION OF CAUSATION Treatise I,iii,14: We never have any impression, that contains any power or efficacy : This is wrong, if we can have impressions of causations. It may be true that we cannot have an impression of a necessary connection between distinguishable items (Treatise, Prometheus ed. p. 162: Suppose two objects to be presented to us, of which the one is the cause and the other the effect; tis plain, that from the simple consideration of one, or both these objects we never shall perceive the tie, by which they are united, or be able certainly to pronounce, that there is a connexion betwixt them. ). But this does not mean that we can have no impression of causation. Hume argues: we cannot see in one thing that it is necessarily connected to another thing. All general ideas are derived from particular impressions, and no particular impression can be of this necessity. But the necessity associated with causation is of a different kind. It is not the necessity of a connection between one thing and other; it is the necessity of a thing to have a certain form. This necessity can be shown a priori: if there
is to be a thing of kind K, it must be F. Hume tries to explain how and to what degree we can establish necessary connections between separate items. Kant solves the problem by focusing on internal necessity: not the necessity of a connection, but the necessity of a kind-specific form. Causations are processes of a certain kind (or perhaps processes of any kind, as opposed to processes of no kind); and they have, as such, essences. (In order to sort things into kinds, we must observe more than one. But this does not mean that their essence is relational. The necessity with which their essential features belong to them is not a necessity of relation.) Argument Treatise I,iii,14 (Prometheus ed. p. 166): 1. we can have no impression of a necessary relation between distinct items. 2. the idea of necessity arises from a repetition of the union of two items. 3. this repetition only concerns the mind and does not change the repeated items. 4. causality is not objective (something like a secondary quality). Here, step 2. only applies if necessity = necessity of connection. HAVING A CAUSE = NOT INDIFFERENT In Treatise I,iii,9, Hume argues that chances is the opposite of causation (chance is merely the negation of a cause ); hence: (1) A is caused iff A does not happen by chance He also argues that what happens by chance must be entirely indifferent; that is, there can be no greater or lesser chance. Whenever there is a difference in chances, there must be a cause. That is: (2) Whenever some A is more probable than some B, there must be a cause for A.
Everything that can be calculated must have a cause. This means that whenever a process takes a determinate ( determined) course, it must have a cause. That we cannot reliably infer effects from causes means that when looking only at the beginning of a process, we cannot tell what course it will take. This means that we cannot with certainty say about anything that it is, e.g. falling over, burning down, crushing, flying home, etc. for other achievement verbs. For as long as something is falling over etc., it has not fallen over, and according to Hume, we cannot know whether it will have fallen over as long as this is still in the future. But if we cannot know whether it will have fallen over, it seems, we can also not know whether it is falling over. All we can possibly know is whether something has fallen over.