Hume, On causation


Here some extracts from David Hume’s Treatise of Human Nature, (1739) Book 1, part III about cause and effect.

2: Probability, and the idea of cause and effect

“To begin in an orderly fashion, we must consider the idea of causation and see from what origin it is derived.
• It is impossible to reason soundly without understanding perfectly the idea about which we reason;
• and it is impossible to understand an idea perfectly without tracing it back to its origin and examining the primary impression from which it arises.
• The examination of the impression gives clearness to the idea,
• and the examination of the idea gives a similar clearness to all our reasoning.

Take any pair of objects that we call cause and effect, and turn them on all sides in order to find the impression that produces this prodigiously important idea. I see straight off that I mustn’t search for it in any of the particular qualities of the objects: whichever of these qualities I pick on, I find some object that doesn’t have it and yet does fall under the label of ‘cause’ or ‘effect’. And indeed everything that exists, whether externally or internally, can be considered as either a ‘cause’ or an ‘effect’, though it is plain that no one quality universally belongs to all beings and gives them a title to that label.

So the idea of causation, ·since it doesn’t come from any•quality·, must be derived from some •relation among objects; and that relation is what we must now try to discover.

The first thing I find is that only contiguous pairs of objects are considered as cause-effect related and that nothing can •operate at a time or in a place other than —even if extremely close to— the time and place that it •exists in. It sometimes seems that one object acts on another that is at a distance from it, but they are commonly found on examination to be linked by a chain of causes, with each link contiguous to the next, and the end links contiguous to the distant objects; and in any particular case where we can’t discover such a chain we still presume it to exist.
So we can take it that contiguity is essential to causation; (…)

The second relation that I shall claim to be essential to causes and effects is not so universally acknowledged ·as contiguity·, being a subject of some controversy. It is the relation of the cause’s priority in time to the effect.
Some claim that it is not absolutely necessary for a cause to precede its effect and that any object or action can in the very first moment of its existence exert its productive quality, giving rise to another object or action that is absolutely simultaneous with it. But experience in most instances seems to contradict this opinion, and anyway we can may establish ·the essentialness of· the relation of priority by a kind of inference or reasoning, ·as follows·. It is an established maxim, both in physics and the human sciences, that an object O1 that exists for some time in its complete state without producing another object O2 is not the sole cause of O2 ·when it does occur, but is assisted by some other factor that pushes O1 from its state of inactivity and makes it exert the energy which it secretly possessed.
Now if any cause could be absolutely simultaneous with its effect, it is certain, according to this maxim, that all causes must be simultaneous with their effects; for any one of them that holds back its operation for a single moment doesn’t exert itself at the very time at which it might have operated, and so it is not the whole cause of the effect.
The consequence of this would be nothing less than the destruction of the succession of causes that we observe in the world—indeed, the utter annihilation of time. For if one cause were simultaneous with its effect, and this effect with its effect, and so on, there would plainly be no such thing as succession, and all objects would be coexistent.

If you find this argument satisfactory, good! If not, I ask you to allow me the same liberty that I took in the preceding case, of supposing it to be satisfactory. You will find that the affair is of no great importance.

Having thus discovered or supposed the two relations of contiguity and succession to be essential to causes and effects, I find myself stopped short: this is as far as I can go if I attend only to single instances of cause and effect.

When bodies collide, we think that the motion in one causes motion in the other;

but when we consider these objects with the utmost attention, we find only that one body comes up to the other, and that the former’s motion precedes the latter’s, though without any interval that we can perceive. It does no good for us to rack ourselves with further thought and reflection on this individual case: we have said all we can about it.

You might want to stop looking at particular cases and define ‘cause’ as ‘something that is productive of something else’; but this doesn’t say anything. For what would you mean by ‘production’? Could you define it except in terms of causation? If you can, please produce the definition. If you can’t, you are here going in a circle, producing merely one synonymous term instead of a definition.

Shall we then rest contented with •contiguity and•succession as providing a complete idea of causation? By no means! One object can be contiguous and prior to an- other without being thought to be its cause.

There is also a •necessary connection to be taken into account, and that relation is much more important than either of the others.


When I cast my eye on the known •qualities of objects, I immediately find that the relation of cause and effect doesn’t depend in the least on them. When I consider the •relations between them I can find only contiguity and succession, which I have already regarded as imperfect and unsatisfactory. Should I despair of success, and accept that what I have here is an idea that is not preceded by any similar impression?

That would be strong evidence of light-mindedness and instability, given that the contrary principle has already been so firmly established as to admit of no further doubt — at least until we have more fully examined the present difficulty.

So we must proceed like someone who, having searched for something and not found it where he expected, beats about all the neighbouring fields with no definite view or plan, hoping that sheer good luck will eventually guide him to what he is looking for. We have to leave the direct survey of this question about the nature of the necessary connection (…) and try instead to find some other questions the answering of which may afford a hint on how to clear up the present difficulty.

I shall examine two such questions:

  1. What is our reason for holding it to be necessary that everything whose existence has a beginning also has a cause?
  2. Why do we conclude that causes of kind K1 must necessarily have effects of kind K2, and what is going on when from the occurrence of a K1 we infer that a K2 will occur, and how does it happen that we believe the predictions generated by such inferences?

Before going further, I should remark that although the ideas of cause and effect are derived from impressions of reflection as well as of sensation (…): So the same relation of cause and effect that belongs in the external world belongs in the mind as well.

6: The inference from the impression to the idea

(…) If we consider these objects in themselves and never look beyond the ideas we form of them, we shall find that none of them implies the existence of anything else. Such an inference —·based purely on the ideas·— would amount to knowledge, and would imply the absolute contradiction and impossibility of conceiving anything different,·that is, of conceiving the predicted effect not to follow·. But clearly there can’t be any impossibility of that kind, because all distinct ideas are separable. Whenever we pass ·inferentially· from a present impression to the idea of some other object, we could have separated the idea from the impression and have substituted any other idea in place of it.

So it is purely by experience that we can infer the existence of one object from that of another. The experience goes like this:
We remember having had frequent instances
of the existence of one sort of object, and also remember that individuals of another sort have always gone along with them, regularly occurring just after them and very close by. Thus we remember seeing the sort of object we call ‘flame’ and feeling the sort of sensation that we call ‘heat’. We recall also their constant conjunction in all past instances —·always flame-then-heat·. Without more ado we call the one ‘cause’ and the other ‘effect’, and infer the existence of the heat from that of the flame.
In all the instances from which we •learn the conjunction of particular causes and effects, both the causes and effects have been perceived by the senses and are remembered; but whenever we •reason about them, only one is perceived or remembered, and the other is supplied on the basis of our past experience.

Thus, in moving on through our topic we have suddenly come upon a new relation between cause and effect —finding this when we least expected it and were entirely employed on another subject. This relation is the constant conjunction of cause with effect. Contiguity and succession are not sufficient to make us regard two objects as cause and effect unless we see that these two relations are preserved in a number of instances. Now we can see the advantage of leaving the direct survey of the cause-effect relation in order to discover the nature of the necessary connection that is such an essential part of it.

But, to tell the truth, this newly discovered relation of constant conjunction doesn’t seem to take us far along our way. ·Here is an expansion of that pessimistic thought·:

The fact of constant conjunction implies only that similar objects have always been placed in similar relations of contiguity and succession; and it seems evident that this can’t reveal any new idea; it can make our ideas more numerous, but can’t make them richer. What we don’t learn from one object we can’t learn from a hundred that are all of the same kind and are perfectly alike in every detail.
Our senses show us in one instance two bodies (or motions or qualities) in certain relations of succession and contiguity, and our memory presents us with a multitude of cases where we have found similar bodies (or motions or qualities) related in the same ways.

The mere repetition of a past impression —even to infinity— won’t give rise any new original idea such as that of a necessary connection; and the sheer number of impressions has in this case no more effect than if we confined ourselves to one only.

But although this reasoning seems sound and obvious, it would be folly for us to despair too soon. So I shall continue the thread of my discourse: having found that after the discovery of the constant conjunction of any objects we always draw an inference from one object to another, I shall now examine the nature of that inference, and of the transition from the impression to the idea.

Perhaps we shall eventually find that •the necessary connection depends on the inference rather than •the inference’s depending on the necessary connection!

It appears that the transition from an impression that is present to the memory or senses (and said to be of a ‘cause’) to the idea of an object (which is said to be an ‘effect’) is founded on past experience, and on our memory of their constant conjunction.

So the next question is:
how does experience produce the idea ·of the effect·?
Is it done by the •understanding or by the •imagination?
Are we caused to make the transition by •our reason or by •some ·non- reasoned· association and relation of perceptions? ·

I shall start with the former suggestion, giving it about a couple of pages·.


If reason did the work, it would have to be relying on the principle that Instances of which we haven’t had experience must resemble those of which we have; the course of Nature continues always uniformly the same.

In order to clear this matter up, therefore, let us consider all the arguments that might be given to support such a proposition. They will have to be based either on •·absolutely certain· knowledge or on •probability;
so let us look into each of these degrees of certainty, to see whether either provides us with a sound conclusion along these lines.

My previous line of reasoning will easily convince us that no demonstrative arguments could prove that instances of which we have had no experience resemble those of which we have had experience. We can at least conceive a change in the course of Nature; which proves that such a change is not absolutely impossible. To form a clear idea of anything is an undeniable argument for its possibility, and can all on its own refute any claimed demonstration against it.

Probability doesn’t concern the relations of ideas as such, but rather the relations among objects; so it must be based in some way on the impressions of our memory and senses, and in some way on our ideas.

If our probable reasonings didn’t have any •impressions mixed into them, their conclusions would be entirely chimerical:
and if there were there no •ideas in mixture, the action of the mind in observing the relation —·that is, in taking in that such-and-such makes so-and-so probable·— would strictly speaking be sensation, not reasoning.
In all probable reasonings, therefore, there is•something present to the mind that is either seen or remembered, and from this we infer •something connected with it that is not seen nor remembered.

The only connection or relation of objects that can lead us beyond the immediate impressions of our memory and senses is that of cause and effect, because it is the only one on which we can base a sound inference from one object to another.
The idea of cause and effect is derived from experience, which informs us that certain specific kinds of objects have always been constantly conjoined with each other;
and as an object of one of these kinds is supposed to be immediately present through an impression of it, we on that basis expect there to be an object of the other kind.

According to this account of things—which I think is entirely unquestionable—•probability is based on •the presumption that the objects of which we have had experience resemble those of which we have had none; so •this presumption can’t possibly arise from •probability.
One principle can’t be both the cause and the effect of another. This may be the only proposition about the causal relation that is either intuitively or demonstratively certain!


You may think you can elude this argument. You may want to claim that all conclusions from causes and effects are built on solid reasoning, saying this without going into the question of whether our reasoning on this subject is derived from demonstration or from probability. Well, please produce this reasoning so that we can examine it.


Thus, not only does •our reason fail to reveal to us the ultimate connection of causes and effects, but even after experience has informed us of their constant conjunction we can’t through •our reason satisfy ourselves concerning why we should extend that experience beyond the particular instances that we have observed. We suppose, but can never prove, that objects of which we have had experience must resemble the ones that lie beyond the reach of our discovery. I have called attention to •certain relations that make us pass from one object to another even when no reason leads us to make that transition; and we can accept as a general rule that wherever the mind constantly and uniformly makes a transition without any reason, it is influenced by•these relations. That is exactly what we have in the present case. Reason can never show us a connection of one object with another, even with the help of experience and the observation of the objects’ constant conjunction in all past instances. So when the mind passes from the idea or impression of one object to the idea of or belief in another, it isn’t driven by reason but by certain forces that link the ideas of these objects and unite them in our imagination.
If among•ideas in the •imagination there were no more unity than the •understanding can find among •objects, we could never draw any inference from causes to effects, or believe in any matter of fact. The inference, therefore, depends solely on the ·unreasoned· union of ideas.


Ideas are indeed subject to a uniting force that may at first sight seem different from any of these, but will be found ultimately to depend on the same origin.

Thus, because a particular idea is commonly attached to a particular word, nothing is required but the hearing of that word to produce the corresponding idea; and this transition will be one that the mind is hardly able to prevent, however hard it tries. In this case it is not absolutely necessary that on hearing the sound we should reflect on past experience and consider what idea has usually been connected with the sound. The imagination, unaided, takes the place of this reflection; it is so accustomed to pass from the word to the idea that it doesn’t delay for a moment between hearing the word and conceiving the idea.

But though I acknowledge this to be a true principle of association among ideas, I contend that it is the very same as that between the ideas of cause and effect, and is an essential part of all our causal reasonings.

The only notion of cause and effect that we have is that of certain objects that have been always conjoined together, and in all past instances have been found inseparable. We can’t penetrate into the reason for that conjunction. We only observe the fact itself: from constant conjunction, objects acquire a union in the imagination. When the impression of one becomes present to us, we immediately form an idea of whatever usually accompanies it; and consequently we can lay this down as one part of the definition of opinion or belief, that it is an idea related to or associated with a present impression.

Thus, though causation is a •philosophical relation— because it involves contiguity, succession, and constant conjunction—it’s only in its role as a •natural relation that it produces a union among our ideas and enables us to reason on it and draw inferences from it.



Effettua il login con uno di questi metodi per inviare il tuo commento:


Stai commentando usando il tuo account Chiudi sessione /  Modifica )

Google+ photo

Stai commentando usando il tuo account Google+. Chiudi sessione /  Modifica )

Foto Twitter

Stai commentando usando il tuo account Twitter. Chiudi sessione /  Modifica )

Foto di Facebook

Stai commentando usando il tuo account Facebook. Chiudi sessione /  Modifica )

Connessione a %s...