Causation: A Very Short Introduction (Very Short Introductions)

by Stephen Mumford

There are a number of views. Some think the basis of causation is regularity: that one thing or event is constantly conjoined with another.

Another theory is that the existence of the effect depends on the existence of the cause; but does this adequately distinguish causes from other, related phenomena?

Do causes produce their effects by guaranteeing them? Do causes have to precede their effects? Can causation be reduced to the forces of physics? And are we right to think of causation as one single thing at all?

And anyone who makes a causal claim must have such a theory, whatever it may be, otherwise the claim would be empty.

One such theory—not a very sophisticated one—could be based on the fact that there was no disease in town before the rats came, therefore the rats caused it. The theory coming out of that observation might be formulated as follows: a cause is a new factor that has been introduced and that precedes a conspicuous change.

What is causation? It is initially a conceptual question: What do we mean by causation? But it could easily advance from that to a question of what is the real-world essence of causation, which is more of an ontological one: What is causation?

Empirical testing remains the mark of science and is thought the ultimate tribunal of scientific truth.

Traditionally, however, it is understood to be non-empirical in that philosophical truths are not settled primarily through recourse to experience.

This ethical theory is called utilitarianism. The point is that the evidence of our senses doesn’t seem to help us decide whether this theory—of what the good consists in—is right or not.

How do we decide on such questions? A traditional answer is that we use our reason to explore and settle philosophical questions.

Some experience of the world is of course needed in order to acquire our basic concepts and be able to talk about anything at all. But once this has been acquired, it seems that we are able to reason about it in fairly abstract terms.

This is the kind of approach that will be taken in this book. We will try to use our thinking to reason through such questions as whether causes must always occur before their effects. Is there any absurdity in supposing so; or any in supposing not so?

To break, upset, wake, and drill are all causal verbs that we use to make specific claims about causation. All of them seem to involve one thing making another happen.

The immediate lesson seems to be that causation requires more than one thing being followed by another. A man might take a pill and then die, or touch a rat and then die, but for us to say that the pill or the rat caused the man’s death, we need something more. It is this something more—the causal connection—that we will be investigating in the rest of the book.

David Hume (1711–76) promoted an idea, which continues to attract adherents, that there is something elusive about causation that makes it a particularly difficult matter to know. This is a claim that could be challenged but first we should try to understand it.

The problem when we start wondering about whether any of those events are causally connected is that the supposed causal connection is not itself part of our experience.

How much simpler would it be to make causal claims if it was just a matter of seeing the causal connection tying the two events together, like a rope? Instead, all we see are the two events, the striking of the match and its lighting.

The causal connection itself seems unobservable. It hides away and we have to infer its presence from other factors of the situation.

To a large extent it is a vast scientific endeavour to figure out what causes what and even when we think causation has been established t...

Because causation doesn’t exactly hit you in the face, there are even some who deny that it exists at all. There are weaker and stronger ways of stating this view.

The reductionist is not denying that there is causation, but they are denying that it is an additional thing in the world, over and above other, more familiar elements.

The stronger kind of strategy, however, could be called eliminativist. The idea here is to find some reason to eliminate a certain category of thing from our considerations altogether. The claim applied to the current topic would be that causation does not exist at all.

As an example of such eliminativism we might take Bertrand Russell’s classic 1913 paper ‘On the Notion of Cause’. Russell (1872–1970) pointed out that although we conceptualize the world in causal terms, if we instead defer to the way physics understands things, we will see that causation has no place.

Russell noted that in science, asymmetric causal relations don’t appear at all.

The causal conceptualization of the world is thus an ignorant and prescientific one and, in a famous passage, Russell says: ‘The law of causality, I believe, like much that passes muster among philosophers, is a relic of a bygone age, surviving, like the monarchy, only because it is erroneously supposed to do no harm.’

Russell’s view still has some adherents in philosophy but why has it not prevailed?

The point is that 2 + 2 can equal only one sum, 4; whereas 4 can be the sum of several combinations (2 and 2, 1 and 3, 10 and minus 6, and so on). And in this respect there is at least some asymmetry.

From a specific value of m and c, one can derive only one value for E in E = mc2. But for any value of E, there is an infinite number of values for m and c that would produce an equivalence.

Second, Russell’s account was based on his understanding of the physics of 1913. There have been a number of attempts by physicists to put asymmetry back into physical theory. One such notion is entropy, which is an irreversible thermodynamic property.

But we should not forget that physics is the representation and should not be mistaken for the world itself.

In that case, if we felt that physics had left out of its representation a central datum about the world—the asymmetry of causation—then we might be entitled to ask for a better physics.

A belief in causation may well be a philosophical one—a metaphysical one, even—but there may be occasions where we can ask for a better physics: one that reflects our metaphysical commitments.

A belief in causation, we would maintain, is a very central one.

Without causation, nothing in our universe would seem to hang together. Hume even called it ‘the cement of the universe’ (‘Abstract of the Treatise’, 1740).

Now if we were to jettison causation from our web of belief, so much else would have to go with it. It would require an entire reconceptualization of the world and almost everything we have ever believed about it.

Why do we have these regular associations of types of event? Why do they pair up? There is a very obvious and natural answer: it is because the first type of event is a cause of the second.

We shouldn’t say that one kind of thing regularly follows another because they are causally connected. Rather, we think that one thing causes another only because one regularly follows another.

But if we cast aside the presuppositions of common sense, our experience shows us only that one thing follows another; never that one thing is connected with another, compels it, produces it, or makes it happen.

Why, then, should we think that one kind of event caused another if all we know is a sequence of events? Hume has an answer. We could not know that one thing caused another just from observing a single instance, such as just one ball touching another and the second moving away. But in our experience we have no doubt seen many more than just one such instance.

Hume’s theory was that it is only through observation of regularities that we get any idea at all of causation. Repetition is the key. One type of event is always followed by another and this is what leads us to believe that the first type of event caused the second.

On Hume’s empiricist view, therefore, we have no reason to even believe in a causal connection, since there is no sense impression of it.

Instead of types of tile, the world consists of types of event, according to Humeans. And our beliefs about causation are determined simply by discovering such patterns in nature.

Hume’s view is often called a constant conjunction account. For A to cause B is for there to be a constant conjunction between A and B (plus a few other conditions, which will be discussed in the next chapter).

But Humeans have an answer to this view. Although we often believe there to be a mechanism that is responsible for a constant conjunction or regularity, the workings of this mechanism themselves can only be explained in terms of further regularities.

The regularity view offers us no explanation of why this should be, however, so the inductive inference looks to be entirely groundless. Hume was not happy with this implication, but he thought it inescapable.

Modern-day proponents of Humeanism, such as the aforementioned David Lewis, see the Humean mosaic in a slightly different way, concerned more with the metaphysical whole of reality rather than just our temporal perspective on it. When such modern metaphysicians speak of the regularities, they think of the regularities to be found in all the world’s events, from start to finish, omnitemporally.

What count as genuine regularities are those that are constant conjunctions throughout all of history, from the Big Bang to the Big Crunch, if these are the world’s first and final events.

The world will have been unkind to us if it misleads us in this way. But for all genuine causes, the future instances will be like the past. They will be the ones that have genuine constant conjunction every time.

The problem of induction just relates to our knowledge, therefore, and our limited perspective on the world. It is not a problem of causation itself, on this omnitemporal version of Humeanism.

However, whether this white ball caused the red one to move is, according to Humeanism, dependent on every other collision of balls in the whole history of the world, presumably including some that occur a thousand years later. So, in general, A causes B only relative to many other events. Can this really be so?