Causation: A Very Short Introduction (Very Short Introductions)

by Stephen Mumford

Perhaps this is a harmless abstraction but only if there is nothing complexity brings to causation that we would miss if we try to ignore it.

John Mackie (1917–81) noted that for almost every effect, there could be multiple causes at work. Suppose someone drops a cigarette in a house, which subsequently burns down. The house is unlikely to have burnt down just from the dropping of the cigarette. It also needed the presence of flammable materials, such as furniture, and plenty of oxygen so that the flame could take hold.

Furthermore, this whole set of factors, while being sufficient for the fire, was not itself necessary for it.

[a cause] is an insufficient but non-redundant part of an unnecessary but sufficient condition [for the effect]: it will be convenient to call this (using the first letters of the italicized words) an inus condition. (Mackie, The Cement of the Universe, 1980: 62)

One reason why causation interests us is because it is related to the important issue of human agency. When we act, we cause things to happen. We are the originators of new causal chains, or so we would like to think.

A source of resistance to the necessitarian view of causation is, however, that it could threaten our free will.

But if that decision was caused by other events outside that person—causes that necessitated the decision—then in what sense is it truly free?

Some think that a way out of this problem is to distinguish between ordinary event causation and the sort of causing that agents do. The latter can be called agent causation.

What use is it then for your mind to make free decisions if the actions of your body are necessitated through causation?

But here is one major problem. Just as necessity seems to impinge on our freedom, so too does contingency. Suppose your action is uncaused, or has some contingent element such as chance or randomness. That doesn’t seem to make you free. On the contrary, it makes you lose control.

There seems to be no free will if all is necessary; but no free will if all is contingent either.

Suppose a gust of wind comes just at the time of the striking. It seems that there can be failures of causation, not because something is missing from the causal set-up, but because something else has been added, such as a gust of wind or water.

The world is regular and predictable, but it is less than perfectly so. It may be regular enough for us to know what we should try to do to bring about certain effects, but we also know that it is quite possible for our expectation to be disappointed.

There are two ways in which a machine, or any kind of causal set-up, can fail. One would be if a part failed. Suppose a vital cog falls out of the machine, for instance, leaving a gap in the mechanism. The chain of causes that travels through the machine fails at that point. This is an instance of what we could call subtractive interference, which is the taking away of something from the cause, which prevents it having its usual effect.

In the second kind of case, we leave all parts of the machine intact but now we add some further element. Perhaps the cog in the machine gets covered in dust, which causes it to jam. Again the machine fails: not because something has been taken away but because something has been added. We will call this additive interference.

When we say that one thing, A, necessitates another, B, then as necessity is usually understood, that would mean that whenever A occurs B must occur. This seems to follow in other cases of necessity.

if it’s Wednesday and Barack Obama is president, then tomorrow is still Thursday, and so on for anything else we can add.

So if we want to know whether one thing, A, necessitates another, B, we would want to know whether if there is A plus some other thing, X, for any X we care to name, there is still B. This is a test we could adopt for judging whether causes necessitate their effects.

This is what writers mean when they say that the notion of cause involves the idea of necessity. If there be any meaning which confessedly belongs to the term necessity, it is unconditionalness. That which is necessary, that which must be, means that which will be whatever supposition we make with regard to other things. (Mill, A System of Logic, 1843: III, v, 6)

Now if it is possible to have additive interference, causation fails the test for necessity.

For all we know, the potential number of additive interferers is infinite. We cannot rule them all out in a finite list, insisting on their absence, for instance. In that case, how can we say that the cause necessitated the effect?

The question is whether the link between beheading and death, although it is clearly causal, is a matter of necessity. And it looks like it is not, even if everyone beheaded has died thus far.

The argument from additive interference has a broad significance. Even in those cases where a cause successfully produced its effect, it did not do so by necessitating it.

What this suggests is that, even for anti-Humeans, a distinction should be made between the notions of causal production and causal necessitation.

Elizabeth Anscombe (1919–2001) challenged the claim that necessity was a part of our notion of causation. To say that A caused B, she argued, was not the same as to say that A necessitated

It seems possible to have a case of causation that is probabilistic, which is not to say that it is completely random.

A cause, then, could be something that raises the probability of an effect, and sometimes successfully produces it, but without ever ensuring it.

If we can make sense of a notion of indeterministic causation, then Anscombe was right that necessitation is not part of the concept of cause.

But even if necessitarianism is wrong, this still does not mean that we have to accept the completely contingent picture of the Humean mosaic. There could yet be other options.

Causes, we might say, make a difference to what happens. They are difference makers. Without a particular cause occurring, history would have been different.

President Lincoln was killed by an assassin’s bullet, by which we mean that the gunshot caused his death. A way of understanding what this amounts to is that without the shot, Lincoln would have survived that trip to the theatre.

There are lots of other things that could have been otherwise but which did not make a difference to his death. Had he been watching a different play, for instanc...

The one thing that made the difference to him living or dying was undoubtedly the gunshot.

In looking for the cause, therefore, and in saying what it is for something to be a cause, the suggestion is that we should look for a difference maker. And a way of thinking about this is to imagine what would have happened if such-and-such had not been there.

one sues a company for damages, for instance, one would need to show that the subject of the suit made a difference. If one suffers an illness, one would have to show that one wouldn’t have suffered it but for the subject’s actions or negligence.

In philosophical terms, this would be known as a counterfactual dependence test of causation.

And there is a philosophical theory that this is more than just a test: that causation itself consists in counterfactual dependence between events.

To get a theory of causation as a counterfactual dependence, we need to add more elements. First, we are speaking not just of a counterfactual assumption but also a dependence on that assumption. You might say that if the Sun goes out it will get very cold here, stating that one thing depends on the other. And you will probably be quite happy to grant this conditional true even if its antecedent (its ‘if …’ clause) is false. It being very cold here counterfactually depends on the Sun going out.

There can be counterfactual dependences that are not causal. If this month is June (which it’s not), then you know that next month is July. Next month being July counterfactually depends on this month being June: but it’s not caused by this month being June.

One idea is that the counterfactual dependence has to be between separate events. Causation needs to be a natural relation, concerning events that are happening in the world, rather than what Hume called relations of ideas.

relation connects two or more things and a question for the theory of causation is what those things are. Humeans tend to say that they are events whereas Aristotelians tend to think of them as substances: individual objects, for instance.

An Aristotelian would point out that it is the elk, an individual biological entity, that is causing the lateness of the train. A Humean, on the other hand, would say it is the event of the elk being on the rail that causes the event of the train being late.

But what makes them both Humean is the idea that causation is a contingent relation between distinct events.

The difference between the two theories is whether that relation is one of constant conjunction or counterfactual dependence.

There is an elk on the line and the train is late. There is no strong connection between these two events. The elk does not make the train late, necessitate it, or even tend towards

It is therefore the counterfactual dependence part of the theory that does the work. There are lots of cases where one event follows another. What makes some cases causal is that, in those, had the first event not occurred, the second would not have occurred either.