Causation: A Very Short Introduction (Very Short Introductions)

by Stephen Mumford

A distinction is sometimes drawn between this relational kind of view and singularist accounts of causation.

In contrast, Humeans are effectively claiming, as an essential part of their theory, that one thing causes another only if that sequence is an instance of a covering law.

There is, however, a further weakness of a constant conjunction view that opponents have claimed. Arguably, the theory has no resources to distinguish between causes and coincidences.

The Humean would have to say, consistently with their theory, that since causation consists just in regularity, the lucky tie caused each time the winning of the bet.

If one thinks about the issue of causes and coincidences, it brings to light a seemingly paradoxical result of the Humean regularity theory. It will be easier for an accidental constant conjunction to occur the fewer instances it has. If A occurs only five times in the history of the world, and is followed by B on each such occasion, then it counts as an exceptionless regularity. But it seems quite possible, given the few instances, that it could be a mere coincidence that the Humean mistakes for a cause.

More detail is to be added. We will see that regularity is not quite all that we need in order to say that one thing causes another. But regularity is a major part of the notion of cause.

In addition to constant conjunction, Hume thought that our notion of cause included the ideas of temporal priority and contiguity. Temporal priority means that causes must precede their effects in time. Contiguity means that causes and effects must be at places next to each other.

There are, though, some important metaphysical questions to ask about causation too, which are not just questions about our concept of cause.

What is so appealing about temporal priority? Our experience seems to show us that causes come first, and effects some time after.

And from this, we can use the temporal ordering of events to distinguish the causes from the effects when there is a regularity.

Suppose we were to find that happy people tend to be friendly. There is a correlation between happiness and friendliness. We might decide that there is a causal connection between these two factors, but which was the cause and which was the effect?

When we accept that causes are temporally prior to their effects, something useful comes out of it. If A caused B, then B did not cause A. The acceptance of temporal priority might explain this asymmetry.

Temporal priority would deliver an asymmetry to causation that neither constant conjunction nor contiguity could.

This asymmetric temporal priority seems very important to our notion of causation. It adds something crucial to regularity. It gives it a direction. Let us now turn to contiguity.

Hume has an answer to this. We think that causation does not occur over a distance: not immediately so, in any case. The cause of the lighting—being struck against a rough matchbox—must be at the same place as the lighting.

You cannot strangle someone without touching them; and you cannot be nourished by food at a distance.

One might think that there are some easy counterexamples to Hume’s principle of contiguity. There can be at least some action at a distance. Suppose Peter sees Jane at the other end of the street and calls to her. She turns and sees him. Peter has affected Jane over a distance.

In all such cases, it seems plausible that the cause works on the effect via a chain of intermediate causes and effects, where each such link in the chain involves contiguous action.

In each such case, we can find that causation has travelled from one place to a distant one only by effects at each of the intervening points. Even when Peter calls to Jane, the sound—vibrations in the air—disturbs also the places in between Peter and Jane.

The idea of the causal chain is an important one. Causes can line up in a row and follow one after another, creating effects at a much later time and distant place.

Sometimes finding an originating cause is more illuminating than discovering what was the immediately preceding cause.

Causal chains can thus be spread over decades, centuries, and even millennia, reaching far-flung places. But each link could still involve contiguity and immediate temporal priority.

And given that this happens at a distinct moment in time, or slightly extended period of time, then it suggests that the cause cannot occur before the effect. The demands upon causation of both temporal priority and contiguity seem to be in tension, therefore.

An option, suggested by the billiard case, is to say that causes and effects are simultaneous. This would allow us to retain Hume’s other commitment, to contiguity.

Or consider a house of playing cards. Two cards lean against each other, each simultaneously keeping the other up. If either is removed, the other falls down.

The causation with respect to the dissolving of sugar occurs only once the sugar is in the tea. Someone placing it in the cup is just the explanation of how it got there. It’s not really the cause of it dissolving.

Now Hume had said, of course, that temporal priority was part of the very notion of cause. If he was right on this, we could not even entertain the idea of simultaneous causation. It would be like a contradiction in terms. But given that the examples mentioned above look perfectly coherent, then it seems more likely that Hume was wrong.

Even if simultaneity is conceivable, that’s not enough to guarantee that causation works that way, however. What, for instance, of the extended causal chains?

There are a number of possible replies. One would be to say, as Kant did, that some causation is simultaneous but not all is. Another Kantian thought is to make a distinction between instantaneous and simultaneous causation. Just because causes and effects can occur simultaneously does not mean that they occur instantaneously. Some causes have their effect over a lengthy period of time. Kant gave the example of a stove gradually heating a room.

But what one might say is that each link in a causal chain involves only simultaneity of cause and effect, some of which involve temporally extended processes. The chain can thus take time.

We should not assume that Hume’s notion of contiguity goes unchallenged either. Again, there’s an argument that he was wrong to insist upon it as a conceptual truth.

Physicists discuss cases of quantum entanglement.

Apart from that, however, what is significant is that it seems we are able to conceive of causation occurring non-locally, that is, without contiguity. Some call this the quantum non-locality of causation.

How would we distinguish the cause from the effect? And how can we tell when we have genuine causes if we have nothing but regularity to go on? These are still good questions and anyone who rejects Hume’s answers will have to find better ones of their own.

A common anti-Humean thought about causation is this. When causation is in play, the effect is more than just a mere possibility among many others. There is a good reason why the specific effect is produced.

The Humean mosaic suggests that, in theory, anything could follow anything. But anti-Humeans reject this. They argue for realism about causation, which they take to mean that causes are genuinely productive of their effects in a stronger sense than Hume allows.

What do we mean by necessity and contingency?

By necessity, they may mean that something is strictly entailed, that it has to be the case, or that it i...

By contingent, they may mean something that could be true or not, that it might be the case, or that it is true in some but not all possible worlds. It may be thought necessary that 2 + 2 = ...

And within the debated category, we can also put causes. Is it necessary or contingent that water dissolves sugar, that paracetamol relieves pain, and struck matches light?

Hume considered necessity as a possible fourth element to the idea of cause, alongside the three ideas we have already discussed: regularity, temporal priority, and spatial contiguity.

His argument was that a single instance of causation did not reveal any evidence to us of necessity.

And when we see more matches struck and lit, we are seeing just more of the same, where each individual striking also fails to reveal necessity. The Humean view is, then, that regularity fails to reveal any necessity in causation.

Real necessity, as far as Humeans are concerned, resides only in ‘relations of ideas’; 2 + 2 = 4 is necessary only because its truth is contained in the meanings of the ideas involved.

The attraction of the necessitarian view is that it takes seriously the feeling of compulsion in causation.

Regularity would thus be symptomatic of causation: a good way of identifying it.

The case for necessity in beheading looks strong. The necessitarian will say that the same is true of all other cases of causation.

There remains the Humean challenge of how we could know from our experience that necessity was involved in any case of causation. A necessitarian might respond, however, that Hume’s empiricist project is too restrictive over what counts as admissible evidence.

Hume could not distinguish between accidental and genuinely causal constant conjunctions, for instance. But the realist about causation can indeed make that distinction. Some regularities will be mere coincidences; others necessitated causally.

There is a variation on the necessitarian view that respects the complexity of causal situations. Complexity is no doubt a pervasive feature of causal set-ups.