Causation, further themes

DOI: 10.4324/9780415249126-N114-1
Version: v1,  Published online: 2005
Retrieved August 14, 2020, from

3. Causation as counterfactual dependence

Turning to the conceptual analysis of causation, the idea that a cause makes a difference to its effects motivates two broad classes of theory: counterfactual theories and the probabilistic theories (see Eells 1991). Here we shall limit our discussion to counterfactual theories (see Counterfactual conditionals).

David Lewis’s (1973, 1986; see also Lewis, D. K.) well-known counterfactual theory of causation under the assumption of determinism explains the idea of making a difference in terms of counterfactual dependence: one event e counterfactually depends on another event c just in case if c had not occurred then e would not have occurred. Lewis believes that the meaning of these counterfactuals can be made clear in terms of relations of similarity between the actual world and merely possible worlds. However, he is careful to stipulate that in normal circumstances where backwards causation is ruled out, the most similar worlds are ones which hold fixed the past background conditions. Accordingly, in a situation in which c and e are joint effects of a common cause d, one is prohibited from reasoning that if c had not occurred, d and so e would not have occurred. In this situation one is supposed to hold fixed the fact that d has already occurred, reasoning that if c had not occurred, e would have occurred anyway due to d’s occurrence.

With great ingenuity, Lewis developed his theory on the basis of the idea that causation is intimately linked with counterfactual dependence. A great deal has been learnt about causation through the detailed exploration of the counterfactual approach conducted by Lewis and his students (see Collins, Hall and Paul 2004). Nonetheless, as Lewis himself acknowledged, his counterfactual theory encounters considerable difficulty explaining our causal judgements about pre-emption. In pre-emption examples, there are two or more candidate causal pathways to some effect, but only one goes through to completion with the others being cut short.

The following is a familiar example of pre-emption. Billy and Suzy throw stones at a bottle. Suzy’s gets there first, shattering the bottle. Billy’s arrives a split second later, encountering nothing but air and flying shards of glass. But Billy’s throw was accurate so that his stone would have shattered the bottle if Suzy’s had not. The problem posed by the example is that Suzy’s throwing her stone caused the bottle to shatter, but the bottle’s shattering does not counterfactually depend on her stone-throwing. For, due to Billy’s throw, the bottle would have shattered even if Suzy had not thrown her stone.

A natural thought is that the theory can be modified to deal with pre-emption examples. But it has proved difficult to do so effectively. In order to deal with a different kind of pre-emption case, Lewis’s original theory required that a cause be connected to its effect by a chain of counterfactual dependences. But this sophistication does not help with the example under consideration. Even if we introduce an intermediate event between Suzy’s throwing of the stone and its shattering the bottle – say, her stone being in mid-air – it is not possible to establish the last link in the chain of counterfactual dependences. For again it is true that even if Suzy’s stone had not been in mid-air, the bottle would still have shattered due to Billy’s stone.

Another thought is that the time of the bottle’s shattering depends on Suzy’s throw: if Suzy had not thrown her stone when she did, Billy’s would have hit and shattered the bottle later. This idea forms the basis of Lewis’s last (2000) version of his counterfactual theory, which requires that there should be a rich pattern of counterfactual dependences connecting the time and manner of the cause’s occurrence with the time and manner of the effect’s occurrence. But it is possible to modify the example to falsify this version of the theory too. Suppose that Suzy and Billy are throwing not stones, but sophisticated electronic devices which can detect the trajectory of other objects in their vicinity. Suppose that Billy’s device is programmed to cut out just when it detects that another projectile is on course to hit the bottle at the same time, from the same angle, and with the same force as it would. Once more, Suzy’s missile caused the bottle to shatter, but there is no rich pattern of counterfactual dependences connecting the time and manner of her throw with the time and manner of the bottle’s shattering, since Billy’s would have shattered the bottle at the same time and in the same manner if Suzy had not thrown hers.

The difficulties that pre-emption examples pose for the counterfactual approach are significant. Our intuitions about such examples seem to track our judgements about processes rather than counterfactual dependences. But, as we saw in our discussion of empirical approaches, there are many processes in the world and determining which are causal and which are not is no trivial matter.

Citing this article:
Menzies, Peter. Causation as counterfactual dependence. Causation, further themes, 2005, doi:10.4324/9780415249126-N114-1. Routledge Encyclopedia of Philosophy, Taylor and Francis,
Copyright © 1998-2020 Routledge.

Related Articles