## Lewis on Causation

June 2, 2008

Now one thing I used to be sure of is that causal dependence between two events was sufficient for causation (say that e causally depends on c just in case, if c hadn’t happened e wouldn’t have either.) But even that, I’m beginning to doubt now. Imagine you are playing football, and the goalie is defending in such a way that you can’t make the direct shot. You end up making an unlikely goal by bouncing it off both posts (or something like that.) Now, if the goalie hadn’t been defending, you could have made the direct shot – and that would have been a completely different goal from the one you in fact made – there would have been no elaborate bounces from post to post, it would have just gone straight in. Thus, if the goalie hadn’t been defending, that goal would not have been scored. Sure, a different goal would have been scored, but not that one. Now my intuition is that that the goalies defending didn’t cause the goal. She prevented the direct goal, but that doesn’t mean she caused the elaborate goal which in fact occurred. Maybe my intuitions have just gone astray here – what do other people think?

1. Andrew,
The problem I think you are encountering in your conception is the following: the goalie does not cause the goal to happen, just as the effect does not cause the cause to happen. The analogy you are drawing is backwards, in other words. Some particular event E preventing some other specific event D from happening does not mean that E caused D, or would have caused C – some other specific event which did in fact follow from E’s prevention of D.
Events which occur are not caused in virtue of other events being prevented. Lewis’ counterfactual analysis just means that the real story behind causation is that one event relies upon another in order for it to occur, and they do so infallibly (in a certain sense). Certain aversions might figure into a particular casual story, but it is not the aversion which does the causal linking.
In short, I think the thought experiment is misleading.

2. I’m not quite sure how that is relevant?

The counterexample I had in mind was very simple. Let c be the goalie defending, and let e be the goal which resulted (the elaborate goal that bounced off the posts.) The point was “if c hadn’t happened, e wouldn’t have happened”. It sounds like you share my intuition that the goalie didn’t cause the goal, so we have a case of causal dependence (see the definition), without causation. Nothing more or less than that.

3. The point is that the goalie’s defending – his simply being intimidatingly present at the goal – is not a sine quo non for the erratic goal-post shot being made. The shot e might have been made without c, since we have no reason to think that the goalie had any sufficient role in the way the kicker acted. Perhaps he would have still shot the e-shot, regardless of who was there – just for kicks. In other words, take away the goalie and you do not necessarily take away the style of kick.

A counterfactual analysis might run something like this: the kicker was in the mental state M of thinking that if he were to kick the ball off of both goal posts, he could avoid the goalie’s defense and so, the particular erratic shot E he in fact kicked would not have happened had the kicker not been in M. Therefore, M caused E.

It’s not that external influences cannot cause an agent to do something (though Leibniz would disagree), but in the case of the goalie, he has no causal connection to the kicker’s action. He might figure into a story about why the kicker had some mental state which led him to make the shot, or why the kicker was in some physical state of readiness and position to kick the particular shot he in fact did – but the particularity of the shot does not rely on the goalie per se.

Is this any clearer?

4. Hi Christopher.

You’re denying the counterfactual:

* If the goalie had not been defending, the e-goal would not have been scored.

Now, certainly there *are* situations in which the kicker would have kicked the elaborate shot, even if the goalie hadn’t been defending. For example, if she was trying to show off, or whatever. But that was not the situation I was describing. In the situation I was describing things were such that, if the goalie hadn’t been defending, the kicker would have made the direct shot. Perhaps she doesn’t care too much about showing off, or is disposed to always go for the safest shot. The situation I described sounds consistent, maybe you have reasons to think it isn’t. But giving me a situation in which the counterfactual fails doesn’t establish that.

5. I deny the counterfactual because it doesn’t represent a well-formed counterfactual case of causation, ala Lewis. Not every counterfactual that you conjure will count as being sufficient for causation. For instance, consider the counterfactual: If the cue did not break, the player would have made the shot. There is no causal relationship between the cue not breaking and a particular shot being made. Perhaps the cue did not break – but having a non-broken cue is not causally sufficient for making a shot.

Lewis’ account of counterfactual causation is meant to explain what it is we mean and what it really means to say that one thing caused another. To stipulate that in the situation you were describing “things were such that, if the goalie hadn’t been defending, the kicker would have made the direct shot” is, while a well-formed counterfactual statement, simply one which could not occur.

No goalie’s being present has any causal connection to the way in which a kicker makes a shot. In short, you can state the ‘goalie-counterfactual’ and it is coherent in a common-sense way (that is, the claim ‘Man, if that goalie hadn’t been there, he would have just kicked it straight in’ seems reasonable enough), but this is just not enough to secure a real causal connection between the events.

6. Not every counterfactual that you conjure will count as being sufficient for causation.

Look, that’s why this is a counterexample to Lewis’s theory of causation. BTW, the case you gave I think would be counted by Lewis as a case of causation – can you give me references to where he says otherwise?

“things were such that, if the goalie hadn’t been defending, the kicker would have made the direct shot” is, while a well-formed counterfactual statement, simply one which could not occur.

I know you think this, but like I said in my last post you haven’t given me reason to think it couldn’t occur. Prima facie, the situation I described is utterly common place – I’m going to need a good reason before I start believing it is false.

No goalie’s being present has any causal connection to the way in which a kicker makes a shot.

Remember, I was giving an example of *causal dependence* without causation – I don’t see why you think I’d necessarily disagree. I think I may know where you’re getting confused though: the term ‘causal dependence’ is being used here in a technical sense. An event e causally depends on an event c iff *if c hadn’t happened, e wouldn’t have*. Lewis is jumping the gun a bit here by calling this *causal* dependence. Similarly arguing that there is no causal connection between the goalie and the kicker is not sufficient to show there is no causal dependence in this precise sense. The semantics for counterfactuals, in Lewis’s theory, are given in terms of similarity to the actual world, and does not presuppose causation.