Non-causal counterfactuals

In Discussion on March 22, 2010 by Joshua Smart

[For the record, I’m rather angry at WordPress that I lost this post the first time around.]

I have no doubt that Lewis has considered the following objection, but either I missed it or he does so elsewhere. Any suggestions on what response he makes/might make?

Suppose that c (alone) causes both e and f and that c []–> ec []–> f~c []–> ~e, and ~c []–> ~f. It would appear that f []–> e and ~f []–> ~e follow. But if that’s the case, then doesn’t Lewis have to say that that f also caused e?

An example in English. Suppose that my genetic makeup is such that in each possible world in which I exist my heart beats irregularly. Suppose further that there is only one medication that can fix this, however it is a side effect that my ears swell up to a size they never would have otherwise. On Lewis’ theory, it seems as though we have to say that my heart beating rhythmically caused my ears to swell since, had my heart not beat rhythmically, my ears would not have swollen.

A few possible responses that don’t seem right:

  1. Lewis seems at pains to reject backtracking through counterfactual dependence. So we can’t say something like, “if my heart hadn’t beat rhythmically then I wouldn’t have taken the medicine and then the medicine wouldn’t have caused my ears to swell.
  2. He might try to appeal to a causal chain, but one can imagine a case in which e and f are both immediate effects of c (e.g. are two adjacent neurons simultaneously stimulated by a third neuron).
  3. He always could bite the bullet, but I doubt he does here given how much he avoids it throughout “Postscripts”.

6 Responses to “Non-causal counterfactuals”

  1. Good question, Josh.

    My only thought is that if we say that (f []–> e) then we would have to say that (e []–> f), too (for if e hadn’t happened, f wouldn’t have happened, either). But this is problematic because we don’t want to say that something is caused by the thing that causes it and vice versa. So Lewis could respond to your complaint by saying that when we get mutual causation of this sort, we should throw the case out because obviously, causation doesn’t work that way.


  2. Exactly.

    Principled throwing out is certainly an option, but Lewis doesn’t seem to be into those sorts of solutions. I think he would say that there is a clear commonsense judgement here, and if any analysis doesn’t follow it, then we should disregard said analysis.

  3. Hey guys,

    I have to admit that this stuff has been throwing me for a loop. However the problem here seems to be mercifully addressed by Peter Menzies in his SEP article “Counterfactual Theories of Causation”:

    “Thirdly, the counterfactuals that are employed in the analysis are to be understood according to what Lewis calls the standard interpretation. There are several possible ways of interpreting counterfactuals; and some interpretations give rise to spurious non-causal dependences between events. For example, suppose that the events c and e are effects of a common cause d. It is tempting to reason that there must be a causal dependence between c and e by engaging in the following piece of counterfactual reasoning: if c had not occurred, then it would have to have been the case that d did not occur, in which case e would not have occurred. But Lewis says these counterfactuals, which he calls backtracking counterfactuals, are not to be used in the assessment of causal dependence. The right counterfactuals to be used are non-backtracking counterfactuals that typically hold the past fixed up until the time at which the counterfactual antecedent is supposed to obtain.”

    So it seems that Lewis just throws out these possible counterfactuals by banning backtracking outright (Josh’s Response #1). I’m not sure though if Lewis has an independent reason for banning backtracking or if it simply is because doing so blocks such spurious dependencies.

  4. This seems right in most cases. However, I’m worried that it doesn’t work for my case. If, in every possible world in which e occurs f also occurs, then couldn’t we say e []-> f, regardless of what else is going on? Or am I missing something about counterfactuals here?

  5. I thought Lewis provided a response to precisely this sort of case is “Causality”. It appears to be the epiphenomena case he discusses towards the end of the paper. He admits that given the counterfactuals above it would appear that “we have a spurious causal dependence of f on e”. He then mentions one possible solution is to require that a cause must always proceed its effect, but he rejects that response. Lewis then argues that instead we can simply reject the counterfactuals that are causing the trouble. In order to determine whether e causes f, we have to go to the closest possible world in which e does not occur. However, Lewis argues, the closest possible world where e does not occur is not one where c does not occur. Instead, the closest possible world where e does not occur is a world in which c still occurs, but the laws are slightly different such that its occurence does not guarantee the occurence of e. In other words, in evaluating whether e causes f, we ask about the closest world in which ~e. But that world, according to Lewis is one in which c still occurs, but does not necessitate e’s occurence. Therefore, in that world f will (most likely) still be caused to occur by c. Thus, the counterfactual if ~e, then ~f does not hold and so e does not cause f.

    I’m not sure I buy that response, but it is Lewis’ attempt to provide a solution to precisely this problem.

  6. I agree with Leo and Collin that Lewis will take Josh’s 1st option: simply denying the counterfactual.

    Lewis seems to have a criterion of selecting counterfactuals: do not “diverge at all from the actual course of events until just before the time of e”. I don’t quite understand this criterion, but I learn from Collin’s post. It seems that Lewis can say that when we construct the counterfactual, in the world where e does not occur, the only change is the causal chain from c to e, but the causal chain from c to f doesn’t change. In that case, ~e does not imply ~f. Even though e and f could be immediate effects of c, they don’t need to occur (or not occur) together. Timing doesn’t matter.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: