Topics Language

Question Why are counterfactual claims taken seriously by philosophers? Aren't they just an imaginative way of thinking and talking? For example, why is a counterfactual of the form "If it had been the case that A, then it would be the case that C" supposed to have truth conditions? For if causal determinism is true, then there is a complete specification W of the history of world w in which A would occur such that W entails either the truth of C or the falsity of C, making the counterfactual either vacuously true or a contradiction (and this is so for all possible deterministic worlds which include A); whereas if causal determinism is not true, then the history of w cannot be fully specified because A depends on non-deterministic processes, and the truth or falsity of the counterfactual is not determined. And for a non-deterministic world of which the history is fully specified (i.e. W includes the outcomes of non-deterministic processes) in which A occurs, the vacuous/contradictory result again obtains. Isn't the Lewis/Stalnaker account thus just an attempt to apply logical rigor to a non-rigorous way of thinking-- namely, the everyday talk of counterfactuals that invokes a sort of imaginary time travel?

Accepted:
July 11, 2013

Comments

A very sophisticated question! In short, philosophers take counterfactual conditionals seriously at least partly because everyday language and thought take them so seriously. Entire legal regimes, such as the negligence regime in tort law, use confident judgments about counterfactuals -- "Had you not acted negligently, the plaintiff wouldn't have been injured (then and there)" -- in ways that matter hugely to people's lives. Indeed, it's hard to imagine a way of living that doesn't sooner or later involve counterfactual reasoning. The reliance on counterfactuals probably extends to all of natural science too, because explaining how or why a phenomenon occurred commits one to counterfactuals about how things would have gone had the "explanans" for the phenomenon not occurred.

Your points about determinism and indeterminism are good ones. Theories of counterfactuals are supposed to work regardless of whether determinism is true. If determinism is true, then a counterfactual of this form must be true: "Had you not done what you did, the conditions (or laws of nature) prior to your action would have been different." Even if determinism is false, robust counterfactuals are supposed to be true: "Had you not acted negligently, the plantiff wouldn't have been injured (then and there), even if the plaintiff could have been injured (then and there)."

As you probably know, the literature on counterfactuals is enormous. For a very interesting recent challenge to our confident belief in counterfactuals, have a look at Alan Hájek's paper "Most Counterfactuals are False," available here.

The most obvious reason why counterfactual talk is taken seriously by philosophers is that it's virtually impossible to avoid it. We constantly find ourselves asking -- for good reason -- what would happen in certain circumstances, and so understanding more deeply what that sort of talk might amount to seems to be a reasonable project.

You offer a dilemma. We consider a counterfactual "If A were the case, then C would be the case." You then give us a choice between determinism and indeterminism. So suppose determinism is true. Then even if 'A' is false as things are, the deterministic story you're imagining can still be applied in a hypothetical case in which A is true. After all, we do that sort of thing all the time when we solve physics problems! If the result of applying the theory is that C also turns out to be true, then it's true as things actually are that if 'A' were true, 'C' would be true as well. Why is that vacuous? It's certainly not trivial; otherwise physics itself would be trivial.

On the other hand, if assuming 'A' rules out 'C,' then it's false as things actually are that if 'A' were true, 'C' would also be true. That's not vacuous, and it doesn't make the counterfactual a contradiction. Keep in mind: the laws of nature are contingent truths.

But in fact, we're over-simplifying. Suppose the world is deterministic. Suppose Johnny is about to strike a match. Will it light? Our two assumptions don't answer the question. Whether the match would light depends not just on the laws but also on the background conditions, as you're aware. But notice: when I say "If Johnny were to strike a match, it would light" I'm saying (on a Lewis/Stalknaker-type account) that in the non-actual situations that most closely resemble how things actually are except that Johnny strikes a match, the match lights. That's something I could well be wrong about, or right about, consistently with determinism and with the actual laws of the world. Whether Johnny's match lights in the nearest possible situations where he strikes isn't just obvious. It depends (among other things) on all sorts of contingent facts about the actual world, and these are facts about which I might well be mistaken. The point of spelling out truth conditions is to give an account of what being right or wrong would amount to.

Things don't change if we consider indeterministic worlds. One reason is that even if things aren't fully deterministic, there would still be true counterfactuals. Some aren't so interesting. For example: if I were 6' tall, I'd be over 5' 10." That's true, even if it's true as a matter of logic/mathematics. Others wouldn't have to be so trivial. There could be cases where strict causal relations hold even if not all events have strict causes. But suppose everything is, so to speak, loose and separate. Then it might be that all counterfactuals that aren't true as a matter of logic or math are false. (False, by the way; not indeterminate.) That would be a big deal but it wouldn't make the non-logical counterfactuals vacuous and it wouldn't make then contradictory. It would just make them false. It would also leave us with a lot of true "might"-counterfactuals. For example: if Johnny were to strike the match, it might light, and it might not. Lewis's account spells out truth conditions for "might" counterfactuals, and also allows us to state truth conditions for "would" counterfactuals in terms of "might." From "If it were the case that A then it might be the case that not-C," it follows on Lewis's account that "If it were the case that A then it would be the case that C" is false.

As for rigor: the everyday use of counterfactuals may lack rigor in various sorts of ways, but this isn't as bad as it might sound. The everyday use of language in general lacks various sorts of rigor, but that doesn't make the study of semantics pointless. And it's also worth keeping in mind: Lewis saw it as a virtue of his theory that it can take straightforward account of certain kinds of lack of rigor. You say there's no one answer to questions about which possible worlds are nearest? Lewis would agree. He'd point out, however, that once you're settled on the criterion of closeness that fits your purposes, you can apply his apparatus.

A closing thought: suppose the reply to my comments is that I still haven't addressed the issue about applying rigor to the non-rigorous. (It's not clear to me that your original worry amounts to this, but no matter.) Even to apply Lewis' apparatus contextually goes beyond anything we can do with absolute rigor, or so it could be argued. But now the criticism proves too much. We're almost never in a position to apply physics (or any other science, for that matter) with the sort of rigor that criticism has in mind.

Theories in philosophy are often like theories in science, or so I'd suggest: they're more or less useful intellectual tools. My own take on what Lewis and Stalnaker have bequeathed us is that this intellectual tool has more than proved its usefulness. That's not to say it's beyond criticism or will never be replaced. But it's a considerable accomplishment.