If you're interested in classic semantic accounts of counterfactuals using possible worlds, Stalnaker's 1968 Theory of Conditionals and Lewis' 1973 Counterfactuals will help.

Counterfactuals are interesting and difficult, but not mysterious. They can be given reasonable treatments that don't commit to bizarre metaphysics.

The counterfactual scenario is completely inaccessible. For example if I say "If the Germans had won WW2" How is it possible to say anything true about this scenario? There is no truth of the matter because X didn't happen. — Andrew4Handel

The classical view is that you can order possible world-states along some contextually determined relation of metaphysical similarity, and that any counterfactual with the antecedent 'If the Germans won world war 2...' would be true just in case the consequent is true relative to all of the worlds closest to the actual world in which 'the Germans won WW2' is true. The difficulty is then the determination of the similarity relation.

I've myself wondered if a robust theory of truth such as the correspondence theory can adequately incorporate counterfactual statements into their stable (not to mention certain types of future-tensed statements). — Arkady

Counterfactuals aren't a problem for any theory of truth so far as I can tell. Modal truths are in a sense 'about' non-actual world states, but they are evaluated nonetheless relative to the actual world, and possible world-states 'accessible' from it, in the sense of modal logic. So it can be an 'actual fact,' for example, that something can or could happen, since what can or could happen is determined by the actual abilities of individuals, or metaphysical possibilities that are actually in place.

Counterfactuals are interesting and difficult, but not mysterious. They can be given reasonable treatments that don't commit to bizarre metaphysics.

Lewis' 1973 Counterfactuals will help.

I'm not familiar with either the literature on counterfactuals or Lewis, but I've heard the whispers, and isn't Lewis like the ne plus ultra of committing to bizarre metaphysics to explain counterfactuals? Which of course wouldn't meant that a particular essay he wrote on the topic wouldn't be helpful, even if you don't agree with his metaphysical stuff - but then, if he presents things clearly, in that work, one would imagine he'd also have presented things clearly to himself - so then why the weirdness?

Or is just that the whispers are wrong (I don't think I've read a word of Lewis in the original, or a page of secondary literature directly treating his work)

Also: part of the above is that it kind of usea particular individuals to illustrate general laws (i.e. if any individual a with power b does c, then d). The reason the sentence in my example is true has absolutely nothing to do with alex. — csalisbury

Your example then would be an example of a counterfactual conditional statement used to specify what it is for individuals of a specific kind (e.g. human beings) to have the power to benchpress 400 pounds, or to lack this power. Do we have a problem with the semantics of this statement in the actual case where the antecedent is false (i.e. in the actual case where Alex has the power to benchpress 400 pounds?

Rather than being faced with a counterfactual conditional statement where the antecedent describes an unactualized power, as my initial suggestion was meant to be dealing with, we now have an antecedent that describes a power not being possessed by an individual, in the counterfactual case, that he actually possesses. The statement that you propose then is a logical consequence of a partial definition of what it is for individuals of a specific kind not to possess a specific power. This definition could be construed as a partial specification of what it is for actual human beings no to possess the power to lift 400 pounds. It would go something like this: "Someone who raises the bar no more than 3 inches when she attempts to benchpress 400 pounds lack the power to benchpress 400 pounds." The deflationary explanation of the truth of this partial definitional statement would be: "Someone who raises the bar no more than 3 inches when she attempts to benchpress 400 pounds lacks the power to benchpress 400 pounds." is true if someone who raises the bar no more than 3 inches when she attempts to benchpress 400 pounds lacks the power to benchpress 400 pounds.

In line with my previous suggestion, the counterfactual conditional statement regarding Alex can be regarded to be true on account of the fact that it is a logical consequence of this partial definition of the lack of a power to benchpress 400 pounds.

Lewis himself was a modal realist, which is a bizarre metaphysics, but his formal treatment of counterfactuals probably could be adapted into any metaphysics. The moniker 'possible worlds' is unfortunate. You have to get a feel for what they do in the formalism to get what they are, and how to implement them in some metaphysical hypothesis or other.

The book I mentioned is for the most part a semantic/logical exercise. It's a lot more enlightening than abstract ruminations on the nature of possibility and truth are ever going to be. I would even say they make those ruminations look confused in retrospect – it's one of those works that can make people have conversion experiences to analytic philosophy. Worth a read.

I'll check it out. I think I'm undergoing something like that kind of conversion myself right now, through Sellars. I've realized I've wasted a lot of time constructing myself the left-out arguments in continental philosophy, and it's so refreshing to read people who spell it out. (That said, I still think many of the continentals make extremely good points and have a better synoptic vision. I would like to read them in conjunction.)

I've realized I've wasted a lot of time constructing myself the left-out arguments in continental philosophy, and it's so refreshing to read people who spell it out. (That said, I still think many of the continentals make extremely good points and have a better synoptic vision. I would like to read them in conjunction.) — csalisbury

I quite agree with you on both counts, regarding the complementary strength/weaknesses of both traditions.

The formalization, more than just explicitness, gives a sense that there are actual stakes to what's being done – because if you need your models to produce certain results, and they don't, you've failed, and in a concrete way, and this failure leads to a possible metric of improvement.

The modal realism is uninteresting precisely because it is informal – it's still not clear to me what motivates it or what it's trying to solve, and it seems to make mince-meat of the ordinary truth conditions of counterfactuals anyway. It seems to me a deeply confused position, but the beauty of it is that it's literally a position that doesn't matter, so it can be ignored, whereas the formal treatment of counterfactuals does.

The formalization, more than just explicitness, gives a sense that there are actual stakes to what's being done – because if you need your models to produce certain results, and they don't, you've failed, and in a concrete way, and this failure leads to a possible metric of improvement. — The Great Whatever

I remember either Timothy Williamson or Scott Soames (or maybe both) making this exact same methodological recommendation. David Wiggins also sometimes formalises some of his arguments in a very precise fashion. But he also proposes the methodological principle that, in order to pass a necessary sanity test, arguments that are couched in technical or semi-technical terms must make sense when rephrased in plain English (or whatever your native language is).

Williamson and Soames are unusually parochial, but yeah, their point can't be denied.

The formalization of arguments is less interested than the formalization of logics, grammatical fragments, and so on. If you see philosophy as a sort of conceptual engineering, you can have genuine results, improvements, and so on. Arguments are a bit pointless because you can just deny a premise, or just create a new distinction to resolve a contradiction. But you can't deny the efficacy of a model in producing certain results – so as long as it's agreed that such a result is desirable, there's a metric of improvement, and so long as it's not clear what's desirable, there are no serious stakes anyway.

'Continental' methodologies are generally constructed to expand the palate, cause sea-changes in world-view, harmonize historical trends, and so on. That's fine and all, but it gets boring. You want to work with your hands, and you just can't do that with that sort of philosophy. There's nothing to hit resistance against, nothing to build.

Thanks for that breakdown, that makes a lot of sense. I suspected something like that, but you've spelled it out nicely. (re: the continental/analytic split - I'm currently convinced that Sloterdijk and Sellars dovetail verynicely, and fill out complementary gaps, but I'll need to read much more of both to cash that out. I also think that much of what Sellars say is essentially Derrida with all the bullshit left out. But again, still haven't read enough.)

I think I've mentioned it before, but that aspect of formalization is something I encountered while dabbling with programming, and it's just extremely satisfying to have that kind of concrete push-back when you're doing the wrong thing (e.g., in this case, the program crashes or gives bunk output.) All that said, though, I think it's ok to keep an eye on a Big Picture so long as you do so warily.

No, what makes the first statement true is not some "power" that Peirce has. Rather, it is the fact that there is a real tendency in the universe for things with mass (such as a stone and the earth) to move toward each other in the absence of some intervening object (such as a man's body). — aletheist

I would differ with this. He had to lift the stone into the appropriate position, and hold it there with the possibility of dropping it. So, what makes the statement true, really is some power that Peirce has. Nature may set up some rocks on the side of a cliff, but it really doesn't have the power to drop the rock whenever it wants. Nature doesn't have the capacity to propose "if I drop the stone".

This misses the whole point of the example. In context, Peirce was illustrating for his audience that laws of nature are real generals; it had absolutely nothing to do with his "power" to let go of the stone. If it helps, we can change the subjunctive conditional to eliminate that aspect: "If my hand were to disappear magically, then the stone would fall to the ground."

Again, that is the wrong kind of subjunctive conditional. — aletheist

It's not. The discussion is explicitly about counterfactuals. The exact example in the OP is "For example if I say 'If the Germans had won WW2' How is it possible to say anything true about this scenario?"

Sure, how would you distinguish between the accidental and the necessary when dealing with particular conditionals? Especially when the Peircean view - now backed by quantum theory - sees the world as irreducibly spontaneous (because never completely constrained by its own habits).

So you have to take the probabilistic big picture view - as in, Popperian falsification. Pragmatism only claims to minimise our uncertainty about some proposition. In that sense, absolute verification is a naive realist's pipedream. — apokrisis

You said it was verificationism. Now you're saying it's falsification?

But even if it's falsification, how do you falsify the counterfactual "if X had happened then Y would have happened"?

It's partly language game.

how do you falsify the counterfactual "if X had happened then Y would have happened"? — Michael

My point was that counterfactuality amounts to having some theory in play. You can be sure of X because you are sure of what would count as not-x. So counterfactuality becomes the basis on which we can verify or falsify.

You are thinking of counterfactual conditionals- a strictly logicist issue. I'm talking about the place of counterfactuality in pragmatic or scientific reasoning.

Remember that I was replying on your specific question about Schrödinger's cat/Peircean epistemology. So I'm talking about counterfactuality in the context of what QM would call counterfactual definiteness.

Remember that I was replying on your specific question about Schrödinger's cat/Peircean epistemology. So I'm talking about counterfactuality in the context of what QM would call counterfactual definiteness. — apokrisis

My question about Schrödinger's cat was directed at aletheist's claim that counterfactual claims can be said to be true (or false) because the laws of nature necessitate a particular outcome given the antecedent. But there are situations where this doesn't work. So given a counterfactual claim such as "had I opened the box at this particular time I would have found the cat to be dead", something other than a reference to the laws of nature must be used to explain its truth value (assuming it has one).

And I don't really understand how your approach solves this problem.

My point was that counterfactuality amounts to having some theory in play. You can be sure of X because you are sure of what would count as not-x. So counterfactuality becomes the basis on which we can verify or falsify. — apokrisis

Let's ignore verification, which is irrelevant and impossible, but the counterfactual nature of falsification is even worse (or better if you like them).

For a falsification to occur, a certain task (the test resulting in a falsifying outcome) must be possible if the theory is false.

Thus testing is doubly counterfactual!

So, whether you like counterfactuals or not, they really seem more important than a mere linguistic curiosity.

Interestingly, the Lewisian or Everettian take on the ontological status of counterfactuals (i.e. that if they obey the laws of physics, they are real) doesn't work for the principle of testing.

So given a counterfactual claim such as "had I opened the box at this particular time I would have found the cat to be dead", something other than a reference to the laws of nature must be used to explain its truth value (assuming it has one). — Michael

For any statement to be true, it must agree with the laws of physics. This goes for factual and counterfactual statements.

A logicist concerned with the deductive truth of a conditional arguement might well think that there is a problem here in drawing a true conclusion from a false premise. The Peircean view then instead seeks to make sense of why people in fact quite routinely make use of counterfactual conditionals in their reasoning. Counterfactuality is in fact necessary for our ideas to be testable. It is the opposite of a bad practice when truth is the outcome not of deduction but an inquiry motivated by abductive/retroductive thought.

QM then is a further complication here as it says that even when the choice of future outcomes is constrained to be bivalent, all you might be able to say by way of prediction is something probabilistic. But even classically, that is the case with a coin toss - ahead of the flip, you know the outcome is going to be heads or tails, but your guess is 50/50.

So the OP raised a concern about valid deduction using counterfactuals. But they have other uses in reasoning. And even modal logic tries to get at that in imagining ensembles of worlds where it is as if some experiment has been run using an infinity of slightly different conditions.

For any statement to be true, it must agree with the laws of physics. This goes for factual and counterfactual statements. — tom

So given the counterfactual statements "if I had opened the box at time t I would have found the cat to be dead" and "if I had opened the box at time t I would have found the cat to be alive", which agrees with the laws of physics?

And what of statements like "bachelors are unmarried men" and "1 plus 1 equals 2"? Surely their truth has nothing to do with the laws of physics.

Also, what do you mean by the laws of physics? Are you referring to our models? Because our models have been wrong before (and some of our current ones are probably also wrong).

I just said that QM gives you a probabilty of either statement being the true one. So physics says counterfactuality is not just epistemology but looks to be ontology. Nature itself takes all its future conditionals into account in a way that is robustly measurable.

So to do physics now, we actually have to be able to sum up possibilities in concrete fashion. Counterfactuals are real not just for general physical laws (altheist's point), they are real for individual quantum events.

Meaning that the object being measured might have been otherwise, right? So, not determinate, in that sense, not subject to 'the laws of physics', because at that level, it's a probability.

I just said that QM gives you a probabilty of either statement being the true one. — apokrisis

So we have to abandon the principle of bivalence (in at least some occasions)? Some statements do not have exactly one truth value? We can't say either that "if I had opened the box at time t I would have found the cat to be dead" is true and "if I had opened the box at time t I would have found the cat to be alive" is false or that "if I had opened the box at time t I would have found the cat to be dead" is false and "if I had opened the box at time t I would have found the cat to be alive" is true.

So given the counterfactual statements "if I had opened the box at time t I would have found the cat to be dead" and "if I had opened the box at time t I would have found the cat to be alive", which agrees with the laws of physics? — Michael

It depends how precise you insist on being with your language and the particular history you are in.

The laws of physics state that the proportion of the instances of you that become correlated with the dead or alive cat varies with time, so any statement that agrees with that is true.

Yeah sure. Just like a coin toss. Because we can only give a probability of heads vs tails, we must abandon foolish notions about there being heads or tails. :-}

Really, you just appear to be being argumentative and not even trying.

So we have to abandon the principle of bivalence? Some statements do not have exactly one truth value? — Michael

Please don't abandon reason! You just need to index yourself against the outcome of a quantum measurement. The subjective perspective of a particular index, renders the other outcomes counterfactual, not false.

Yeah sure. Just like a coin toss. Because we can only give a probability of heads vs tails, we must abandon foolish notions about there being heads or tails. :-} — apokrisis

So a statement like "if I had opened the box at time t I would have found the cat to be dead" could be true even though the laws of nature do not determine that this would have been the outcome?

Then something other than the laws of nature must be used to explain how such a counterfactual claim can be true. aletheist's answer to the problem doesn't work.