Wednesday, October 6, 2010


I want to open a combination lock, but I don't know the combination. Can I do it? There is some sequence of numbers which I can enter, and which is such that if I enter them it will open. So it seems I can. On the other hand, I can't fly a plane. Yet there is a sequence of button presses and movements of levers which I can make, and which is such that if I enter them the plane will fly.
It seems, thus, that the thing to say is that there is something which I can do, which is such that if I did it, the lock would open, and would open as a result of what I did, but nonetheless I can't open the lock.
But now suppose that I try to open it. I enter a sequence of numbers at random. And I get lucky: the sequence is right, and the lock is open. But if what I said above is correct, then it seems that I should say: "I opened it, even though it wasn't the case that I was able to open it." And that sounds weird.
Dan Johnson suggested to me that "can" is context-sensitive, and the context shifts when you open it. Maybe I should say: "Yes, I was able to open it, given my luck"?
Or maybe the thing to say is this. The subjunctive conditional
  1. Were I to try to open the lock, I would happen to enter the correct combination
is true at some worlds and not true at others. At the worlds where it's true, I can open the lock, by using the obvious method for opening the lock: trying. At the worlds where it's false, I can't open the lock. The world where I do enter the correct combination when I try to open the lock is always a world where (1) is true. If subjunctive conditionals have non-trivial truth values (which basically means: either Molinism or determinism holds), then some of the worlds where I don't try to open the lock are worlds where (1) is true and some of them are worlds where (1) is false. So at some worlds I can open it and at others I can't, and before I try, I can't tell which world I'm in. On the other hand, if subjunctive conditionals don't have non-trivial truth values, and my choice of numbers would be indeterministic, then in the worlds where I don't try, (1) is not true (either false or nonsense), and in those worlds I can't open the lock.


Heath White said...

I tend to think that in cases like this, "can" means roughly "can intentionally." You can accidentally open the lock, but you can't intentionally open the lock, since you don't know what you are doing, in the relevant sense.

BTW, the different varieties of "can" have interesting consequences for moral principles like "ought implies can". There would be as many such principles as there are "can"s and some would be more plausible than others.

Andrew said...

There seems to be something in the general area of Aristotle's 1st and 2nd potentiality. Can I speak German? Yes on one level, and no on another.

Alexander R Pruss said...

Unsurprisingly it looks like the case has been discussed in the literature. See here for references.