Sunday, February 21, 2010

Prospects for prospective possibilism

It's difficult to come up with simple objections to Zimmerman. He's quite good at covering all the angles. I have to resort to a complicated objection. Here goes.

The prospectivist thinks that you ought to do what is prospectively best. You ought to A iff it is the option that is prospectively best (i.e., maximizes expectable value).

The possibilist thinks that you ought to do what is the best you could do rather than do what will be best. You ought to do A iff it is the best thing that could happen, not the best thing that will happen depending upon what you decide.

What happens when you combine the prospectivist view and the possibilist view? Of course, you get prospective possibilism. That's the obvious part. The tricky part is that I think you get problems. As a prospectivist, you are supposed to cash out 'S at t1 is obligated to A at t2' in terms of the expected value of A-ing at t2 in light of the evidence you have at t1. What happens when you have reason to believe at t1 that at t2 you won't do what you'd have to do to carry out successfully the plan you put into action at t1? That depends upon why you won't do it. If you won't do it because you freely choose not to, that doesn't matter. If you won't do it because you'll freely do something that prevents you from doing it or knowing how to do it, that doesn't matter. If you won't do it because you'll find that you cannot do it but not because of some free action that has this as the predictable consequence, that does matter.

Suppose you have to decide whether to build your mine in Whiskey River or Dry Gulch. You know that in Whiskey River there's whiskey and there's nothing to drink in Dry Gulch. There are snakes in Dry Gulch and you've been assured that as a result, some of the miners will have to suffer from very painful snake bites. Nothing life threatening, mind you, just every so often one will be bitten and need medical attention. There's nothing like that in Whiskey River. Other things equal, I think it's reasonable to prefer to set up your mine in Whiskey River. And, all else is equal. Except this. You know that people who go to Whiskey River drink more than they do in Dry Gulch. You know that there's a very good chance that you'll start to drink in Whiskey River and this will affect your job. You'll start to forget important details about the mining operation that could lead to a loss of life. So, you know there's a very high chance that by setting things up in Whiskey River you'll encounter a three-option mine shaft case where you fail to know where the miners are because you didn't know something you were supposed to. You know you'll encounter similar situations in Dry Gulf. Indeed, you'll face them with the same frequency, but as you won't have taken to drinking more than you ought, you'll always know where the miners are. (Okay, maybe not always but more often than you would in Whiskey River.)

The worry is this. It seems intuitive, to me at least, that knowing what you know now, you know you should set up shop in Dry Gulch. However, we have something like a three-option case. The options:

1. Set up shop in WR and refrain from drinking. No snakes. Workers are as safe as they'd be otherwise in DG.
2. Set up shop in DG. Snakes. Apart from the snakes, workers are as safe as they'd otherwise be if you didn't take to drink and set up shop in WR.
3. Set up shop in WR and take to drinking. No snakes. Workers are put at great risk thanks to your ignorance which was the predictable side-effect of your freely taking to drink.

You could freely bring it about that 1 happens, but you know that's it's very unlikely that you will and so quite likely that if you choose to go to WR, you'll bring it about that 3 happens rather than 1 or 2. To my mind, that's like throwing away information in the 3 option mine shaft case. I take it that a prospective actualist would say that you ought to go for 2 and that seems right to me. I take it that the prospective possibilist (or, one prospective possibilist) would say go for 1. That seems wrong. To me, the intuition here is worth taking seriously and it seems like a bad idea for someone who makes such hay about mine shaft cases in beating up on objectivism about 'ought' to not deal with this sort of intuition as well.

Yes, actualism is supposed to be quite bad. I agree. I'm not urging anyone to go actualist. I'm concerned about the combination of prospectivism and possibilism. Independently, they look quite good but in combination, I'm not entirely sure I like what you get.


Richard said...

(i) "you should set up shop in Dry Gulch"

(ii) "Independently [of prospectivism], [possibilism] looks quite good"

How do you reconcile these two claims? It seems we can naturally extend the problem scenario beyond prospectivism -- just suppose that there's not just "a very good chance that you'll start to drink in Whiskey River", but that this will be the actual consequence of your setting up in WR. It seems pretty plausible, then, that you should set up in DG, no?

Such cases serve to show that possibilism is nuts. What does prospectivism have to do with it?

Clayton said...

For (ii), I was thinking that there's something right about Z's attack on actualism in his 1996 book where he didn't try to combine possibilism and prospectivism. That's what made possibilism look good (by comparison). However, since his version of prospectivism seems to give different roles to outcomes depending upon how those outcomes depend upon later decisions of yours, I worry that there's an unnatural fit and this is particularly worrisome given that Z's motivation for possibilism seems to be cases that elicit the same sort of intuition I'm trying to elicit here. (My guess is that you agree that there's something strange about saying you ought to go to WR?)