The Puzzle
A nearly omniscient being is going to put money in two boxes on Wednesday. No one will touch the boxes after that. Suppose you KNOW that the nearly omniscient being is 99% accurate in predicting whether the person on Thursday will open both boxes or just one box. The omniscient being will make a prediction on Wednesday and then place money in the boxes based on the prediction. Here’s is what the nearly omniscient being (hereafter NOB) will do.
BOX A
NOB will place $1,000 in Box A no matter what the being predicts.
BOX B
NOB will place $1,000,000 in Box B only if the being predicts that you will only take Box B. NOB will leave Box B Empty if NOB predics that you will take both.
It’s now Thursday. You may take either of the boxes you wish. You may also take both. What should you do?
Here are the camps divided into what people typically think you should do…
One boxers say…
Take Box B. Somehow, you’re decision will cost you $1,000,000 if you do anything else.
Two boxers say…
Come on…the money has already been put in the box. Surely, your decision NOW won’t affect anything. The practically rational thing to do is take both boxes…
My Favorite Response: I’m a ONE-BOXER
(I’m told William Lane Craig offers roughly this response…)
Here is my case for being a one-boxer. Our evidence RIGHT NOW might point to the fact that there is no way that our decision on Thursday affects the contents of the box. That’s why some have the intuition to be two boxers. However, if we imagine ourselves in a scenario where we know a person has been 99% accurate in predicting, we would have to ask what the best explanation of that phenomena is. You might think if you were in such a far fetched scenario that you would have some empirical evidence for a range of far fetched hypotheses that do entail your decision on Thursday affects the being decision on Wednesday. Here are two…
ETERNAL BEING – Perhaps the NOB has reliable immediate access to all truths past, present, and future. NOB could be outside of time or get reports from something that is outside of time. NOB (or at least NOB’s source of information) sees me make my choice on Thursday when I do it, and bases his decision on that choice.
TIME TRAVELING BEING – NOB can time travel and at least glance into the future. The being bases his Wednesday decision on what he sees me do on Thursday.
In both cases, we have conceivable scenarios in which what I decide on Thursday has some direct causal bearing on the past. I’m not saying that these are even remote possibilities for something that could happen in the actual world. All I am saying is the following conditional is true.
(A) If you were in a scenario where you KNEW a nearly omniscient being was 99% accurate in determining what your future actions would be, then the hypothesis that your actions in the future affect what the being does in the past would be a LIVE OPTION.
Objection One: These aren’t plausible theses.
Response. It might not be plausible to suppose these are actually true, but they would certainly be live options if you knew there was a being with this remarkable ability. When we imagine we are in possible scenarios where our knowledge is radically different from the actual world, we have to make sure shift what our total evidence is. Given that they are live options in the scenario, you shouldn’t act as if your decision on Thursday has no impact on the contents of the box.
Objection Two: there would be better explanations of the beings ability than that it was eternal or that it time traveled…The being could be a really good cognitive psychologist with accurate brain reading software, and it could be that whenever brains of a certain arrangement play the game, they choose both. Whenever they do not exhibit that arrangement they choose one box.
Response. There might be a better explanations, but the probability of TIME-TRAVELER and ETERNAL BEING alternatives might be sufficiently high that you shouldn’t rule them out as a possibility when making practical decisions…You may not be justified in believing that the being was ETERNAL or TIME-TRAVELLER. I’m just saying it would be enough of a live option and the consequences so severe that it’s not rational to risk that one of these explanations in not true.
I’m told…that this is roughly what William Lane Craig has said in response to the problem. I’m wondering if anyone is aware of any responses to this solution?
The response is no good. It just misses the point. Suppose we stipulate that you accept that it’s just the cognitive psychology thing going on. What should you do then? (This is the essential problem.) In other words, we could phrase the problem this way, and Craig’s “solution” will just have no application: What is the rational thing to do, for someone who believes that it’s just the cognitive psychology thing going on?
The problem still arises, since this person would continue to have a good argument that he should one-box, but also a good argument that he should two-box.
I’m not sure how this responds to my response.
My claim is that if they were in a situation where they KNEW that they were dealing with a NOB, then they would have good empirical evidence so that it would not rationally support accepting that it was merely the cognitive psychology thing that was going on.
We’d have to set up the case more carefully than merely stipulating that they accept that it is the cog sci thing. The fact that they accept that it’s the cog sci thing doesn’t give them good reason to make the two-box argument.
I’d have to see the scenario fleshed out so that they had STRONG evidence that it was the cog sci thing that also RULED OUT the non cog sci options before I’d be convinced that they had a good two-box argument.
Thanks for the comments…
(Do you have thoughts on how to flesh out the scenario?)
Hi, you’re welcome.
Oh, in that case, I think that you (or perhaps Craig) are not really a one-boxer. Your position, rather, is that Newcomb’s problem cannot really get started, because no one could (justifiably) believe that a mere cognitive scientist could predict that accurately. That’s how I would categorize this solution anyway.
A genuine one-boxer, in contrast, would say that, even if we grant that nothing but the cog sci thing is going on, there would still be no route to two-boxing!
This is why this thing about fleshing out the scenario struck me as being beside the point. “I’ll grant you that nothing but the cog sci thing is going on. It is still irrational to two-box.” That’s the real one-boxing position, so far as I understand Newcomb’s problem.