I The Actual Sleeping beauty Problem

  • Thread starter Thread starter JeffJo
  • Start date Start date
  • Tags Tags
    Beauty
Click For Summary
The Sleeping Beauty problem, originating from Arnold Zuboff's work, involves a subject (SB) who is awakened either once or twice based on a coin toss, raising the question of how SB should assess the probability of the coin landing on heads upon awakening. Adam Elga later simplified the problem, proposing a two-day scenario with awakenings contingent on the coin's outcome, leading to a debate on SB's credence regarding the coin's state. The discussion highlights that while some argue SB has no new information upon awakening, the introduction of a second coin (Dime) in an alternative model clarifies that SB can update her credence to 1/3 for heads based on the possible states of the coins. Critics of the Halfer perspective assert that the probability distribution remains consistent, as SB's knowledge of the awakening does not negate the relevance of the coin's outcomes. Ultimately, the problem illustrates the complexities of belief and probability in scenarios involving memory and information.
  • #31
Moes said:
the question she is asked is what is her credence of heads conditioned on the fact that the coin was flipped and that she is now ( at a time where it’s possible for her to be thinking about this) awake.
Yes; I did not include "the coin was flipped" in the conditions before, but adding it to the conditions is fine and does not change anything I've said.

Moes said:
Whether the coin landed heads or tails she would now be awake
But whether the coin landed heads or tails does change the number of times she is awakened. And since when she is awakened she does not know how many times she will be or has been awakened, she has to include all the possible awakenings when she computes the conditional probability of heads given that she has been awakened.

Moes said:
I don’t see how this condition could change the probability.
See above.
 
Physics news on Phys.org
  • #32
Moes said:
I don’t see how this condition could change the probability.
The phrase "change the probability" is misleading. No probability "changes". There are simply different conditional probabilities based on different conditions. Conditioning on Sleeping Beauty being awakened and the coin having been flipped is different from conditioning on just the coin having been flipped.
 
  • Like
Likes sysprog and Dale
  • #33
Moes said:
I believe I agree with the references you gave
Then the betting follows.

There has been some comments that the betting doesn’t cause the credence, which is true: the betting measures the credence. And since Sleeping Beauty is described as rational she won’t take bets that she expects to lose and it is that rationality which then constrains her credence. Because she is rational we can infer her credence from the expected value of a wager.
 
Last edited:
  • Like
Likes PeterDonis
  • #34
PeterDonis said:
That's not the relevant "prior" because Sleeping Beauty already knows on Sunday Night that she will not be awakened at all in the (H, H) case, so that case drops out of any calculation of probabilities conditioned on her being awakened. And she knows that the question she is going to be asked is about a probability conditioned on her being awakened (since she knows she will only get asked the question on being awakened).
One of the interesting aspects of Probability is that there can be many ways to create a sample space. If I roll two six-sided dice, I can create a sample space with 36 pair-combinations, or 11 sums. It's just easier to assign probabilities to the pair-combinations.

You can claim that "a relevant prior" is the state of the coin(s) that SB can predict, on Sunday Night, will exist when she awakes. That's completely impractical, since it is also the posterior she needs to use. So of course there is no "new information."

But there is no such thing as "the relevant sample space." You are using circular reasoning, by claiming I can only use the same space for both prior and posterior. And quite wrong.

My prior sample space, which I find to be more relevant than yours, is {(H,H), (H,T), (T,H), (H,H)}. The probability for each one-outcome event is 1/4.
  • This is what the experiment moderator sees when he decides whether or not to wake SB.
  • The resultant event space exists whether or not he wakes her.
  • It continues to exist whether or not he wakes her.
  • By waking SB, he does not create a sample space. He makes her a conditional observer of the one that exists.
  • When she becomes an observer, she can deduce that (H,H) is no longer a possibility.
  • THIS IS NEW INFORMATION.
Please note that SB knows, before being put to sleep, how each part of this process can go. She knows the coins will be tossed. Se knows that there are four possible combinations the moderator will look at. She knows what will happen in each contingency. She knows she will be asked a question in only three of the four.
PeterDonis said:
But she already knows that this will be the case on Sunday Night.
Yep. What has that to do with whether, in my prior (which has the advantage that the probabilities are not controversial), (HH) was a possibility and now it isn't?

PeterDonis said:
And she also knows that she will not be awakened at all in this state, so she knows this state is irrelevant for computing probabilities conditional on her being awakened, which are the relevant probabilities for questions she will be asked when she is awakened.
But the point you seem to overlook is that the state exists whether or not she is made an observer.

So change it. Wake her before and after you reverse the dime (or on both Monday and Tuesday in the original). If both coins are showing Heads in my version (or the coin landed Heads and it is Tuesday), ask her whether she likes Gone With the Wind more than The Wizard of Oz. Otherwise, ask her for her credence in Heads. But now she has what I hope you would call "new information." The prior is the one I called relevant, and the posterior is the one you said was the relevant prior.

She can say her confidence in Heads is 1/3.

But now, why does it matter if you lied, and had no intention of waking her if you already knew her preference about movies? The state that determines what type of - or even if - an observer she becomes is a valid prior.

PeterDonis said:
No, I'm asking you, because you made a claim about what a halfer would say. If you're not going to back up that claim with details, you should not have made it in the first place.
And I'm saying that, just like I can't understand why you want to call what is clearly as posterior "the relevant prior," I have never seen it justified. If you want to see an example of it, google for "double halfers."
 
Last edited:
  • Like
Likes Dale
  • #35
I am also not convinced by the “no new information” claims. I would like to see an actual calculation from information theory to support that claim. I did such a calculation previously and got 1 bit of information, which seems too high. So I am not confident in my result, but I really don’t think that the 0 information claim is convincing.
 
  • #36
Dale said:
I am also not convinced by the “no new information” claims. I would like to see an actual calculation from information theory to support that claim. I did such a calculation previously and got 1 bit of information, which seems too high. So I am not confident in my result, but I really don’t think that the 0 information claim is convincing.
The problem with the Sleeping Beauty Problem, as it is usually presented, is that the breaks the paradigm under which other probability problems are solved. The point is, you have to develop new rules to cover it, and the controversy always revolves around those rules. That's what I tried to address with my version (it doesn't add new information, btw, it adds the same information Elga added to it but in a more conventional way).

I haven't looked at your proof (I find applying information theory to be a bit of a kludge anyway), but based on past (and present, unfortunately) experience I suspect that those who don't like your answer will invent a different way to apply it that gets their answer.
 
  • #37
JeffJo said:
The point is, you have to develop new rules to cover it
I don’t think that is the case. The standard rules applied in a straightforward manner lead to a well defined result. It isn’t that new rules are needed, it is just that some people trust their intuition more than the rules.
 
  • #38
PeterDonis said:
That's not the relevant "prior" because Sleeping Beauty already knows on Sunday Night that she will not be awakened
It's not the prior that you claim is relevant. It is a prior, and has relevancy somewhere. You just won't say what you think makes it relevant, or not, to this problem. You just insist on this one.

Do you have such criteria? What are they? They seem to be based on what she knows on Sunday night, which will prove why you are wrong. (I'll include the days, which actually are not relevant, to aid exposition).

On Sunday Night, SB learns:
  1. She will be put to sleep soon.
  2. Then two coins, a Quarter and a Dime, will be flipped and then left untouched for 24 hours.
    1. At this point, Sunday SB knows what the probability distribution will be. Each of the outcomes in {(H,H), (H,T), (T,H), (T,T)} will have a probability of 1/4. Call this distribution A.
    2. This is a valid prior for something.
  3. On Monday Morning, the coins will be examined.
    1. Sunday SB knows that distribution A will still apply at this time
  4. If either coin is showing Tails, SB will be wakened. If that happens:
    1. Sunday SB knows this result depends on the state of the coins. She knows that distribution A will be used to determine if it happens, and knows that it will not happen if the state is (H,H).
    2. Monday SB will be asked for her credence in Quarter=Heads. Let's delay talking about her answer, because she doesn't know she is Monday SB.
    3. Monday SB will be put back to sleep, with amnesia.
  5. The Dime will be turned over on Monday Night, and then left untouched for 24 hours.
    1. Even though the state changes, distribution A still applies to the new state.
  6. On Tuesday Morning, the coins will be examined.
    1. Sunday SB knows that distribution A will still apply at this time.
  7. If either coin is showing Tails, SB will be wakened. If that happens:
    1. Sunday SB knows this result depends on the state of the coins. She knows that distribution A will be used to determine if it happens, and knows that it will not happen if the state is (H,H).
    2. Tuesday SB will be asked for her credence in Quarter=Heads. Again, delay talking about her answer.
    3. Tuesday SB probably should be put back to sleep, with amnesia, but this does not affect anything except symmetry.
  8. On Wednesday, SB is wakened and the experiment ends.
Based on this, Sunday SB knows that:
  • Both Awake Monday SB and Awake Tuesday SB will have the same knowledge.
  • In fact, it is the same knowledge Sunday SB has about their situation.
  • So I can just call her Awake SB, and base any answer on this shared knowledge.
  • Awake SB knows that her existence, as an Awake SB, depended on distribution A.
  • But Awake SB also knows that (HH) can no longer be part of the distribution. THIS IS NEW INFORMATION.
  • Awake SB updates her distribution to {(H,T), (T,H), (T,T)}, with each outcome having probability 1/3.
Of course, all of this was pretty trivial from the beginning. But I'd be interested in knowing why you think Awake SB does not know both distributions, A and B, or why her existence does not depend on the change from A to B.
 
Last edited:
  • #39
JeffJo said:
You are using circular reasoning, by claiming I can only use the same space for both prior and posterior.
No, I'm not. I'm simply saying that the information you claim is "new" is not, since Sleeping Beauty already has it on Sunday, before the experiment is started. All the reasoning you go through to purportedly claim that "the sample space changes" or "new information is gained", or however you want to phrase it, is reasoning Sleeping Beauty herself can go through on Sunday. Basically you are arbitrarily saying Sleeping Beauty cannot reason from all the information she has on Sunday, but has to wait until she is awakened. Sorry, not buying it.

JeffJo said:
Awake SB also knows that (HH) can no longer be part of the distribution. THIS IS NEW INFORMATION.
No, it isn't, because Sunday SB already knows that Awake SB will only occur if (H, H) does not occur, and Sunday SB already knows that she will only be asked questions as Awake SB. So Sunday SB is already in the same logical position as Awake SB.
 
  • #40
Dale said:
I don’t think that is the case. The standard rules applied in a straightforward manner lead to a well defined result. It isn’t that new rules are needed, it is just that some people trust their intuition more than the rules.
There actually is a way to use conventional probability for Sleeping Beauty. It does require using a modified a definition that, while it applies to conventional problems, adds an unorthodoxy that is needed to include the separate the possible awakenings, as Elga did. And it is what halfers specifically deny is valid.

This isn't going to be rigorous, so leave me some slack. An outcome in a discrete, finite sample space is "trivial" if it (well, the event containing just it) has zero probability. I mean, you can include 3.5 as an outcome for rolling a six-sided die, but it is trivial. Outcomes with positive probability are non-trivial. A sample space is non-trivial if it is a valid sample space (mutually exclusive and collectively exhaustive) that contains only non-trivial outcomes.

"New information" traditionally occurs when you learn something that makes some outcomes in a non-trivial sample space become trivial. To update the probabilities, you make a new ("posterior") sample space out non-trivial outcomes, and divide each prior probability by the sum of all of them. While worded in an unorthodox way, I don't think anybody will disagree so far.

In the traditional Sleeping Beauty Problem, a non-trivial sample space on Sunday Night is {H,T}, with probabilities 1/2 for both. But Sunday SB also see that there are two days, which are needed to describe the state at any point in time (this is regardless of whether she is awake). So there are four possible outcome descriptions {H&Mon, H&Tue, T&Mon, T&Tue}. These are used in the Sunday Night sample space, since H&Mon and H&Tue represent the same future, and so are not mutually exclusive.

But I could use {H&Mon, T&Mon} as a non-trivial sample space, each with having probability 1/2.

But when SB is awake during the experiment, she needs an outcome description that includes the day. H&Tue is not needed, because it is trivial. But T&Tue, which in the prior sample space was the same outcome as T&Mon and had probability 1/2, is now disjoint. The space {H&Mon, T&Mon} is no longer a valid sample space, so it is also not a non-trivial sample space. This was exactly my definition for new information. We need to use {H&Mon, T&Mon, T&Tue}. Each had probability 1/2 in the prior. Their sum is 3/2, and we update each to (1/2)/(3/2)=1/3.

The objection halfers have - that Peter insisted I identify earlier while he wouldn't identify his reasons - is that T&Mon and T&Tue still represent the same outcome to Awake SB. Their logic is that the day of the week is an "indexical" quality that can't be used as an outcome description. I still don't understand this logic, which is why I suggested he ask a halfer. Better yet, google for "SLEEPING BEAUTY INDEXICAL."
But when SB is awake,
 
  • Skeptical
Likes weirdoguy
  • #41
JeffJo said:
all of this was pretty trivial from the beginning.
I agree that getting the correct answer is trivial, but we disagree on what the correct explanation of the correct answer is, so I think you need to re-think this claim.

JeffJo said:
I'd be interested in knowing why you think Awake SB does not know both distributions, A and B, or why her existence does not depend on the change from A to B.
I have never made either of these claims. You are attacking a straw man.

Awake SB knows the same things that Sunday SB knows. Both of their knowledge includes both "distributions". But both of their knowledge also includes the fact that only one distribution is relevant, because both of their knowledge includes the fact that awakening only occurs if (H, H) is not the state of the coins.
 
  • #42
JeffJo said:
This isn't going to be rigorous, so leave me some slack.
For someone who claims to have a wonderful new way of resolving all controversy about the Sleeping Beauty problem, this request is, to say the least, dubious.
 
  • #43
JeffJo said:
H&Mon and H&Tue represent the same future
No, they don't, because as the problem is specified, H&Mon is possible but H&Tue is impossible, and is already known to be so on Sunday. So including H&Tue in the sample space is already wrong on Sunday; it's already known to be "trivial" (in your idiosyncratic terminology) then.
 
  • #44
JeffJo said:
The objection halfers have - that Peter insisted I identify earlier while he wouldn't identify his reasons - is that T&Mon and T&Tue still represent the same outcome to Awake SB.
Please give a reference for this claim. Telling me to Google is not enough. If you are going to make a claim about what halfers say, you need to actually give a reference where a halfer is saying it. Those are the rules you signed up to when you joined this forum. Either follow them or your thread will be closed.
 
  • #46
After moderator review, the thread will remain closed.
 

Similar threads

Replies
126
Views
8K
  • · Replies 57 ·
2
Replies
57
Views
6K
  • Poll Poll
  • · Replies 601 ·
21
Replies
601
Views
40K
  • · Replies 16 ·
Replies
16
Views
3K
  • · Replies 2 ·
Replies
2
Views
5K
  • · Replies 9 ·
Replies
9
Views
2K
  • · Replies 2 ·
Replies
2
Views
3K
  • · Replies 2 ·
Replies
2
Views
2K
  • · Replies 7 ·
Replies
7
Views
2K
  • · Replies 10 ·
Replies
10
Views
7K