Proof of Immortality III

[Jabba]

Dave,
- I'm just saying that OOFLam refers to everyone's sense of self -- but, it implies that whoever has an immaterial sense of self is wrong...

Then the likelihood of a self given OOFLam should be calculated the same way as the likelihood of Mount Rainier.

Dave,
- I see your point, but this is one of the wheres where probability can become so confusing...

- If OOFLam stipulated material selves, who would argue?
- But since
1) it doesn’t, and...

I thought it did:

Dave,
- Mmm.
- My basic claim has been that the hypothesis that "we each" have but one, finite, life to live is wrong. "We each" refers to our personal "selves."
- And, one problem we're having in our discussion is that we seem to be referring to different concepts -- one of those concepts involves what appears to be immateriality, the other apparently does not. My claim at least included the apparently immaterial self.

- Anyway, in a sense, I'm claiming that if an immaterial self is reasonably possible, it must be a reality, as my current existence is so unlikely if there is no immaterial self.
- Consequently, I need to define "reasonably possible." In this case, the prior probability of immaterial selves simply needs to be greater than the likelihood of my current existence minus the existence of immaterial selves. And that likelihood is unimaginably small. So, if that likelihood is not unimaginably small, the posterior probability of OOFLam is.

 

[Loss Leader]

Dave,
- Mmm.
<snip>.

Everyone knows your argument. You don't need to restate it. You cannot show that a thing is possible by saying something else is unlikely. You actually need some EVIDENCE. You haven't provided that.

Is the rarity of a royal flush a reason to bet on the zzibblety-shmedlin of elves?

 

[caveman1917]

@caveman1917:

I'm inclined to agree with JayUtah that there are qualitative differences between deductive arguments and probabilistic ones. For example, if hypothesis H implies that observation X is impossible, and X is observed, then H is false with certainty. However, if hypothesis H implies only that observation X is improbable, and X is observed, then not only does this not imply that H is false, it doesn't even imply that H is probably false. So here we have an argument of a particular from that is valid in a purely deductive context not in a probabilistic context. I'd call that a qualitative difference.

It is valid in a probabilistic context. P(H|X) -> 0 as P(NOT X|H) -> 1. All deductive arguments are probabilistic arguments. And even if you'd specifically define "probabilistic arguments" as "not deductive arguments" then there would still be (for every possible deductive argument) an uncountable number of counterexamples to JayUtah's claim.

 

[godless dave]

Dave,
- Mmm.
- My basic claim has been that the hypothesis that "we each" have but one, finite, life to live is wrong. "We each" refers to our personal "selves."
- And, one problem we're having in our discussion is that we seem to be referring to different concepts -- one of those concepts involves what appears to be immateriality, the other apparently does not. My claim at least included the apparently immaterial self.

- Anyway, in a sense, I'm claiming that if an immaterial self is reasonably possible, it must be a reality, as my current existence is so unlikely if there is no immaterial self.

Why would your current existence be unlikely if there is no immaterial self? If the self is entirely material, then wouldn't you calculate the likelihood of a particular self existing the same way you calculate the likelihood of Mount Rainier existing in its current form, or any other material phenomenon?

 

[jt512]

jt,
- I see your point -- we have no idea how often X occurs in general, so "hardly ever" doesn't really mean anything about the posterior probability of H...
- I've been claiming that the likelihood of X being unimaginably small -- given H -- has negative mathematical implications re H. At this point, I should probably change that to an indefinite mathematical implication re the posterior probability of H. The other variables in the formula determine the direction of the implication.- My ultimate claim is that the value of the different variables in this case "add up to" an unimaginably small posterior probability of H.

- How's that?

Yes, the posterior odds of a hypothesis are the product of two factors: (1) the (marginal) likelihood ratio, ie, the ratio of the probability of the observation under H to the probability of the observation under the complement of H, and (2) the prior odds of H. Just knowing that the observation is improbable under H doesn't imply that H itself is improbable. It could be that the observation is even more improbable under ~H, or that the prior odds of H is too low to be overcome by the observation.

BTW, the only element of Bayes' Theorem that you get to "make up" is the prior, because the prior can be your own subjective belief about the hypothesis before taking the observation into account. It is normal for different people to have different priors. However, in a serious application of Bayes' Theorem, the other elements, ie, the likelihoods, would be objective values given the two competing hypotheses, or models. Any two people should agree on the likelihoods, because they are completely determined by the observation and the models.

 

[caveman1917]

What the hell is all the giant red text for?

Good question. I just interpret it as a request for decent standards to be applied to the claim expressed as such.

 

[JayUtah]

I need do no such thing.

You refuse to show relevance. That is how we distinguish whether you have a material contribution, as opposed to your typical irrelevant and pointless badgering.

*plonk*

 

[JayUtah]

...an uncountable number of counterexamples to JayUtah's claim.

...none of which approach the probabilistic argument in question.

 

[JayUtah]

Anyway, in a sense, I'm claiming that if an immaterial self is reasonably possible, it must be a reality, as my current existence is so unlikely if there is no immaterial self.

No. This is just assuming your desired conclusion. That's circular reasoning.

You want to show something is reasonably possible because you believe something else is vanishing improbable. That's a false dilemma.

Restating your argument does not fix these fatal flaws.

 

[caveman1917]

You refuse to show relevance. That is how we distinguish whether you have a material contribution, as opposed to your typical irrelevant and pointless badgering.

*plonk*

Lol

In no way, under any circumstances, by any means, can a probabilistic argument be proof of an allegation of fact except for the uncountable number of ways, circumstances and means by which it can.

ftfy

 

[caveman1917]

...none of which approach the probabilistic argument in question.

Still having trouble with this I see:

A: No X are Y.
B: All Y are X. NOT(No X are Y).
A: Prove "This specific X is Y".

Can you articulate the error A is making here?

 

[JayUtah]

]BTW, the only element of Bayes' Theorem that you get to "make up" is the prior, because the prior can be your own subjective belief about the hypothesis before taking the observation into account. It is normal for different people to have different priors.

To incorporate this into my real-world example of spam filtration, the prior I cited was my belief in the hypothesis H ("The message is spam") based on worldwide rates. Instead I might suppose that a small regional ISP receives less spam than large international ISP. My prior probability might be smaller than the one suggested by global data. That constitutes my belief that a message is spam before conducting any tests. A different operator may legitimately believe in a different value depending on his circumstances.

 

[JayUtah]

Anyway, in a sense, I'm claiming that if an immaterial self is reasonably possible, it must be a reality, as my current existence is so unlikely if there is no immaterial self.

No. You have given no reason to believe that your current existence is vanishing improbable if there is no material self. You keep stumbling over the Mt. Ranier analogy as if to illustrate this fundamental error. This is why your critics keep returning to it, and they will likely keep doing so until you see why the analogy works.

Nor have you shown that any notion of an immaterial self is reasonably possible.

Consequently, I need to define "reasonably possible."

No, you can just use the normal definition of "reasonable," which requires an appeal to reason. You have used typical fringe-argument equivocation to try to convert abstract possibility to reasonable possibility. Abstract possibility arises out of the impropriety in declaring something impossible in a dearth of evidence. Naturally your critics will not prematurely assert as fact the non-existence of a soul, merely cite the lack of evidence that there is or needs to be one. Fringe claimants use this as an excuse to argue the abstract possibility of their claim. The disqualification of abstract possibility is why an argument from ignorance is almost always a fallacy.

Skeptics acknowledge that there may someday arise evidence for a soul. But there currently is none. You are trying to force your critics to choose between a closed-mindedness they do not exhibit and some sort of concession that your belief is true. You have only abstract possibility -- the lack of provable impossibility. If you want to show that a soul is reasonably possible, you have the burden to show actual reasons.

In this case, the prior probability of immaterial selves simply needs to be greater than the likelihood of my current existence minus the existence of immaterial selves. And that likelihood is unimaginably small.

No, you may not cobble up a private definition of "reasonable" out of more pseudo-statistical handwaving, especially when it's just a statement -- in statistical terms -- of your persistently circular argument.

 

[I Am The Scum]

All deductive arguments are probabilistic arguments.

I beg your pardon? I would agree only in the sense that 1 and 0 are probabilities, but that's typically not what we mean with the term.

 

[Jabba]

jt,
- I see your point -- we have no idea how often X occurs in general, so "hardly ever" doesn't really mean anything about the posterior probability of H...
- I've been claiming that the likelihood of X being unimaginably small -- given H -- has negative mathematical implications re H. At this point, I should probably change that to an indefinite mathematical implication re the posterior probability of H. The other variables in the formula determine the direction of the implication.
- My ultimate claim is that the value of the different variables in this case "add up to" an unimaginably small posterior probability of H. - How's that?

Yes, the posterior odds of a hypothesis are the product of two factors: (1) the (marginal) likelihood ratio, ie, the ratio of the probability of the observation under H to the probability of the observation under the complement of H, and (2) the prior odds of H. Just knowing that the observation is improbable under H doesn't imply that H itself is improbable. It could be that the observation is even more improbable under ~H, or that the prior odds of H is too low to be overcome by the observation.

BTW, the only element of Bayes' Theorem that you get to "make up" is the prior, because the prior can be your own subjective belief about the hypothesis before taking the observation into account. It is normal for different people to have different priors. However, in a serious application of Bayes' Theorem, the other elements, ie, the likelihoods, would be objective values given the two competing hypotheses, or models. Any two people should agree on the likelihoods, because they are completely determined by the observation and the models.

jt,
- What are your thoughts about my ultimate claim?

 

[Loss Leader]

I beg your pardon? I would agree only in the sense that 1 and 0 are probabilities, but that's typically not what we mean with the term.

That was my take as well. It seems to be a mix up in definitions. If you consider "Definitely" and "Definitely Not" as probabilities, then deduction can be called probabilistic. Otherwise, a person cannot deduce probability greater than zero or less than 1 merely from the fact that something did or did not happen.

Knowing that the damn Yankees won the penant doesn't give us any information as to how likely their win was. We'd need to know how many teams they played, their pitching staff, injury reports and much more about what was happening Before they won.

Jabba has only one data point - he exists. He can't deduce a probability from that.

 

[Loss Leader]

jt,
- What are your thoughts about my ultimate claim?

He just wrote it in the post you quote. It's nonsense.

 

[jond]

jt,
- What are your thoughts about my ultimate claim?

Jabba: why do you think an immaterial soul (for which there is no evidence, and further would have to physically interact with a material brain) is more likely than the existence of human beings whose brains produce the sense of self?

A while back I posted about this and you ignored it. Why? Could it be that you have no way of pulling numbers out of your nether regions that could possibly support this?

 

[JoeMorgue]

That was my take as well. It seems to be a mix up in definitions. If you consider "Definitely" and "Definitely Not" as probabilities, then deduction can be called probabilistic. Otherwise, a person cannot deduce probability greater than zero or less than 1 merely from the fact that something did or did not happen.

It's yet another variation on the "You can't know everything, therefore everything is fair game" argument, just with mathematics.

 

[caveman1917]

I beg your pardon? I would agree only in the sense that 1 and 0 are probabilities, but that's typically not what we mean with the term.

As I stated, if you use probabilistic logic in which the probabilities of the propositions are either 0 or 1 (a perfectly valid probabilistic argument) then you retrieve the inference rules of classical deductive logic. Every deductive argument is formally equivalent to a probabilistic argument. I've not seen any definition of a probabilistic logic which explicitly excludes arguments in which all probabilities are either 0 or 1.

But as I also stated, even if you explicitly define "probabilistic argument" as "at least one probability which is not 0 or 1" - which seems like special pleading to me - then there are still an uncountable number of counterexamples to JayUtah's claim. Namely those probabilistic arguments in which all probabilities are within epsilon of either 0 or 1, where epsilon is any non-zero real number such that you consider the allegation of fact to be proven if it is within epsilon of having probability 1. Such epsilon must exist since even using deductive logic you can never be exactly sure of the premises you use (see my example of the customer in the spam mail example having a non-zero probability of being wrong).