- Contemporary messages sorted: [ by date ] [ by thread ] [ by subject ] [ by author ] [ by messages with attachments ]

From: Jesse Mazer <lasermazer.domain.name.hidden>

Date: Mon, 20 Jun 2005 21:26:33 -0400

Stathis Papaioannou wrote:

*>
*

*>Hi Jesse,
*

*>
*

*>I was still trying to put some sort of reply together to your last post,
*

*>but I think your water analogy is making me more rather than less confused
*

*>as to your actual position on these issues, which is obviously something
*

*>you have thought deeply about. With the puzzle in this thread, I was hoping
*

*>that it would be clear that the subject in the room *has* to experience the
*

*>light changing colour every 10 minutes, and therefore can draw no
*

*>conclusion about which state is the high measure one. It seems that many on
*

*>this list would indeed say that running a mind in parallel increases its
*

*>measure, and some would say (eg. Saibal Mitra in recent discussions - I
*

*>still have to get back to you too, Saibal) that the subject would therefore
*

*>find himself continually cycling in the 10^100 group.
*

*>
*

*>To summarise my position, it is this: the measure of an observer moment is
*

*>relevant when a given observer is contemplating what will happen next. If
*

*>there are 2N successor OM's where he will experience A and 3N successor
*

*>OM's where he will experience B, then he can assume Pr(A)=0.4 and
*

*>Pr(B)=0.6. Only the ratio matters. Moreover, the ratio/ relative measure
*

*>can only be of relevance at a particular time point, when considering the
*

*>immediate future. To say that an individual will not live to 5000 years
*

*>even though there exist OM's where he is this age, because his measure is
*

*>much higher when he is under 100 years of age, makes no sense to me.
*

*>
*

*>Now, minimising acronym use, could you explain what your understanding is
*

*>of how measure changes with number of copies of an OM which are
*

*>instantiated, and if it doesn't, then how does it change, and when you use
*

*>it in calculating how someone's life will go from OM to OM.
*

Well, see my last response to Hal Finney; I don't have a definite answer to

how the number of physical copies (as viewed from a third-person persective)

affects measure, because my idea is just to start with the set of all

possible distinct observer-moments and try to derive a unique

self-consistent absolute and relative measure on them, and thenmy hope is

that somehow you could derive the appearance of a regular physical universe

from the measure on observer-moments. The measure on the set of all unique

observer-moments is really the fundamental thing, physical notions like

"number of copies" are secondary. But I have speculated on the

"anticipatory" idea where multiple copies affects your conditional

probabilities to the extend that the copies are likely to diverge in the

future; so in your example, as long as those 10^100 copies are running in

isolated virtual environments and following completely deterministic rules,

they won't diverge, so my speculation is that the absolute and relative

measures would not be affected in any way by this.

*>Also, you have talked about memory loss, perhaps even complete memory loss,
*

*>while still being you: in what sense are you still you? Isn't that like
*

*>saying I am the reincarnation of Alexander the Great or something?
*

If we have a formal theory of the "similarity" between observer-moments, it

seems unlikely to me that there would be a sharp cutoff where I could have a

nonzero similarity to observer-moment A but 0 similarity to observer-moment

B which differs only in the tiniest way from A. Certainly, the similarity of

my current OM to an OM of Alexander the Great would seem to be greater than

the similarity between my current OM and an OM of a bat or an alien or

something; even if the details of our life histories are different, many of

our implicit memories (how various sensations feel, how to do things like

running or catching without thinking about it) would probably be pretty

close. So the difference in similarity between my current moment and a

Jesse-moment one hour from now vs. my current moment and an

Alexander-the-Great moment would seem to be a matter of degree. Also, even

if the probability of my "next" observer-moment being an Alexander-the-Great

moment would be very tiny because of the lack of sufficient similarity, we

could probably imagine a continuous transformation between my current OM and

any other OM, such that each OM in the sequence would be pretty similar to

the next one even though the endpoints are quite dissimilar.

There is the question of what it is, exactly, that's supposed to be moving

between OMs, and whether this introduces some sort of fundamental duality

into my picture of reality. I think of position as a kind of panpsychic

monism rather than a type of dualism, but I don't really have a perfect

philosophical justification for this; maybe it would be better to say the

fundamental entities in my picture of reality are *strings* of OMs rather

than individual OMs, which sort of removes the need to picture any sort of

entity "moving" between OMs. But I don't know if that's really satisfactory.

Anyway, it seems like any "first-person" TOE potentially suffers from the

charge of being dualist rather than monist, because you can always point to

the difference between mathematical descriptions of OMs (assuming you have a

mathematical theory of consciousness/OMs) which can be understood in

third-person terms, and the qualia or "what-it-is-likeness" of those OMs.

For example, Chalmers ( http://consc.net/chalmers/ ) sometimes says that he

is a "dualist" for exactly this reason, although I think he also sometimes

calls it a "dual-aspect theory".

In a way I don't think such philosophical definition-wrangling is all that

important, if one could actually come up with a mathematical theory that

uniquely assigned each OM an absolute probability and also gave conditional

probabilities for transitions between OMs, and this in turn allowed you to

make predictions about what type of "physical universe" a typical observer

would experience as well as the subjective probabilities he'd experience in

various copying experiments, then we could all test such a theory against

our own experience without having to worry too much about the philosophical

meaning of it all--a sort of "shut up and calculate" approach to a

first-person TOE.

*>You say we need a theory of consciousness to understand these things, but
*

*>don't you mean a theory of personal identity?
*

To me it's all part of the same thing--if you're going to "take

consciousness seriously" as Chalmers says, then one aspect of consciousness

I think you need to take seriously (and formalize in terms of a mathematical

theory) is the feeling of a *stream* of consciousness, rather than just an

isolated observer-moment with memories of the past and anticipation of the

future. But it might be possible to take seriously other aspects of

consciousness, but say this feeling of time passing is a sort of illusion--I

discussed this a long time ago in my '3 possible views of "consciousness"'

thread at http://tinyurl.com/74qb4

Jesse

Received on Mon Jun 20 2005 - 21:27:46 PDT

Date: Mon, 20 Jun 2005 21:26:33 -0400

Stathis Papaioannou wrote:

Well, see my last response to Hal Finney; I don't have a definite answer to

how the number of physical copies (as viewed from a third-person persective)

affects measure, because my idea is just to start with the set of all

possible distinct observer-moments and try to derive a unique

self-consistent absolute and relative measure on them, and thenmy hope is

that somehow you could derive the appearance of a regular physical universe

from the measure on observer-moments. The measure on the set of all unique

observer-moments is really the fundamental thing, physical notions like

"number of copies" are secondary. But I have speculated on the

"anticipatory" idea where multiple copies affects your conditional

probabilities to the extend that the copies are likely to diverge in the

future; so in your example, as long as those 10^100 copies are running in

isolated virtual environments and following completely deterministic rules,

they won't diverge, so my speculation is that the absolute and relative

measures would not be affected in any way by this.

If we have a formal theory of the "similarity" between observer-moments, it

seems unlikely to me that there would be a sharp cutoff where I could have a

nonzero similarity to observer-moment A but 0 similarity to observer-moment

B which differs only in the tiniest way from A. Certainly, the similarity of

my current OM to an OM of Alexander the Great would seem to be greater than

the similarity between my current OM and an OM of a bat or an alien or

something; even if the details of our life histories are different, many of

our implicit memories (how various sensations feel, how to do things like

running or catching without thinking about it) would probably be pretty

close. So the difference in similarity between my current moment and a

Jesse-moment one hour from now vs. my current moment and an

Alexander-the-Great moment would seem to be a matter of degree. Also, even

if the probability of my "next" observer-moment being an Alexander-the-Great

moment would be very tiny because of the lack of sufficient similarity, we

could probably imagine a continuous transformation between my current OM and

any other OM, such that each OM in the sequence would be pretty similar to

the next one even though the endpoints are quite dissimilar.

There is the question of what it is, exactly, that's supposed to be moving

between OMs, and whether this introduces some sort of fundamental duality

into my picture of reality. I think of position as a kind of panpsychic

monism rather than a type of dualism, but I don't really have a perfect

philosophical justification for this; maybe it would be better to say the

fundamental entities in my picture of reality are *strings* of OMs rather

than individual OMs, which sort of removes the need to picture any sort of

entity "moving" between OMs. But I don't know if that's really satisfactory.

Anyway, it seems like any "first-person" TOE potentially suffers from the

charge of being dualist rather than monist, because you can always point to

the difference between mathematical descriptions of OMs (assuming you have a

mathematical theory of consciousness/OMs) which can be understood in

third-person terms, and the qualia or "what-it-is-likeness" of those OMs.

For example, Chalmers ( http://consc.net/chalmers/ ) sometimes says that he

is a "dualist" for exactly this reason, although I think he also sometimes

calls it a "dual-aspect theory".

In a way I don't think such philosophical definition-wrangling is all that

important, if one could actually come up with a mathematical theory that

uniquely assigned each OM an absolute probability and also gave conditional

probabilities for transitions between OMs, and this in turn allowed you to

make predictions about what type of "physical universe" a typical observer

would experience as well as the subjective probabilities he'd experience in

various copying experiments, then we could all test such a theory against

our own experience without having to worry too much about the philosophical

meaning of it all--a sort of "shut up and calculate" approach to a

first-person TOE.

To me it's all part of the same thing--if you're going to "take

consciousness seriously" as Chalmers says, then one aspect of consciousness

I think you need to take seriously (and formalize in terms of a mathematical

theory) is the feeling of a *stream* of consciousness, rather than just an

isolated observer-moment with memories of the past and anticipation of the

future. But it might be possible to take seriously other aspects of

consciousness, but say this feeling of time passing is a sort of illusion--I

discussed this a long time ago in my '3 possible views of "consciousness"'

thread at http://tinyurl.com/74qb4

Jesse

Received on Mon Jun 20 2005 - 21:27:46 PDT

*
This archive was generated by hypermail 2.3.0
: Fri Feb 16 2018 - 13:20:10 PST
*