Conscious States vs. Conscious Computations
Jason
strings of data representative of a universe or of an observer's mind
are enough to create consciousness or whether it is the act of
computation that creates observer moments. Recently I had a thought
that both are required together for the following reason:
A given piece of data can represent an infinite number of different
things depending on the software that interprets it. What may be an
mp3 file to one program may look like snow to an image editor.
Therefore a bit-string only has meaning in the context of the program
operating upon it, without a corresponding program (i.e. physical
laws) associated with a state (state of the universe or observer's
mind) there is no consciousness. This means that a simple program
counting to infinity, although iterating over every possible bit-
string would not create universes or self-aware-substructures as the
software is too simplistic and the "meaning" of any one state is just
a counter's value.
Instead something like the Universal Dovetailer is required, in which
the states created have different meanings depending on program to
which it belongs. In some programs, states may be interrelated to
produce illusions of time to observers, observers exist in interactive
environments, etc. In other words an OM requires more than the data
describing the mind, it requires a specification of a state machine
and the state which corresponds to the OM.
Jason
Hal Finney
> A given piece of data can represent an infinite number of different
> things depending on the software that interprets it. What may be an
> mp3 file to one program may look like snow to an image editor.
I'm doubtful that you could find a string of any significant length which
both sounds like sensible music and looks like a realistic picture. I'm
even more doubtful that the enormous length of the data that would
represent the brain activity associated with an observer-moment could
be meaningfully interpreted as anything else.
My guess is that sufficiently long, meaningful data strings have
their meaning implicitly within themselves, because there is no
reasonable-length program that can interpret them as anything else.
Hal Finney
Jason Resch
Jason writes:
> A given piece of data can represent an infinite number of different
> things depending on the software that interprets it. What may be an
> mp3 file to one program may look like snow to an image editor.
I'm doubtful that you could find a string of any significant length which
both sounds like sensible music and looks like a realistic picture.
When I said snow I meant like snow on a TV (random garbage)
I'm
even more doubtful that the enormous length of the data that would
represent the brain activity associated with an observer-moment could
be meaningfully interpreted as anything else.
My guess is that sufficiently long, meaningful data strings have
their meaning implicitly within themselves, because there is no
reasonable-length program that can interpret them as anything else.
Hal Finney
I agree that the data encoding an Observer Moment contains many instances of self-relational and internally meaningful data structures, here is my scenario: Some piece of advanced technology maps out the neural network of one's brain, including which neurons are firing at the instance the brain was scanned and then saves it as a file. Does this file on the computer constitute an observer moment? Does duplicating this file increase that observer moment's measure? Or for it to constitute an observer does some software have to load the file and simulate future evolutions of brain states in a manner consistent with how a real brain would to create a valid observer moment?
If you believe the file alone creates an OM then a program which counted to infinity would indeed iterate over all possible observer moments, the problem is some program can interpret any data file as being any other observer moment given the interpretation software uses the appropriate "encryption" algorithm which could decrypt that data to be any other observer moment, which is why I see data alone as meaningless without the context of an algorithm that created/uses/processes it.
Jason
Vladimir Nesov
I'm not yet qualified to engage in in-depth discussion on this list, but re this point: what is 'reasonable-length'? Why is interpreter supposed to be limited? If it is, how should it be limited? If interpreter is just 'assumed' and not encoded in any form, can't it be an arbitrary thing, up to containing all the knowledge you need for any resulting interpretation?
--
Vladimir Nesov mailto:robo...@gmail.com
Russell Standish
>
> My guess is that sufficiently long, meaningful data strings have
> their meaning implicitly within themselves, because there is no
> reasonable-length program that can interpret them as anything else.
>
> Hal Finney
>
This is patently false, as otherwise unbreakable ciphers would not
exist. The "anything else", of course is junk.
However, it does seem to be true that not all strings are capable of
being interpreted as an observer moment of a conscious observer. This
is kind of a flip side of what you're saying. I'm not aware of this
being proven, though, or even the similar (and weaker under COMP)
statement that not all strings are capable of being interpreted as
self-describing Turing machine. Note that this feeds in Chalmers
conscious rocks argument.
Cheers
--
----------------------------------------------------------------------------
A/Prof Russell Standish Phone 0425 253119 (mobile)
Mathematics
UNSW SYDNEY 2052 hpc...@hpcoders.com.au
Australia http://www.hpcoders.com.au
----------------------------------------------------------------------------
Youness Ayaita
The first problem is to understand why and how observers interpret
data in a meaningful way despite of the fact that the data has no
unique meaning within itself.
On 26 Sep., 21:09, Jason <jasonre...@gmail.com> wrote:
> A given piece of data can represent an infinite number of different
> things depending on the software that interprets it. What may be an
> mp3 file to one program may look like snow to an image editor.
If we invited an inhabitant of a strange universe to our universe
(e.g. to an interuniversal conference), he would most probably
perceive nothing but random noise (if his senses allow him to perceive
anything at all). He would feel like the image editor confronted with
an mp3 file. Though, the fact that we being humans perceive something
useful is self-evident since we are a product of evolution within our
universe. Useful interpretation of the environment has been a
necessary condition for survival. The successful analogy between an
observer and a computer program shows that the process of observation
has a computational character: The observer 'calculates' a meaning for
his perception in a systematic way (which was elaborated
evolutionary). We can formalize this similar to Russell and introduce
the map from descriptions to meanings as a property of the observer.
The second problem you address in your message concerns the embedding
of the observer in the universe's description (you write of "self-
aware substructres"). You give a very nice example:
> Some piece of advanced technology maps out the neural network of
> one's brain, including which neurons are firing at the instance the brain
> was scanned and then saves it as a file. Does this file on the computer
> constitute an observer moment? Does duplicating this file increase that
> observer moment's measure? Or for it to constitute an observer does some
> software have to load the file and simulate future evolutions of brain
> states in a manner consistent with how a real brain would to create a valid
> observer moment?
Before I'm writing an uncompleted answer, I'd prefer to read what the
long-time participants (Russell, Bruno and others) are thinking about
this point.
Youness
John Mikes
*
Remember, when the human mind was a "steam-engine"? then later it was a "telephone-switchboard"? Now it is "computer" with the image in mind of that embryonic contraption IBM et al. fabricate in binary primitiveness. Would have Leibnitz, or Plato imagine a computer-based OM? of course not. Why do we assume that NOW we have reached the ultimate in omniscience? that our present toy is representing "all"? Do we have any criticism for the new gadget coming about in the 25th c.? Are we denying any advancement?
(This list 'humbly' agreed in views representing a century ahead, Bruno's: 2 centuries, - the reason why I suggested the 25th c. 'new gadget' which may be just as unforeseeable for people before its arrival as was a computer and its workings before Charles Babbage.)
*
Those 'strings' in a 'software' are our present limitations for (wider?) meanings that may go way beyond the 'perceived reality' of today. Which is itself partial and incomplete.
I appreciate the 'present level' inventiveness and the discussions about (logical?) incompleteness found in such, but always in mind that the 'position' is early 21st c. and prone to changing.
*
Jason wrote:
"A given piece of data can represent an infinite number of different
The unidentified 'information: "bit" in our binary machine, (0 or 1).
Strings identify it better, still applicable to any relation if in 'reasonable' length.
As Hal wrote:
their meaning implicitly within themselves, because there is no
reasonable-length program that can interpret them as anything else."
Infinite length strings maybe a solution, but maybe also an impractical cop-out.
John Mikes
Bruno Marchal
Le 27-sept.-07, à 02:09, Russell Standish wrote to Hal Finney:
> However, it does seem to be true that not all strings are capable of
> being interpreted as an observer moment of a conscious observer.
I still would appreciate what do you mean by a string being interpreted
as an observer moment of an observer, without comp.
Without comp, I don't find any clear definition of "interpretation"
which gives sense in such a context.
(Wait perhaps my comment of the post you have addressed to me ...).
Bruno
Russell Standish
information processing, which in turn means operating on strings
written in some alphabet (WLOG binary).
But such information processing needn't imply that only computable
functions are used. So COMP is not necessarily true.
Cheers
Bruno Marchal
Le 26-sept.-07, à 21:09, Jason a écrit :
> Instead something like the Universal Dovetailer is required, in which
> the states created have different meanings depending on program to
> which it belongs.
Yes, different meaning from the inside points of view, which defined
different measure on the relative set of possible computational
histories.
From the third person point of view we just cannot know which histories
support us, and to get the proportions right we have to bet we belong
at least on a continuum (from the 1-pov view).
Bruno