Stathis Papaioannou wrote:
>
>
> Brent Meeker writes:
>
>> >> > I make the claim that a rock can be conscious assuming that >>
>> computationalism > is true; it may not be true, in which case neither
>> >> a rock nor a computer may be > conscious. There is no natural
>> syntax >> or semantics for a computer telling us > what should count
>> as a "1" or >> a "0", what should count as a red perception, and > so
>> on. These >> things are determined by how the computer is designed to
>> interact > >> with its environment, whether that mean outputting the
>> sum of two >> numbers to > a screen or interacting with a human to
>> convince him that >> it is conscious. But what > if the environment is
>> made part of the >> computer? The constraint on meaning and > syntax
>> would then go, and >> the vibration of atoms in a rock could be
>> implementing > any >> computation, including any conscious
>> computation, if such there are.
>> >> > > John Searle, among others, believes this is absurd, and that >>
>> therefore it disproves > computationalism. Another approach is that it
>> >> shows that it is absurd that consciousness > supervenes on physical
>> >> activity of any sort, but we can keep computationalism and > drop
>> the >> physical supervenience criterion, as Bruno has.
>> >> > > Stathis Papaioannou
>> >>
>> >> I have a view that seems to me to be slightly different. >>
>> Consciousness requires interaction with an environment; consciousness
>> >> implicitly requires a distinction between "I" and "the world". So
>> >> when you attribute consciousness to a rock, incorporating "the
>> world" >> as part of the rock, while the remainder of the rock is
>> "conscious" >> that raises problems. We can say that this part of the
>> rock is >> conscious of that part; making some arbitrary division of
>> the rock. >> But then it's not conscious in/of our universe. > >
>> That's right: if it's conscious, then it's conscious in its own
>> isolated > virtual universe. It's another means to a many worlds theory.
>> > >> When you say there is no canonical syntax, which is what allows
>> >> anything to be a computation of anything else, I think that is >>
>> overstates the case. Suppose a particular pair of iron atoms in the
>> >> rock are magnetically aligned and the syntax counts that as "0"
>> while >> anti-aligned counts as "1". Then what computation is
>> implemented by >> "0000000..."? The arbitrariness of syntax
>> supposedly allows this to >> be translated into "27" or some other
>> number. But then the >> translation has to have all possible words in
>> it and the relational >> meanings of those words; including the words
>> for all the numbers in >> that world. This places a pretty strong
>> restriction on the size of >> the rock-world - there are only some
>> 10^25 atoms to do all this >> representing.
>> > > The rock can make numbers and universes as big as you want through
>> the > method of parallel computing. Suppose the rock had only a few
>> distinct > physical states. If we place no restriction on what these
>> states can > represent, then they can represent multiple binary
>> strings or finite > numbers or sentences or whatever - all in
>> parallel. Any serial > computation can be made up of multiple parallel
>> computations,
>> I find that doubtful - do you have a reference? Isn't it the
>> definition of "incompressible" computation that there is no way faster
>> than executing each step in sequence.
>
> I'm not referring to speed, just to doing it. For example, a serial
> stream of consciousness can be emulated by multiple shorter parallel
> streams; there is no way of knowing whether you're being run in serial,
> parallel, how fast the real world clock is running, etc.
I wasn't referring to speed either, but rather to needing to calculate states 1...n, before knowing how to calculate n+1... Are you thinking the computation is like a recording so you can play n+1... and then 1...n?
>> >and vice > versa. You can't say, aha, we've used that string for
>> "dog" so we can't > now use it for "cat", because who is going to
>> patrol the universe to > enforce this rule? This is what you are left
>> with if you eliminate the > constraint that the computation has to
>> interact with an external observer.
>>
>> I think my objection is different. You are assuming there is a "we"
>> too whom these strings represent something, but in the closed
>> rock-world there is no "we". The representation must be this
>> conscious-part of the rock representing that other-part of the rock.
>> But how can one part represent the other - by having a dictionary that
>> translates states of one into the other. But how is the dictionary
>> encoded? It seems there's an infinite regress.
>>
>> How is this infinite regress avoided in our world? By consciousness
>> not representing the rest of the world. The world is what it is and
>> representation is not essential. I suppose this is somewhat like
>> Peter's "primitive substance" whose only function is to distinguish
>> things that exist from their representation.
>
> Then there must be a way to distinguish true reality from virtual
> reality. This is something like Colin Hale's idea that the environment
> participates in the brain process that produces consciousness (which he
> aims to prove by experiment, a staggering achievent if he succeeds), so
> that an isolated virtual reality is impossible.
> Of course, in the evolution of brains and in the design of digital
> computers, the semantics is provided by the real world; similarly in
> human language, the semantics is provided by the language speakers.
> Without the provision of a semantic context it's gibberish. But there is
> a special quality that consciousness has which other systems lack: it is
> conscious in and of itself. Take the following steps:
>
> 1. A conscious computer is built to interact with its environment, say
> digging holes.
>
> 2. A co-processor is added (which need itself not be conscious) to
> provide the computer with a virtual environment in which to dig the
> holes, so that all the input that came from the real environment comes
> from this co-processor and all the motor output from the computer goes
> to the coprocessor.
> 3. The computer and coprocessor are connected together and sealed in a
> capsule with power input from a solar panel.
>
> 4. Everybody dies and there is no record left which will give a clue as
> to what the original computer was meant to do or how it was designed.
> I claim than at step 4 it is impossible to know that the computer was
> designed to dig holes, but that it will still be digging holes in its
> own mind (maybe the same holes after a while) until it breaks down. Do
> you think it stops being conscious at one of the steps 2 to 4?
>
> Stathis Papaioannou
I don't know. But I think I can sharpen the problem a little. In your example we know the computer is conscious in the beginning. How can we know this?...By its behavior. But then, when it's encapsulated with a simulated environment, it has no external, discernible behavior. So after everyone that knew the code dies, it's impossible to know that the computer is conscious of digging a hole. But is it?
I guess I'm not convinced that there's no way to discern from the outside that it is a simulation of something. This connects the problem to the question of whether any process (including a null one) executes any arbitrary computation. Obviously if the encapsulated system is executing the null process and that also counts as an execution of the simulated-hole-digging+consciousness, it will be impossible to discern that it is a conscious system. But we do have theories of information and entropy that should be applicable to any computation and would tell us whether anything is being computed. For example if we look at the vibrational states of the atoms in a rock we see that they are distributed according to a maximum entropy principle and hence there is no significant information flow from one part of the rock to another.
In the thought experiment you propose, we would observe information flow between parts of the system and an accumulation of memory in the conscious part. Whether we could discern what was being simulated is an open question. It seems very hard, but I don't see how to prove it is impossible. It seems to me that it is hard on the same order as discovering/inventing the laws of nature.
Brent Meeker
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "Everything List" group.
To post to this group, send email to everything-list.domain.name.hidden
To unsubscribe from this group, send email to everything-list-unsubscribe.domain.name.hidden
For more options, visit this group at
http://groups.google.com/group/everything-list?hl=en
-~----------~----~----~----~------~----~------~--~---
Received on Tue Jan 16 2007 - 17:36:24 PST