Re: COUNTERFACTUALS

From: Marchal <marchal.domain.name.hidden>
Date: Tue Jun 29 11:59:24 1999

GSLevy wrote:

>I think the requirement for computationalism for the existence of
>consciousness is a red herring. The real requirement is rationality of
>consciousness as well as anthropic and rational constraints in the flow of
>consecutive events. Computationalism and physical laws are derived from
>these
>anthropic and rational constraints. (Maybe this is just a question of
>terminology, some of you may equate computationalism with rationality)

Even without the comp hypothesis humans are (at least) Universal Turing
Machines. With comp we can look at things with a *Universal Turing
Machine - tropic* constraint. We can introduce rationnality by
restricting these machines with self-referential correctness, for
exemple. And we can use logic
and computer science to study the possible discourse of these machines.

Nevertheless, I'm not sure rationality is necessary for consciousness.
Well, it depends on the levels.

>GSLevy:
>Similarly recording the activities of neurons and then playing them back
>results in no consciousness at playback time but consciousness certainly
>exists at recording time.

You say that, I guess, because the "activities of the neuron" are
counterfactually correct (sensible to environment change). And the
playback is only accidentally correct.
The problem is the following one. As Maudlin has shown, you can
transform an accidentally correct computation into a counterfactually
correct one WITHOUT adding any physical activity. (And this is what I do
with the filmed graph argument, the *crackpot proof* in Mallah's term).
This make the attribution of consciousness to the "physical recording"
problematic.

>GSL:
>Now lets make it more interesting and say that the playback machine is
>implemented simply by rewinding the universe to some past state and
> allowing
>the same events to happen again.

By taking the rewinding of the entire universe you got the situation of
the dreaming brain, or any closed universal machine, and with comp this
change nothing. But here you say:

>then in my opinion the two
>consciousness are actually one and the same because they view the world in
>exactly the same way with the same rationality.

And then you quote Hal and Jacques M Mallah:

>Hal:
>>(My solution is, as I said earlier, that the question isn't meaningful,
>>because it is at best another iteration of an already-produced
>>calculation,
>>and it doesn't matter if a conscious calculation is instantiated multiple
>>times.)
>
>Jacques:
>>I obviously reject that.

And you say:
>And this is the point where the idea of measure comes in.

Well. I agree with you (if I understand you). In fact I agree with
Hal when he says <<that the question isn't meaningful, because it is at
best another iteration of an already-produced calculation>>
This is without doubt very important. It helps to explain the role of the
depth of the computation (in Bennett sense). But I disagree with Hal when
he adds:
<< and it doesn't matter if a conscious calculation is instantiated
multiple times>>.

In fact, it will matter in the case those computations can potentially
diverge or (I still don't know) actually will diverge (in the UD*). In
such a way a
*relative* measure will appears (and be completley defined in the limit).
And consciousness depends here and now on this limit, because, as you say:

>Consciousness is playback independent and is also time independent. You
>could
>implement it on a time share computer and it would not know the difference
>even if the active time intervals last one microsecond every billion years.
>As long as the rational link between the time segments is maintained.

But you should realize at this stage that consciousness
is no more linkable to the actual physical activities of the neurons.
Quite the contrary, it is the physical activities of the neurons (and of
the stars BTW) wich emerges from the possible experience of consciousness.
The physical emerge from the computationnal, as kind of sharable dreams
between rational (sure!) machines.

>This however is only my opinion. How to prove this and make scientific
>predictions as Jacques demands can only be done if one is brave enough and
>willing to go through Tegmark's experiment. Eventually we will all have a
>chance to do so and maybe Jacques will then be convinced :-). However if he
>is right we won't be around to change our mind :-(

Another way to make this more "scientific" is to take Church's thesis
seriously, to defined consciousness by something like the truth the
machine can know (infer?) and cannot prove (a little child of Godelian
Consistency), and then to use computer science to isolate and make
precise the relative measure from which we can derive qualitative aspects
of the physical world, like the MW aspect. And then compare with nature.
Immortality/MW are only one consequence among others of the hypothesis
that we are Turing emulable.

Bruno.
Received on Tue Jun 29 1999 - 11:59:24 PDT