From: Matt Mahoney (matmahoney@yahoo.com)
Date: Wed Apr 02 2008 - 14:11:42 MDT
--- Lee Corbin <lcorbin@rawbw.com> wrote:
> Matt writes
>
> > Lee wrote:
> >
> >> Matt writes
> >>
> >> > Suppose you have a 3 way Turing test between a
> >> > human, a GLUT, and a human whose memory is
> >> > set to read-only at the start of the test
>
> I guess I didn't understand that part. I thought that
> you meant that for the duration of a test, e.g. between
> times t1 and t2, no *new* memories may be acquired.
> That sure *sounds* like what you meant! No?
That is what I mean. After the test, the person would have no memory of
taking part in it.
> >> > (of equivalently, whose mental state is reset after
> >> > each question). Is the read-only human conscious?
>
> So I responded
>
> >> Of course the read-only human is conscious.
> >> Suppose it's only Homer, and he's only reciting
> >> the Iliad around a campfire. We adjust his nervous
> >> system so that *during the narration*, on which
> >> he is totally focused, neither the campfire, nor
> >> any of the sounds, nor any other of his senses
> >> are leaving any record whatsoever in his memory.
> >> (Naturally we have to idealize this a bit in order
> >> to address, in principle, the answer to your
> >> idealized "read-only human" question.)
> >
> > That is not the same. He will still remember remembering.
>
> Your brief reply fails to specify where my example goes
> wrong. *During the test*, while Homer is reading, he is
> of course recalling his past life (he has to, for among other
> reasons, to be able to recite the Iliad). But he is forming
> no *new* memories, nor is he going to be able later to
> remember remembering, where the latter "remembering"
> specifies a process occuring between t1 and t2. That
> remembering did indeed happen, but later he won't
> remember that it did.
>
> In fact, during the narration he doesn't happen to recall that
> a moment earlier he had said such and such---by my
> stipulation, he is able nonetheless to recite the whole thing
> the same way that you, while reciting the alphabet, do not
> remember (unless you try) reciting the letter 'C' when you
> have got to 'Q'.
Normal memory is never read-only. I will still remember reciting the
alphabet. I will know that I said 'C' because if I had skipped it I would
have remembered doing so. Each time I recall a sequence like the alphabet or
a phone number, I remember that I recalled it and it also reinforces that
memory. What we call "conscious experience" is a recallable, continuous time
sequence of perceived events, both perceptions and thoughts.
Consider the case of Homer without a hippocampus who is able to recall the
Iliad word for word (which he learned before his surgery) but has no short
term memory. He will not remember anything that happened after his surgery
and will not form any new memories. After reciting the Iliad he will not
remember that he had just done so. Is he conscious?
> >> It may be easy for some to say that there is no such thing as
> consciousness,
> >> but someday we'll have much more thorough characterizations of it than
> >> we do now, and the argument that they're wrong will be all the more
> solid.
> >
> > If you define consciousness as a sequence S of algorithmically similar
> states
> > (K(S_n+1|S_n) = O(1)) then certainly it exists.
>
> First, I don't define consciousness as a sequence of states,
> similar or not. You could aim your question at Stathis, for
> example. Moreover, I don't follow your mathematical
> notation at all. Is the O supposed to be Landau's "big-oh"
> notation? (I gather that K is Kolmogorov complexity, though.)
That is what I meant. But choose your own definition if you prefer. If we
could agree on a definition then the rest would be easy.
-- Matt Mahoney, matmahoney@yahoo.com
This archive was generated by hypermail 2.1.5 : Wed Jul 17 2013 - 04:01:02 MDT