Re: Statistical Pattern Recognition

<richardalanforrest@xxxxxxxxxxxxxx> wrote:

On Dec 28, 5:54 am, j.wilki...@xxxxxxxxx (John Wilkins) wrote:
rich hammett <bubbaric...@xxxxxxxxxxxx> wrote:
Minun olisi pitänyt tietää, olisi pitänyt tietää,
olisi pitänyt tietää KUKA SINÄ OLET, r norman:

William Calvin, being a neurophysiologist, is automatically a "Good
Guy" and therefore beyond criticism.

What we learn by trial and error is how to compute the movement of an
arm and hand so as to release a projectile with the proper initial
velocity (magnitude and direction) so as to strike a target after
traveling through a trajectory. That small brains can do so does not
mean it is not done.

I almost certainly misinterpreted your statement about computing
tides. That the tides rise and fall does not mean that the earth-moon
system is a computer computing when the tides should rise and fall.

I'm not sure I'm getting your exact definition of "compute." And
I think your sentence becomes an obvious truism if you replace "should"
with "do."

In fact, although I haven't read Wilkins' words very deeply, yet,
I'd say that we use simulations to _analyze_ a physical process.

I agree - here the simulations are processes occurring *outside* the
phenomena being analysed. The mistake I think is being made here is to
think that our analysis is at one with the physical process occurring
underneath the phenomena. It's a kind of anthropomorphising of nature.



But from both of you, I'd like some sort of moderately closed definition
of "computer" that includes most artifical digital and analog computing
machines, but excludes the human brain. Or, alternatively, a definition
that includes the modern 8086 von Neumann-type architecture, and excludes
the human brain. That one should be fairly easy.

A computer is a symbol manipulation process that closely instantiates a
Turing machine in its internal workings. That is to say, the physical
thing must *work* like a computer. It is my view that the brain only
partially does this and that, if connectionism is right, much of it is
rather unlike a Turing machine (although of course you can simulate a
connectionist system on a Turing machine, like any physical process).



I count this as part of my brain therapy, by the way.

Me, of my old age and general decrepitude.

To throw an uncontrite spanner into the works....

Roger Penrose, in "The Emperor's New Mind" argues that the human brain
does things which *cannot* be computed. Mind you, this was some time
ago, and I know that Wolfram has claimed that the functioning of the
brain can only be modeled by cellular automata. There is also, of
course, the school of thought that the whole universe is a computer
running a single program whose output is the whole universe from
beginning to end, but that strikes me as one of those pragmatically
useless ideas which lead nowhere.

I know a lot of people make this claim but I have never seen any
convincing argument for it. Usually someone rattles on about Gödel, but
it seems to me that all that is proven thereby is that for a given
formalisation, some accepted truths of mathematics are not provable.

I think that if there is order, a physical system can be computed. The
mind is an ordered activity of the brain, ergo it can be computed (to
some degree of precision, running up against a wall with quantum
physics).

Then there is the theory that the earth was constructed as computer to
find the ultimate question of life, the universe and everything to
which the answer is 42, which is no doubt where Sean's
superintelligent mice come in.

But with the mice you need a larger Hilbert manifold to do the
computing.

Oh, and having just finished the latest PTerry (6/10 - not one of his
better works), I am reminded of the wonderful device which was built
by the Treasury which modeled the flow of money in the economy as
water flowing through tubes controlled by valves.
http://en.wikipedia.org/wiki/MONIAC_Computer

Analogue computers are fun.

Indeed they are. And notice the reverse inference in that book, that if
you modify the analogue computer, you'll modify the thing it is
modelling? That's a case of the kind of error I think occurs in general
when thinking about simulations.
...
--
John S. Wilkins, Postdoctoral Research Fellow, Philosophy
University of Queensland - Blog: scienceblogs.com/evolvingthoughts
"He used... sarcasm. He knew all the tricks, dramatic irony, metaphor,
bathos, puns, parody, litotes and... satire. He was vicious."


--
John S. Wilkins, Postdoctoral Research Fellow, Philosophy
University of Queensland - Blog: scienceblogs.com/evolvingthoughts
"He used... sarcasm. He knew all the tricks, dramatic irony, metaphor,
bathos, puns, parody, litotes and... satire. He was vicious."

.

Relevant Pages

  • Re: OT: An end to this God debate
    ... To think that the human brain and mind are the be-all-end-all of the ... physics, poker, etc. ... There could be races of aliens in the universe ... These debates remind me of the "Got You Last" games played by ...
    (rec.gambling.poker)
  • OT: An end to this God debate
    ... To think that the human brain and mind are the be-all-end-all of the ... physics, poker, etc. ... There could be races of aliens in the universe ... be-all-end-all of perception is arrogant at best, and terribly foolish ...
    (rec.gambling.poker)
  • Re: Statistical Pattern Recognition
    ... connectionist system on a Turing machine, ... brain can only be modeled by cellular automata. ... the school of thought that the whole universe is a computer ...
    (talk.origins)
  • Re: Poll: Are PCs Turing Machines?
    ... unbounded number of digits. ... dependent on physics. ... Whether a PC simulates a Turing machine is a question unrelated to ... possible worlds because an infinite universe is physically possible, ...
    (sci.math)
  • Re: Statistical Pattern Recognition
    ... partially does this and that, if connectionism is right, much of it is ... connectionist system on a Turing machine, ... brain can only be modeled by cellular automata. ... the school of thought that the whole universe is a computer ...
    (talk.origins)