Newsgroups: comp.ai.alife,comp.ai.philosophy,comp.ai,alt.consciousness
Path: cantaloupe.srv.cs.cmu.edu!rochester!galileo.cc.rochester.edu!prodigal.psych.rochester.edu!stevens
From: stevens@prodigal.psych.rochester.edu (Greg Stevens)
Subject: Re: Thought Question
Message-ID: <1995Jan6.172134.149@galileo.cc.rochester.edu>
Sender: news@galileo.cc.rochester.edu
Nntp-Posting-Host: prodigal.psych.rochester.edu
Organization: University of Rochester - Rochester, New York
References: <3dfhkq$gov@news.worldlink.com> <Shanks.179.2EFFA0C1@us.net><3dprpk$6a@bbs.pnl.gov> <34@reservoir.win-uk.net> <3ebc5a$bc8@ixnews2.ix.netcom.com> <3ec1o8$sno@agate.berkeley.edu> <d-sears1.2.000D2126@uiuc.edu> <3edkgf$mmb@agate.berkeley.edu> <1995Jan4.171133.19446@galileo.cc.rochester.edu> <3eg1c1$oli@agate.berkeley.edu> <1995Jan5.133627.1006@galileo.cc.rochester.edu> <3ehbfp$8oh@agate.berkeley.edu> <1995Jan5.191042.14803@galileo.cc.rochester.edu> <3ej99b$6do@agate.berkeley.edu>
Date: Fri, 6 Jan 95 17:21:34 GMT
Lines: 69
Xref: glinda.oz.cs.cmu.edu comp.ai.alife:1675 comp.ai.philosophy:24390 comp.ai:26282

In <3ej99b$6do@agate.berkeley.edu> <jerrybro@uclink2.berkeley.edu> writes:
>stevens@prodigal.psych.rochester.edu (Greg Stevens) wrote:

>> This is the case with the computation of a brain driving.  It receives
>> input at a certain rate and needs to process it at a certain rate.
>> If it gets behind, the output (your behavior as a driver) changes the
>> input (where you see you are on the road) and if you get behind and
>> can't react in time, you crash.  But it's all still computation.

>I suspect that your notion of "computation" is, like Neil Rickert's,
>a lot broader than what I had intended by the term.  You introduce
>"probabilistic algorithms", and if you can introduce those, you
>can eventually introduce just about anything I should think.  

Like religion and magic?  No, I think probabilistic algorithms are perfectly
valid as computational constructs.

>To use
>Neil Rickert's example, if we consider a car brake as a computational
>device, then we are for one thing departing the realm of Turing
>machines, since what makes a car brake a car brake is that it
>actually brakes the car's motion (this is the "output"), whereas
>insofar as the brake is a kind of Turing machine (if it indeed
>can be interpreted as one), then the output being braking
>the car, as opposed to printing numbers on a tape, is not a factor.

If you symbolically represent the car itself, and even the surroundings,
as sequences of numbers on tape, a T-machine braking system can brake a
T-machine car.  You just have to be extansive and consistent with your
model and interpretation.  A simulated tornado DOES blow down houses --
the simulated ones.

>... Nor do treatments of the
>brain as a Turing machine, since this ignores the form of the
>input and output. 

Well, I'd agree with going against Turing machines, but saying that something
can not be treated as a Turing machine is not saying that it is not
computational.  Computer science has evolved away from that stance, as far
as I have seen.

Secondly, I would not object to the Turing machine for the same reason as you.
The form of the input is irrelevant, really, because it can all be construed
as information and symbolically represented.  If you are into the whole
symbolic representation IDEA with computation to begin with, why be half-
hearted about it?  Symbolically represent the light waves the hit our eyes,
the floor we walk on.  The actual substance doesn't matter.


>But to give my two cents, I really don't see it
>as particularly useful, though perhaps it is exciting, for
>people who work with computers as they exist today, and who
>understand them as they exist today, to attempt to annex such
>a broad range of things to the notion of "computation" that
>the applicability of today's knowledge to all that range of
>things.

But I don't think most of the researchers who use the term today use
it that way.  Perhaps this is a problem with popularizing cognitive science
and AI in the media?  Most scientists I know and know of think of computation
as a formal thing, as Neil Rickert has described.  Once, perhaps, "lay-people"
hear that the brain is performing "computations" they think of their Apple
Macintoshes and imagine people like Data wandering around saying "I do not
understand."  But that is simply not what is MEANT within the field.

Greg Stevens

stevens@prodigal.psych.rochester.edu

