Newsgroups: comp.ai.alife
Path: cantaloupe.srv.cs.cmu.edu!das-news2.harvard.edu!fas-news.harvard.edu!newspump.wustl.edu!news.ecn.bgu.edu!psuvax1!news.pop.psu.edu!news.cac.psu.edu!howland.reston.ans.net!spool.mu.edu!umn.edu!news
From: "Robert Warren Subiaga, Jr." <subia001@maroon.tc.umn.edu>
Subject: Re: Thought Question
To: comp.ai.alife
Message-ID: <86044.subia001@maroon.tc.umn.edu>
X-Minuet-Version: Minuet1.0_Beta_15
Sender: news@news.cis.umn.edu (Usenet News Administration)
Nntp-Posting-Host: dialup-5-132.gw.umn.edu
X-Popmail-Charset: English
Organization: University of Minnesota, Twin Cities
Date: Thu, 12 Jan 1995 03:09:08 GMT
Lines: 91


Newsgroups: comp.ai.alife,comp.ai.philosophy,comp.ai,alt.consciousness
Subject: Re: Thought Question

>.In article <3ec1o8$sno@agate.berkeley.edu>, <jerrybro@uclink2.berkeley.edu> writes:
>> In <34@reservoir.win-uk.net> shane@reservoir.win-uk.net (Shane McKee) 
>> writes: 
>> 
>> >I still haven't had anybody (including Roger Penrose) give me a
>> >good reason why the human brain can't be regarded as a computer.
>> >OK, so the mind sometimes doesn't work algorithmically, but its
>> >constituent molecules do obey known (or at least simulable)
>> >physical laws. Why -can't- intelligence be an emergent property of
>> >an underlying algorithmic process involving these constituents?
>> 
>> You're confusing two separate issues.  One is, whether the brain
>> can be regarded as a computer.  The other is, whether the brain
>> can be simulated on a computer (because physical reality can).
>> An atomic explosion can be simulated on a computer, but it does
>> not follow from this that the atomic explosion can be regarded
>> as a computer.
>> 
>David Sumpter  <djts@dcs.ed.ac.uk> wrote
>Fri, 6 Jan 1995 11:20:48 GMT
>
>--But Penrose does actually argue that a brain can NOT be simulated by a 
>computer, letalone the computer be considered a brain.

This is true, but the specific issue raised by Penrose IS an extension of 
the idea that simulation is not reality--Penrose only goes beyond to give a 
physicists's concrete reason WHY this is so.  In particular, Penrose does 
not start out from the point of dealing with the GENERIC issue of the 
physiology of Life or even nervous systems. Rather, he starts 
out at the upper level, in "consciousness." 

It is obvious from the way logicians, mathematicians and scientists are 
able to formalize Godel Undecidability that they are not vulnerable 
TO that Undecidability. (That is, unless they simply got very 
lucky, in which case they would still be vulnerable to the "lethal" effects 
of some other Godel argument). Computers (at least those that operate at 
their fundamental levels by ANY discrete algorithmic procedure) ARE--by 
DEFINITION--vulnerable to the Godel argument. Ergo, human beings have a 
fundamental level of "reasoning" more expansive than computers.

In simplified shorthand, Penrose's point is analagous to:
                NO human is bulletproof (by definition)
                        [NO computer is immune to Godel Undecidability]
                SUPERMAN is bulletproof (by empirical observation)
                        [Since human beings can see Godel Undecidability
                         --and they needed to to formulate it!--humans
                         ARE immune to Godel Undecidability]
                -->SUPERMAN is NOT merely human
                        [Human beings are not mere computers]

Other livings things also do not seem to "hang up" on any specific 
situation at all that might suggest Godel Undecidability, making us 
skeptical that a Godel argument exists for them either--and increasing the 
plausibility that our immunity to Godel undecidability is something 
physiological that we have in common with other living things. 

More expansively, PHYSICAL dynamics (other than those that have been 
designed as Turing machines) do not "hang up" either, suggesting that the 
fundamental nature of physical laws is also non-algorithmic; an APPARENT 
ability at present to "algorithmize" them (is that a word?)--which by the 
way is a presently UNVERIFIABLE assumption--is likely also a mere illusion.

**************************************************************************

In fact, the issue of a car's brakes being a "computer" raised in other 
"Thought Questions" in this newsgroup is interesting. Assuming the Turing 
Machine paradigm of ultimate algorithmic computability applies to physics, 
we WOULD call the car's brakes a "computer"--but the Turing machine model 
works both ways. By its extension it would demand that any OTHER Turing 
machine's function should able to be simulated on those brakes (if we 
ignore time constraints). You could run your MS-Word program if your Intel 
chip crashes! Most of us would surmise that a car's brakes are too "simple" 
a computer for this to hold true, however, even in PRINCIPLE. (Well, all of 
us who haven't converted to a MacGyver-cult.)  

Similarly, the discrete, "algorithmic" level of "computation" is the 
poorer, simpler level of explanation--IN PRINCIPLE, not just practice--that 
cannot encompass the broader domain of either modern mathematics or actual 
physical phenomena. That we are astonished at how a car's brakes by far are 
the BEST device to stop the vehicle does NOT mean that "brakes" are THE 
ubiquitous tools all other tools should be modeled after. In the same 
way, the profound ability of algorithmic computation to perform CERTAIN 
functions says nothing about the method's capailities across ALL possible 
functions. (Anybody remember Renaldo Nehemiah's lack of dexterity and short 
NFL career?--Being the fastest runner does not make one the best 
ATHLETE in any sport but track. In the realm of mentality computers make 
great sprinters and distance-runners--but really shitty decathletes.)
