IUBio Biosequences .. Software .. Molbio soft .. Network News .. FTP

About building CONSCIOUS COMPUTERS... [open letter to Marvin Minsky]

CYBERYOGI =CO= Windler windle_c at informatik.fh-hamburg.de
Thu Nov 6 13:04:51 EST 1997

For the newsgroup replyers:

Please don't remove any newsgroups from the header of your reply as long you
don't explicitely answer only about one particular newsgroup.I have wisely 
though about which newsgroups I posted this to and I am not able to read all
the entire newsgroups to those I posted this article seperately,though I won't
be able to find the answeres when someone removes some newsgroups from his re-

This is an open letter to the well known artificial intelligence and brain re-
searcher Marvin Minsky.

If someone doesn't know who he is or what "Extropianism" means,one can find
lots of material about this by asking AltaVista or some other search engines.

Marvin Minsky's homepage is <http://www.ai.mit.edu/people/minsky/minsky.html>,
e-mail: <minsky at media.mit.edu>.

One of the most important newsgroups I intended to post this to is "alt.extro-
pians".Unfortionately the German news host I use <news.uni-hamburg.de> has no
access to it.Please inform me whether this article has been forwarded to that
newsgroup by other servers or not,and please forward it there if it hasn't 
reached it yet.

This article has also been posted to

alt.extropians (not arrived?),

but my news host seems to ignore articles posted to too many newsgoups at 
once,so I had to post multiple copies of this text to different newsgroups.


When you REPLY,ALWAYS add the newsgroup <alt.consciousness> to your reply, 
otherwise I won't find it!

Dear Marvin Minsky,

Like you I am very interested in brain research and conscious computers.

But in your theories about the nature of the mind it is a basical misconcep-
tion to believe that software and consciousness would be identical and though 
that it would be possible to transfer a conscious mind from a brain into a 
computer while keeping it still conscious,simply by exactly simulating the in-
ner working of that brain by evaluating digital equations.

I may appear as a heretic to you,but PLEASE read this e-mail to the end and
don't delete it unread - remember that some of your theories appear hereti-
cally to many other people,too.

Software and consciousness are not identical;they are just interacting with 
eachother within the brain.
When you write a book,you already transfer software(information) from your 
brain into the book,but you don't automatically become this book.And also when
you would have enough time and ability to write down everything you know,you 
feel and you wish into a giant book,you still wouldn't be this book.And even 
when you had the power to command a sort of slave or machine to behave from 
now on as exactly as possible according to all the rules and the entire know-
ledge you wrote into that book,you still wouldn't become the slave with the 
book and you wouldn't percept and feel what he does.And writing computer pro-
gramms is basically nothing else than writing books.Though also if you had the
technology to scan the structure of a conscious brain into binary data,atom 
for atom,particle for particle,and feed it into a brain simulating computer 
program and start the program,you still wouldn't become the program more than 
you become any books you write.mav1.txt

This is not a gradual problem of the quantity or complexity of any software 
taken from a brain and stored elsewhere,but an essencial problem of the nature
of consciousness itself.
The underlying assumption of your behavioristic view of the world is that eve-
rything that behaves equally would be the same.Though you conclude that when a
digitally simulated neuron behaves like a biological neuron,than any circuit 
of digital neurons must do the same and though be the same as its biological 
equivalent and while a brain consists of neurons,everything it does must be 
transferable into a computer program.But the premiss of this conclusion is
wrong.Every simulation is just a model.And any model never does excactly the
same thing as the real object(otherwise it would be the object itself) but 
just something equivalent.Because models are designed to serve a certain pur-
pose,they were proved to do the right thing to serve this purpose,but nothing
else.Though the digital models of neurons have been proved to be able to simu-
late some sorts of learning.But in what case a behavior of a real object is 
provable to be comparable with a model?! - Its only in the case where it re-
acts deterministically.A biological neuron has an operating point(excitation) 
where it will definetly begin to "fire" a signal and another one where it de-
finetly won't fire.These 2 signal levels have a deterministic behavior,as mo-
delled in neuronal computer programs.But what is between these 2 defined le-
vels?! There the behavior of a biological neuron is non-deterministic and 
though it is impossible to be modelled digitally in a way to do "the same" as
the real neuron.Nobody can explain why a biological neuron in this case de-
cides to fire in a certain moment.White(i.e. official) science though uses 
its priest handicraft of statistics to claim it would be just "random" and 
though far of any real interest.The model programmer then inserts a pseudo 
random number generator into the programm and claims the model must be still 
correct,because in the deterministic(the "important") cases it does the same 
and the simulated neuron networks still learns something and though it can't 
be wrong.

In 1948 the Great Master Norbert Wiener asked the heretical question:
"Is the man a message?!" 

More information about the Neur-sci mailing list

Send comments to us at biosci-help [At] net.bio.net