From newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!uunet!cis.ohio-state.edu!rutgers!micro-heart-of-gold.mit.edu!news.media.mit.edu!minsky Tue Jun  9 10:05:53 EDT 1992
Article 5998 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!helios.physics.utoronto.ca!news-server.csri.toronto.edu!rpi!usc!cs.utexas.edu!uunet!cis.ohio-state.edu!rutgers!micro-heart-of-gold.mit.edu!news.media.mit.edu!minsky
>From: minsky@media.mit.edu (Marvin Minsky)
Newsgroups: comp.ai.philosophy
Subject: Re: Homunculus and the witch's brew
Keywords: computation, transduction, homunculus, sensorimotor physiology
Message-ID: <1992Jun1.054526.8610@news.media.mit.edu>
Date: 1 Jun 92 05:45:26 GMT
References: <l2iea9INN44p@exodus.Eng.Sun.COM> <1992May31.212826.1778@news.media.mit.edu> <1992Jun1.020633.15541@Princeton.EDU>
Sender: news@news.media.mit.edu (USENET News System)
Organization: MIT Media Laboratory
Lines: 32

In article <1992Jun1.020633.15541@Princeton.EDU> harnad@phoenix.Princeton.EDU (Stevan Harnad) writes:
>In article <1992May31.212826.1778@news.media.mit.edu> minsky@media.mit.edu (Marvin Minsky) writes:
>>
>>I have the same problem.  SH: are you using the term "transducer" in a
>>non-standard way?  The brain has lots of internal information and
>>processing capability, whereas the term transducer is used normally to
>>refer to a (many-one) transformation; if it has much internal state,
>>then this term is inappropriate because your assertion sounds totally
>>Homuncular.  And your thesis makes no sense that.  way.  Perhaps you
>>should withdraw this term before it is too late -- or else present a
>>definition that does not require reading many difficult references.
>>Otherwise, we'll have to refer to H(T)HH -- Harnad's (Total)
>>Homuncular Hypothesis.
>
>Marvin, I'm not an engineer, so if I've picked a technical term that is
>at odds with my intended meaning, I'd be happy to susbstitute the
>correct one for it (what is the correct one?). I had no idea that a
>many-to-one mapping was criterial for a transducer. I would have wanted
>a term that leaves room for 1:1 analog transformations and through-put
>too, as the dictionary definitions below do. Physiologists, I believe,

Omigosh.  Many-one is a mathematical term that includes 1-1 as a special
case; the point is that it excludes anything else being added.  But
the brain does a great deal of internal computation, even when the
inputs are cut off, so the transducer idea seems eccentric.  Perhaps
you're saying something like the brain is a finite state machine?  Or
that evrything it does is determined by its input history?  Or that it
produces nothing that is not a direct consequence of its history?  Or
what?  I can't tell -- except that "transducer" is surely the wrong
word.  

.


