From newshub.ccs.yorku.ca!ists!torn.onet.on.ca!utgpu!news-server.csri.toronto.edu!rutgers!mcnc!aurs01!throop Mon Jun 15 16:04:36 EDT 1992
Article 6199 of comp.ai.philosophy:
Path: newshub.ccs.yorku.ca!ists!torn.onet.on.ca!utgpu!news-server.csri.toronto.edu!rutgers!mcnc!aurs01!throop
>From: throop@aurs01.UUCP (Wayne Throop)
Newsgroups: comp.ai.philosophy
Subject: Re: Transducers
Message-ID: <60799@aurs01.UUCP>
Date: 10 Jun 92 15:43:44 GMT
References: <1992Jun9.051649.9894@cs.ucf.edu>
	<BILL.92Jun8213911@ca3.nsma.arizona.edu>
	<1992Jun08.225734.32166@spss.com>
	<BILL.92Jun8150837@cortex.nsma.arizona.edu>
	<60795@aurs01.UUCP>
	<BILL.92Jun9131210@ca3.nsma.arizona.edu>
Sender: news@aurs01.UUCP
Lines: 87

>,>>> bill@nsma.arizona.edu (Bill Skaggs)
>> throop@aurs01.UUCP (Wayne Throop)

>>> A question such as, "Scrunch up the palm of your
>>> hand, and describe the folds that you see,"
>> "I regard that question as cheating. [..or..]
>> I can't answer that, my arms were blown off in a mining accident,
> This objection is very powerful but futile, because *in fact* there is
> nothing to prevent the judge from making a decision on the basis of
> such questions.  (Remember that the aim of the human participant is to
> win, not to be fair.)

Agreed.  Still, note that such questions can easily lead the judge to
judge *incorrectly* (as in the second case, a control who is forced to
respond that the control *can't* answer).

And even in the first case, it is pretty clear that questions about the
appearance of the testee itself are not kosher in the TT, since the
whole setup is arranged to keep such details from the judges.  If
moderated "correctly", I'd expect the moderator to censor the query,
just as the moderator should censor attempts by the judges to peek
behind the partitions.

With that said, lets agree for a moment with Bill that a clever TT
judge/questioner, even without recourse to asking testee-appearance
questions, might eventually foil most any non-humaniform testee (unless
the testee was perternaturally intelligent).

But this leads me to the question... why then the TTT?  We've just
shown that the TT is (in the limit) equivalent to it.

And on the other hand, does anybody really think that pressing
questions about immediate sensual experience and manipulation
capabilities is getting at the heart of the matter?  Again, I point out
to you that such a "clever" judge will often wind up ruling deaf, or
blind, or "smell-blinded", or "kinesthetically challenged" *human*
testees *non* human.  That is, such a judge will cause the TT testee to
WIN because the judge DOESN'T DISTINGUISH BETWEEN the TT and the control.

( Perhaps here is where such a TT test is different from a TTT test:
  The TTT test essentially rules out any controls with any sensory deficits
  (compared to some average or mean).  But this seems an awful lot like
  seeing a bunch of computers potentially being allowed into the
  intelligence club, and responding by raising the standards for computer
  applicants, but retaining the old standards for human applicants. )

( If you have read Sacks' "The Man Who Mistook His Wife for a Hat",
  consider how that man would fare even as a TTT testee.  His reactions
  to standard neurophysical tests were extraordinary, interesting,
  and thought-provoking.  "Seeing Voices" is also chock full of
  interesting observations about linguistic development as related
  sensual perception.  Foo, let's just say that nobody can go wrong 
  in reading ALL of Sacks' books. )

On the other other hand, I see no reason why a computer supplied
with (say) files of visual images to study is any different than a
robot with optical sensors.  Each would have the "sensation of sight"
in that they are grounding their analysis of the images in actual
"transduced input" from the real world.  Either analysis would BE a
transduction of input from the real world.  The only difference is that
the entire robot entity is present to capture the real-world
sensations, while the entire computer/sensor complex is not.  Why such
a difference should be relevant is beyond me.  (Or in other words, I
don't see why such a robot or computer would not BE a transducer.)

And whether the robot/computer difference is relevant or not, such a
computer could, it seems to me, potentially stand up against any
questioning probing sensual capabilities in visual and aural areas at
least, and probably do pretty well with touch also.  (Chemosensors,
probably not, or not as easily.  But again, does anybody really think this 
is relevant?  IS smell part of the TTT, for either judge or testee?  "I'm 
just a caveman, the ways of your modern world confuse and frighten me.")

( Example: how would such a computer deal with the "scrunch up your
  hand and describe the lines" question to Bill's satisfaction? It could
  simply say to the moderator "My questioner is asking about the appearance
  of my hand.  Would you hold up YOUR hand to a scanner, scrunch it up,
  image it, and then make the file available to me?" in about the same
  way that a human testee might say "My questioners have been typing so
  fast that my Carpal Tunnel Syndrome is acting up again as I try to
  keep up.  Would you go down to a pharmacy and get some ibuprofin for 
  me, and a glass of water to take it with, so I can go on with the test?". )

( Hmmm.  And I only intended to respond with the first two paragraphs
  of this message when I started typing.  Sigh. )

Wayne Throop       ...!mcnc!aurgate!throop


