Newsgroups: comp.ai.philosophy
Path: cantaloupe.srv.cs.cmu.edu!das-news.harvard.edu!news2.near.net!MathWorks.Com!europa.eng.gtefsd.com!howland.reston.ans.net!cs.utexas.edu!utnut!utgpu!pindor
From: pindor@gpu.utcc.utoronto.ca (Andrzej Pindor)
Subject: Re: Is Common Sense Explicit or Implicit?
Message-ID: <CwE3Gq.3xH@gpu.utcc.utoronto.ca>
Organization: UTCC Public Access
References: <357e2u$f8k@mp.cs.niu.edu> <35alp4$4r9@agate.berkeley.edu> <Cw8oK7.BB7@gpu.utcc.utoronto.ca> <35d6i2$5fd@agate.berkeley.edu>
Date: Mon, 19 Sep 1994 18:31:38 GMT
Lines: 152

In article <35d6i2$5fd@agate.berkeley.edu>,
Gerardo Browne <jerrybro@uclink2.berkeley.edu> wrote:
>Andrzej Pindor (pindor@gpu.utcc.utoronto.ca) wrote:
>: Since, as have already been remarked by other people, the same actions
>: observed can be explained by different assumptions about desires (or beliefs,
>: purposes etc.) claiming that they are objectively real is risky. Maybe
>: it depends on what you consider as 'real'. Is there 'really' color red?
>: You will hopefully agree that there is a limit to how good can we get at 
>: predicting individual's behavior by honing our theory of her/his desires.
>
>Dogs are real, even though we can't use them in a perfect theory to
>predict something with total accuracy.  So the condition that the
>postulated entity must be usable in this way is incorrect.  What I don't

But aren't beliefs postulated so that human behavior seems understandable?
Where else do they enter?
This is not a case for dogs, is it? I totally miss the point of your 
argument.

>like is the scientist's way of saying that nothing is real except what
>fits into his theories.  Thus we hear things like, "all we *really* are
>is a bunch of particles in a void."  As if by discovering *certain*
>entities which obey strict rules, we had the right to conclude that
>everything else is only a rough way of looking at this pristine world,
>that these other entities are ghosts, epiphenomena.
>
It seems to me that you do not fully appreciate what scientists call 'real'
and what not. Not that this is a simple, unambiguous designation. It has 
already been pointed out by someone else that it much depends on a model
of the world you are using.

>Tell me this:  how do we know that the records of the past really
>record anything?  If we take our knowledge of the world today and
>trace it back in a Cray supercomputer, following the zillions of
>particles backwards through time, we will not arrive at a coherent
>past, but will instead be subject to the law of entropy.  That is,
>the simulation will pretty quickly stop meshing, simply because
>it won't be perfect.  And from what I heard, the laws of QM work
>backwards as well as forwards, so there may be a fundamental block on
>the complete knowledge of the past.  I propose to you, then, that
>the only way we can trust our historical records is if we postulate
>the existence of things other than particles in the past.  Such as,
>beliefs.  If we refuse to suffer the presence of these partly
>lawless entities in the halls of the intellect, then we will be
>subject to an immense historical amnesia.
>
You have lost me here. How do problems with creating a model of the past
require postulating beliefs? Do you mean that we can only believe that past
looked this way and not that way, since we can not determine it exactly?
We seem to be talking about different things.

>: So we can't get better and better at such predictions and your criterion for
>: objective reality will not work, will it?
>: You admit that you construct desires on the basis of people's behavior, to
>: be used to predict other behavior - this simply means that there are some
>: regularities in the behavior.
>
>And we construct electrons on the basis of their behavior.  We don't witness
>electrons except by means of the interpretation of data.  But the
>interpretation of these data depends entirely on the assumption that
>electrons have the properties we ascribe to them.  Electrons are
>considered real because they *make sense* of the data.  This is the
>basis for our acceptance of the reality of beliefs and desires.
>
Note that electrons were postulated to explain some phenomena (e.g. 
electricity). They had to have certain properties to provide such explanation.
It was later found that other, unrelated phenomena, could also be explained
using such entities (e.g. optical spectrum of a hydrogen atom). Then even
more phenomena were found explainable by postulating electrons
(with the same properties). Contrast this with the case of beliefs and
desires - they are postulated to explain behavior, but they do not seem, as
far as I know, show up anywhere else. I may be mistaken, perhaps they do.
If so this would certainly strengthen the claim that they are more than just
a figment of imagination of (some) psychologists. 

> Your 'desires' are simply names for those
>: regularities, nothing more, unless you have some independent way of checking
>: those 'desires'.
>
>This is simply wrong.  You could with equal justice say the same thing
>about electrons.  In fact, I would argue that you could say the same
>thing about *anything at all*. This is what allowed Nietzsche to say
>that a thing is the sum of its effects on other things.
>
See above. You cannot reliably claim existence of  a 'thing' on a basis of
its effects on only one thing.

>..........
>: >Sadly circular!  It'sn not enough to say circular.  You have obviously
>: >not read any of the articles in which I show awareness of this problem
>: >and attempt to demonstrate that one can develop refutable theories
>: >concerning beliefs.
>: >
>: You are right, I haven't. Could you please give an example of how one can
>: refute such a theory?
>
>I have a theory that beliefs tend to stick unless the person sees something
>that conflicts with the belief.  I realize that this theory is already
>not quite right.  It has been refuted.  Well, there you are.  What you
>are demanding is that the *existence* of beliefs be refutable, right?
>Well, can the existence of dogs be refuted?  I don't know.  For that
>matter, can the existence of electrons be refuted?  One can imagine
>physicists slowly changing the postulated properties of electrons until
>they resemble nothing like our present conception.
>
One can hardly imagine existence of electrons to be refuted, since so many
_different_ phenomena require electrons for their explanation. You would 
have to find an alternative explanation (the same for dogs). If you look 
into books on history of physics you will find examples of refutation of
entities postulated on basis of one type of phenomena, but found untenable
when a wider range of phenomena was considered (e.g. ether). Theory of
beliefs would be refutable if their consequences had to show in some other
range of observations - if these observations indeed confirmed what 
existence of beliefs required, it would be a good argument for them, if
the observations did not provide such a confirmation, it would constitute
a tentative refutation.
>......
>: Couldn't I preprogram the robot with such knowledge? I do not see any reason
>: why not.
>
>I think you're going out on a limb if you suppose you can preprogram
>a robot to be adaptable the way a human is.
>
I am not supposing that I can, I only ask you if you have any good reason to
state firmly that I could not. Of course I mean 'in principle', I personally
certainly could not but this is not the point. Since most likely you do not 
(you would certainly make a splash in AI community if you could), your argument 
is no argument at all. For me it is enough to say that the adaptability of 
robots being built increases and no clear barrier seems to be in sight. Saying
"no robot can achieve _ever_ adaptability of a human being because it is
_only_ a robot" is not good enough.
.........
>: >
>: Microorganisms which mutate to adjust themselves to changing environment
>: do not seem to require beliefs and desires, do they?
>
>So beliefs are not the only way something can adjust itself.  Fuzzy
>logic chips do some adjusting too.
>
Exactly! So another type of phenomena - adaptability to changing 
conditions - does not seem to require beliefs.

>-epf
>
>
Andrzej

-- 
Andrzej Pindor                        The foolish reject what they see and 
University of Toronto                 not what they think; the wise reject
Instructional and Research Computing  what they think and not what they see.
pindor@gpu.utcc.utoronto.ca                           Huang Po
