Selected Newsgroup Message
From: "Sergio Navega" <snavega@ibm.net>
Subject: Re: Neurons and consciousness
Date: 13 Jul 1999 00:00:00 GMT
Message-ID: <378ba2d9@news3.us.ibm.net>
References: <37790B77.CF1B5B3C@travellab.com>
<g9ae3.2900$c5.749565@news1.usit.net> <3779420F.B83925B6@travellab.com>
<378928aa.275952835@netnews.worldnet.att.net>
<3789F5D6.5D1041AC@sandpiper.net> <7mdsnj$dn4@dfw-ixnews10.ix.netcom.com>
<378A7FE3.54AB8622@sandpiper.net> <378b3b7a@news3.us.ibm.net>
<378B967B.FFF8643@sandpiper.net>
X-Notice: should be reported to postmaster@ibm.net
X-MimeOLE: Produced By Microsoft MimeOLE V4.71.1712.3
X-Complaints-To: postmaster@ibm.net
X-Trace: 13 Jul 1999 20:34:33 GMT, 166.72.21.136
Organization: Intelliwise Research and Training
Newsgroups: comp.ai.philosophy,sci.physics
Jim Balter wrote in message <378B967B.FFF8643@sandpiper.net>...
>Sergio Navega wrote:
>>
>> Jim Balter wrote in message <378A7FE3.54AB8622@sandpiper.net>...
>> >[snip]
>> >None of this is to the point. But I will mention that it is not
>> >at all unscientific to assume that something not proven is not the
>> >case; in fact, Ockhams' Razor is at the heart of science and its
>> >method.
>> >
>>
>> Induction, live long and prosper!
>> ;-)
>
>Not induction, but rather inference to the best explanation.
>See David Deutsch's _Fabric of Reality_ for an excellent
>explanation of the difference, and a clear demonstration that
>Popper solved the "problem of induction" (Hume's observation that
>induction is not logically valid).
>
It's funny, Jim, I think we had this conversation before.
Sure, induction is not exactly the same as inference to the
best explanation, but both are known as weak methods, with
the latter known as abduction. They are in the opposite
end of the spectrum in relation to deductive methods.
About Deutsch's book, I bought it some months ago (because
of an indication of yours) and I'm very glad I did it, it is
really highly legible and well put. It changed some prejudices
I had with Popper (I'm better with him now). However, Deutsch
lets unanswered the same question forgotten by Popper, "where
do hypotheses come from?". This is a too "cognitive" question
for philosophers to worry about, but it is still unanswered
and is *very* important.
I agree that science progresses through falsification of
proposed models and theories in the light of evidences and this
*cannot* be an inductive process. But I propose that hypotheses
"are born" inside our mind as a result of inductive methods.
The fact that this process usually happens "under the rug" only
adds to its elusiveness.
However, even being weak, induction (and abduction, to a certain
extent) are the only ways of obtaining *more* consequents than
*antecedents*, which is exactly what we need during the
generative phase of our thought. After this process takes place,
then falsification will have plenty of material to work with.
Regards,
Sergio Navega.
From: "Sergio Navega" <snavega@ibm.net>
Subject: Re: Neurons and consciousness
Date: 14 Jul 1999 00:00:00 GMT
Message-ID: <378cd950@news3.us.ibm.net>
References: <37790B77.CF1B5B3C@travellab.com>
<g9ae3.2900$c5.749565@news1.usit.net> <3779420F.B83925B6@travellab.com>
<378928aa.275952835@netnews.worldnet.att.net>
<3789F5D6.5D1041AC@sandpiper.net> <7mdsnj$dn4@dfw-ixnews10.ix.netcom.com>
<378A7FE3.54AB8622@sandpiper.net> <378b3b7a@news3.us.ibm.net>
<378B967B.FFF8643@sandpiper.net> <378ba2d9@news3.us.ibm.net>
<378BD721.E8D187B2@sandpiper.net>
X-Notice: should be reported to postmaster@ibm.net
X-MimeOLE: Produced By Microsoft MimeOLE V4.71.1712.3
X-Complaints-To: postmaster@ibm.net
X-Trace: 14 Jul 1999 18:39:12 GMT, 166.72.21.163
Organization: Intelliwise Research and Training
Newsgroups: comp.ai.philosophy,sci.physics
Jim Balter wrote in message <378BD721.E8D187B2@sandpiper.net>...
>Sergio Navega wrote:
>
>> About Deutsch's book, I bought it some months ago (because
>> of an indication of yours) and I'm very glad I did it, it is
>> really highly legible and well put. It changed some prejudices
>> I had with Popper (I'm better with him now). However, Deutsch
>> lets unanswered the same question forgotten by Popper, "where
>> do hypotheses come from?".
>
>Deutsch is quite explicit about this; *it doesn't matter*,
>at least as far as explanatory power goes.
>
At least as far as explanatory power goes, yes, it doesn't
matter. But if seen under the light of epistemology, this
situation is vexingly incomplete.
>> This is a too "cognitive" question
>> for philosophers to worry about, but it is still unanswered
>> and is *very* important.
>
>It's important if your concern is cognition, how the brain works,
>but that wasn't the issue I was addressing, nor Deutsch or Popper.
>They don't answer the question of what causes Alzheimer's, either,
>but these simply aren't the issues at hand.
>
>And it is way too premature for us to be asking where hypotheses
>come from, in terms of cognitive processes, when we don't yet
>have answers to much more basic questions like "how does memory
>work?".
>
It is not premature if both processes must be understood in
tandem. What use could we have for the knowledge of how memory
works if not to allow us to understand what it is used for?
>> I agree that science progresses through falsification of
>> proposed models and theories in the light of evidences and this
>> *cannot* be an inductive process. But I propose that hypotheses
>> "are born" inside our mind as a result of inductive methods.
>
>The evidence is that they are born in all sorts of ways,
>including random juxtaposition. The idea that they result
>from induction seems to be a just so story, as Rickert has
>reiterated so many times.
>
The question here is evaluating induction purely on the
basis of high-level reasoning. I agree that in this regard,
induction is not only weak, but conducts to serious errors.
But this is not enough to dismiss inductive processes on a
more primary, elemental level, which is what I believe
is the responsible for the "mysterious" origins of hypotheses.
>> The fact that this process usually happens "under the rug" only
>> adds to its elusiveness.
>
>Virtually all your cognitive processes happen "under the rug".
>
>> However, even being weak, induction (and abduction, to a certain
>> extent) are the only ways of obtaining *more* consequents than
>> *antecedents*, which is exactly what we need during the
>> generative phase of our thought. After this process takes place,
>> then falsification will have plenty of material to work with.
>
>Flipping coins is also an effective way of producing consequents,
>which puts the lie to the Penrose/Lucas/Godel silliness.
>Again, as Deutsch makes *very* explicit, *it doesn't matter*
>where a hypothesis came from, it only matters whether it explains.
>
Flipping coins can, indeed, be a reasonable way to produce
antecedents for testing. But you want a slightly more efficacious
method? Why don't use the things that worked well in the past as
a "first guide"? That, in my book, is the name of induction: a
better than chance guess.
Regards,
Sergio Navega.
From: "Sergio Navega" <snavega@ibm.net>
Subject: Re: Neurons and consciousness
Date: 15 Jul 1999 00:00:00 GMT
Message-ID: <378dd768@news3.us.ibm.net>
References: <37790B77.CF1B5B3C@travellab.com>
<g9ae3.2900$c5.749565@news1.usit.net> <3779420F.B83925B6@travellab.com>
<378928aa.275952835@netnews.worldnet.att.net>
<3789F5D6.5D1041AC@sandpiper.net> <7mdsnj$dn4@dfw-ixnews10.ix.netcom.com>
<378A7FE3.54AB8622@sandpiper.net> <378b3b7a@news3.us.ibm.net>
<378B967B.FFF8643@sandpiper.net> <378ba2d9@news3.us.ibm.net>
<378BD721.E8D187B2@sandpiper.net> <378cd950@news3.us.ibm.net>
<7mjbl0$96u@dfw-ixnews10.ix.netcom.com>
X-Notice: should be reported to postmaster@ibm.net
X-MimeOLE: Produced By Microsoft MimeOLE V4.71.1712.3
X-Complaints-To: postmaster@ibm.net
X-Trace: 15 Jul 1999 12:43:20 GMT, 200.229.243.40
Organization: Intelliwise Research and Training
Newsgroups: comp.ai.philosophy,sci.physics
Steven B. Harris wrote in message <7mjbl0$96u@dfw-ixnews10.ix.netcom.com>...
>In <378cd950@news3.us.ibm.net> "Sergio Navega" <snavega@ibm.net>
>writes:
>
>>>Flipping coins is also an effective way of producing consequents,
>>>which puts the lie to the Penrose/Lucas/Godel silliness.
>>>Again, as Deutsch makes *very* explicit, *it doesn't matter*
>>>where a hypothesis came from, it only matters whether it explains.
>>>
>>
>>Flipping coins can, indeed, be a reasonable way to produce
>>antecedents for testing. But you want a slightly more efficacious
>>method? Why don't use the things that worked well in the past as
>>a "first guide"? That, in my book, is the name of induction: a
>>better than chance guess.
>>
>
> But one leads to the other, depending on the size of the coin you
>flip. As Dawkins or somebody points out, one of the major ways
>evolution explores "design space" is not totally randomly, but by
>putting together different *blocks* of things that worked before, in
>much the same way that an expert solves problems by combining
>sub-routines and tricks and so on that have previously been successful.
>Not everything is varied randomly. In mating, chromosomes and chunks
>of chromosomes are exchanged. That shuffling has consequences far more
>serious than tiny mutations of one base pair in one gene.
Thanks for supporting my prior assertion so emphatically: induction,
the reuse of what worked well in the past.
>
> This kind of thing, of course, goes on in every industry in
>engineering. Look at any new device, and if you know enough
>engineering you will see that unless it's a very simple thing or the
>inventor is a very, very unusual (and inefficient) mind, the guts of it
>have recognizable chunks that came from other, earlier things. They
>are just modified appropriately and combined in new ways. But the
>recombination is done modularly, on many, many levels.
>
Most of the times, yes. And that's why induction is very useful.
But there are problems, too. When the new design involves things
that *cannot* be built just by the proper combination of previous
designs, then it is time to be creative. Induction alone is not
enough, it is just part of a greater mechanism.
Hence, what I'm saying is that one cannot criticize induction just
because it is not the *whole* mechanism. It is just an important
part of it.
> Machines are just the physical manefestations of ideas, and Dawkins
>calles the modules that make up hypotheses and ideas, "memes" (idea
>genes, if you will). Physical hypotheses (physics theories) have
>modules too, and most of them are mathematical. It all starts with
>Descartes and Newton, and their calculus and analytical geometry (what
>intellectual giants these were) and proceeds from there through
>statistics and linear algebra and tensor analysis to the present day,
>when people confronted with some new reaction in particle physics start
>asking about what symmetry group this new process might be modeled by.
>Modules and modules.
But do you see that among the evolution of these modules, there
were a lot of "unconventional" steps? Much of what has been created
stemmed on inductive reuse, but this was also done in a creative way.
Creativity is also fundamental in this regard and, although one
may say that it depends heavily on randomness, I find that it
also demands "perceptual efficacy", the ability to recognize
what is relevant from what's not.
Regards,
Sergio Navega.
From: "Sergio Navega" <snavega@ibm.net>
Subject: Re: Neurons and consciousness
Date: 16 Jul 1999 00:00:00 GMT
Message-ID: <378f3857@news3.us.ibm.net>
References: <37790B77.CF1B5B3C@travellab.com>
<g9ae3.2900$c5.749565@news1.usit.net> <3779420F.B83925B6@travellab.com>
<378928aa.275952835@netnews.worldnet.att.net>
<3789F5D6.5D1041AC@sandpiper.net> <7mdsnj$dn4@dfw-ixnews10.ix.netcom.com>
<378A7FE3.54AB8622@sandpiper.net> <378b3b7a@news3.us.ibm.net>
<378B967B.FFF8643@sandpiper.net> <378ba2d9@news3.us.ibm.net>
<378BD721.E8D187B2@sandpiper.net> <378cd950@news3.us.ibm.net>
<7mjbl0$96u@dfw-ixnews10.ix.netcom.com> <378dd768@news3.us.ibm.net>
<7mlmc1$na6@dfw-ixnews11.ix.netcom.com>
X-Notice: should be reported to postmaster@ibm.net
X-MimeOLE: Produced By Microsoft MimeOLE V4.71.1712.3
X-Complaints-To: postmaster@ibm.net
X-Trace: 16 Jul 1999 13:49:11 GMT, 200.229.240.76
Organization: Intelliwise Research and Training
Newsgroups: comp.ai.philosophy,sci.physics
Steven B. Harris wrote in message <7mlmc1$na6@dfw-ixnews11.ix.netcom.com>...
>In <378dd768@news3.us.ibm.net> "Sergio Navega" <snavega@ibm.net>
>writes:
>>
>>But do you see that among the evolution of these modules, there
>>were a lot of "unconventional" steps? Much of what has been created
>>stemmed on inductive reuse, but this was also done in a creative way.
>>Creativity is also fundamental in this regard and, although one
>>may say that it depends heavily on randomness, I find that it
>>also demands "perceptual efficacy", the ability to recognize
>>what is relevant from what's not.
>>
>>Regards,
>>Sergio Navega.
>
>
> Sure. Creation is an iterative process, and as far as I can
follow
>it down into the mind, it seems to be very much like evolution by
>natural selection. You take random variations of things that have
>worked in the past, combine them, and evaluate them by figure of merit
>criteria. One of the mechanisms by which "forms" or rules in art cause
>some of the most creative work, is that by trying to stay within the
>box (find a rhyme for that sonnet, find a variation for that rhondo,
>etc) you pick variations you would not otherwise have considered, and
>some of them, on evaluation, turn out to be really good. Not something
>you'd have come up with in ordinary circumstances.
>
That's a good part of it, but is not enough. If our creative process
used only random variations, mindlessly, we would be losing precious
time with the zillions of ineffective combinations. What appears to
happen is the random exploration of the *vicinity* of things that did
work well in the past, much like wandering aimlessly around and over
a mountain. We may spend more time at the top of this mountain, but
often we go to the base to explore the vicinity.
Two things are obtained by this process: we occasionally give a chance
to "chance", but we give more privileges to the things that worked
well before (top of the mountain).
> Obviously I haven't figured all of this out, or am I likely to, but
>there are two observations I want to make. One is that inductive
>"learning" or something very like it, is possible in systems that
>reproduce, so long as there is ANY system for recording experience and
>dupblicating it. If you throw streetcats into cold warehouses to catch
>mice, as was done some time ago in (I think) Chicago, after a time of
>this rather severe selection pressure you get a sub-breed of very
>efficient hunter cats that has long thick coats and very short tails.
>No individual cat "learns" that these things will be useful in its
>environment, but the "species" or genome has, in a sense, made an
>inductive change. All such systems end up adapting to present
>circumstances, and when they reproduce, that behavior amounts to
>assuming that the future will be like the past. That's inductive.
>Because it is, it's not infalable. You can take such cats to the
>tropics and the kittens will still have long coats and short tails, and
>so this is a case where induction fails. The time specific regularity
>of nature, however, is great enough that a certain amount of induction
>at the genomic level is conducive to species survival, but only so
>much. Species with no ability to mutate eventually find themselves
>faced with a changed environment, and that is the end of their genes.
Something like that may also happen within our brain. A child
that grows in icy countries may have difficulties to "feel good"
in tropical countries.
>
> The second observation I want to make, is that when this process
>happens with memes instead of genes, and generates ideas and thoughts,
>it happens at an unconscious level where it's not accessable. There
>are many stories of people solving problems this way. I once heard for
>a guy who had had to unscramble the word "foobanutics" for an
>intelligence test, and couldn't do it. He thought no more about it,
>but some weeks later, it popped into his mind unbidden one day that the
>word was "obfuscation." Some part of his brain had been workiing on
>the problem, unbidden and unseen, for some or all of that time, and had
>finally found the answer and presented it.
>
In a way, I find that what we must understand is this mechanism.
Our conscious thought, the process that AI usually tried to model,
is just the tip of a gigantic iceberg. But in order to understand
what's going on in unconscious realms, we've got to use scientific
methods. That's the reason to study neuroscience, implicit learning,
cognitive neuroscience and stuff like that.
> We know this kind of thing happens when we dream. One part of
the
>brain (not necessarily one physical part, but some of the circuitry)
>builds sets and props for the other part (the ego or conscious "I"
>which experiences the dream), which part then explores these "sets",
>and is sometimes even surprised, frightened, or shocked by what it
>finds. In my dreams I'm astounded by what I see and feel and hear
>while exploring that old house or talking to those old classmates.
>Quite obviously, some part of the mind runs along under the surface,
>making up stores and modeling the behavior of the world and other
>people, using memories freely to do it, and all the time presenting its
>conclusions as emotions or other odd "ESP-like" intuitions for which no
>rational explanation can be given by the "talking mind" (the one we all
>think of as "us.") Since a great many creative thoughts come out of
>this subconscious or submind (or set of subminds), the whole process is
>very mysterious.
It is mysterious indeed. But it is also an invitation to mysticism.
One of the greatest dangers that we're subject to today is the
unscientific exploration of this immense repository. Take "dream
interpretation", for instance.
There's no evidence to support the kind of conclusion that some
"psychologists" draw from the interpretation of dreams. Dreams may
be just a reflex of random processes happening over things that we've
been in contact during our day. To say that, for instance, if you
dreamed of "whales" then you've got to leave your job and look for
another one, is a senseless, if not hazardous, conjecture. Unfortunately,
there are a lot of therapists that go through that way (and publish
books about it!). And the general public love it. Pity.
Regards,
Sergio Navega.
Back to Menu of Messages Sergio's Homepage
Any Comments? snavega@attglobal.net