k.p.collins at worldnet.att.net
Mon May 26 14:13:56 EST 2003
"BilZ0r" <BilZ0r at TAKETHISOUThotmail.com> wrote in message
news:Xns9387A8C16FE35BilZ0rhotmailcom at 18.104.22.168...
| I was wondering if anyone knew a good basic web site on cortical
| I need information to answer questions such as:
| "Discuss the role of "cortical columns" in the generation of
| "Discuss the experimental evidence supporting the hypothesis that
| columns are the fundamental functional modules for processing all
| sensory information."
| Yes, you are helping me with study for exams, thats why I don't
| answers (although they would be appriciated :) ). Yes, any basic
| or good textbooks would be of great help...
It's been in AoK, Ap6 all along.
reposted from comp.ai.philosophy [but I don't suggest you use this
stuff in response to an exam Q. It all stands Verified, but, because
it will probably be 'unfamiliar' to your Prof, and, unless your Prof
is otherwise 'prepared to comprehend what's here, your Prof will tend
to 'blindly' and automatically have a 'moving away from' reaction to
it, and your grade suffer the consequences of that]:
From: "KP_PC" <k.p.collins at worldnet.att.net>
Subject: Re: Marvin Minsky 'blasts' AI research in Wired
Date: Sunday, May 25, 2003 10:30 PM
"KP_PC" <k.p.collins at worldnet.att.net> wrote in message
news:2YGza.170247$ja4.8572801 at bgtnsc05-news.ops.worldnet.att.net...
| "Marvin Minsky" <minsky at media.mit.edu> wrote in message
| news:f04e2625.0305232019.15984557 at posting.google.com...
| | rstevew at deeptht.armory.com (Richard Steven Walz) wrote in message
| news:<3ec1f67a$0$1097$8eec23a at newsreader.tycho.net>...
| | > [...]
| | As usual, I agree with much of what Steve Walz
| | says. In this case I think I see a reason why the
| | bottom-up approach to neurology has not
| | much progressed. We know a great deal about
| | how individual neurons and synapses work. And
| | generally, that knowedge appears to show that
| | those low-level mechanisms are much the same
| | in all animals, back to fish and even invertebrates.
| | This suggests to me that our 'higher-level' thought-
| | processes grew from our having evolved of ways
| | to 'insulate' those processes from most of those
| | low-level physiological details, more often than
| | from exploiting them. In short, this is how we
| | became more symbolic and almost
| | Turing-programmable.
| | Accordingly, what I think we need is more
| | knowledge about the intermediate structures that
| | achieve this marvelous insulation. Most notably,
| | the cerbral cortex is largely composed of the
| | so-called' columns' -each of which appears to be
| | a special kind of neural network composed of
| | several hundred neurons and tens of thousands
| | of synapses. The trouble is that, so far as I know,
| | the brain-science community does not know much
| | about what these do. (I suspect that I may be out
| | of date, but the best I've seen are some theories
| | of Grossberg about how these might work.) But
| | I haven't seen anyt theories of high-level thinking
| | based on good intermediate theories. Instead, we
| | still see a great deal of talk that assumes almost
| | direct connections between thinking and the
| | functions of stuff at the chemical level-like someone
| | being disturbed because of deficiencies in
| | serotonin, etc.
| | In other words I think we need theories based on
| | intermediate structures and functions. Could
| | some of those cortical columns possibly be
| | parts of K-lines, or slots in frames, or elements
| | of semantic networks? I read Science and
| | Nature frequently, yet I don't believe that I've ever
| | seen any mention of such things in any brain-
| | science article. This suggests that there is
| | here a forty-year gap; I recently met a functional
| | MRI neuroscientist who had actually heard of the
| | Newell-Simon idea of a GPS-but had not
| | considered experiments to look for signs of their
| | Difference-Detectors at high cognitive levels.
| | As for higher level functions, the functional MRI
| | kind of approach is providing a gredat deal of
| | evidence. However, most research keep trying
| | to interpret that data in terms of old deas from
| | our everyday folk-psychology-which, in my view
| | is chock-full of words that describe little of what
| | brains actually do, but just parts of obsolete
| | popular psychological models. I would love to
| | see some researchers try to interpret that data
| | from the point of view of more modern
| | architectural views (e.g., the theories of, say,
| | Aaron Sloman, or even of mine).
| | Marvin Minsky
| Dr. Minsky, Everything you call for was Reified
| in Neuroscientific Duality Theory [NDT] decades
| ago. The things I will discuss in this
| reply were all Reified for the first time in NDT.
| For instance, with respect to the so-called
| "cortical columns", it's first necessary to
| discuss cortical architecture in general.
| Cortex, as a whole, constitutes a
| "crumpled-bag nucleus". Curmpled-bag
| nuclei occur repeatedly and hierarchically
| within nervous systems, two other stand-outs
| are the cerebell[u]m and the inferio[r] olivary
| To see a rough analogue of a crumpled-bag
| nucleus, take a brown paper grocery bag and
| crumple it :-] then uncrumple it, and shap it into
| a 'spherical' crumpled 'state', with its openig
| gathered together, a bit, but left wide enough
| for all of the information-processing I/O
| communication ["pathways"] to pass through
| it in a structurally-organized fashion.
| Why this neural architecture is of special
| significance is that crumpled-bag nuclei
| always serve the same purpose - they
| constitute the architecture of minimal circuit
| length [and, therefore, of minimial energy
consumption, and, most-importantly, of
| minimial response latencies, all of which
| are Crucial with respect to survival within
| the essentially-predatory environments
| in which nervous systems evolved]. [All of
| the nervous system is like this. Every
| twist and turn in the neural architecture
| constitute rigorous Topological functions
| that literally-embody evolutionarily-derived
| functionality which can be =read directly=
| from the Topology.]
| The easiest way to understand the
| functionality of crumpled-bag nuclei
| is to use the inferior olivivary nucleus
| as a teaching example. What it does
| is receive joint-receptor information and
| relay it to the cerebellum via the very-
| potent "climbing fiber" pathway. What
| it does is "translate" [as in Maths
| "coordinate system translation"] body-
| conformation variations so that the
| cerebellum [and, in the main via the
| cerebellum, the rest of the nervous
| system, can 'see' a 'stable' world-image
| with respect to survival-dependent
| Thus, the activation received via sensory
| receptors of a hand held palm-down =above=
| a hot or sharp object will result in different
| efforctor activations than will the activation
| of the =same= receptors when the hand is
| held palm-up =beneath= a hot or sharp
| The Neocortex itself is 'just' another
| crumpled bag nucleus. It performs the
| extraordinarily-generalized coordinate-
| system "translation" with respect to the
| totality of the neural activation
| that occurs within a nervous system.
| There are only a few cortical fe[a]tures
| that are of particular significance. One of
| these is the 'orthoganally-interlaced',
| 6-layered, nature of the cortical interneuron
| architecture. Another is that stochastic
| activation from the "reticular system" is
| intermingled [layer 5[(?)]] within this layered,
| interlaced network.
| The result is that "loop circuits" can be
| set-up dynamically within cortex. It's through
| the variation of the circuit lengths, with
| experientially-correlated convergence
| upon minimal loop-circuit lengths, that
| cortex carries out it's generalized
| "translation" function. Such tuning of
| loop-circuit lengths is what allows the
| "type-II synchronization" [like gears in
| a clock, vs. the regimented 'marching'
| of "type-I synchronization] of cortical
| outputs to the effectors [and at distance
| within cortex, and subcortical strucures]
| is what allows the coordinated activation
| of the effectors in which directed, purposful,
| behavior is manifested.
| Everything important in 'momentary'
| cortical function derives in the
| tuning of such loop-circuit lengths.
| [Of course, trophic modification
| accompanies neural activation in a
| way that's rigorously determined
| by the neural activation that actually
| occurs in the network. This "learning"
| also occurs in rigorous accord with
| convergence upon minimal circuit
| lengths [energy consumption, response
| In addition to this, the only other really-
| important feature of cortex is it's
| sensory-motor-association modality-
| mapping, including its mapping to
| subcortal structure.
| Within all of this, the so-called "cortical
| columns" arise as an 'artifact' of the
| way that the nervous system drives
| the effectors so that stuimlus sets that
| are being 'attended to tend to 'always' be
| oriented front-centered.
| The 'cortical column' artifact results
| from the fact that such front-center
| orientation tends to result in cortical
| activation 'hovering' about cortical loci
| that are correlated to optimal
| front-center orientation, and maximally
| converged upon minimal loop-circuit
| length [energy consumption, response
| latency] trophic dynamics just grow the
| columnar structure as an otherwise
| architecture. That is, all the meaning
| is in the topologicallt-distributed
| trophic-minimization of
| circuit lengths, etc.
[posted in a follow-up in comp.ai.philosophy. but re-posted from a
saved copy, so it's not 'threaded']
BTW, this stuff has been discussed in AoK,
Ap6 all along. The existence of interdigitating
"cortical stripes" is also explained in Ap6.
If anyone wants a few crude visualization
aids with respect to the
"overlaying sliding fields" that are
responsible for the existence of
the ubiquitous "cortical cloumn" artifacts
of "front-center orientation", poke around
at these Google URLs:
The whole thread:
The "INFOCALC.BAS" code is really all you need.
This code is =really= old, and 'quick-&-
dirty'. It was always for my own purposes only.
It was originally done on an 8-bit ascii-only
machine. The code still serves it's
visualazation-aid purpose. The
'editor' is a 'pain'. It was a one-shot app
I used to design 5 x 5 arrays of minus signs
and asterisks, the asterisks constitution the
'info'. I just did an alphabet and some other
images. It produces a random-access file,
so it's necessary to use the editor to build the
image file. You'll probably want to enhance it
so that the images in "STIMULIxx.DAT" are
read/write, rather than just "write" :-]
If anyone wants a copy of an 'image' dta file,
msg and I'll send it as an attachment. Better
yet, I'll attach one to this msg.
I still use QBasic[tm[ for stuff like this
because it ships with Windows[tm] so there
are millions of copies out there, which
makes it possible for many folks to see
this or that that's in the code.
Further discussion of the "front-center
It's Fundamental within development,
and begins prenatally. It's hierarchical
from global-system to ionic conductances,
and includes myriad factors, such as
limb-orientation 'probabilities' [not really
"probabilities", but actual-experience
'echoes'], and abstract cross-correlation
'echoes' [AoK, Ap5; "biological mass" in
all of its myriad information-processing-
If you read AoK, and take the 'time' to
read some of the refs cited in AoK, you'll
get all of this stuff. I'm writing, here, for
folks who are already familiar with the
fuctional Neuroanatomy, but who
don't yet grasp it's overall integration.
K. P. Collins
[end copy of un-'threaded' follow-up]
| This overly-simplified discussion
| provides minor insight into NDT.
| NDT does the same, at a much
| greater level of detail, with respect
| to every major nuclear group within
| nervous systems, providing concrete
| biological mechanisms for [explanations
| of], among other things, creativity,
| curiosity, and volition. NDT flat-out nails
| cognition and consciousness. Everything
| in NDT reduces directly to the Proven
| Neuroscience experimental results.
| NDT is equivalently-stated, and
| verified, in 5 independent-perspectives,
| one of which is computer scientific.
| I ask [consider it a Formal Challenge
| to MIT] that I be allowed to present an
| =overview= of NDT at MIT [or where
| ever you are these days], in forum that
| is Open to the Public at a minimal cost
| to the attendees [basically, only to
| cover MIT's costs - lighting, etc.] Folks
| in Computer Science, Physics,
| Mathematics, Behavioral Science,
| Sociology, and Philosophy will all find
| newness that will be of significant
| Usefulness to them if they attend such
| a presentation.
| NDT is the "intermediate" stuff that
| you call for in your post, quoted above, Sir.
| It constitutes nearly 32 years of Devoted
| Work. Its implications with respect to Human
| Survival are Immense, which is why I am
| Obliged to work to achieve its generalized,
| Open, communication.
| I Hope MIT will Accept this Formal
| Challenge, with keen Cricticality,
| ]and] with strong 'heart'.
| I'm in Massachusetts so I'll be able to
| travel there with relative ease [I'll need
| small considerations like free parking
| during the process, though, and sufficient
| lead-'time', and knowledge of MIT's
| expectations, so I can specifically
| address them. And a Student-Guide
| would be helpful, if possible and practical.
| I've worked 'outside of the system', and
| am aware that I remain 'naive' with
| respect to expectations of 'the system'.
| The Student-Guide would be my
| temporary 'crumpled-bag' "translation"
| 'nucleus' :-]
| Sincerely, K. P. Collins
| "Schmitd! Schmitd! Ve vill build a Shapel!"
More information about the Neur-sci