Richard,
I'm not sure your substitutions of words constitute a valid
criticism of the computational neuroscience field. Sure, I may
concede that some of the categories they introduce may suffer
from some serious flaws. But we cannot forget that we still don't
know how the whole neural system works (as opposed to our very
good understanding of how a bunch of transistors give rise to
the US Air Traffic Control System). The way computational
neuroscientists introduce concepts (and often mix levels of
analyses) should be seen as the way scientific reasoning works
in fields with a lot to discover. The scientific method allows
the construction of such abstract notions, because it has
some chance of being appropriate and if they are not, the method
will eventually discard or substitute them for better notions.
History of science is filled with such things. In transistors,
stochastic behavior is not important. In neurons, it seems an
important factor. But what is the influence of noise in these
systems? What good does it do? We will only discover by
postulating theoretical constructions capable of generating
predictions and letting the evidences guillotine the bad models.
Sergio Navega.
----- Original Message -----
From: "Richard Loosemore" <[EMAIL PROTECTED]>
To: <[email protected]>
Sent: Wednesday, October 11, 2006 11:47 AM
Subject: Re: [agi] Fwd: Articles in this week's Science
[Warning. The following is short and brutally frank.]
This first article (which overviews the others) is a typical piece of
neuroscience BS - at least from the point of view of a substantial number
of people in the cognitive science community.
To get an idea why, let me quote to you three excerpts from the piece, but
translating them into claims about high level aspects of COMPUTING,
instead of high level aspects of COGNITION.
Rememember, these neuroscience folks are trying to sell the idea that they
are making progress towards understanding how the entire brain works as a
mind, not as a bunch of neurons. They don't claim to be doing
neurophysiology, they claim to be making a link to cognition, thought,
consciousness and the like. So, in my edited versions below, I have
changed references to high level cognition to become references to large
software systems. With this change, the essential vacuousness of the
claims sticks out like a sore thumb.
****************************************************
"Understanding the dynamics and computations of single [transistors] and
their role within [computer circuits] is at the center of neuroscience.
How do single-[transistor] properties contribute to information processing
and, ultimately, to [the behavior of extremely complex software systems]?
What level of description is required when modeling single [transistors]?
Herz et al. (p. 80) review single-[transistor] models..........."
"Single [transistors] are part of larger networks. Destexhe and Contreras
(p. 85) review advances in the computations created by stochastic input in
[transistors] and networks of [transistors]. They emphasize the importance
of irregular activity in [transistor] computations............"
"On a higher processing level, computational neuroscience based on [an
approximate circuit diagram (with only 1% of the wires transistors
actually showing)] of the [Intel Xeon] can help us understand the
complexities of [the highest-level aspects of the software that runs the
US Air Traffic Control System]. O'Reilly (p. 91) reviews developments in
models, of [the software that runs the US Air Traffic Control System]. He
develops the idea that the [floating point unit in an Intel Xeon]
represents a synthesis between analog and digital forms of
computation.........."
****************************************************
If someone said that there was an entire field that was making progress in
understanding the issues involved in the design and functioning of the
largest software/hardware projects on the planet, and if the people in
this field grabbed an issue of Science to present shining examples of
their best work, as above, what would you think?
The emperor has no clothes.
Richard Loosmore
Ben Goertzel wrote:
There's a special section in this week's Science called "Modeling the
Mind"
that should be of interest to many denizens of this list. Here are the
titles:
Of Bytes and Brains
Peter Stern and John Travis
Science 6 October 2006: 75.
<http://www.sciencemag.org/cgi/content/summary/314/5796/75> Summary >|
PDF <http://www.sciencemag.org/cgi/reprint/314/5796/75.pdf> >|
News
An Enterprising Approach to Brain Science
Greg Miller
Science 6 October 2006: 76-77.
<http://www.sciencemag.org/cgi/content/summary/314/5796/76> Summary >|
Full Text <http://www.sciencemag.org/cgi/content/full/314/5796/76> >|
PDF
<http://www.sciencemag.org/cgi/reprint/314/5796/76.pdf> >|
Vision's Grand Theorist
Ingrid Wickelgren
Science 6 October 2006: 78-79.
<http://www.sciencemag.org/cgi/content/summary/314/5796/78> Summary >|
Full Text <http://www.sciencemag.org/cgi/content/full/314/5796/78> >|
PDF
<http://www.sciencemag.org/cgi/reprint/314/5796/78.pdf> >|
Reviews
Modeling Single-Neuron Dynamics and Computations: A Balance of Detail and
Abstraction
Andreas V. M. Herz, Tim Gollisch, Christian K. Machens, and Dieter Jaeger
Science 6 October 2006: 80-85.
<http://www.sciencemag.org/cgi/content/abstract/314/5796/80> Abstract >|
Full Text <http://www.sciencemag.org/cgi/content/full/314/5796/80> >|
PDF
<http://www.sciencemag.org/cgi/reprint/314/5796/80.pdf> >|
Neuronal Computations with Stochastic Network States
Alain Destexhe and Diego Contreras
Science 6 October 2006: 85-90.
<http://www.sciencemag.org/cgi/content/abstract/314/5796/85> Abstract >|
Full Text <http://www.sciencemag.org/cgi/content/full/314/5796/85> >|
PDF
<http://www.sciencemag.org/cgi/reprint/314/5796/85.pdf> >|
Biologically Based Computational Models of High-Level Cognition
Randall C. O'Reilly
Science 6 October 2006: 91-94
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/[EMAIL PROTECTED]
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/[EMAIL PROTECTED]
--
No virus found in this incoming message.
Checked by AVG Free Edition.
Version: 7.1.408 / Virus Database: 268.13.2/471 - Release Date: 10/10/2006
-----
This list is sponsored by AGIRI: http://www.agiri.org/email
To unsubscribe or change your options, please go to:
http://v2.listbox.com/member/[EMAIL PROTECTED]
