[Comp-neuro] "realistic models"
reeke at mail.rockefeller.edu
Tue Aug 19 18:02:43 CEST 2008
Dear Jim, et al,
Re the question of whether the brain is too complex to understand
its own workings, I refer interested readers to the writings of
Colin McGinn, a philosopher who has taken and defended this position.
>From his entry in Wikipedia:
> Although McGinn has written dozens of articles in philosophical logic,
> metaphysics, and the philosophy of language, he is best known for his
> work in the philosophy of mind. In his 1989 article "Can We Solve the
> Mind-Body Problem?", McGinn speculates that the human mind is innately
> incapable of comprehending itself entirely, and that this incapacity
> spawns the puzzles of consciousness that have preoccupied Western
> philosophy since Descartes. Thus, McGinn's answer to the hard problem
> of consciousness is that humans cannot find the answer. This position
> has been nicknamed the "New Mysterianism". The Mysterious Flame:
> Conscious Minds in a Material World (2000) is a non-technical
> exposition of McGinn's theory.
Needless to say, I do not myself accept this position, but it is
interesting to try to refute it.
George Reeke, Ph.D.
Head, Laboratory of Biological Modelling
The Rockefeller University
1230 York Avenue
New York, NY 10065
email: reeke at rockefeller.edu
On Mon, 2008-08-18 at 12:41 -0500, james bower wrote:
> Bard and everyone else:
> On Aug 17, 2008, at 8:30 AM, G. Bard Ermentrout wrote:
> > Carson Chow, a former colleague, has an interesting summary of this
> > doscussion on
> > sciencehouse.blogspot.com
> Yes, a well written summary of several of the points -- and the
> introduction of an idea that, in fact, does lie somewhere near the
> foundation of this debate. The question as to whether the brain can
> be represented by a structure (whatever it is) less complex than the
> brain itself -- formally, this moves us into complexity theory and a
> Kolmogorovian (sic) framework for thinking about levels of
> complexity. I actually like the characterization "Kolmogorov
> Complexity Complete" (KCC). And yes, I do suspect that the brain is
> KCC - and more formally, that the brain approaches in its complexity
> the complexity of the problem(s) it evolved to solve.
> Which brings us to a specific aspect of Kolmogorov complexity which is
> directly relevant to neuroscience, and that is the relationship
> between the complexity of the solution to a problem and the intrinsic
> complexity of the problem itself. In his book "Vision", Marr proposed
> (in what he referred to actually as a 'bottom up' approach to
> understanding the nervous system), that one must first understand the
> nature of the computational problem being solved, and then consider
> the set of algorythms that could solve the problem and then and only
> then, look at the particular instance of that set implemented in the
> brain (under the assumption that the brain was not KCC). Complexity
> theory provides a formal framework (oh that again) for considering the
> relationship between the inherent complexity of a particular problem,
> and the complexity of the solution to that problem -- I believe (and
> someone will surely correct me if I am wrong), there is a fundamental
> principle that the complexity of the problem sets a kind of floor for
> the complexity of the solutions to the problem. That is, you can
> find more complex solutions to a problem -- but you can't find a
> solution with less complexity than the problem itself, if you did,
> then you could recast the original problem in a less complex form.
> Accordingly, if the brain is KCC, then, by definition, solutions to
> real brain problems involving 4 input 'neurons', 10 in the hidden
> layer, and 3 output 'neurons' must be underestimating the real nature
> of the problem the brain solves. Or in other words, if the solution
> to the problem is less complex than the brain, then one has
> misunderstood the problem.
> In this context, Todd Troyer's post today makes the rather important
> point, that probably our greatest deficiency in studying the brain is
> that our understanding of the complexities of natural behavior
> significantly lags even our understanding of the structure of neurons.
> And yes, neuroethology is the branch of neuroscience that has made the
> effort to try to link behavior (and even natural behavior) to the
> brain. Linking to a previous post, most (although not all)
> neuroethologists study "simpler" systems. Furthermore, the roots of
> neuroethology are european, and place much more emphasis on innate
> patterns of behavior, than did american behaviorists, who did think
> that with the right combination of m and m's, animals could
> significantly stretch what they associated with what.
-------Rest of Jim's post deleted to reduce inbox clutter, GNR-------
More information about the Comp-neuro