IUBio Biosequences .. Software .. Molbio soft .. Network News .. FTP

Book Review - The Cerebral Code

Harlan Harris hharris at cs.uiuc.edu
Mon Feb 17 12:03:56 EST 1997

In article <5e8ou2$23e at lace.colorado.edu>, Danny Yee <danny at cs.su.oz.au> wrote:
>     title: The Cerebral Code
>          : Thinking a Thought in the Mosaics of the Mind
>        by: William H. Calvin
> publisher: The MIT Press 1996
>     other: 256 pages, glossary, references, index, US$22.50

I'd like to follow up on Danny Lee's review with a couple of comments and
questions. I too enjoyed reading the book (Calvin is one of the better "pop"
science writers I've read, and has very creative new ideas as well), but am
not yet convinced. Certainly the neuroscience is not yet available to either
confirm or deny more than a few issues related to Calvin's theory, but I'm
disappointed that apparently no work on computational modeling of these 
processes have been done. The thought-experiments are interesting and 
promising, but a very clear goal for future research seems to me to be 
evaluating the theoretical power of the mechanism proposed. Can this model do
everything it is asked to do? Are there homunculus problems with setting up
the Darwinian landscape (the barriers Calvin talks about)? Humans, with 
training, can do formal logic; can this model do that? I've been unable to 
find anything on Calvin's web page, or elsewhere, about these types of 
questions. Does anyone know if that type of research is ongoing anywhere?

On another, more general note, I'm very very pleased to see a model that
describes how a neural system could run what Dennett and others have called
a "symbolic virtual machine". While connecting neurons to do feedforward 
symbolic/logical calculations (by building logic gates, etc) is trivial, it'
s also pretty clearly not what happens given the time scales involved (right?). 
Calvin's system does moderately complex (for some as-yet-unknown definition 
of "complex") symbolic computations without trying to deal with the very 
subtle issues of the neurochemistry of individual neurons, LTP, etc. This 
seems very promising to me in a general sense. Does anyone know of other 
attempts to do something similar? That is, to build a virtual machine out of 
neural components?


Harlan Harris                 Univ. of Illinois at Urbana-Champaign
hharris at cs.uiuc.edu           (217) 244-5974

More information about the Neur-sci mailing list

Send comments to us at biosci-help [At] net.bio.net