A recurring theme in philosophical discussions is the relationship  
between the Classical (facts and logic) and the Romantic (art and  
beauty). Zen and the Art of Motorcycle Maintenance is perhaps the most  
popular book on this subject (and possibly the only book I've read on  
it, I don't remember!). The difficulty in understanding one from the  
other was, if I remember, not well resolved.

This may be at the heart of recent disagreements on this list.

I am poorly qualified to speak on the issue myself, except that I am  
equally poor at science (despite a PhD) and art (despite now making a  
living from photography, and playing the horn). But I have a theory.

Many of us have some understanding of logic and the way that digital  
computers work. It is fundamentally very simple, just 1's and 0's, but  
the overall result can be very complex. Immensely useful, but I'd say  
that it is dry, predictable, capable of simulating or imitating  
creative acts, but incapable of what most of us mean by creativity.  
They are mere tools, the Classical at gigahertz speeds.

But there is another form of computing, the way our own brains work.  
I'd say few of us have a good idea about how this form of computing  
works. Someone too caught up in the Classical can easily dismiss  
"wetware" as imprecise, unpredictable, not easily programmed, not  
capable of "real" computing. But neural networks are vastly more  
capable in processing complex inputs to produce useful outputs.  
Digital computers are tinker toys in comparison. Even simple  
artificial neural networks (a few hundred connections) can decipher  
complex behavior that will flummox the Classical approach of a digital  
computer.

I got a sense of how this works many years ago when analyzing some  
complicated experimental results using an artificial neural network. I  
quickly got a plausible result that had eluded me for months using  
regression analysis. The nature of these networks is that it is  
difficult or impossible to understand what a trained network is doing,  
how it learned to do what it does, or whether it is doing it  
correctly. That is, the two forms of computing - digital and neural  
network - do not connect well. Even for a simple network. Extrapolated  
to human scale, with a trillion connections, and you have a way to  
view the difference between the Classical and the Romantic - they are  
the results of two very different forms of computing. Different in  
kind, and also in scale - the neural network is vastly larger and  
immensely more capable (although sometimes rather slow, certainly for  
me and my third finger).

Do not regard what I'm saying as putting a dry explanation to Beauty -  
our neural networks are *huge*, as profound as all the stars and  
planets of the universe. Amazing, incredible, awesome, unfathomable.  
And every one is different, a result of its exposure to many others in  
a complex environment. The kicker: just to encourage us to behave in  
certain ways, our neural networks let us know when we encounter Beauty  
and other good things (some bad things, too, like French fries). You  
might say that this is, in fact, the definition of beauty. Doesn't  
matter.

Finally, my conclusion: do not forgo the full power of your brain or  
the brains you are connecting to. Our "left brain" does a remarkable  
job simulating a digital computer, and we sure need it to read musical  
notation. But that is not where the Romantic comes from. 
_______________________________________________
post: [email protected]
unsubscribe or set options at 
https://pegasus.memphis.edu/cgi-bin/mailman/options/horn/archive%40jab.org

Reply via email to