Wednesday, March 16, 2005

Hum-Drum Gum

Listening to The Hounds Of Love Very Loud.

Tomrrow's People has turned out to be better than the reviews suggested. The first chapter was very Ghee-Whiz and I wonder if one of the reviewers got beyond it. However, I still see some unexplained examples where further clarification is required. For a long and dense book, such things seem important. I did keep feeling that the social and political implications of all her Brain-based IT were glossed over but maybe all that comes later in the book.

I can remember thinking of so many strange things that I wanted to mention here but as usual they have flitted away leaving only a strange empty feeling like a half-remembered dream. In fact I can remember the exact point on the journey home yesterday when I thought of one of them and now all I have is the mental image of the view through the windscreen. It was something deep about consciousness or physics. Phooey!

I have just remembered something from the Greenfield book. She says that a question that has plagued her is why nerves connected to a particular part of the brain give rise to audio processing and others linked to another part of the brain cause visual thoughts or smell etc. My immediate though (before she went on to mention it) was about synesthesia. Is it possible that the answer to the question is that the brain processes these impulses in a way which suits what the impulses are? Cross-over the connections and the brain will eventually learn to process the impulses correctly. I may have fallen asleep during later discussions but this seems obvious. The nerves to the brain have evolved in specific locations for all people but the brain is a more generalised organ which can deal with anything you throw at it. Greenfield also dismisses the idea of being able to produce implants to the brain which will allow external technical manipulation of brain states. In some distant improbable case, is it not possible that you could implant what you want and the brain would eventually learn to make some sense of it? Those slugs that developed some form of visually sensitive tissue all those years ago had to have a brain which could learn to make sense of the images. I don't want to go into all the Richard-Dawkins style discussions of how each tiny part of form and function evolves as a result of external stimuli, but having a trainable brain in place ready to accept the input of any further evolving structures seems elegant. I feel I must have missed something here and expect a call of clarification at any moment.

So, despite the reviews, I am aiming to finish this book and then go onto The Private Life Of The Brain which is entirely within Susan Greenfield's area of expertise. I still want to write a neural net. I am contemplating something to recognise characters but that may be a little ambitious. Even to get something trainable would be fun. So on to the stack of unfinished projects is popped neural nets. Note the correct use of the word popped - when did you last have to bother with stacks and queues? I don't think I have used any recursive procedures for years as all the stuff you have to do with them is already handled by either the OS or the development environment. I miss those five minutes of commercial machine code I did all those years ago. Actually I think it was probably Assembler if you want to be pedantic but it was still instruction - data - instruction - data and it was at the sharp end of a financial system. Probably still in place somewhere.

Who said I have a crystalline brain?

No comments: