Hi, This article
http://sciam.com/article.cfm?chanID=sa006&colID=1&articleID=0006AD38- is interesting; it describes new neuroscience results regarding the existence of communication networks among glia in the brain. This kind of discovery reinforces my feeling that, for AGI work these days, it's best to steer clear of neuroscience except as a general source of conceptual inspiration. We just don't understand how the brain works very well yet. This year it's intercommunicating glia --in 2005 or 2006, who knows? If someone wants to base an AGI system on how the brain works, they really just have to wait another N years to get started ... they'd be better off to focus on building narrow-AI systems to help neuroscientists analyze their data... However, this of course doesn't mean it's useless to create computational models inspired by the brain. "Neural net" computer programs have been very interesting even though they bear only slight resemblance to biological neural networks. I recall that in my speculative model of neurocognition, Hebbian Logic, http://www.goertzel.org/dynapsyc/2003/HebbianLogic03.htm I hypothesized a kind of reinforcement learning mechanism operating at the level of *neuronal cluster* rather than individual neurons. I called this "semi-local learning." Hypothetically, this newly discovered glial communication network could play a role of enabling semi-local learning vaguely similar to what I hypothesized in my paper. -- Ben G ------- To unsubscribe, change your address, or temporarily deactivate your subscription, please go to http://v2.listbox.com/member/[EMAIL PROTECTED]
