X-Message-Number: 441 From: att!fernwood.mpk.ca.us!alc!lovejoy Wed Sep 11 19:28:33 PDT 1991 To: fernwood!att.att.com!whscad1! Subject: Re: Consciousness Since none of Searle's neurons are conscious or know english, it follows by his own argument that he is not conscious and does not know english. Consider the following question: is the information recorded on a videotape by a video camera symbolically encoded? If so, then the information in neurons (synapses, brain cell membranes, etc.) is also symbolic. All of it. I agree that the "information-theoretic" approach taken so far by "AI" researchers is way off course--but not because it relies on symbolic computation. The problem is that the semantic level is far too high. Neurons don't deal in high level concepts for which human languages have words. Their semantic space is far, far more primitive. Human consciousness and intelligence are functions of complex, hierarchically organized neural networks of billions of neurons. Attempting to model or simulate such a system using a machine to evaluate words and sentences is like trying to predict the weather using gross measurements of temperature, humidity and barometric pressure. Such abstractions are too coarse by orders of magnitude to be really effective. Words and sentences no more cause intelligence and consciousness than clouds cause weather. These things are effects, not causes. --alan Rate This Message: http://www.cryonet.org/cgi-bin/rate.cgi?msg=441