Good point, Ron. Each of those neurons are acting in parallel, all functioning simultaneously. Trying to simulate this with a single threaded machine is just not practical. Another sort of architecture is required, like maybe using carbon rather than silicon *grin* Sort of like nanotechnology. When all you have is a hammer everything looks like a nail.
I remember as an undergradute speculating on a mixed mode analog/digital machine, as a precursor to a dedicated AI engine. You might be able to prototype such a machine on one of those fancy programmable chips they have now (ASIC?) Something simple but massively paralleled is essential. BTW, I have no philosophical problem with implementing consciousness on silicon rather than carbon, I think it just emerges naturally from certain types of massively parallel computing systems. Like life itself, it just _has_ to happen, it can't _not_ happen if you got the right circumstances and enough time. K. -----Original Message----- From: Ron Wormus [mailto:[EMAIL PROTECTED] Sent: Tuesday, May 17, 2005 12:02 PM To: [email protected] Subject: Re: SciAm article on brain A nice overview (~200pages) of the brain and the emerging science of consciousness is: "Stairway to the mind" By Alwyn Scott; 1995 Copernicus in which he discusses the non-linear emergance of the mind from brain function. paraphrasing... .......How is the observed activity of the brain related to the activity of its consitituent neurons? The neo cortex is composed of ~ 10 billion neurons, each of which has 10,000 input connections which equates to an immense number (10exp110) raised to the10exp16 th power or the immense number (10exp110) multiplied by itself ten thousand trillion times! This is a combinatorial barrier that is much larger than those between physics & chenistry or between chemistry & biochemistry........ --On Tuesday, May 17, 2005 9:55 AM -0400 Jed Rothwell <[EMAIL PROTECTED]> wrote: > The article I referred to yesterday is K. Boahen, "Neuromorphic > Microchips," Scientific American, May 2005, p. 56. The relevant quote > is on the first page: > > "The brain does not execute coded instruction; instead it activates > links, or synapses, between neurons. Each such activation is > equivalent to executing a digital instruction, so one can compare how > many connections a brain activates every second with the number of > instructions a computer executes during the same time. Synaptic > activity is staggering: 10 quadrillion (10^16) neural connections a > second. It would take a million Intel Pentium powered computers to > match that rate -- plus a few hundred megawatts to juice them up." > > So computers are already within a factor of 1 million. Perhaps they > will have to come within a 3 to 5 orders of magnitude before they > begin to look intelligent to us. I think they will also need > radically new software. My sense is that programs like Cyc will not > cut the mustard. I have no idea how long it will take. Anywhere from > 50 to 500 years, I suppose. Fortunately, the interim devices will be > profitable, so progress toward intelligent machines seems inevitable. > > - Jed > > > >
