Do I really have to say this again? Have you not read what I have been saying?
Whatever my tech reasons are... They don't matter! Because you and nobody else scientifically knows that you don't have to. To prove you don't have to do what I suggest you actually have to do it, and contrast with computers. Not assume it. You don't know. I don't. Nobody knows. That's my point. This "I can't see why ...." is not an argument. Do the science and come back with "I have done the experiment and now I know that computing models of intelligence and natural intelligence are identities under all conditions." Please go back through the relevant threads and see this position stated time and time again. Your position has no science basis. You don't know. I don't know. Nobody knows. The science is not done. To do the science is what I suggest. And I also suggested how the premise you have can be false. Do I have to say this again for anyone? Do you not see how this problem works? In asking me what you have you deliver scientific evidence of the problem. Cheers Colin -----Original Message----- From: "Benjamin Kapp" <[email protected]> Sent: 14/05/2015 9:13 AM To: "AGI" <[email protected]> Subject: Re: [agi] Re: Starting to Define Algorithms that are MorePowerfulthan Narrow AI Why can't you model the thing your building in a computer? Why can't you model the fields of the mind? Do you believe there is something necessarily material about minds? Isn't this disproved by cybernetic implants which have provided people with restored cognitive functionality despite being of different material? On Wed, May 13, 2015 at 7:00 PM, Colin Hales <[email protected]> wrote: Hi Rob, I had this realisation in 2001/2. At that point I put my ear to the ground and I looked. Hard. In 2004 in frustration I joined academia and had unlimited access to the entire world's publications. Doing a PhD, I kept up with folk at the bleeding edge in lectures, seminars and workshops and conferences.....I soaked through all sorts of 'tech announcement' analysis and 'breaking' stories. My ear is still on the ground. I wait. Today I still wait. Sometimes in some weird materials lab someplace an announcement is made that has keywords that might be construed as along the lines of my proposition. I then look and have, so far, found nothing. You know what usually happens? "Breakthrough X happens in materials or quantum mechanics. Woohoo!" Headline. Then everyone gets excited and says wow! "We may be able to solve the AI problem when we build a new computer with it" ...... and thus they throw a potential solution at failure. Time and time and time again. Vanadium Dioxide is my favourite exemplar. Recent materials for memristors another. There was yet another of these literally this week! http://www1.rmit.edu.au/browse/RMIT%20News%2FNewsroom%2FNews%2FMedia%20Releases%2Fby%20date%2FSep%2FTue%2030/ Go and look, excited...and yet again...no cigar. Science has begun to make materials that can solve the AGI problem. What they are not applying it to is that right solution. There are materials that can do what I want to do. Vast nonlinear control systems. But nobody ever chooses to solve the problem with it. Instead everyone thinks "lets build a computer" Fine. Computing is great. It's just not the solution to AGI! >60 years of trained-in habit entails a systemic blindness to the way science >was traditionally operating: by building it to understand it By knowing that >unless you could build it you don't understand it. And by 'building it' I do >not mean use a computer! Instead of letting nature do computation computers >and computing model it. Not the same thing. As I write this I can hear the >reader's brain grind on my words. How can they be different, you think? Well >in exactly the way I have described in all these posts. and nonstop for over >10 years. So ironically now we are overprepared for real AGI and the only thing stopping this happening is us. We keep choosing not to solve the problem. Instead it's if only we had a computer powerful enough" and "Moore's law blah blah...." receding rainbow of failure. I intend to write a book on this issue! It's bizarre. Maybe someplace there is a lab that does my proposition and it's all tucked away. I doubt it. You know why? Because bodies like DARPA keep throwing $gazillions at doing it with computers (this includes all existing neuromorphic chips of any kind.....where models of reality stand in for reality). Unless this is a massive smokescreen or unless left hands and right hands are not talking at a breathtaking level (conspiracy theory bollocks).... then this indicates that out here in the real world of people and a world overdue and in desperate need of it, we have literally programmed ourselves (in tacit culture) to fail in AGI and appear to have actually locked ourselves in a failure loop.... Then I turn up, after decades of thinking about robots and doing control systems in business.... and because I am old enough to have seen how it used to be....and.... because I was not in science, I had none of the programming. And I say "hey guys why don't we try this?! (how it was done for 350 years before computers)." And guess what? Here I am in 2015 saying the same damned thing. And all it is is what the original cybernetics folk would have done had computers never been built. And by now AGI would be real had they continued (the likes of Ashby et. al.) without computers and with the neuroscience we have now. Indeed today, neuroscience itself would look entirely different had this happened. So the damage is not just confined to AGI. People have been hurt because of the lack of knowledge. The failure to look at solving AGI without computers has actually hurt people. Sick people. Some days I wish I had never seen anything, and taken the blue pill and re-joined everyone in the matrix. So here I am, some kind of Morpheus with a red pill and ... yeah metaphor overdose. You get the picture. cheers Colin. (This email or something like this will appear in the new book) On Wed, May 13, 2015 at 5:36 PM, Nanograte Knowledge Technologies <[email protected]> wrote: Colin Fascinating thread and subject matter. Just a general question please. How certain are you that some governmental scientists somewhere have not already done this research and constructed such bio-machines? You may be surprised, or disappointed even, to find that you're not the only person on this list who thinks along these lines. Publications, as indicators of technological progress, usually are a few years behind the actual times and hardly-ever reflect the true state-of-the-art research, e.g., the GRAPE system. I'm asking specifically, because I noticed quantum-detailed publications in this field around 2009/2010, which trends well with field-test-ready prototypes for 2014/2015. The other reason I'm asking is because I've been studying a particular phenomenon, which systemic behaviour you might actually be describing to me. Unfortunately, details are subject to a commercial NDA, etc. Looking forward to your reply. Rob Date: Wed, 13 May 2015 09:11:33 +1000 Subject: Re: [agi] Re: Starting to Define Algorithms that are More Powerfulthan Narrow AI From: [email protected] To: [email protected] On Tue, May 12, 2015 at 7:24 PM, Steve Richfield <[email protected]> wrote: Colin, Two quick thoughts: 1. Your description of ion channels sounds a LOT like a Hall-effect device. I suspect that ion channels may be **VERY** sensitive to magnetic fields!!! Aside from implementing natural compasses, Hall-effect may be a part of their computational functionality. Note in passing that Hall-effect devices are FAST, so it may not be beyond reason that there might be some really high-speed analog computation going on in ion channels!!!. Individual channels have a (relatively) slow stochastic nature. You need about 10 tightly bunched. All 'computation' then sits atop that overall average regularity, resulting in both types of signalling that then do all relevant computations. See the book HILLE Ion Channels of Excitable Cells. I don't have to bother with the stochasticity. I can build filamentary currents that get straight to work fast. Currents that then produce the same 2 signalling types. 2. You might be able to model some of the things your are thinking about with a fish tank full of salty water and structures made of Play Dough. You will also need a battery, a voltmeter, and some insulated wire with exposed ends. Electrolytic tanks have been used to model many complex EM things. Fishtank full of Gatorade and playdoh and radioshack toy instruments.... bliss!!!! Yay!!! I knew this had to become fun eventually!! Can I use a 3D printer too? :-) Steve On Tue, May 12, 2015 at 2:04 AM, Colin Hales <[email protected]> wrote: Hi again, Yes the potential drops off as 1/r and the dipole as 1/r^2 as you say. Not the field intensity. That is 1/r^2 and 1/r^3 resp. But this is irrelevant. Don't confuse potentials with the fields. I wrote an article on this Hales, C. G. and S. Pockett (2014). "The relationship between local field potentials (LFPs) and the electromagnetic fields that give rise to them." Frontiers in Systems Neuroscience 8: 233. http://journal.frontiersin.org/article/10.3389/fnsys.2014.00233/full The line source you mention doesn't actually contribute to the field system in any functional sense for subtle reasons. This is another broken aspect of the thinking. You have to deal with the actual physics of ions in water and in ion channel pores in space and the details of the charge transport as applied through Maxwell's equations,,..NOT the physics of a model. Just because a resistor is in a model and predicts voltages correctly does not mean that the fields in nature are the fields of a resistor. In general: the physics of the field system is not the field system of the circuit element models. The same total current has 3 lives: 1) Intracellular 2) Transmembrane and 3) extracellular. In terms of contribution to the actual functional field system (2) Dominates both (1) and (3). To see this: The ion transit speed and transport dynamics in the extracellular space and intracellular space is 10000-50000 times slower than transmembrane and radically diffuse and diluted. Almost non existent as a charge density. It is the electric field that matters and when you do the math the field due to the axial current (line source) is negligible because the current does not involve a functional charge density even though the total current is the same. ergo negligible E field contribution. In contrast, the transmembrane portion (of the exact same total current) is radically confined to an Angstrom-level pore-width and along a path length in a very particular direction 20-50 times longer than anywhere else in tissue (through the thickness of the membrane). The transmembrane ions are like bullets from hundreds of parallel machine guns in comparison to traffic in the extracellular space and the intracellular space, where ions are confined by water to almost zero path length and bounce in totally randomised directions. None of this detail is in any circuit element model. It is charge density and current density (not current) that matter for field generation. Charge density and current density are radically different in each phase of ion transport (1), (2) and (3). Hence they produce different fields. I am doing the full convective simulations of this over the next few months. The failure, over decades, to look at the actual ion transport mechanisms in the ECS and ICS and contrast them with the transmembrane ion channel current has caused yet another stuff-up in understanding the field system. The only people that actually know this are in microfluidics and it is a modified form of microfluidics equations that I will solve (with the water flow velocity set to zero). When you actually compute the magnitude of the real electric field produced by the transmembrane ion traffic as totalled by tens of 1000s of cells within in a 500um radius sphere they can easily add up to that needed to effect each other even though the field drops off as 1/r^3. This is a very short distance. It is the gradient of the potential, not the potential that matters. The E field is a very complex vector sum that dominates even though it drops off faster with distance. The E field in the Lorentz force does the work. You can choose a million exotic circuit elements and find a part of a neuron who's potentials may be modelled with it. That does not mean that the neuron 'is' one of those things. Its not diodes yet there's lots of diode like things going on. It's not a resistor yet there are lots of behaviours that obey resistor-like laws. You can view neurons through a model-lens made of SR or bar fridges and hockey sticks and igneous rocks that produces the same voltages and current. .... and on and on and on.....and you are welcome to do that to suit what you are doing. In none of it does it tell you what the actual natural material is doing in relation to EM fields. That is why I build what I will build. I build what the brain does, not what a model of the brain does. I can't help it if this is the way the brain is. If I found anything different I'd be building that instead. When I compute (1), (2) and (3) I'll send the results to the list. It'll be a while. Congrats! My work here of showing you the potholes on the road to understanding EM field origins is done. :-) I think we are officially grokked out. cheers colin On Tue, May 12, 2015 at 3:23 PM, Steve Richfield <[email protected]> wrote: Colin, You have described regenerative operation, which is a near-field sort of thing and not capable of sensing small things at a distance where signals drop off as r^2, HOWEVER, I just realized that the field from a line (rather than a small dipole) source, like from an axon rather than an ion channel, drops off LINEARLY with distance. Hence, at distances that are short compared with axon length, regeneration might be enough to work. I just didn't see any need to stick with a purely regenerative model, when SR completely sidesteps the limits of regeneration AND there is plenty of evidence of SR in neurons. Regarding the past tense of grok - it becomes past tense when you can no longer grok - like when you get Alzheimer's or die. Until then it is an active sort of thing, like your fields, and so remains in the present. Steve On Mon, May 11, 2015 at 8:09 PM, colin hales <[email protected]> wrote: Hi Steve, The fields originate in a dissipative evanescent dipole that exists as long as the action potential transmembrane current exists. EM field feedback is in modulation of distant network signal timing and propagation phenomena. Positive, negative whatever. It emerges at a higher organizational level that has nothing to do with the physics originating the fields. The magnetic field comes from a brief transmembrane current. The electric field is a result of a battle between diffusion and electromigration in the immediate vicinity of the ends of the very same transmembrane current. If the transmembrane current is large and long enough (requiring lots of collocated ion channels)... Then this causes a depletion of ion charge on one side and accretion on the other....dipole big enough to contribute to signaling at distance. It exists as a dissipative cascade that is momentary, stops and then equilibrium is chemically restored. Think of it as a capacitor discharge, stop, recharge. In the EM field feedback the moment of discharge is determined in part by impinging E field from elsewhere in the tissue. That may constitute a positive feedback from distances a long way away. Positive feedback also exists within the longitudinal propagation of the action potential. That is regenerative. Models usually depict this as resulting from potentials and currents. I suspect that it's actually the magnetic field that is very strong at distances of um. That magnetic field tickles distant ion channels located in [The entire original message is not included.] ------------------------------------------- AGI Archives: https://www.listbox.com/member/archive/303/=now RSS Feed: https://www.listbox.com/member/archive/rss/303/21088071-f452e424 Modify Your Subscription: https://www.listbox.com/member/?member_id=21088071&id_secret=21088071-58d57657 Powered by Listbox: http://www.listbox.com
