Colin, OK, I think I see our point of confusion. I covered this point briefly before. I'll try again...
You seem to be stuck on a particular explanation involving exquisite sensitivity that is NOT needed to make super-sensitive systems, Further, exquisite sensitivity has some operational issues that seem to make it impractical. Even if it were there it would slow the system WAY down (over other better methods) as the point of exquisite sensitivity is searched for. Yes, things can be made super-sensitive as you describe, like standing a pencil on its point. This is called "regeneration" because in some way 99.999% of the output is fed back into the input (with the pencil it is 100.000%, which is why you can't stand a pencil on its point). In the case of the pencil, the energy that is used to regenerate is gravity. The problem with regeneration in neurons is that it is slow (like the time it takes to stand a pencil on its point) and not readily adaptable to responding to signals, e.g. specific correlations, frequencies, etc., though regenerative receivers have been made. However, super-regeneration (SR) sidesteps the limitations of regeneration, and does NOT need any sort of exquisite adjustment to achieve its nearly limitless sensitivity. It is the adjustment that makes regeneration slow, so SR can respond immediately without the "settling time" of regeneration. In SR, the pencil is let drop, then quickly stood up and let drop again, and again, and again. The output is the time to fall, rather than slowly searching for the point of perfect balance. Microscopic changes that would be seen as shifts in the pencil's balance instead become changes in the time to fall in its rapidly repeating cycle. It isn't just cells that repetitively fire. Dendritic structures have also been seen to do this. I suspect that this extends all the way down to individual ion channels, but this would be SO low-level that no one would have noticed this if they weren't specifically looking for it. Suppose for a moment that the "output" of a repetitively firing structure like a neuron (or dendrite or ion channel) is NOT its voltage or firing rate, but rather is the VARIATION in its output or firing rate. This would dovetail perfectly into SR methodology. Integrating would provide the signal without an absolute reference, like capacitor coupling, which would be NO problem where the signals represented the logarithms of things like probabilities. Neurons are well known for dynamically adjusting their sensitivity. I suspect that this is NOT looking for some point of exquisite sensitivity, but rather to normalize their firing rates, etc., to avoid metabolic problems attendant in firing too fast, and response time problems of firing too slow. Anyway, SR is but an implementation detail overlaying your basic theory. For me, it makes your theory believable, and provides a general framework for the multi-level channel-synapse-dendrite-cell-column-region-hemisphere-brain-family-society emergence of intelligence. Please grok SR before dismissing it. Steve ======================= On Sat, May 9, 2015 at 5:27 PM, colin hales <[email protected]> wrote: > Ok gotcha. > > 2 issues. > > 1) the EM field coupling has at least some empirical verification in the > lab. I can cite the papers if you want although I suspect you know them. > So there is your first glimmer of proof that the fields are strong enough > to be causally active. > > 2) the effect is 'butterfly'. That is, 1:10^5 to 1:10^6 will, if > persistent, rapidly involve themselves in entrainment of neuron firing. > This occurs at the axon hillock poised right on the verge of firing. > Hypersensitised. > > My own calc show that only certain cell densities will achieve this. Only > certain bran regions and morphologies will do it. > > This is about state trajectories around complex time-varying stable and > unstable equilibrium points where sensitivities are exquisitely high. That > is where EM coupling comes into causal range... When 50000 cells all near > each other cohere even briefly and create a 1 v/m field on top of the > 10,000,000 v/m field in the initial segment/hillock. It only has to > perstently shift (advance/retard) firing times by a tiny amount to cause a > massive change in state trajectory. Such things as Hopf bifurcation points > do this kind of radical state change with tiny changes. > > Finally, to go into the more esoteric end of my proposal.... Even when the > fields are too small to do local change, they are vastly greater than all > the chemical noise of cell componentry. > > And I hypothesise that they deliver consciousness from the 1st person > perspective of being the fields. I am looking at how a 'virtual' feedback > loop, at the level of whole percepts, can additionally impact the state. > > Imagine a visual scene that is expected and the actual visual scene > superposed on each other and that mismatch. That mismatch is a distributed > thing involving many neurons that may collectively be biased in the manner > already described. > > What I am looking at is tissue elegance at a spectacular level. Tissue can > contribute to conscious AND be/not be causally efficacious > > Or > > Tissue need not contribute to consciousness at all AND yet be causally > efficacious. All based on cell density and morphology.... > > Once you see how this can work and then do the numbers, finding it > possible.... > > Then you end up in my position, embarked on testing it inorganically. > > This is vastly more complex than any previous view of the brain. The > overlapping, superposing dynamic interference patterns in millions of cells > contains a vast amount of information that is currently thrown out. > > Imagine how much information there is merely to specify the exact > morphology of a cell. Now imagine how the fields embody, literally, that > information. > > Now imagine 10,000 cells near each other. 10,000 _factorial_ different > field systems not counting any modulation by even 1 synapse. Just by the > cell firing. Not including subthreshold oscillations. > > Can you see how vastly underestimated brains have been? Is it any wonder > the story of the last half century of AGI might have underestimated brain > tissue complexity? > > I hope you can now see this. > > The good news is that, if you stop using computers, accessing to the same > complexity is straightforward. > > And ... To hark back to the Wright bros even tho I promised I wouldn't, > if, in a hundred years we have a computer that could compute a model of it > all .. It would be an AGI simulator, not an AGI. Because the fields are > INPUTS. Just like a flight simulator traces you about flight but is not > flight. > > That is how this could all turn out. > > Damn. Batteries. Gotta go. > > Cheers > Colin > > > -------- > ------------------------------ > From: Steve Richfield <[email protected]> > Sent: 10/05/2015 6:06 AM > > To: AGI <[email protected]> > Subject: Re: [agi] Restating Colin's Hypothesis > > Colin, > > I think you missed my point. I'll try again... > > The "information content" in the EM field is, except for that portion of > the EM field that is generated by adjacent structures, VERY low-level - too > low for it to be able to have much effect on ion channel chemistry > **UNLESS** there is some sort of positive feedback mechanism. To > illustrate, extracellular electrodes tend to only see what they are next > to, with everything else being apparent "noise" superimposed on that signal. > > Hence, if your EM theory is correct, then there absolutely **MUST** be > positive feedback to get enough gain to be able to make sense the "noise" > subtleties in the EM field. > > However, positive feedback is fundamentally an unstable thing, unless > there is some "clever" mechanism that sidesteps the complex stability > issues. Automatic Gain Control (AGC) works well in systems that do NOT have > positive feedback, but the sudden onrush to lockup as feedback approaches > 100% makes AGC approaches unworkable for positive feedback systems. > > Super-regenerative (SR) approaches sidestep these problems AND exhibit > properties well known in neurons, like spiking. SR operates just BEYOND > 100% feedback, squelching high-level oscillation when it occurs, with the > "output" being the rate of squelching (spikes in the case of neurons). > > Yea, people associate SR with glassical electronics, but it is still alive > and well in modern devices now being manufactured, e.g. keyless security > systems. A Google search finds more recent applications than glassical > applications. > > The reason SR never really dominated the shortwave receiver market was > that its operation only reflected its input over a small portion of its > "cycle" of repetitive oscillations, so while it was able to produce full > output regardless of how weak input signals might be, as input signals got > weaker, the output got noisier FASTER than competing approaches like > superheterodyne designs. Of course superheterodyne systems utilizing AGC > were MUCH more complex, but people gladly paid for a few more tubes to > better hear signals from around the world. > > So, how can a seemingly simple 2-tube radio possibly drive a speaker from > micro-volt signals received from around the world? THAT is the sort of > combination of operational simplicity combined with limitless sensitivity > that ion channels MUST have to work as you envision. > > So, it seems to me that you must make up your mind here. If ion channels > are able to respond to subtleties in EM fields then there MUST be some sort > of positive feedback mechanism that operates more or less independently of > input amplitude, or the presence of high-level signals that are NOT of > interest to a particular ion channel. If you don't think it is SR, then > what else DO you think it might be? > > I don't (yet) see any other candidates. > > Thoughts? > > Steve > ================= > > On Sat, May 9, 2015 at 2:13 AM, colin hales <[email protected]> wrote: > >> I remember those circuits.... You sometimes can find them in old books >> that also speak of the 'aether' [image: 😊] >> >> It's not a route to brain-mimetic fields. >> >> All we really have to do is replicate the fields of the membrane. I >> already know how they are generated. They 100% consistent but non-uniquely >> related to all the familiar compartmental equivalent- circuit models. The >> potentials they express have complex Hodgkin -Huxley and Fitzhugh-Nagumo >> non-linearity. Paradoxically you can make these nonlinearities with >> old-school valves! >> >> But the fields are out in the space between cells and have complex power >> law spatiotemporal dynamics and vast information content. >> >> I'm concerned more with the system underlying action-potential signaling >> at the moment. Without the fields. Designed in a way that makes the >> addition of the fields a relatively simple thing. >> >> If only things were as simple as those old circuits! >> >> Cheers >> Colin >> >> >> ------------------------------ >> From: Steve Richfield <[email protected]> >> Sent: 7/05/2015 7:15 PM >> To: AGI <[email protected]> >> Subject: Re: [agi] Restating Colin's Hypothesis >> >> Colin, >> >> Are you familiar with super-regenerative receivers? These are simple >> circuits - well within what we already know neurons can do, that extract >> subtle features of EM fields - like the energy at a particular frequency. >> This concept could be adapted to extract all sorts of subtle features from >> EM fields. >> >> These were in common use around WW1, and have returned in the form of >> various wireless devices. >> >> I thought maybe that understanding these might get your reverse >> engineering the brain "juices" flowing. >> >> Steve >> =================== >> >> On Wed, May 6, 2015 at 8:21 PM, Colin Hales <[email protected]> wrote: >> >>> Hi all, >>> I'm struggling to find time to attend here. My appearances will be >>> patchy. >>> >>> Your restatement of where I am is heading in the right direction. That >>> is, you're grappling with the beginnings of the right ideas. Which is good >>> to see. My goal here is simply to get the general idea of AGI as a >>> computer-less adaptive control system based on brain physics implemented in >>> inorganic crystalline solid form. Artificial brain tissue. >>> >>> Imagine a cascade of intricately dynamically nested resonating loops >>> each triggering downstream cascades that branch and converge ... a massive >>> parallel state machine where the entire thing is intrinsically dynamic and >>> its primary physics is EM fields. There are a virtually infinite number of >>> ways that any dynamic cascade can happen. All because the of EM field >>> physics. >>> >>> Whatever it does as a form of 'computation' is *emergent*. I actually >>> don't care ahead of time what that 'computation' looks like. The only thing >>> I know for sure is that it's a product of a massive quadrature resonance. 2 >>> axes. Action Potential (is the main slow strong loop) and EM field coupling >>> (that weakly links loops orthogonally through space at the speed of light). >>> I am replicating both these axes, not computing any model of any >>> computation it might appear to produce. The dynamic interaction between >>> these two axes (both of which are actually implemented in the one >>> single unified physical EM field sytem)* and its self-modification of >>> its own dynamics* that is the reason why this is better classed as an >>> adaptive control system. >>> >>> With respect, Feymann diagrams are irrelevant here unless you can tell >>> me the science of '*what it's like to be a Feynmann diagram*' from a >>> 1st person perspective. It's just the virtual photon exchange diagram of >>> EM fields. Nothing to it. There's no esoteric quantum states or other >>> magic. Classical wave mechanics. Of course it's all ultimately quantised EM >>> fields at the finest scales. But I don't care - as long as I recreate the >>> brain's fields the way the brain creates them and uses them (causally) >>> ....voila ....I get the QM/space deep structural fabric contribution for >>> free. I am, in effect, involved in an experiment that this very fabric is >>> important (indeed essential) and I am do it to find out why it is important >>> *without >>> knowing fully *how it works or what it is. Like we do in science. >>> Building it to understand it. Make fire to understand combustion. Like that. >>> >>> The most important concept of all is that what we humans are, literally, >>> is 'being' this field system. We are not 'being' any apparent computation >>> that someone might characterise, by observing the fields. My hypothesis is >>> that 'being' the field system in the process of complex resonances is >>> essential and non-optional in cognition and intelligence *in the same >>> way that air/flight surface interaction is essential to flight and fuel >>> chemistry is important in combustion*. I do not claim to know the truth >>> of this yet. But I also know that *nobody else does either*, including >>> everyone on this list. If this hypothesis is upheld then it means that >>> computer-based cognition has the same relation to AGI as a flight simulator >>> has to flight - that is, it is a way of designing it and understanding it >>> but it is not flight. It's just that it's obvious when flight is absent >>> (CRASH!) or underperforming. A mature understanding of this >>> underperformance is what I seek. >>> >>> Whatever it is that is in the brain, from a 3rd person perspective its >>> just Maxwell's equations doing their thing. What no physics currently >>> explores is the 1st person perspective of *being *the field system. I >>> have worked out a way of viewing the 3rd person membrane-centric fields >>> from the 1st person and how it might be ;like something'. I published it >>> here by Trojan Horse in the last section of >>> >>> Hales, C. G. (2014). "The origins of the brain's endogenous >>> electromagnetic field and its relationship to provision of consciousness." >>> *Journal >>> of Integrative Neuroscience* *13*(2): 313-361 >>> >>> Using this technique it is possible to 'be' the membrane and 'see' a >>> massive blizzard of highly structured interference pattern of 'virtual >>> bosons'. Frome the 3rd person all it looks like is membrane and ion >>> channels doing their thing. It's a perspective shift, nothing else. >>> >>> But to get my basic intent you don't have to bother with this idea. >>> >>> The brain goes to spectacular levels of trouble to create this massive >>> unified EM field system and all we have done for 60 years is *throw it >>> away at the get-go *with software and lumped-element electrical circuit >>> models of voltages and currents. The *existence *of the brain's field >>> system is as old as any measurement of any potential in any tissue. It >>> blasts out the scalp as witnessed in EEG and MEG. It has 8 orders of >>> magnitude of spatiotemporal structure. It has a causal role as verified in >>> recent wet neuro experiments. Yet 'being' the fields is an aspect >>> unexplored. That is what my experiment aims to verify. That the fields are >>> not optional. >>> >>> Ion channels >>> * in localised bundles called synapses -- chemical and electrical/gap >>> junction >>> * in localised bundles called the axon hillock or initial segment >>> * in localised bundles called the node of Ranvier >>> .... originate the EM field system in both supra-threshold and >>> sub-threshold dynamics ... thereby cause transmembrane electric field >>> dipoles AND a membrane-plane circulating magnetic field. The cascade of >>> looping *down the membrane* is Action Potential signalling >>> (longitudinal). The expression of field orthogonal to the membrane is the >>> EM field coupling (transverse field). It is the Lorentz force action in the >>> transverse axis that gets called 'ephaptic' coupling by some. >>> >>> So the bottom line is that I will throw myself at the mercy of the >>> experimental test of this hypothesis and live with what it tells me. You >>> just have to get used to the idea that I can't talk about computation or >>> algorithms or software. This is AGI done without any such things. They are >>> meaningless in the context of the experiment and the design, I can >>> understand how the mainstream computer-based AGI movement may find their >>> eyes bouncing off the idea. Culturally foreign ground. Maybe in time the >>> two approaches will sit alongside each other in a more mature understanding >>> of AGI. We'll see. >>> >>> Better go. >>> >>> cheers >>> colin >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> >>> On Wed, May 6, 2015 at 6:07 PM, Steve Richfield < >>> [email protected]> wrote: >>> >>>> Colin, >>>> >>>> I'm going to take a shot at restating your hypothesis in a more >>>> physics-tractable form. The remainder of this posting are what I think you >>>> are trying to say: >>>> >>>> Colin in effect says that the computational unit is NOT the synapse, >>>> but rather is the ion channel. These are MUCH more numerous than synapses. >>>> While the voltages seen in extracellular recordings are quite low, the >>>> field GRADIENT near an active ion channel is HIGH - enough to have major >>>> effects on nearby/contacting structures. Brains are a lot like a bowl of >>>> spaghetti, and every place the "noodles" touch becomes a point of high >>>> field interaction. We don't yet know what those interactions do, but we DO >>>> know that there are a lot of synapses that interconnect contacting neurons, >>>> so at minimum such points of contact are probably capable of spawning >>>> synapses, if the "data" indicates a synapse would be useful. >>>> >>>> Then there is the far-field effects from neurons that are near but NOT >>>> in contact. The activity (or lack thereof) should be an important parameter >>>> to use in development, because it is an indicator of just how successful >>>> learning has been throughout the entire system. Where learning has been >>>> UNsuccessful, neurons should probably be more plastic in their >>>> functionality. >>>> >>>> Ion channels are capable of fairly complex computation, including >>>> memory (from ion accumulation and physical alterations), nonlinearities, >>>> etc. It has previously been presumed that ion channels are just "pumps" >>>> that keep neurons doing what neurons do, but the prospect for ion channel >>>> computation can NOT be ignored. >>>> >>>> When a neuron becomes active, its many ion channels radiates complex >>>> patterns of field-gradients, which could affect the operation of other >>>> nearby neurons, especially if the ion channels in the other neuron were to >>>> align themselves with a radiating neuron. >>>> >>>> While I now grok the importance of field gradients generated by ion >>>> channels, I still don't see how/why this should affect consciousness any >>>> more than it affects the many other functions of a neural systems. I am not >>>> yet even convinced that consciousness exists - except in our minds as a >>>> simplistic model for whatever happens behind our eyeballs. How do you link >>>> consciousness (over other neural functions) with EM fields? >>>> >>>> C'mon; help me put Colin's hypothesis into a solid physics form. >>>> >>>> Steve >>>> >>>> >>>> >>>> >>>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >>>> <https://www.listbox.com/member/archive/rss/303/11721311-f886df0a> | >>>> Modify <https://www.listbox.com/member/?&;> Your Subscription >>>> <http://www.listbox.com> >>>> >>> >>> >>> [The entire original message is not included.] >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> >>> >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > <https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac> | > Modify > <https://www.listbox.com/member/?&;> > Your Subscription <http://www.listbox.com> > -- Full employment can be had with the stoke of a pen. Simply institute a six hour workday. That will easily create enough new jobs to bring back full employment. ------------------------------------------- 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
