Mike, Colin, et al, I still see absolutely NO evidence that fields, Hall effect, etc, etc., aren't just part of a computational process, just like the nonlinearities and thresholds of FETs, etc., are just part of conventional computers.
A more appropriate comparison would be to Analog Computers, that produce VERY precise results despite using very imprecise components having all sorts of strange electrical characteristics. WAY back at the 1st NN conference in San Diego I published an article explaining how dendrites operating independently could by performing as part of an probabilistic OR of probabilistic ANDs function - that a particular neuron could be performing several unrelated functions - one for each dendritic structure. Since most propositions are almost always FALSE, the noise introduced by presenting the probabilistic OR of several such functions would be washed away when the outputs are then ANDed with other such functions. The benefit here would be a LARGE reduction of metabolism - simple power and space reduction. Implementation detail. However, I don't see any benefit for AI/AGI systems copying this approach - unless there ends up being a really high overhead for bus drivers. However, this discussion DOES address how some functions that we KNOW are being performed might be working despite our not finding any mechanism when looking through microscopes. A good example is mutual inhibition, that the Hall effect would explain nicely. Another good example is stabilization - with SO many interconnected high gain components, why don't brains simply lock up or oscillate uncontrollably (which some do, which we call epilepsy). Field effects would explain this nicely. So, I am taking a middle position - that field effects have a BIG part of neurological computing, but I still believe the result to be fairly precise analog computing and NOT some sort of field process that doesn't involve fairly precise computing. Look at these effects as components and NOT part of something mystical. Of course this means that there is more functionality than many people thought was there, but I don't know ANY researcher who thought that we already had most of the pieces of this puzzle. This stuff DOES affect AGI, albeit slightly, in that AGI will have to deal with many/most of the same issues if/when it progresses close enough to optimized functionality to experience lockup, oscillations, etc., and the operation of fields, Hall effect, etc., will probably provide some useful programming clues - to programmers are astute enough to pay attention. Am I alone in this belief? *Steve* ================= On Wed, Mar 15, 2017 at 5:10 PM, Mike Archbold <[email protected]> wrote: > jeez, talk about a sketchy conjecture: > > "Mehta adds that the fact that dendrites are about 100 times larger in > volume than somas, it's possible that our brains have 100 times more > capacity to compute information than previously believed." > > I've never quite bought into the brain being a "computer that is > processing information" either. > > I've always liked reading Colin's and Dorian's myth busting comments > over the years. Keep it going.... > > Mike A > > On 3/15/17, Colin Hales <[email protected]> wrote: > > All good with that, Ben. This is a situation that is not going to correct > > itself overnight. All I can do is be its champion and raise awareness. > > > > There's nothing major left to come out that I can think of. I'm sure > > there'll be minor refinements and special subtleties. Nothing major. > Happy > > to be wrong tho. All the major things I've expected have now come true > and > > reached the literature. Which is a relief. Happy to hear that I've missed > > something but all the really big stuff is done. > > > > BTW, ironically the one time the AGI conference is in Melbourne I'm on > the > > other side of the planet in Sweden. Otherwise I perhaps could have > > presented something on this. Another day. > > > > cheers > > colin > > > > > > > > On Wed, Mar 15, 2017 at 5:19 PM, Ben Goertzel <[email protected]> wrote: > > > >> > >> Obviously, these interesting new neuroscience results should be exciting > >> and disturbing to anyone attempting to build AGI based on the > >> "conventional" perspectives of neuroscience ... but for those of us > >> building AGI based on non-neuroscience-oriented considerations, they're > >> not > >> terribly relevant... > >> > >> It has been my view since forever that we don't know enough about the > >> brain to meaningfully base our AGI designs on current theories of brain > >> function... these new findings about dendrites validate this view for > me, > >> but certainly don't change the situation (i.e. this is not the last > major > >> upheaval neuroscience is gonna experience en route to a full > >> understanding > >> of the brain) > >> > >> -- Ben > >> > >> On Wed, Mar 15, 2017 at 2:13 PM, Colin Hales <[email protected]> > wrote: > >> > >>> AGI winter? > >>> > >>> No way! Summer is afoot. :-) > >>> > >>> We just have to step back and question 60 years of presupposition and > be > >>> open to the reality of the answer. There's enough evidence now to > >>> justify > >>> being deeply suspect of the current approaches. > >>> > >>> Not only is ephaptic coupling (EM field-based neural coupling) the > >>> second > >>> kind of brain signalling, a blizzard of DAP (dendritic action > >>> potentials) > >>> joins SAP (synaptic action potentials), which is equally capable of > >>> expressing massive quasi-static EM field wave mechanics and its own > >>> ephaptic coupling in neuropil. > >>> > >>> We've been working with only a tiny part of what the brain is doing for > >>> 60+ years. All the projections by Kurzweil? Out by 3 orders of > magnitude > >>> at > >>> least. > >>> > >>> Dorian and I might have different approaches to how AGI will come about > >>> (his organic, mine inorganic), but we both share a common and simple > >>> principle: the replication of the exact brain physics responsible for > >>> brain > >>> signalling. Not computing our way around a model of brain signalling. > >>> The > >>> latter is a chip with the physics of a model of the brain, not a chip > >>> with > >>> the brain physics in it. Totally different things that a different in a > >>> way > >>> that almost entirely neglected by science. > >>> > >>> In my (inorganic) case, it has only been since the noughties that we've > >>> had the tech to put the brain's signalling physics on 3D chips and let > >>> it > >>> interact with itself the way it interacts in the brain. We have no > >>> excuses > >>> for not doing this any more. We know what that physics is. > >>> > >>> Doing real AGI is a 'CERN-supercollider' scale outcome, but there's no > >>> winter implied in this. All it needs is $ and the will. Dorian is up > for > >>> it. I am too. I'm testing the design of such a thing (simulation) as we > >>> speak. If, as I suspect is the case, replicating the brain's signalling > >>> physics is the real answer, then the actual winter (failing for 60+ > >>> years > >>> to fully replicate all brain functions responsible for intelligence), > >>> may > >>> be seen in hindsight to have been in place all along and only now to > >>> have > >>> ended. That era produced all manner of wonderful useful things. But > not > >>> AGI. > >>> > >>> All the old EM field guard have died off. E Roy John, Pribram, Freeman. > >>> Sue Pockett has retired. JJ McFaeedn has left the building. I'm 61. > >>> Dorian > >>> and I seem to be the only representatives left alive that have the will > >>> and > >>> the knowledge to do what is needed: the resurrection of cybernetics as > >>> it > >>> was left for dead in the 1950s when computers were invented. That is > the > >>> real AGI summer. AGI done without computers. It's been sitting there > >>> waiting all that time. > >>> > >>> Maybe the $ will come. I am self funding but I can only go to phase II. > >>> Phase III (a few years away) needs a bucket o'cash and a chip foundry. > I > >>> have the ideas for the necessary robotics nursery and how to test it > >>> all. > >>> Maybe out there there's someone else thinking along the same lines. I > >>> don't > >>> know. But there's a way forward and it's different. A green field left > >>> fallow by the old cyberneticists. > >>> > >>> So no wintery thoughts, please! :-) > >>> > >>> cheers > >>> colin > >>> > >>> > >>> > >>> > >>> On Wed, Mar 15, 2017 at 4:31 AM, justcamel <[email protected]> > wrote: > >>> > >>>> Another AGI winter? > >>>> > >>>> > >>>> On 14.03.2017 03:11, Colin Hales wrote: > >>>> > >>>> Hi, > >>>> > >>>> Over the last 15 years, every 3 years or so we get yet another paper > >>>> that takes us all towards the centralisation of brain signalling on > the > >>>> EM > >>>> fields. Not _away_ from such an idea. _Towards_ that idea. > >>>> > >>>> This is merely the latest in that long vector towards EM fields as > >>>> central to brain operation. > >>>> > >>>> ============================================ > >>>> Moore, J.J., Ravassard, P.M., Ho, D., Acharya, L., Kees, A.L., Vuong, > >>>> C., and Mehta, M.R. (2017). Dynamics of cortical dendritic membrane > >>>> potential and spikes in freely behaving rats. Science. > >>>> earlier Arxiv version > >>>> http://biorxiv.org/content/early/2016/12/28/096941 > >>>> > >>>> See http://science.sciencemag.org/content/early/2017/03/08/s > >>>> cience.aaj1497 > >>>> and commentary > >>>> "Why our brains may be 100 times more powerful than believed" > >>>> here: http://newatlas.com/brains-more-powerful/48357/ > >>>> ==================================================== > >>>> > >>>> Guess what?: 'Dendritic Action Potentials' (DAP) must now become a > >>>> thing. All your simulation packages? Just got old. All your > >>>> neuromorphic > >>>> chips? Has-beens. > >>>> > >>>> SOMATIC AP is 60 years old. > >>>> DENDRITIC AP now comes home. > >>>> > >>>> Neocortical sub- and suprathreshold dendritic membrane potential (DMP) > >>>> breaking out into localised firing within the dendrite structure. > >>>> Dendrite > >>>> firing has been observed for a long time, but this is the first time > >>>> anyone > >>>> has seriously accessed its origins and correlated it with behaviour. > >>>> Collectively the DMP are very strong (as represented by voltage > measured > >>>> in > >>>> tissue: Higher than somatic action potentials!) This is because neural > >>>> tissue is 90% dendritic and there are collosal numbers of > post-synaptic > >>>> densities (synapses). > >>>> > >>>> The implication ... you guessed it .... the brain is not a computer > >>>> (analog or digital) but a system of interacting fields who's > >>>> long-distance > >>>> outward signs ... the tip of the iceberg ... are soma-related action > >>>> potentials. I reckon it's at least 3 orders of magnitude more complex, > >>>> not > >>>> just two....because it's totally spatialised and interacting at > distance > >>>> at > >>>> near the speed of light. > >>>> > >>>> It is physically impossible for any signals to operate chemically > >>>> (ion-channel ion transport leading to extracellular ion motion = > >>>> currents) > >>>> on the fast timescales found to actually operate in the dendrites. > Ions > >>>> can > >>>> barely move a nanometer on those timescales. There are NO currents at > >>>> all! > >>>> No current can possibly be the origin of collective signalling of this > >>>> kind. > >>>> > >>>> The fields? No problem. Action at a distance. Speed of light. > >>>> Remotely-activated modulation of remote transmembrane fields (in this > >>>> case > >>>> the post-synaptic density of ion channels, advancing and retarding > >>>> signal > >>>> events). Easy. Plain old classical physics of the Lorentz force. > Field > >>>> systems exactly of the kind I did in my PHD thesis. > >>>> > >>>> And exactly the same thing in my chip design ... what I am > >>>> experimentally working on already ... my proposed system does this > >>>> naturally. This is because I have no neurons in my design. I merely > >>>> have > >>>> loci of signalling that does the same thing dendrites/soma/axons do. > >>>> > >>>> This seems like a big deal to me. Is that 'ol penny gonna drop this > >>>> time? How much evidence can a system ignore before it goes bang and > >>>> shifts. > >>>> (Reminds me of a certain political context ... let's not go there) :-) > >>>> > >>>> Back to testing. > >>>> > >>>> cheers > >>>> > >>>> colin > >>>> > >>>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > >>>> <https://www.listbox.com/member/archive/rss/303/23508161-fa52c03c> | > >>>> Modify <https://www.listbox.com/member/?&;> Your Subscription > >>>> <http://www.listbox.com> > >>>> > >>>> > >>>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > >>>> <https://www.listbox.com/member/archive/rss/303/11721311-20a65d4a> | > >>>> Modify <https://www.listbox.com/member/?&;> Your Subscription > >>>> <http://www.listbox.com> > >>>> > >>>> > >>> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > >>> <https://www.listbox.com/member/archive/rss/303/19237892-5029d625> | > >>> Modify <https://www.listbox.com/member/?&;> Your Subscription > >>> <http://www.listbox.com> > >>> > >> > >> > >> > >> -- > >> Ben Goertzel, PhD > >> http://goertzel.org > >> > >> “Our first mothers and fathers … were endowed with intelligence; they > saw > >> and instantly they could see far … they succeeded in knowing all that > >> there > >> is in the world. When they looked, instantly they saw all around them, > >> and > >> they contemplated in turn the arch of heaven and the round face of the > >> earth. … Great was their wisdom …. They were able to know all.... > >> > >> But the Creator and the Maker did not hear this with pleasure. … ‘Are > >> they > >> not by nature simple creatures of our making? Must they also be gods? … > >> What if they do not reproduce and multiply?’ > >> > >> Then the Heart of Heaven blew mist into their eyes, which clouded their > >> sight as when a mirror is breathed upon. Their eyes were covered and > they > >> could see only what was close, only that was clear to them.” > >> > >> — Popol Vuh (holy book of the ancient Mayas) > >> *AGI* | Archives <https://www.listbox.com/member/archive/303/=now> > >> <https://www.listbox.com/member/archive/rss/303/11721311-20a65d4a> | > >> Modify > >> <https://www.listbox.com/member/?&;> > >> Your Subscription <http://www.listbox.com> > >> > > > > > > > > ------------------------------------------- > > AGI > > Archives: https://www.listbox.com/member/archive/303/=now > > RSS Feed: https://www.listbox.com/member/archive/rss/303/ > 11943661-d9279dae > > Modify Your Subscription: > > https://www.listbox.com/member/?&; > > Powered by Listbox: http://www.listbox.com > > > > > ------------------------------------------- > AGI > Archives: https://www.listbox.com/member/archive/303/=now > RSS Feed: https://www.listbox.com/member/archive/rss/303/10443978-6f4c28ac > Modify Your Subscription: https://www.listbox.com/ > member/?& > Powered by Listbox: 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
