--- Log opened Fri Apr 24 00:00:03 2020 00:09 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 265 seconds] 00:36 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 01:55 -!- Urchin [~urchin@unaffiliated/urchin] has quit [Ping timeout: 264 seconds] 02:07 -!- Urchin [~urchin@unaffiliated/urchin] has joined ##hplusroadmap 03:04 -!- twoplustwo [dce9c7cd@205.199.233.220.static.exetel.com.au] has joined ##hplusroadmap 03:19 < twoplustwo> hi everyone 03:19 < twoplustwo> does science have an idea how the brain stores memories chemically? 03:19 < twoplustwo> can someone point me to a book or a research paper that could give me some info? 03:28 < darsie> short and long term memory might work differently. Like SDRAM and ROM. 03:28 < nsh> nobody has any idea 03:28 < nsh> as far as i can tell 03:28 < nsh> i have ideas but they are not scientific, in the sense that they are ill-formed and not even testable conjectures 03:28 < nsh> or worse, metaphysical 03:29 < nsh> i suppose examination of failure modes of memory creation and access might shed some light 03:29 < nsh> but it's crass to talk about it terms of chemicals, or even neurons 03:30 < nsh> like talking about prison discipline in terms of the metals in the alloy in the badges on the prison guards 03:30 < nsh> sure these things are there but they are not necessarily the most explicative concepts 03:31 < nsh> actually the failure mode may be less revealing than the superlative function modes 03:31 < nsh> people with eidetic memory, etc. 03:32 < darsie> "nobody has any idea" as in deep learning AI learning how to play space invaders? 03:32 < twoplustwo> lol 03:32 < nsh> i can fold a piece of paper so that when you pull a tail the wings flap 03:32 < nsh> i wouldn't say the paper has a memory physiological function 03:32 < nsh> it's just a property of the dynamics instilled into the system 03:33 < nsh> memory *of 03:33 < twoplustwo> I can come up with lots of funny absurd ideas 03:33 < nsh> i don't think computers, even in the limit of big data and computes are a great model for studying life yet 03:33 < nsh> we can study crude caricatures of life, sure 03:33 < twoplustwo> but I'm really curious about what the current understanding is 03:34 < darsie> AI builds connectomes too complex to understand. 03:34 < nsh> .t https://www.biophysics.org/blog/understand-the-regulation-of-learning-and-memory-formation-from-a-molecular-prospective 03:34 < saxo> Understand the Regulation of Learning and Memory Formation from a Molecular Prospective 03:34 < nsh> .t https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5142816/ 03:34 < saxo> A biological perspective on memory 03:34 < nsh> .t https://www.sciencedirect.com/science/article/pii/S0092867414002906 03:34 < saxo> The Molecular and Systems Biology of Memory - ScienceDirect 03:34 < nsh> .t https://www.ncbi.nlm.nih.gov/pmc/articles/PMC4246028/ 03:34 < saxo> The neurobiological bases of memory formation: from physiological conditions to psychopathology 03:34 < twoplustwo> ok, thanks 03:34 < nsh> darsie, i can tangle up a rope so it's too complex to understand. it's still not doing cognition 03:35 < nsh> actually probably you could do similar things to neural networks through tangling and toplogy 03:35 < nsh> *topology 03:35 < nsh> arguably there's a reduction between the two 03:35 < twoplustwo> isn't logic gates enough? 03:36 < twoplustwo> my understanding is very primative 03:36 < nsh> if everything is information and all physics is computation on that information 03:36 < nsh> and we believe that classical information suffices then logic dates are enough 03:36 < nsh> but there's room for doubt on all three points 03:36 < nsh> *gates 03:38 < nsh> it might be just as sensible to say that memory is a property of all physical systems and the interesting phenomenon is the coupling of that capacity to physiological behaviours 03:38 < nsh> then you have to consider the substrate/interface, and grope about with things like action potentials in neurons and dendrites and try to abstract from reductive models of those 03:39 < nsh> then someone posits that there's subcellular complexity that's effective and you have to try and abstract more 03:39 < twoplustwo> https://www.youtube.com/watch?v=H6u0VBqNBQ8 03:39 < nsh> eventually someone just makes a thing that convinces us it remembers the meaningful way we're convinced other people remember 03:39 < nsh> .t 03:39 < saxo> The Origin of Consciousness – How Unaware Things Became Aware - YouTube 03:40 < twoplustwo> This is the Kurzgesagt YouTube channel 03:40 < nsh> well it's a notch above peppa pig 03:41 < twoplustwo> :-) 03:44 < twoplustwo> do you have any interesting links to people with eidetic memory? 03:44 < twoplustwo> I've read about John von Neumann 03:45 < twoplustwo> https://en.wikipedia.org/wiki/List_of_people_claimed_to_possess_an_eidetic_memory 03:45 < twoplustwo> There's Highly Superior Autobiographical Memory 03:47 < twoplustwo> I've come across Kim Peek too 03:48 < darsie> darsie, i can tangle up a rope so it's too complex to understand. it's still not doing cognition <-- Looks like you agree that neural networks easily can be too complex to understand. Those can be cognitive. 03:50 < darsie> Like, it would be hard to tell which characters have been trained on a text recognition AI by looking at the connectome. 03:52 < darsie> But what we can say is it's in the weights of the connectome. 03:52 < nsh> the jump from 'complex' to 'cognitive' is what i question 03:52 < twoplustwo> https://en.wikipedia.org/wiki/Cyc 03:53 < nsh> if the same computation can be performed by two children playing cats cradle 03:53 < twoplustwo> I suspect you've come across Cyc, before 03:53 < nsh> then we are in a weird place 03:53 < nsh> where any complex information processing no matter how it is physically embedded is cognitive 03:53 < nsh> that's a big metaphysical ask 03:53 < nsh> read Permutation City by Greg Egan if you haven't already 03:53 < nsh> he explores this 03:54 < darsie> What I meant was we may not figure out how our memory works, simply because it's too complex, but we can tell the basic building blocks. 03:55 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 272 seconds] 03:56 < darsie> And I guess short term memory uses dynamic excitation of neural networks (neurons firing), rather than changing weights of connections. 03:57 < darsie> Hence my comparison to SDRAM, which uses flip flop circuits. 04:22 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 04:44 < nsh> [ohai FBI] on a scary thought tip: how difficult is it to culture coronaviruses ex-vivo if you somehow have a source of cells they can infect 04:44 < nsh> ie an immortal cell line that you can cause to develop ACE2 receptors 04:45 < nsh> or something like that 04:45 < nsh> is this a risk? or would "bad people" find it easier to cultivate in (unwitting) human candidates 04:45 < nsh> darsie, it may not decompose that way 04:46 < nsh> memory might be a feature of continuum dynamics rather than an emergent property of discrete interactions yielding states 05:57 -!- twoplustwo [dce9c7cd@205.199.233.220.static.exetel.com.au] has quit [Remote host closed the connection] 06:21 < kanzure> nmz787: another topic, but i mean why not ask while doing so 06:22 < kanzure> why didn't you guys send him to my /papers2/neuro/memory/ thing 06:37 < gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=f2331c92 Michael Folkson: Add Socratic Seminar on Grokking Bitcoin >> http://diyhpl.us/diyhpluswiki/transcripts/london-bitcoin-devs/2020-04-22-socratic-seminar/ 06:37 < gnusha> https://secure.diyhpl.us/cgit/diyhpluswiki/commit/?id=368ebbc6 Bryan Bishop: Merge pull request #99 from michaelfolkson/add-london-socratic >> http://diyhpl.us/diyhpluswiki/ 06:44 < fltrz> kanzure, is your server indexed by some search engine? 06:48 < kanzure> http://diyhpl.us/~bryan/papers2/neuro/memory/ 06:48 < kanzure> https://diyhpl.us/~bryan/papers2/neuro/memory/ 06:54 -!- N-time [~Mark@212.225.172.60] has joined ##hplusroadmap 07:15 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 265 seconds] 07:33 -!- N-time [~Mark@212.225.172.60] has quit [Quit: Leaving] 07:43 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 07:45 < nsh> live in 15m - panel discussion on privacy concerns around contact-tracing technologies for pandemic mitigation: https://www.disruptionlab.org/fridays 07:49 < kanzure> nsh: https://github.com/ChristopherA/Lists-of-High-Signal-Low-Noise-Links/blob/master/ImmunityCredentials-VerifiableClaims-COVID19.md 07:56 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 265 seconds] 07:59 < nsh> ty kanzure 08:02 -!- yashgaroth [~ffffffff@172.58.19.137] has joined ##hplusroadmap 08:10 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 09:07 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has joined ##hplusroadmap 09:27 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has quit [Ping timeout: 264 seconds] 09:29 < fenn> yeah nsh that was pretty bad quantum woo level talk 09:29 < fenn> we know a lot about memory formation, it's just incorrect to say "nobody has any idea" 09:32 < kanzure> there has been substantial work on memory in snails 09:32 < kanzure> with optogenetics. 09:33 < nsh> we know a lot of about the biological/neurological/molecular-biological/biophysical interface and we have the working assumption that the interface is the substrate 09:33 < nsh> the latter is not well-founded 09:33 < nsh> when i say 'we don't know' i'm not talking in absolute but in relative terms 09:33 < nsh> just as almost all mathematics is undiscovered 09:33 < nsh> but we know a lot of mathematics 09:33 < nsh> these are not incompatible observatoins 09:34 < nsh> if you want to focus positively on the knowledge accumulated, that's fine 09:34 < nsh> i elect to counterbalance that tendency by focussing on the lacunae 09:34 < nsh> this is a dialectic 09:35 < nsh> to be woo i would have had to posit definitely. i just proposed hypothetically 09:35 < nsh> but again it's elective whether you find that to be useful or not 09:35 < kanzure> nsh do i need to assign a biology textbook to you 09:37 -!- aeiousomething [~aeiousome@unaffiliated/aeiousomething] has joined ##hplusroadmap 09:46 < fenn> nsh when we rebuild the structure we observe in silicon, it does useful things that we expect of brains 09:46 < fenn> what more do you want? 09:48 < fenn> the main problem seems to be getting good enough observations 09:49 -!- mauz555 [~mauz555@2a01:e0a:56d:9090:2876:e83c:cef1:e3fa] has quit [Ping timeout: 272 seconds] 09:50 -!- mauz555 [~mauz555@88.125.182.66] has joined ##hplusroadmap 09:51 < nsh> from the American Chemical Society: 09:51 < nsh> "Thalidomide exists in two mirror-image forms: it is a racemic mixture of (R)- and (S)-enantiomers. The (R)-enantiomer, shown in the figure, has sedative effects, whereas the (S)-isomer is teratogenic." 09:51 < nsh> almost complete fidelity can be great, or it can be disastrous 09:51 < nsh> machines running the planet might make put life on earth on vastly different trajectories for the want of the enantiomeric equivalent of a soul 09:51 < nsh> but i don't wanna all elietzer 09:51 < nsh> i can be safely ignored most likely 09:56 -!- mauz555 [~mauz555@88.125.182.66] has quit [Ping timeout: 256 seconds] 09:57 -!- mauz555 [~mauz555@88.125.182.66] has joined ##hplusroadmap 09:58 < yashgaroth> enantiomers don't have 'almost complete fidelity' to each other unless you're an 1800s chemist, surely there's a better metaphor 10:00 < yashgaroth> also per your earlier question, there are already a number of cell lines that you can grow covid in, it wouldn't be hard if you can do mammalian cell culture and have access to the virus 10:19 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has joined ##hplusroadmap 11:02 -!- urkk [~urkk@45.134.239.188.dynamic.jazztel.es] has quit [Ping timeout: 250 seconds] 11:03 -!- urkk [~urkk@4.142.239.188.dynamic.jazztel.es] has joined ##hplusroadmap 11:21 -!- prometheus [~root@88.230.142.165] has joined ##hplusroadmap 11:22 < prometheus> this message is test 11:29 < kanzure> didn't work 11:46 -!- prometheus [~root@88.230.142.165] has quit [Quit: leaving] 11:53 < kanzure> hey is c2wiki still a thing? 11:53 < kanzure> is there a backup anywhere? 12:12 -!- mauz555 [~mauz555@88.125.182.66] has quit [Read error: Connection reset by peer] 12:26 -!- dongcarl [~dongcarl@unaffiliated/dongcarl] has quit [Quit: Ping timeout (120 seconds)] 12:26 -!- dongcarl [~dongcarl@unaffiliated/dongcarl] has joined ##hplusroadmap 12:27 -!- balrog [~balrog@unaffiliated/balrog] has quit [Quit: Bye] 12:27 -!- balrog [~balrog@unaffiliated/balrog] has joined ##hplusroadmap 12:49 -!- mauz555 [~mauz555@88.125.182.66] has joined ##hplusroadmap 12:54 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has quit [Ping timeout: 260 seconds] 12:54 -!- mauz555 [~mauz555@88.125.182.66] has quit [Ping timeout: 256 seconds] 12:58 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has joined ##hplusroadmap 13:02 -!- prometheus [~root@88.230.142.165] has joined ##hplusroadmap 13:29 < fenn> kanzure: http://archive.org/details/c2.com-wiki_201501 and it's still "a thing" just not a usable thing 13:33 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Quit: WeeChat 2.7.1] 13:35 -!- mauz555 [~mauz555@2a01:e0a:56d:9090:ec47:4cfe:fdff:656c] has joined ##hplusroadmap 13:39 -!- mauz555 [~mauz555@2a01:e0a:56d:9090:ec47:4cfe:fdff:656c] has quit [Ping timeout: 272 seconds] 13:40 -!- prometheus [~root@88.230.142.165] has quit [Quit: leaving] 14:23 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has quit [Ping timeout: 256 seconds] 14:30 -!- mauz555 [~mauz555@2a01:e0a:56d:9090:ec47:4cfe:fdff:656c] has joined ##hplusroadmap 15:05 -!- darsie [~kvirc@84-114-73-160.cable.dynamic.surfer.at] has quit [Ping timeout: 265 seconds] 15:35 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 15:49 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has joined ##hplusroadmap 16:19 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 256 seconds] 16:33 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 16:56 -!- Malvolio [~Malvolio@unaffiliated/malvolio] has quit [Quit: BEWARE THE LIVING] 17:51 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has quit [Ping timeout: 240 seconds] 17:53 -!- mauz555 [~mauz555@2a01:e0a:56d:9090:ec47:4cfe:fdff:656c] has quit [Remote host closed the connection] 18:04 -!- justanotheruser [~justanoth@unaffiliated/justanotheruser] has joined ##hplusroadmap 18:27 -!- balrog [~balrog@unaffiliated/balrog] has quit [Quit: Bye] 18:39 * nmz787 bought a TIG welder and plasma cutter today... one step closer to building my own vacuum chamber for chemistry/semiconductor synthesis stuff 19:21 -!- balrog [~balrog@unaffiliated/balrog] has joined ##hplusroadmap 19:56 -!- turona [~quassel@2a01:c22:a47b:e700:cc0b:7e12:bb0a:a2f4] has quit [Ping timeout: 272 seconds] 19:59 -!- turona [~quassel@2a01:c23:6451:a500:6a8d:b6d9:a5b9:9ba4] has joined ##hplusroadmap 20:04 < fenn> should be useful for a lot of other stuff too 20:04 < fenn> have you got a local scrap yard to go treasure hunting in? 20:15 -!- ra_ [~ra@ool-18baef17.dyn.optonline.net] has quit [Ping timeout: 264 seconds] 20:54 -!- yashgaroth [~ffffffff@172.58.19.137] has quit [Quit: Leaving] 21:47 -!- darsie [~kvirc@84-114-73-160.cable.dynamic.surfer.at] has joined ##hplusroadmap 22:31 < adlai> zounds, i missed the bull session about NNN learning... my understanding, inspired somewhat by the "behavioralism" dogma, is that the main mechanism is cascades of resonance between positive feedback loops within the brain, that eventually hit the motor cortices 22:31 < adlai> sadly, from the perspective of "it's all a giant tensorflow", this is nigh indistinguishable from a neverending epileptic fit 22:35 < adlai> in the absence of a relatively compact critereon, such as "gives student's exam a failing grade if the sum of points per question is less than minimum passing grade", although preferably questions slightly more interesting than that, there isn't much to cognate about 22:37 < adlai> perhaps a critereon that better selects against epilepsia is proficiency at capacitative musicianship 22:37 < adlai> .wik theremin 22:37 < saxo> "The theremin (/ˈθɛrəmɪn/; originally known as the ætherphone/etherphone, thereminophone or termenvox/thereminvox) is an electronic musical instrument controlled without physical contact by the thereminist (performer). It is named after its inventor, Léon Theremin, [...]" - https://en.wikipedia.org/wiki/Theremin 22:39 < adlai> .wikt eidetic 22:43 < adlai> "However, when the experts were presented with arrangements of chess pieces that could never occur in a game, their recall was no better than the nonexperts, suggesting that they had developed an ability to organize certain types of information, rather than possessing innate eidetic ability." - no citation, so i won't cite the wikipedia article either :) 22:45 -!- CryptoDavid [uid14990@gateway/web/irccloud.com/x-ausutlsyoqrxbuii] has quit [Quit: Connection closed for inactivity] 22:45 < adlai> the local rumor is that devout hassidim know several texts by heart, and have them reasonably-well cross-indexed, too, although this is probably not much different from the ol' "arma virumque cano, ceterae ceterae ceterae" 22:46 < adlai> "Robert Evans can identify new objects that appear in starfields of 1500 galaxies" wikipedia cites Oliver Sacks for this claim, and i find that capability a bit more relevant here than chess playing 22:51 < adlai> wikipedia also quotes CNN that "Stephen Wiltshire is a prodigious savant, capable of drawing the entire skyline of a city after a helicopter ride.", although nobody seems to care what was the minimal altitude of the perspective sequence used during the flyover 22:57 < adlai> those who probably best understand the dimensionalities of the criteria space for human learning are probably the engineers at robotics companies, although good luck getting them to pay for medical school instead of drawing hitech salaries --- Log closed Sat Apr 25 00:00:04 2020