--- Log opened Wed Jan 25 00:00:24 2023 00:00 -!- RubenSomsen_ is now known as RubenSomsen 01:27 -!- andytoshi [~apoelstra@user/andytoshi] has quit [Ping timeout: 268 seconds] 01:27 -!- soundand1ury [~soundandf@user/soundandfury] has quit [Ping timeout: 268 seconds] 01:43 -!- soundandfury [~soundandf@user/soundandfury] has joined #hplusroadmap 01:44 -!- andytoshi [~apoelstra@user/andytoshi] has joined #hplusroadmap 01:46 -!- Malvolio [~Malvolio@idlerpg/player/Malvolio] has joined #hplusroadmap 02:13 -!- soundandfury [~soundandf@user/soundandfury] has quit [Ping timeout: 260 seconds] 02:13 -!- andytoshi [~apoelstra@user/andytoshi] has quit [Ping timeout: 260 seconds] 02:19 -!- soundandfury [~soundandf@user/soundandfury] has joined #hplusroadmap 02:20 -!- andytoshi [~apoelstra@user/andytoshi] has joined #hplusroadmap 02:45 -!- TMM_ [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.] 02:45 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap 04:04 -!- TMM_ [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.] 04:04 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap 04:36 -!- pasky [~pasky@nikam.ms.mff.cuni.cz] has joined #hplusroadmap 06:42 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::4249] has joined #hplusroadmap 07:25 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Read error: Connection reset by peer] 07:26 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap 07:48 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has quit [Read error: Connection reset by peer] 08:13 < kanzure> "The molecular memory code and synaptic plasticity: A synthesis" https://gershmanlab.com/pubs/memory_synthesis.pdf (2023) 08:29 < kanzure> section 2.2.8 on molecular memory transfer is kinda wild (page 7) 08:43 -!- L29Ah [~L29Ah@wikipedia/L29Ah] has joined #hplusroadmap 09:01 < hprmbridge> lachlan> looks fascinating, sent it to my thesis advisor 09:33 < hprmbridge> nmz787> https://static1.squarespace.com/static/607da9820de541322b95fc19/t/638e30661716fb22115d6c41/1670262889353/program%2B2022-webiste-rev+%28002%29.pdf 09:33 < hprmbridge> nmz787> several hits for DNA 09:33 < hprmbridge> nmz787> "2022 IEEE IEDM Program " 09:36 < hprmbridge> nmz787> https://phys.org/news/2022-12-world-smallest-christmas.html 09:39 < hprmbridge> nmz787> https://heidelberg-instruments.com/product/nanofrazor-explore/ 09:44 < lsneff> "Another... (full message at ) 09:47 < hprmbridge> kanzure> something something RNA packaged into capsids (your message didn't completely relay) 09:47 < hprmbridge> kanzure> "virus-like particles" 10:19 < kanzure> aersolized adenovirus https://twitter.com/DavidAMarkowitz/status/1618279791811235840 10:24 < lsneff> kanzure: that paper was wild 10:24 < lsneff> thank you for finding it 10:27 < muurkha> lsneff: another what? 10:27 < muurkha> 17:44 < lsneff> "Another... (full message at 10:27 < muurkha> ) 10:28 < lsneff> Jeez, what a mess 10:28 < lsneff> "Another remarkable feature of this model system is that the mechanism of transfer has been identified: pathogen avoidance is encoded by a small RNA (P11) packaged into virus-like particles that enable extracellular transmission" 10:28 < muurkha> looks like Matrix has bug 10:28 < muurkha> [a 10:28 < muurkha> *a 10:28 < muurkha> help cliff stoll is jingling his keys against my modem serial line 10:32 < kanzure> yes the memory cannibalism testing was also weird 10:33 < lsneff> and memory transfer in chickens? 10:34 < lsneff> i've heard of RNA/peptide memory as a concept before, but had also heard it was an old, crackpot idea that didn't hold up to modern scrutiny. perhaps that's not the case 10:34 < muurkha> pretty sure immune system memory is DNA? 10:35 < kanzure> maybe it is crackpot, although i'm not impressed by our progress in figuring out biological memory 10:38 < lsneff> The inference vs generative memory is a quite attractive idea 11:48 < fenn> immune system memory is based on B-cell and T-cell activation (boolean) and the number of cells targeting each epitope (variable large integer) 11:49 < fenn> it's not like the immune system is writing down data in DNA; there's just a huge lookup table of random DNA sequence shufflings on the antibody sequence 11:50 < fenn> that random shuffling only happens once during development 12:38 < hprmbridge> nmz787> pretty sure DNA is heredity memory 12:39 < hprmbridge> nmz787> /me end troll 12:39 < lsneff> not in the same way as the above paper, unless instincts are propagated through memories stored within the genome 12:39 < lsneff> oh lol, yes 12:43 < hprmbridge> nmz787> haha 12:43 < hprmbridge> nmz787> your human instinct is strong, young padawan 12:43 < hprmbridge> nmz787> it seems there could be a way to engineer connectome state 12:46 < hprmbridge> nmz787> like, get some rat neurons, seed them on a MEA and then train to play a flight simulator.. then induce some sort of barcoding process... then... induce some process which trades barcodes and assembles them into a linked list structure... then pluck a cell, transfer to a new MEA... then induce a process which uhh, reads the linked-list and the top-of-stack entry is popped off and umm, assigned to a 12:46 < hprmbridge> nmz787> new daughter cell 12:47 < hprmbridge> nmz787> you'd then need some sort of lookup table to correlate which cell-ID to the number of dendrites or something 12:47 < hprmbridge> nmz787> has someone even made a cartoon of what this system might look like, ignoring whether such DNA/protein circuits/processes have been discovered/engineered? 12:48 < hprmbridge> nmz787> this seems like it could/should be a simulator first 12:49 < hprmbridge> nmz787> maybe it would start with a breadth-first search algorithm... starting when one of the trained neurons got blasted with a focused spot of light or a large pulse from the electrode it was growing on 12:49 < hprmbridge> nmz787> i.e. a linked-list nucleation/starting point 12:50 < kanzure> for instinct consider tony zador's "Encoding innate ability through a genomic bottleneck" https://www.biorxiv.org/content/10.1101/2021.03.16.435261v2.abstract 12:56 < kanzure> "Structure induces computational function in networks with diverse types of spiking neurons" https://www.biorxiv.org/content/10.1101/2021.05.18.444689v4.abstract 13:39 < fenn> "writing" a connectome to neural tissue sounds really difficult 13:40 < fenn> it's like going back down to a planetary surface once you're in space; you'd better have a good reason 13:45 < kanzure> well maybe there's a way to encode one genetically 13:52 < fenn> each individual cell (within some radius) would have to have its own growth factor, not just each cell type 13:52 < fenn> that way the dendrites would be able to sniff out their targets and grow toward them 13:52 < kanzure> i think the assumption is more like cellular automaton seeds 13:52 < kanzure> unless you're talking about direct placement techniques 13:55 < kanzure> you could laser sinter to destroy unwanted synapses and dendrite growths 13:55 < fenn> but the cells have to be intertwingled enough to form the desired connection in the first place 13:56 < kanzure> what if you worked only with really really big neurons 13:56 < fenn> if the connectome were more sparse to start with it would be easier, but then it probably wouldn't be a very interesting mind either 13:57 < fenn> the problem is that biology is right up against the limit for how densely neurons can be packed 13:58 < kanzure> an aquatic single-cell algae caulerpa taxifolia grows up to 30 cm long 14:01 < kanzure> "more than 3 meters long" okay 14:02 < kanzure> "Caulerpa species support their large cell size by having the cytoplasm circulate constantly, supported by a network of microtubules. This behavior was known in 1967.[10] The cytoplasm does not leak out when the cell is cut." 14:02 < kanzure> https://en.wikipedia.org/wiki/Cytoplasmic_streaming 14:07 < kanzure> with a big enough neuron you might be able to stitch enormous dendrites into the cell body 14:12 < abetusk> Does anyone in here have a fiber laser? 14:12 < kanzure> here is a 2022 review on "single-cell [micro]surgery" https://www.nature.com/articles/s41378-022-00376-0 14:13 < kanzure> "Laser-induced fusion of human embryonic stem cells with optical tweezers" https://core.ac.uk/download/pdf/38031891.pdf 14:22 < kanzure> page 130 discusses femtosecond laser pulses to attach an axon to a neuron https://era.library.ualberta.ca/items/59376e66-a0aa-4681-847a-1a776bfddc58/download/c46da184-ef93-4096-a69c-4a614eb2ced8 (warning: auto-downloads) 14:22 < kanzure> er, pdf page 157 i mean. 14:24 < hprmbridge> kanzure> https://cdn.discordapp.com/attachments/1064664282450628710/1067932897396396092/image.png 14:26 < kanzure> (from "Femtosecond laser pulse interaction with tissue for attachment of cells and neurons" 2018) 16:41 -!- TMM_ [hp@amanda.tmm.cx] has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.] 16:41 -!- TMM_ [hp@amanda.tmm.cx] has joined #hplusroadmap 17:21 < hprmbridge> nmz787> "laser sinter dendrites" means you also need the connectome stored digitally, which means it needs read-out from the cell 17:23 < hprmbridge> nmz787> Hmm, having dendrites fumble around sounds hard 17:23 < hprmbridge> nmz787> Has anyone engineered pseudopodia control? 17:23 < hprmbridge> nmz787> GMO dancing cells... 17:24 < hprmbridge> nmz787> Bacteria that are responsive to audio (or light), etc 17:24 < hprmbridge> nmz787> I'd hope the plasmid was named pRaver or something 17:56 < hprmbridge> kanzure> maybe with magnetic beads? for your pseudopodae. 18:30 -!- yashgaroth [~ffffffff@2601:5c4:c780:6aa0::4249] has quit [Quit: Leaving] 19:22 < lsneff> Wish we didn’t have to think about how to engineer biological neural networks, it would be easier if we knew how they worked so we could just simulate them accurately 19:43 < muurkha> pretty sure this is a problem with all natural systems 19:44 < muurkha> beyond the very simplest 19:45 < muurkha> turbulence in air or water, crystal growth in rocks or alloys, spectra of elements, behavior of servers under load, etc. even when how they work is sort of known, simulating it accurately involves running the actual system 19:45 < lsneff> This is true 19:47 < muurkha> with enough experience running the actual system we can figure out what to disregard to run a simulation 19:47 < muurkha> efficiently 19:48 < muurkha> in theory you could do this entirely in simulation from first principles but nobody does it that way 19:49 < lsneff> Indeed. I am legitimately worried that we will not have enough introspection on the internals of neurons to determine how they work. 19:55 < muurkha> You'd think the molecular structure of benzene would have been impossible with the tools available in 01865, though. 20:01 < lsneff> How did they do that anyhow? 20:13 < fenn> dream magic, it is told 20:16 < fenn> https://psy-minds.com/kekules-dream-structure-benzene/ 20:21 < hprmbridge> kanzure> think very hard, then write down the answer. 20:37 < fenn> you forgot step 1 20:39 < hprmbridge> kanzure> what was the problem, again? 21:04 -!- deltab [~deltab@user/deltab] has quit [Ping timeout: 260 seconds] 21:08 < muurkha> It's an interesting story. 21:11 -!- deltab [~deltab@user/deltab] has joined #hplusroadmap 22:38 < hprmbridge> nmz787> Muurkha: traffic jams 22:38 < hprmbridge> nmz787> Another of life's great mysteries 23:58 -!- A_Dragon [A_D@libera/staff/dragon] has quit [Ping timeout: 615 seconds] --- Log closed Thu Jan 26 00:00:25 2023