Competing theories for where engrams (physical memories) are stored:
- neuronal network / synapses
- microtubule network in inside the neurons
- now RNA recently joined the competition
As microtubules are protein, so made from RNA, this experiment can not exclude microtubules
My guess: everything plays a role: individual neurons encode internally bits of information (like an inode), while the multidimensional network of neurons creates the structure (like the filesystem), while electrical impulses on the network are commands (like software) both made from files, and accessing some other files.
The more advanced animals have more advanced filesystems with partial RAID1 mirroring (experiences removing fully one hemisphere) and caching (we know that's mostly the cerebellum job for movements).
Sleep is both a log replay/filesystem scrub and defrag (committing memories to long term storage if they are frequently accessed)
I just scratched my head tbh. Maybe they really want organic computing to work just like silicon-electrical computing?
At some point in cybernetic convergent evolution almost in parallel with the technological singularity, there will be IT architecture problems that will need to be addressed, and people on HN will bikeshed if square auxiliary optic nerve connectors or triangular ones are better. Just imagine being hellbanned for bringing-up circular ones, and no one else is able to interact with you. Eeek!
It's a guess like he said and an analogy geared toward a tech audience on a tech forum seems apt. Don't see why there is a need to talk down to him, this isn't a forum aimed at biology.
sleep is also 'reshuffling' of memory representations that's continuously happening to reconcile internal inconsistencies and converge on some coherent state that can help with real-time predictions.
this is likely the actual learning mechanism as opposed to error correction via stimulus-response reinforcement loop, so popular in AI research.
I don't see what the experiment says about memory, it seems to only show that a coarse level of fear/stress in response to stimuli can be communicated. If you have the shock conditional on something like a color of light and could show that the conditional response was communicated, that would be much more interesting.
I don't get the distinction you are trying to make.
Trained snails learned to withdraw into their shells for long period of times after a shock to avoid receiving any more shocks.
That learning was passed along to other snails through RNA.
It's a conditional response to an external stimuli that I would expect to be stored just as any other memory?
This is the part I'm not sure about. The alternate explanation I have goes like this:
- Snail A gets shocked a lot, as a result it gets really stressed out, altering its RNA/epigenetics to produce lots of stress hormones
- Lots of stress hormones lead Snail A to withdraw for a long time when touched
- Snail B gets an injection of Snail A's RNA
- As a result, Snail B starts producing lots of stress hormones
- Lots of stress hormones lead Snail B to withdraw for a long time when touched
This doesn't involve what we would typically regard as memory or learning, and certainly wouldn't generalize to a method of transferring memories in the general case. A modified experiment that I think would show genuine memory transferal would be to occasionally show red and green lights, have snails in group 1 shocked after the green lights only and snails in group 2 shocked after the red lights only, inject their respective RNA, and see if the receiving groups are distinguishable in their responses to red vs green lights.
How is your explanation different than memory propagating through RNA? If memory is being passed through genetics, isn't the employment of epigenetics essentially a tautology?
Saying that the ability to pass memories is limited to a specific type or granularity doesn't negate that a learned response (a memory) was passed on.
buts its not a neurological memory, you're just bridging a physiological response from A to B. B doesn't have any new memories, its just being roped into behaving like A because the genetic expression of A altered, and that got copied to B through RNA.
> This doesn't involve what we would typically regard as memory or learning
But it is memory nevertheless, just a special kind of it. Maybe more ancient kind created for diffuse nervous system with abilities limited by abilities of a diffuse nervous system.
> and certainly wouldn't generalize to a method of transferring memories in the general case.
Yes, I believe, so. One could learn by an injection nothing that couldn't be learned by a hydra.
Or maybe a more complex nervous system could learn something on top of learned by a hydra? I mean all those ancient physiological mechanism are mostly intact in our bodies, they work and body relies on them and uses them. Why not to get more than a simple hydra?
If depends if there is a physiological pathway between the mRNA and whatever neural structure is used for memory. This does not necessarily mean that the same effect can be replicated in mammals, a snail is several magnitudes simpler than your typical mammallia animal. It would nonetheless be quite interesting to study.
Hypothetically, if memory and experiences can be, to use a programmer term, serialized into mRNA, then that would be quite an advancement for brain computer interface technology as the mRNA can then be amplified and sequenced to extract out the information.
There are some theories explaining that memory is stored in Space, and that DNA acts as an “antenna” which can retrieve information via quantum entanglement.
Ok, well IF that’s the case (I’m not saying it is) then this finding wouldn’t be that surprising but rather expected behavior.
So when they modified “antenna” of that snail so that it can access information previously accessed only by the other snail.
Anyways that’s just about entertaining an idea, so don’t take it seriously, please.
I dunno, it kinda sounds like the conditioning of the snails was passed through the RNA. I don't know if this would scale to humans, but that would be a super interesting experiment.
29 comments
[ 118 ms ] story [ 3073 ms ] thread- neuronal network / synapses
- microtubule network in inside the neurons
- now RNA recently joined the competition
As microtubules are protein, so made from RNA, this experiment can not exclude microtubules
My guess: everything plays a role: individual neurons encode internally bits of information (like an inode), while the multidimensional network of neurons creates the structure (like the filesystem), while electrical impulses on the network are commands (like software) both made from files, and accessing some other files.
The more advanced animals have more advanced filesystems with partial RAID1 mirroring (experiences removing fully one hemisphere) and caching (we know that's mostly the cerebellum job for movements).
Sleep is both a log replay/filesystem scrub and defrag (committing memories to long term storage if they are frequently accessed)
At some point in cybernetic convergent evolution almost in parallel with the technological singularity, there will be IT architecture problems that will need to be addressed, and people on HN will bikeshed if square auxiliary optic nerve connectors or triangular ones are better. Just imagine being hellbanned for bringing-up circular ones, and no one else is able to interact with you. Eeek!
Yes they are. They are proteins. Proteins are made from DNA->RNA->ribosomal transcription.
Also microtubules interact with RNA quite a lot. Check the references: https://pubmed.ncbi.nlm.nih.gov/15037309/
this is likely the actual learning mechanism as opposed to error correction via stimulus-response reinforcement loop, so popular in AI research.
Prof. Wilson has a great talk on learning as 'discovery of hidden connections': https://www.youtube.com/watch?t=1147&v=Vf_m65MLdLI&feature=y...
also see his works on 'hidden state inference': http://web.mit.edu/wilsonlab/html/publications.html
This is the part I'm not sure about. The alternate explanation I have goes like this:
This doesn't involve what we would typically regard as memory or learning, and certainly wouldn't generalize to a method of transferring memories in the general case. A modified experiment that I think would show genuine memory transferal would be to occasionally show red and green lights, have snails in group 1 shocked after the green lights only and snails in group 2 shocked after the red lights only, inject their respective RNA, and see if the receiving groups are distinguishable in their responses to red vs green lights.Saying that the ability to pass memories is limited to a specific type or granularity doesn't negate that a learned response (a memory) was passed on.
maybe 'state' would be a better word
But it is memory nevertheless, just a special kind of it. Maybe more ancient kind created for diffuse nervous system with abilities limited by abilities of a diffuse nervous system.
> and certainly wouldn't generalize to a method of transferring memories in the general case.
Yes, I believe, so. One could learn by an injection nothing that couldn't be learned by a hydra.
Or maybe a more complex nervous system could learn something on top of learned by a hydra? I mean all those ancient physiological mechanism are mostly intact in our bodies, they work and body relies on them and uses them. Why not to get more than a simple hydra?
Hypothetically, if memory and experiences can be, to use a programmer term, serialized into mRNA, then that would be quite an advancement for brain computer interface technology as the mRNA can then be amplified and sequenced to extract out the information.
Ok, well IF that’s the case (I’m not saying it is) then this finding wouldn’t be that surprising but rather expected behavior.
So when they modified “antenna” of that snail so that it can access information previously accessed only by the other snail.
Anyways that’s just about entertaining an idea, so don’t take it seriously, please.